{ "version": 4, "as_of": "2026-10-09", "method": "跨材料的编辑思想地图；人物/机构归属分开，不是私人思想或预测排名", "sources": [ { "id": "D01", "title": "ARK Investment Process", "date": null, "url": "https://www.ark-invest.com/investment-process", "access": "官方流程页；无发布日期，2026-10-05读取。不是某次真实投资决策。", "authors": "ARK institutional team", "summary": "连接成本、需求分析与公司估值；公开说明组合研究与调整流程。", "additional_readings": [ { "date": "2026-10-08", "summary": "机构成本曲线与需求弹性框架；五年模型纳入销量、成本、稀释和估值。", "access": "2026-10-08重新打开官方正文；机构作者，非Cathie个人独著。" } ] }, { "id": "D02", "title": "Big Ideas: Investment Opportunity Report 2026 Summary", "date": "2026-09-10", "url": "https://www.ark-funds.com/articles/etf/big-ideas-the-investment-opportunity-report-2026-summary", "access": "官方摘要；未声称审阅完整报告或独立核验底层数据。", "authors": "ARK Invest", "summary": "从技术变化延伸到可投资机会，比较历史投资周期并讨论组合情景。", "additional_readings": [ { "date": "2026-10-08", "summary": "2026摘要把AI成本下降与token增长联系；800/135与“近3倍”表述不一致，应重算而非沿用。", "access": "2026-10-08重新打开官方正文；机构作者，非Cathie个人独著。" } ] }, { "id": "D03", "title": "Morgan Stanley Mobile Internet Report release", "date": "2009-12-16", "url": "https://www.morganstanley.com/press-releases/morgan-stanley-releases-the-mobile-internet-report_4659e2f5-ea51-11de-aec2-33992aa82cc2", "access": "官方发布说明，非完整长篇研报。", "authors": "Morgan Stanley technology/telecom team; Mary Meeker named", "summary": "结合技术融合、区域经验与使用扩散分析移动互联网。" }, { "id": "D04", "title": "Trends — Artificial Intelligence", "date": "2025-05-30", "url": "https://bondcap.relayto.com/e/2025-trends-in-artificial-intelligence-tn9e6r7a9i24j", "access": "BOND署名的托管文本；核查选定段落，非逐图/逐数字全审计。官方PDF超出检索大小。", "authors": "Mary Meeker / Jay Simons / Daegwon Chae / Alexander Krey", "summary": "并置使用、成本、资本投入与商业化问题，讨论平台和专业化路径。" }, { "id": "D05", "title": "On Bubble Watch", "date": "2025-01-07", "url": "https://www.oaktreecapital.com/insights/memo/on-bubble-watch", "authors": "Howard Marks / Oaktree", "summary": "区分技术进步、估值与投资者心理。", "access": "作者官方备忘录；核查价格/心理讨论，不代表科技股持仓建议。", "checked_at": "2026-10-05" }, { "id": "D06", "title": "The Calculus of Value", "date": "2025-08-13", "url": "https://www.oaktreecapital.com/insights/memo/the-calculus-of-value", "authors": "Howard Marks / Oaktree", "summary": "以盈利能力讨论价值，区分价值与价格。", "access": "官方备忘录；核查Value及市场心理段落，不审计历史收益。", "checked_at": "2026-10-05" }, { "id": "D07", "title": "Hard Lessons E4: Stan Druckenmiller — Invest, Then Investigate", "date": "2026-02-27", "url": "https://www.morganstanley.com/insights/videos/hard-lessons", "authors": "Stanley Druckenmiller; interviewer Iliana Bouzali", "summary": "回顾分阶段加仓、仓位规模及旧信号失效。", "access": "节目官方逐字稿E4；一份访谈，回忆不是当年的独立证据，不证明当前持仓。", "checked_at": "2026-10-05" }, { "id": "D08", "title": "Third Point Q1 2025 Investor Letter", "date": "2025-04-30", "url": "https://assets.thirdpointlimited.com/f/166217/x/bd64805fc1/third-point-q1-2025-investor-letter_tpil.pdf", "authors": "Daniel S. Loeb / Third Point", "summary": "政策变化下调整敞口与催化剂机会。", "access": "基金官方署名信，TPIL托管；核查Review and Outlook，不还原完整交易。", "checked_at": "2026-10-05" }, { "id": "D09", "title": "Third Point Q3 2025 Investor Letter", "date": "2025-10-31", "url": "https://assets-malibu-life.s3.us-west-2.amazonaws.com/system/uploads/fae/file/asset/1673/Third_Point_Q3_2025_Investor_Letter_TPIL.pdf", "authors": "Daniel S. Loeb / Third Point", "summary": "讨论HBM、国际估值及AI融资风险。", "access": "基金署名信的托管PDF；核查pp2–5及信用章节，非完整预测业绩审计。", "checked_at": "2026-10-05" }, { "id": "D10", "title": "NVIDIA FY2026 Q2 Earnings Call", "date": "2025-08-27", "url": "https://s201.q4cdn.com/141608511/files/doc_financials/2026/q2/NVDA-Q2-2026-Earnings-Call-27-August-2025-5_00-PM-ET.pdf", "authors": "NVIDIA management; individually attributed analyst questions", "summary": "核查Moore、Arcuri、Muse、Rasgon的具名提问。", "access": "公司托管FactSet更正稿；核查具名Q&A及角色名录，不是分析师研报。", "checked_at": "2026-10-05" }, { "id": "D11", "title": "NVIDIA FY2026 Q3 Earnings Call", "date": "2025-11-19", "url": "https://s201.q4cdn.com/141608511/files/doc_financials/2026/q3/NVDA-Q3-2026-Earnings-Call-19-November-2025-5_00-PM-ET.pdf", "authors": "NVIDIA management; individually attributed analyst questions", "summary": "核查需求、瓶颈、毛利及费用提问。", "access": "公司托管更正稿；逐项定位具名Q&A，不把管理层回答归为分析师观点。", "checked_at": "2026-10-05" }, { "id": "D12", "title": "Micron FY2026 Q3 Earnings Call — public transcript mirror", "date": "2026-06-24", "url": "https://www.roic.ai/quote/MU/transcripts/2026-year/3-quarter", "authors": "Micron management; Arcuri / Moore / Muse questions", "summary": "核查合同价格、押金与资金用途提问。", "access": "公开逐字稿镜像；核查指定Q&A，转录有噪声，未音频验证；数值不是我们的共识预测。", "checked_at": "2026-10-05" }, { "id": "D13", "title": "HPE FY2026 Q3 Earnings Transcript", "date": "2026-09-02", "url": "https://investors.hpe.com/~/media/Files/H/HP-Enterprise-IR/documents/q3-2026/q3-2026-transcript.pdf", "authors": "HPE management; Mark Newman / Bernstein question", "summary": "核查服务器增长中的量价及订单约束。", "access": "公司官方逐字稿pp15–16；具名分析师提问，不是其完整研报。", "checked_at": "2026-10-05" }, { "id": "D14", "title": "Dell FY2027 Q2 Earnings Call — public transcript mirror", "date": "2026-09-01", "url": "https://stockanalysis.com/stocks/dell/transcripts/589603-q2-2027/", "authors": "Dell management; Mark Newman / Bernstein question", "summary": "核查传统/AI服务器中的客户类别。", "access": "公开逐字稿镜像；核查Newman Q&A，未音频验证，不将全部CPU服务器增长归因于AI。", "checked_at": "2026-10-05" }, { "id": "B1D01", "title": "Fundsmith Equity Fund Annual Letter to Shareholders 2024", "date": "2025-01", "date_precision": "month; 原稿January 2025，财务截止2024-12-31", "url": "https://www.fundsmith.co.uk/media/pirmvyly/annual-letter-to-shareholders-2024.pdf", "authors": "Terry Smith, CEO, Fundsmith LLP; Fundsmith署名观点", "summary": "从资本回报与再投资机会检验质量；现金转换下降需追踪CapEx回报。", "access": "官方12页PDF；读取pp5–7、10–11及署名，不声称独立核验底层财务或完整收益归因。", "locator": "PDF pp5–7 (lines179–295); pp10–11 (lines420–462)", "checked_at": "2026-10-06", "retrieval_refs": [ "turn9view1", "turn11view2", "turn14view1", "turn22view3" ] }, { "id": "B1D02", "title": "Fundsmith Equity Fund Semi-Annual Letter to Shareholders 2026", "date": "2026-07", "date_precision": "month; 原稿July 2026，财务截止2026-06-30", "url": "https://www.fundsmith.co.uk/media/lfhpxi1x/fundsmith-equity-fund-semi-annual-letter-to-shareholders-2026.pdf", "authors": "Terry Smith, CEO, Fundsmith LLP; Fundsmith署名观点", "summary": "开放式资金流约束促使调整持有/买入过程，仍保留质量与合理估值原则。", "access": "官方17页PDF；读取pp5–10的流程修订、署名与费用段；未完整审计交易清单或归因。", "locator": "PDF pp5–7 lines158–244; pp8–10 lines246–314", "checked_at": "2026-10-06", "retrieval_refs": [ "turn9view2", "turn14view0", "turn17view3", "turn22view2" ] }, { "id": "B1D03", "title": "Nomad Investment Partnership Annual Letter, period ended 31 December 2008", "date": "2008-12-31", "date_precision": "period_end; 原信发表日未标，不能当作2008-12-31实际发布日", "url": "https://igyfoundation.org.uk/wp-content/uploads/2021/04/Full_Collection_Nomad_Letters_.pdf", "authors": "Nicholas Sleep署名；涉及Qais Zakaria共同管理及团队推理", "summary": "危机下重查价格/价值，保留资本和思考余量；讨论规模利益回馈客户。", "access": "作者基金会批准选集中的独立年度原信；同容器非独立URL。读取PDF pp147–154正文及p155署名/角色脚注；不转载选集、不核验历史业绩。", "locator": "PDF pp147–155 (printed pp147–155): lines5381–5740; scale section p151 lines5571–5607", "checked_at": "2026-10-06", "retrieval_refs": [ "turn9view3", "turn14view2", "turn17view1", "turn22view1", "turn11view3" ], "container_id": "nomad-approved-collection" }, { "id": "B1D04", "title": "Nomad Investment Partnership Interim Letter, period ended 30 June 2009", "date": "2009-06-30", "date_precision": "period_end; 原信发表日未标", "url": "https://igyfoundation.org.uk/wp-content/uploads/2021/04/Full_Collection_Nomad_Letters_.pdf", "authors": "Nicholas Sleep署名；涉及Qais Zakaria共同管理及团队推理", "summary": "延伸规模回馈、长期持有和估值期限；警惕机械套用均值回归。", "access": "同一作者批准选集中的另一独立半年原信。读取PDF pp156–163、p165署名脚注；未据此认证永久个人哲学。", "locator": "PDF pp156–163 lines5775–6054; p165 lines6098–6104", "checked_at": "2026-10-06", "retrieval_refs": [ "turn14view3", "turn17view0", "turn22view0", "turn17view4" ], "container_id": "nomad-approved-collection" }, { "id": "B1D05", "title": "IGY Foundation official home — identity context", "date": null, "url": "https://igyfoundation.org.uk/", "authors": "IGY Foundation", "summary": "官方自述Nick Sleep与Qais Zakaria创办Nomad，IGY为Nick慈善基金会。", "access": "无日期官方身份说明，2026-10-06读取；不计作独立有日期研究材料，也不证明当前证券持仓。", "checked_at": "2026-10-06", "retrieval_refs": [ "turn22view4", "turn7search1" ] }, { "id": "B2D01", "title": "PSCM Annual Investor Presentation", "date": "2026-02-11", "url": "https://assets.pershingsquareholdings.com/wp-content/uploads/2026/02/11144917/2026-Annual-Investor-Presentation.pdf", "access": "官方85页PDF；本次审阅slides 8–16、22–23、38及免责声明，不宣称全报告审阅。", "authors": "Pershing Square Capital Management team；非Bill Ackman独著", "summary": "区分基金NAV和股价回报，展示资本配置、融资结构及核心选股约束。", "role_at_publication": "买方机构；slide 38列Bill Ackman为CEO / Portfolio Manager，Ryan Israel为CIO。", "retrieved_at": "2026-10-06", "evidence_type": "original official presentation" }, { "id": "B2D02", "title": "PSH 2026 Interim Report — Investment Manager’s Letter", "date": "2026-08-13", "url": "https://assets.pershingsquareholdings.com/wp-content/uploads/2026/08/13002306/Pershing-Square-Holdings-Ltd.-June-2026-Interim.pdf", "access": "官方59页PDF；本次审阅Investment Manager’s Report / LETTER TO SHAREHOLDERS及署名（物理PDF第8–10页，索引P7–P9）、公司/资本结构章节。材料列表显示2026-08-12，董事长报告签署2026-08-13，信件无独立日期；不是六月期末发表。", "authors": "William A. Ackman（CEO）与Ryan Israel（CIO）联合署名", "summary": "将业务盈利增长、估值倍数波动及基金折价分开解释；讨论多年跟踪公司后择价买入。", "role_at_publication": "PSCM买方CEO与CIO，见Investment Manager’s Letter末尾联署（物理PDF第10页）；不是卖方观点。", "retrieved_at": "2026-10-06", "evidence_type": "original official investor letter", "date_basis": "Report signed August 13, 2026; official materials index lists August 12, 2026; letter itself has no separate date." }, { "id": "B2D03", "title": "David Einhorn — Sohn Investment Conference, Solve AI", "date": "2024-04-03", "url": "https://www.greenlightcapital.com/Download.aspx?ID=9d7d933c-1141-4e41-a630-7b0b17a13687", "mirror_url": "https://www.scribd.com/document/762996515/Sohn-Investment-Conference-Presentation-April-3-2024", "access": "官方主页公开列出的47页演示链接因octet-stream解析失败；读取Scribd公开逐页原文，封面确认David Einhorn和日期，审阅slides 1–6、40–44。镜像并非机构授权完整性认证；不复制演示。", "authors": "David Einhorn演示；Greenlight机构披露与估算；不是上传者Jeffrey原创", "summary": "从现金回收和周期正常化评价Solvay，区分当期低谷盈利与长期目标。", "role_at_publication": "Sohn官方2024议程明确David Einhorn为President, Greenlight Capital。", "role_source_url": "https://www.sohnconference.org/sohn2024", "retrieved_at": "2026-10-06", "evidence_type": "original presentation inspected through public mirror" }, { "id": "B2D04", "title": "Greenlight Capital Q1 2024 Partner Letter", "date": "2024-04-24", "url": "https://imonkey-files.s3-us-west-1.amazonaws.com/Greenlight%20Letter-Q1.pdf", "access": "公开9页原始信PDF镜像，信头日期与末尾机构落款可见；审阅pp.1–2、5–8。并非Greenlight官网独立认证的副本；不转载含保密声明的原文。", "authors": "落款Greenlight Capital；机构共同意见，个人贡献未披露", "summary": "提出少依赖外部重估、更多依赖可持续回购和分红回收价值的条件。", "role_at_publication": "Greenlight基金管理机构（买方）；个人Einhorn职务由独立2024 Sohn官方议程确认，信件不能改写为其独著。", "retrieved_at": "2026-10-06", "evidence_type": "original institution letter inspected through public mirror" }, { "id": "B2D05", "title": "Greenlight Capital Q3 2025 Partner Letter", "date": "2025-10-14", "url": "https://hedgefundalpha.com/wp-content/uploads/2025/10/Qlet2025-03-1_unlocked_watermark.pdf", "access": "公开9页原始信PDF镜像，有水印和部分字符OCR错误；审阅pp.1–5、7–8日期、署名与披露；未以Hedge Fund Alpha新闻摘要替代信件。", "authors": "落款Greenlight Capital；DME Capital Management LP d/b/a Greenlight机构意见，非Einhorn独著认证", "summary": "质疑AI产业支出融资和收入加总口径；同时承认技术长期影响和投资回报可以不同。", "role_at_publication": "Greenlight买方机构；基金观点带有持仓和利益关系，不是中立产业统计。", "retrieved_at": "2026-10-06", "evidence_type": "original institution letter inspected through public mirror" }, { "id": "B2D06", "title": "PSH 2025 Annual Report — Bill Ackman Letter to Shareholders", "date": "2026-02-18", "url": "https://assets.pershingsquareholdings.com/wp-content/uploads/2026/02/18175039/Pershing-Square-Holdings-Ltd.-2025-Annual-Report.pdf", "mirror_url": "https://alpha-file.com/content/8121", "access": "官网列表确认发布日期及原PDF；原文件13MB超web读取上限，读取Alpha-File公开逐段原文；审阅印刷pp.6–10，末尾William A. Ackman署名明确。未把镜像页面错误的Q4分类日期当发表日期。", "authors": "William A. Ackman个人署名信；报告其余章节有团队/董事会作者", "summary": "个人信讨论市场价格发现与经营价值，使用长期股东视角；并非2026-10本周观点。", "role_at_publication": "买方PSCM CEO/Portfolio Manager；同期官方2026演示slide38核验职务。", "retrieved_at": "2026-10-06", "evidence_type": "original signed investor letter inspected through public mirror" }, { "id": "B2D07", "title": "David Einhorn — Robin Hood Investors Conference, Peloton Stock Pitch", "date": "2024-10-23", "url": "https://www.greenlightcapital.com/Download.aspx?ID=c9a435af-fdc1-4ee3-a2fd-f869be7f6952", "mirror_url": "https://www.scribd.com/document/953310072/Robin-Hood-Investors-Conference-October-23-2024", "access": "官网公开索引列此日期；官方octet-stream读取失败；30页演示公开逐页镜像封面David Einhorn和10/23/24，审阅slides1–2、15–23、27–29。上传者不是原创作者，镜像完整性尚未逐字官方校验。", "authors": "David Einhorn演示；估算属于Greenlight机构", "summary": "分拆硬件和订阅利润，比较成本结构并区分SBC前后盈利；同行基准不保证削减成本可实现。", "role_at_publication": "买方Greenlight；当年Sohn官方议程核验President，Robin Hood官网2024-10-21会前公告确认其Greenlight affiliation。", "role_source_url": "https://robinhood.org/news/robin-hood-launches-pick-a-ticker-competition-to-fuel-poverty-fighting-programs/", "retrieved_at": "2026-10-06", "evidence_type": "original named presentation inspected through public mirror" }, { "id": "B3D1", "title": "Market-Expected Return on Investment: Bridging Accounting and Valuation", "date": "2021-04-14", "url": "https://www.morganstanley.com/content/dam/im/assets/publication/thought-leadership/consilient-observer/article_marketexpectedreturnoninvestment_en.pdf", "authors": "Michael J. Mauboussin / Dan Callahan, CFA; Counterpoint Global Insights", "source_role": "MSIM Counterpoint Global买方研究共同署名；该PDF未注明作者当时完整职称", "access": "官方23页PDF；核查封面、pp2–5、8及15；原文日期2021-04-14，当前PDF页脚©2025不作为新文章日期。非全报告数字/底层数据审计。", "checked_at": "2026-10-06", "summary": "从价格内含的经营预期反推投资回报门槛；区分公司投资回报与投资者证券回报。" }, { "id": "B3D2", "title": "ROIC and the Investment Process: ROICs, How They Change, and Shareholder Returns", "date": "2023-06-06", "url": "https://www.morganstanley.com/im/publication/insights/articles/article_roicandtheinvestmentprocess.pdf", "authors": "Michael J. Mauboussin / Dan Callahan, CFA; Counterpoint Global Insights", "source_role": "MSIM Counterpoint Global买方研究共同署名；不等同Morgan Stanley卖方股票覆盖研报", "access": "官方31页PDF；核查pp1–3、8–10；原文日期2023-06-06，当前PDF页脚©2024不作为新文章日期。未复制或独立核验FactSet底层数据。", "checked_at": "2026-10-06", "summary": "把ROIC、增长投资、现金流和股价内含预期结合；同时讨论无形投资调整与ROIC回归。" }, { "id": "B3D3", "title": "Bayes and Base Rates 2.0: How History Can Guide Our Assessment of the Future", "date": "2026-05-13", "url": "https://www.morganstanley.com/content/dam/im/assets/publication/thought-leadership/consilient-observer/article_bayesandbaserates2_ltr.pdf", "authors": "Michael J. Mauboussin / Dan Callahan, CFA; Counterpoint Global Insights", "source_role": "MSIM Counterpoint Global买方研究共同署名", "access": "官方10页PDF与官方摘要；核查pp1–6。仅方法层面转述，不把文中媒体引用的OpenAI或Anthropic数字作为独立核实的公司财务。", "checked_at": "2026-10-06", "summary": "以合适参照类建立先验，并根据新证据更新；历史罕见不等于未来不可能，增长也不自动创造价值。" }, { "id": "S1D01", "title": "NVIDIA FY2026 Q4 Earnings Call — issuer-hosted corrected transcript", "date": "2026-02-25", "url": "https://s201.q4cdn.com/141608511/files/doc_financials/2026/q4/NVDA-Q4-2026-Earnings-Call-25-February-2026-5_00-PM-ET.pdf", "authors": "NVIDIA management; Vivek Arya / BofA and Harlan Sur / JPMorgan attributed questions; NVIDIA management; Aaron Rakers / Wells Fargo Securities question", "summary": "Arya询问客户资本开支与现金流约束；Sur询问网络收入构成和未来产品。 / 核查Rakers关于Vera独立CPU需求的提问及角色名录。", "access": "公司托管FactSet更正稿；核查p2角色名录、pp8–10两位指定Q&A，不是许可研报全文，未音频复核。 / Additional checked scope: 发行人托管FactSet更正稿；核查PDF pp2,15–16具名问题，非许可研报；管理层技术说法未独立全核。", "checked_at": "2026-10-06" }, { "id": "S1D02", "title": "Lam Research FY2025 Q3 Earnings Call — issuer-hosted edited transcript", "date": "2025-04-23", "url": "https://filecache.investorroom.com/mr5ir_lamresearch2/1443/download/LRCX-USQ_Transcript_2025-04-23.pdf", "authors": "Lam management; Harlan Sur / JPMorgan attributed questions", "summary": "Sur把关税风险连接到制造地点、调配时滞及材料库存成本。", "access": "Lam investorroom公司文件库托管Refinitiv编辑稿；核查p2身份和pp9–10 Sur问答，管理层未披露产地/成本全量细节。", "checked_at": "2026-10-06" }, { "id": "S1D03", "title": "Analog Devices FY2026 Q3 Earnings Call — issuer-hosted transcript", "date": "2026-08-19", "url": "https://investor.analog.com/static-files/ac142f15-c3fb-44b3-9c2c-4274a48d843a", "authors": "ADI management; Harlan Sur / JPMorgan and Vivek Arya / Bank of America attributed questions", "summary": "Sur拆分毛利改善；Arya追问增长中的长期、周期和价格因素。", "access": "公司托管LSEG编辑稿；核查p2角色、pp6–8指定Q&A，不是分析师研报。Arya结构化名录机构别名有噪声；主持人明确Bank of America Securities，保留差异。", "checked_at": "2026-10-06" }, { "id": "S1D04", "title": "Applied Materials analyst coverage — current official role check", "date": null, "url": "https://ir.appliedmaterials.com/stock-information/analyst-coverage", "authors": "Applied Materials Investor Relations", "summary": "当前名单列Bank of America / Vivek Arya和JP Morgan / Harlan Sur。", "access": "官方无发布日期覆盖页，2026-10-06读取。仅核查当前上市公司覆盖关系，不能用作有日期思想材料，也不证明雇佣条款。", "checked_at": "2026-10-06" }, { "id": "S2D-B01", "title": "Cirrus Logic FY2025 Q2 Earnings Call", "date": "2024-11-04", "url": "https://s21.q4cdn.com/307563939/files/doc_events/2024/Nov/04/CORRECTED-TRANSCRIPT_-Cirrus-Logic-Inc-CRUS-US-Q2-2025-Earnings-Call-4-November-2024-5_00-PM-ET.pdf", "authors": "Cirrus management; Christopher Rolland / Susquehanna attributed questions", "access": "发行人托管FactSet更正稿；本地PDF文字提取，核查pp2、6–7本人Q&A；未复核音频，非卖方研报。", "summary": "提问拆解内容、单位、提前出货与财年周数对季节性的影响。", "checked_at": "2026-10-06" }, { "id": "S2D-B02", "title": "Cirrus Logic FY2026 Q1 Earnings Call", "date": "2025-08-05", "url": "https://s21.q4cdn.com/307563939/files/doc_financials/2026/q1/Cirrus-Logic_Q1-FY26-Earnings-Call.pdf", "authors": "Cirrus management; Christopher Rolland / Susquehanna attributed questions", "access": "发行人托管FactSet更正稿；核查pp2、6–7本人Q&A；未复核音频，不代表完整许可研究。", "summary": "区分收入超指引的量/内容/库存/关税提前订单，并追问PC和汽车多元化兑现。", "checked_at": "2026-10-06" }, { "id": "S2D-B03", "title": "Lam Research FY2026 Q1 Earnings Call", "date": "2025-10-22", "url": "https://filecache.investorroom.com/mr5ir_lamresearch2/1467/LRCX%20Q1%202026%20Transcript%2010-22-25.pdf", "authors": "Lam management; Blayne Curtis / Jefferies attributed questions", "access": "发行人托管Refinitiv更正稿扫描PDF；下载后视觉核查p16具名Q&A，未音频复核；网页全文因大小限制。", "summary": "追问投资增长的后半年权重究竟是整个WFE还是具体终端分部。", "checked_at": "2026-10-06", "additional_readings": [ { "date": "2026-10-08", "analyst": "Brian Chin", "locator": "Brian Chin具名问答；本轮读取Roic同场镜像", "summary": "要求拆分AI支出到WFE构成，并检验升级与新建是否并行。", "access": "公开同场转录镜像已读取；issuer PDF直接403，web超体积，未听录音", "url": "https://www.roic.ai/quote/LRCX/transcripts/2026-year/1-quarter" } ] }, { "id": "S2D-B04", "title": "Lam Research FY2026 Q2 Earnings Call", "date": "2026-01-28", "url": "https://filecache.investorroom.com/mr5ir_lamresearch2/1474/LRCX%20earnings%20transcript%201-28-26.pdf", "authors": "Lam management; Blayne Curtis / Jefferies attributed questions", "access": "发行人托管LSEG更正稿扫描PDF；下载后视觉核查p17两次具名Q&A，未音频复核。", "summary": "追问Reliant区域需求构成、NAND升级与洁净室约束及兑现节奏。", "checked_at": "2026-10-06" }, { "id": "S3D02", "title": "NVIDIA FY2027 Q2 Earnings Call", "date": "2026-08-26", "url": "https://investor.nvidia.com/files/content_files/TRANSCRIPT_-NVIDIA-Corp-NVDA-US-Q2-2027-Earnings-Call-26-August-2026-5_00-PM-ET.pdf", "authors": "NVIDIA management; Aaron Rakers / Wells Fargo Securities question", "summary": "核查每GW价值密度、路线图与供给能力的具名提问。", "access": "发行人托管FactSet更正稿；核查PDF pp2,14–15。不是分析师预测、目标价或研报；未将管理层回报主张当实证。", "role_at_publication": "Aaron Rakers, Analyst, Wells Fargo Securities LLC", "checked_at": "2026-10-06", "additional_readings": [ { "title": "NVIDIA Q2 FY2027 Earnings Call", "summary": "具名问题涉及增长桥、供给约束、生态投资与开放模型。", "access": "发行人官方CDN完整逐字稿可访问；已读具名问题及相邻回答，未听音频。", "url": "https://s201.q4cdn.com/141608511/files/content_files/TRANSCRIPT_-NVIDIA-Corp-NVDA-US-Q2-2027-Earnings-Call-26-August-2026-5_00-PM-ET.pdf", "checked_at": "2026-10-07", "same_underlying_artifact": true } ] }, { "id": "B100-BR1", "title": "Technology outlook: AI, macro and valuations", "date": "2024-09-25", "url": "https://www.polarcapital.co.uk/static/documents/Polar_Capital_Global_Technology_Fund_-_Technology_outlook_-_AI_macro_and_valuations.pdf", "access": "2026-10-07读取全部访谈正文；PDF第1页；其余为风险及法律声明。", "authors": "Ben Rogoff; interviewer Vik Heerah", "summary": "Rogoff区分股价波动与模型进展，并以相对估值、增长和资本开支判断AI主题。" }, { "id": "B100-BR2", "title": "Investment Manager’s Report — Annual Highlights FY2026", "date": null, "url": "https://www.pctannualhighlights.co.uk/Investment-Managers-Report/", "access": "2026-10-07逐段读取网页全文；对应截至2026-04-30财年；未核验嵌入图表数值或完整年度财务报表。", "authors": "Ben Rogoff and Alastair Unwin", "summary": "共同署名报告讨论AI时代利润池向模型和基础设施迁移、云厂商资本密集度及软件风险。" }, { "id": "B100-TK1", "title": "At the frontier: Insight on AI and technology stocks", "date": "2025-09-11", "url": "https://www.blackrock.com/us/individual/insights/insight-on-ai-and-technology-stocks", "access": "2026-10-07读取官方嵌入访谈文字；未以音频逐字比对。", "authors": "Tony Kim, Michael Gates; host Oscar Pulido", "summary": "Kim以基础设施、智能模型、应用服务三层描述AI投资，强调资本强度和技术扩散。" }, { "id": "B100-TK2", "title": "Tracking the AI investment opportunity", "date": null, "url": "https://www.blackrock.com/us/individual/insights/tracking-the-ai-investment-opportunity", "access": "2026-10-07读取官方访谈文字正文；涉及2026年技术巡访，页面引注截至2026年7月，未找到明确发布日期。", "authors": "Tony Kim; host Oscar Pulido", "summary": "Kim把数据中心投入换算为功率及供应链需求，并比较三层AI价值捕获；这些是其预测与解释，非已实现事实。" }, { "id": "B100-RC1", "title": "Are tech sector investors too bullish on AI’s promise?", "date": "2025-08-13", "url": "https://www.janushenderson.com/en-ch/investor/article/are-tech-sector-investors-too-bullish-on-ais-promise/", "access": "2026-10-07读取全文Q&A；图表底层数据未独立复算。", "authors": "Richard Clode", "summary": "Clode沿用PE与FCF收益率，认为利润、资产负债表及使用场景比单纯泡沫类比更关键。" }, { "id": "B100-RC2", "title": "Riding the AI wave: How can tech investors harness change and volatility?", "date": "2025-12-04", "url": "https://www.janushenderson.com/en-gb/investor/article/riding-the-ai-wave-how-can-tech-investors-harness-change-and-volatility/", "access": "2026-10-07读取文章全文；团队共同署名，不把所有句子归为Clode个人原话。", "authors": "Alison Porter, Graeme Clark and Richard Clode", "summary": "团队预计推理、记忆与物理AI扩大算力需求，同时关注许可、电力、施工限制、循环投资和亏损科技股炒作。" }, { "id": "B100-PW1", "title": "A deep dive into the global chip shortage — A roundtable discussion", "date": "2021-06", "url": "https://www.columbiathreadneedle.com/uploads/2021/07/99c0a1fbeae011bf80eda27cfde9f89b/hk_j31619-semicondcutor-chip-shortage-apac.pdf", "access": "2026-10-07读取PDF第1至3页全部圆桌问答；图表未复算，第4页为法律说明。", "authors": "Paul Wick, Rahul Narang, Sanjay Devgan, David Egan, Charles Mann", "summary": "Wick解释代工集中与高利用率减少供给余量，短期强需求又会诱发后续产能扩张。" }, { "id": "B100-PW2", "title": "Artificial intelligence and the economy: paradigm shift or hype?", "date": "2023-09-27", "url": "https://www.columbiathreadneedle.com/en/dk/institutional/insights/artificial-intelligence-and-the-economy-paradigm-shift-or-hype/", "access": "2026-10-07读取网页全部圆桌正文，区分三位受访者。", "authors": "Paul Wick, Rahul Narang and Sanjay Devgan", "summary": "Wick将生成式AI视为知识工作工具，与互联网和移动设备的分发作用有所区别。" }, { "id": "B100-AB1", "title": "Information technology sector", "date": "2026-01-07", "url": "https://institutional.fidelity.com/advisors/insights/spotlights/equity-sector-performance-outlook/information-technology-sector", "access": "2026-10-07读取官方全文；持仓说明截至2025-11-30，文内预测保留原时点。", "authors": "Adam Benjamin", "summary": "Benjamin认为持续模型升级支撑AI基础设施，但固定IT预算可能挤压服务支出、模型应用层可能冲击软件。" }, { "id": "B100-AB2", "title": "Fidelity Select Semiconductors Portfolio — Portfolio Manager Q&A as of February 28, 2026", "date": "2026-02-28", "url": "https://institutional.fidelity.com/app/proxy/content?literatureURL=%2FRD_QAA_008.PDF", "access": "2026-10-07读取完整投资方法与问答正文（PDF第2至4页）；未对全9页绩效表与财务数据做独立审计。URL为滚动文件，日期按正文。", "authors": "Adam Benjamin / Fidelity", "summary": "Benjamin承认过去两年非AI复苏判断过早，重新依据库存、需求和供应变化审视周期，并讨论偏离指数的收益代价。" }, { "id": "B100-ABROLE", "title": "Fidelity Select Semiconductors summary prospectus supplement", "date": "2026-07-15", "url": "https://www.sec.gov/Archives/edgar/data/320351/000032035126000312/filing12829.htm", "access": "2026-10-07读取全部17行SEC正文；仅身份更正，不算实质研究来源。", "authors": "Fidelity filing", "summary": "Adam Benjamin不再担任该基金经理；Sonu Kalra自2026年起管理。" }, { "id": "B100-GB1", "title": "Investing Mistakes, Chapter 1001", "date": "2020-08-14", "url": "https://gavin-baker.medium.com/investing-mistakes-chapter-1001-9f0b7dbc6637", "authors": "Gavin Baker", "summary": "自述Intel投资失误：检验10nm时没有充分检验7nm假设；清单不能弥补关键变量遗漏。", "access": "2026-10-07；读取全部文章正文；技术判断仅为2020作者主张。" }, { "id": "B100-GB2", "title": "Gavin Baker and David George on Positional Strategy in AI", "date": "2025-10-30", "url": "https://www.a16z.news/p/gavin-baker-and-david-george-on-positional", "authors": "Gavin Baker; interviewer David George", "summary": "认为AI计算成本改变软件经济，利润率下降须与利润额和资本需求分开判断。", "access": "2026-10-07；官方主持机构编辑转录正文全部读取；未独立审计演讲中技术及市场数字。" }, { "id": "B100-BG1", "title": "Time to Get Fit — an Open Letter from Altimeter to Mark Zuckerberg (and the Meta Board of Directors)", "date": "2022-10-24", "url": "https://medium.com/@alt.cap/time-to-get-fit-an-open-letter-from-altimeter-to-mark-zuckerberg-and-the-meta-board-of-392d94e80a18", "authors": "Brad Gerstner", "summary": "建议削减人员成本并限制资本开支和元宇宙支出，保留AI投资信念。", "access": "2026-10-07；自署公开信全文已读；建议与测算为2022主张，并非实际实现结果。" }, { "id": "B100-BG2", "title": "Altimeter’s Brad Gerstner on Macro, Tech and Startups", "date": "2023-12-05", "url": "https://blog.eladgil.com/p/altimeters-brad-gerstner-on-macro", "authors": "Brad Gerstner; interviewer Elad Gil", "summary": "将技术乐观与交易价格分离；要求估值最终转为现金流，并检验回报概率和证券上行限制。", "access": "2026-10-07；主持人发布访谈转录正文全部读取；日期为发布日期，非确认活动日。" }, { "id": "BX-HS-20240812", "title": "AI Hype vs. Reality: Navigating Tech Investments with Discipline", "date": "2024-08-12", "url": "https://gabelli.com/research/navigating-market-turbulence-strategies-from-portfolio-managers/", "authors": "Hendi Susanto（该署名小节）", "access": "2026-10-07已读取完整Hendi署名小节；其他作者段落不归于他", "summary": "AI可推动多年半导体投资，但传统终端库存调整与过高估值仍能并存；强调竞争结构和安全边际。" }, { "id": "BX-HS-20250422", "title": "Gabelli Japan Research: Finding Technology Leaders", "date": "2025-04-22", "url": "https://gabelli.com/research/gabelli-japan-research-finding-technology-leaders/", "authors": "Ryuta Makino and Hendi Susanto", "access": "2026-10-07读取正文主要研究段落及作者披露；网页图表未独立重算。共同署名，不能拆为Hendi独立研究。", "summary": "通过日本企业调研讨论技术换代、客户集中、库存与利润传导；本文的市场份额和管理层预测均为当时材料陈述，未经当前复核。" }, { "id": "BX-ZWS-20210910", "title": "投资洞见半导体：如何看待近期半导体板块回调？", "date": "2021-09-10", "url": "https://www.cgf.cn/pc/pc/service/articleView?columnId=5180&i_articleid=60534", "authors": "郑巍山", "access": "2026-10-07官方完整正文读取", "summary": "区分局部存储价格变化与全行业景气，关注产品升级、国产渗透，以及设计公司的成本转嫁能力。" }, { "id": "BX-ZWS-20220609", "title": "山哥请回答·5月报：科技、新能源板块一季报", "date": "2022-06-09", "url": "https://www.cgf.cn/pc/pc/service/articleView?columnId=5180&i_articleid=62879", "authors": "郑巍山（官方问答）", "access": "2026-10-07官方完整正文读取；引用券商统计仅作为本人当时采用的证据，未独立重算", "summary": "以收入、利润和利润率检验行业判断；强调代工涨价不意味着全面涨价，产品竞争力影响设计商转嫁成本。" }, { "id": "B71D01", "title": "Berkshire Hathaway 2023 Shareholder Letter", "date": "2024-02-24", "url": "https://www.berkshirehathaway.com/letters/2023ltr.pdf", "access": "官方PDF，2026-10-07读取P0–P3实质正文。", "authors": "Warren Buffett", "summary": "区分经营盈利与未实现投资损益，重视可持续经济优势及高回报增量资本。" }, { "id": "B71D02", "title": "Berkshire Hathaway 2024 Shareholder Letter", "date": "2025-02-22", "url": "https://www.berkshirehathaway.com/letters/2024ltr.pdf", "access": "官方PDF，2026-10-07读取P0–P4实质正文。", "authors": "Warren Buffett", "summary": "承认资本配置与识人错误，强调及时纠错；讨论激励与新增资本的匹配。" }, { "id": "B71D03", "title": "Bill Nygren Market Commentary | 4Q22", "date": "2022-Q4", "url": "https://oakmark.com/news-insights/bill-nygren-market-commentary-4q22/", "access": "本人访谈由雇主发布，2026-10-07读取分析师角色等实质段落。精确日未核，日期为季度标签。", "authors": "Bill Nygren interviewed by John Rotonti", "summary": "分析师应识别共识的重大错误，计算估值影响，并以新信息检验自己或共识的正确性。" }, { "id": "B71D04", "title": "The discipline to stay boring | U.S. equity market commentary 2Q 2026", "date": "2026-06-30", "url": "https://oakmark.com/wp-content/uploads/sites/3/2026/07/The-discipline-to-stay-boring_U.S.-equity-market-commentary_2Q26_vF.pdf", "access": "本人签名官方PDF，2026-10-07读取P0–P3全文论证。", "authors": "William C. Nygren", "summary": "区分AI技术重要性、盈利持续期与股价安全边际；不持有硬件不等于断言其高估。" }, { "id": "EXP200-B28-A", "title": "Philippe Laffont at The Economic Club of New York", "date": "2025-06-12", "url": "https://econclubny.globalnorthstar.net/documents/20124/74672/20250612_Laffont_Transcript.pdf/cfd73cf3-b091-ba07-49a7-ddb48fac86a4", "authors": "Philippe Laffont; moderator Glenn Hutchins", "access": "2026-10-07读取主办方完整33页转录中的实质段落pp.9–18；未听音视频。", "retrieval_level": "primary host full transcript; substantive passages read", "summary": "Laffont以少数长期主题扩展产业链机会，同时描述亏损时提高现金的做法。他把GPU需求延伸到能源、设备和基础设施。其历史交易叙述未独立审计。" }, { "id": "EXP200-B28-B", "title": "Coatue Innovative Strategies Fund Prospectus", "date": "2026-05-01", "url": "https://www.coatue.com/ctek/documents/Coatue_-_CTEK_Prospectus__May_1__2026_.pdf", "authors": "Coatue / Fund institutional disclosure; not a personally authored Laffont essay", "access": "2026-10-07读取完整105页文件中的Coatue Overview（PDF p.28）及Portfolio Management（PDF p.59）等相关条文；并非全文逐页审计。", "retrieval_level": "primary full prospectus; substantive research/risk/PM sections read", "summary": "机构披露把选股、头寸和现金管理结合，Mosaic支持模型和持续监测；Laffont负责基金组合和策略，研究人员仍作投资决策。该文件支持机构流程，不证明每条都是个人独创。" }, { "id": "E7S01", "title": "NVIDIA Q1 FY2027 Earnings Call", "date": "2026-05-20", "url": "https://s201.q4cdn.com/141608511/files/doc_financials/2027/q1/NVDA-Q1-2027-Earnings-Call-20-May-2026-5_00-PM-ET.pdf", "access": "发行人官方CDN完整逐字稿可访问；已读具名问题及相邻回答，未听音频。", "authors": "NVIDIA management; named sell-side questioners", "summary": "具名问题涉及分部边界、独立CPU增量与专用推理产品定位。" }, { "id": "E7S03", "title": "Analog Devices at Nasdaq Investor Conference", "date": "2023-06-13", "url": "https://investor.analog.com/static-files/14a935bd-d7ed-460e-a82c-838ad6ff3d5d", "access": "发行人官方PDF；已读Lipacis关于订单、交期、分销库存的问答。历史身份为Jefferies。", "authors": "Mark Lipacis questions; Prashanth Mahendra-Rajah management answers", "summary": "Lipacis对照订单和交期，并质疑公司渠道叙事与分销商库存报表的差异。" }, { "id": "E7S04", "title": "Texas Instruments Q4 2011 Mid-Quarter Update", "date": "2011-12-08", "url": "https://www.sec.gov/Archives/edgar/data/97476/000009747611000061/transcript.htm", "access": "SEC归档原始逐字稿；已读Lipacis具名问题及相邻回答。历史身份为Jefferies。", "authors": "Mark Lipacis questions; Ron Slaymaker management answers", "summary": "Lipacis追问渠道库存周数与此前基线；管理层未给实时精确数值。" }, { "id": "S36SRC1", "title": "TSMC Q2 2025 earnings call — Charlie Chan Q&A", "date": "2025-07-17", "url": "https://investor.tsmc.com/english/encrypt/files/encrypt_file/reports/2025-07/1f4f86c935f1de837672a6973154e64b26bdae57/TSMC%202Q25%20Transcript.pdf", "access": "公司官网完整PDF正文可检索；已读Charlie Chan问答印刷页pp.8–9（PDF零起始索引7–8）与参会名单，2026-10-07。", "authors": "Charlie Chan (questions); C.C. Wei (answers)", "summary": "Chan把汇率压力与定价联系起来，并追问中国市场变化是否改变AI增长假设。回答属于TSMC管理层。", "retrieval_level": "primary-full-text; relevant Q&A read", "source_type": "official earnings transcript", "additional_readings": [ { "date": "2026-10-08", "scope": "新增阅读Bruce Lu pp.9–10、Gokul Hariharan pp.6–8。", "access": "官方正文重新取得并阅读；与原有同一次电话会去重。", "summary": "新增阅读Bruce Lu pp.9–10、Gokul Hariharan pp.6–8。" }, { "date": "2026-10-08", "analyst": "Mehdi Hosseini", "locator": "pp.17–18", "summary": "追问N2流片增长是否需要更高资本强度。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://investor.tsmc.com/english/encrypt/files/encrypt_file/reports/2025-07/1f4f86c935f1de837672a6973154e64b26bdae57/TSMC%202Q25%20Transcript.pdf" }, { "date": "2026-10-08", "analyst": "Brett Simpson", "locator": "pp.11–12", "summary": "追问结构性成本、涨价与AI效率能否抵消毛利压力。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://investor.tsmc.com/english/encrypt/files/encrypt_file/reports/2025-07/1f4f86c935f1de837672a6973154e64b26bdae57/TSMC%202Q25%20Transcript.pdf" } ] }, { "id": "S36SRC2", "title": "TSMC Q3 2025 earnings call — Charlie Chan Q&A", "date": "2025-10-16", "url": "https://investor.tsmc.com/chinese/encrypt/files/encrypt_file/reports/2025-10/6860312f04fd291d0f26b46c1234f84e6332717e/TSMC%203Q25%20Transcript.pdf", "access": "公司官网完整PDF正文可检索；已读Charlie Chan问答印刷页pp.8–9（PDF零起始索引7–8），2026-10-07。", "authors": "Charlie Chan (questions); C.C. Wei (answers)", "summary": "Chan追问系统创新与先进制程需求如何兼容，并再次检验中国限制与长期AI增长的关系。", "retrieval_level": "primary-full-text; relevant Q&A read", "source_type": "official earnings transcript", "additional_readings": [ { "date": "2026-10-08", "scope": "新增阅读Bruce Lu pp.11–13、Gokul Hariharan pp.6–7。", "access": "官方正文重新取得并阅读；与原有同一次电话会去重。", "summary": "新增阅读Bruce Lu pp.11–13、Gokul Hariharan pp.6–7。" }, { "date": "2026-10-08", "analyst": "Krish Sankar", "locator": "pp.15–16", "summary": "询问1GW AI容量如何映射晶圆需求及GPU/ASIC组合。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://investor.tsmc.com/chinese/encrypt/files/encrypt_file/reports/2025-10/6860312f04fd291d0f26b46c1234f84e6332717e/TSMC%203Q25%20Transcript.pdf" } ] }, { "id": "S51SRC1", "title": "Microsoft FY2025 Q2 earnings call — Keith Weiss Q&A", "date": "2025-01-29", "url": "https://www.microsoft.com/en-us/investor/events/fy-2025/earnings-fy-2025-q2", "access": "Microsoft官方完整HTML转录；已读Weiss提问与Hood/Nadella回答，2026-10-07。", "authors": "Keith Weiss (question); Amy Hood and Satya Nadella (answers)", "summary": "Weiss追问Azure落在指引低端的执行原因。Hood将非AI销售执行与AI容量约束分开解释；这是管理层判断。", "retrieval_level": "primary-full-text; relevant Q&A read", "source_type": "official earnings transcript", "additional_readings": [ { "date": "2026-10-08", "scope": "新增阅读Karl Keirstead与Brent Thill具名问答。", "access": "官方正文重新取得并阅读；与原有同一次电话会去重。", "summary": "新增阅读Karl Keirstead与Brent Thill具名问答。" }, { "date": "2026-10-08", "analyst": "Brad Zelnick", "locator": "Brad Zelnick具名问答", "summary": "询问推理成本下降后产品包装与覆盖客户。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://www.microsoft.com/en-us/investor/events/fy-2025/earnings-fy-2025-q2" } ] }, { "id": "S51SRC2", "title": "Microsoft FY2025 Q4 earnings call — Weiss and Moerdler Q&A", "date": "2025-07-30", "url": "https://www.microsoft.com/en-us/investor/events/fy-2025/earnings-fy-2025-q4", "access": "Microsoft官方完整HTML转录；已读两人提问与相关回答，2026-10-07。", "authors": "Keith Weiss and Mark Moerdler (questions); Satya Nadella and Amy Hood (answers)", "summary": "Weiss问大客户AI实验室日后自建与竞争的风险；Moerdler问SaaS AI计费与利润率。平台学习收益、混合计费的解释来自管理层。", "retrieval_level": "primary-full-text; relevant Q&A read", "source_type": "official earnings transcript", "additional_readings": [ { "date": "2026-10-08", "scope": "新增阅读Kash Rangan具名问答。", "access": "官方正文重新取得并阅读；与原有同一次电话会去重。", "summary": "新增阅读Kash Rangan具名问答。" }, { "date": "2026-10-08", "analyst": "Raimo Lenschow", "locator": "Raimo Lenschow具名问答", "summary": "检验Copilot如何扩展到数据与agent工作流。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://www.microsoft.com/en-us/investor/events/fy-2025/earnings-fy-2025-q4" } ] }, { "id": "S51SRC3", "title": "Morgan Stanley — Will GenAI Turn a Profit in 2025?", "date": "2025-03-03", "url": "https://www.morganstanley.com/insights/podcasts/thoughts-on-the-market/genai-profit-2025-joe-moore-keith-weiss", "access": "机构官方播客转录正文；已读Weiss发言，2026-10-07；未核验音频及事后预测结果。", "authors": "Keith Weiss; Joe Moore", "summary": "Weiss区分研究、产品开发/基础设施建设、商业采用阶段，讨论自动化下按交易或用量计费。不能把Moore的芯片周期判断归给Weiss。", "retrieval_level": "primary-full-text; Weiss transcript passages read", "source_type": "analyst-authored institutional podcast" }, { "id": "S52SRC2", "title": "Adobe Q3 FY2025 earnings call — Mark Moerdler Q&A", "date": "2025-09-11", "url": "https://www.adobe.com/cc-shared/assets/investor-relations/pdfs/adbe-q3fy25-transcript.pdf", "access": "Adobe官网完整PDF转录；已读Moerdler问答p.17，2026-10-07。", "authors": "Mark L. Moerdler (question); Shantanu Narayen (answer)", "summary": "Moerdler检验用量收入能否弥补潜在席位压力及AI规模化时间；席位和自动化同时扩张是管理层的回答。", "retrieval_level": "primary-full-text; relevant Q&A read", "source_type": "official earnings transcript", "additional_readings": [ { "date": "2026-10-08", "analyst": "Brad Zelnick", "locator": "pp.18–19", "summary": "问AEP agents给客户创造的价值如何被Adobe捕获。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://www.adobe.com/cc-shared/assets/investor-relations/pdfs/adbe-q3fy25-transcript.pdf" }, { "date": "2026-10-08", "analyst": "Alex Zukin", "locator": "pp.15–16", "summary": "拆分AI-first采用与传统价量对ARR改善的贡献。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://www.adobe.com/cc-shared/assets/investor-relations/pdfs/adbe-q3fy25-transcript.pdf" }, { "date": "2026-10-08", "analyst": "Tyler Radke", "locator": "p.14", "summary": "要求拆分提价后的真实业务量。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://www.adobe.com/cc-shared/assets/investor-relations/pdfs/adbe-q3fy25-transcript.pdf" } ] }, { "id": "EXP200-S81-A", "title": "DataBank CEO Raul Martynek Talks Data Centers", "date": "2025-06-13", "event_date": "2025-05-29", "url": "https://www.tdsecurities.com/ca/en/tdc-tmt-conference-databank-talks-data-centers", "authors": "Michael Elias (interviewer); Raul Martynek (management)", "access": "官方完整文字转录可访问；已读相关提问和回答。会议与发布日期分开；未复核全部音频。", "retrieval_level": "primary full transcript; relevant exchanges read", "summary": "Elias拆开成本转嫁、融资成本与开发收益率差，追问供给瓶颈；Martynek认为当时收益率不再继续扩大。" }, { "id": "EXP200-S81-B", "title": "QTS On the Data Center Outlook", "date": "2025-08-27", "event_date": "2025-08-12", "url": "https://www.tdsecurities.com/ca/en/tdc-insights-qts-on-data-center-outlook", "authors": "Michael Elias (official host metadata); Tag Greason (management)", "access": "官方完整文字转录可访问；已读相关提问和回答。会议与发布日期分开；未复核全部音频。", "retrieval_level": "primary full transcript; relevant exchanges read; speaker spelling inconsistency documented", "summary": "Elias追问电力宣称与真实可交付量的差异，并指出同一需求被多个申请放大；Greason区分申请与已承诺电力，强调客户及交付约束。" }, { "id": "EXP200-S92-A", "title": "nVent Q1 2026 Earnings Call", "date": "2026-05-01", "url": "https://s22.q4cdn.com/268397047/files/doc_financials/2026/q1/NVT-Q1-Earnings-Transcript.pdf", "authors": "Julian Mitchell (questioner); Beth Wozniak and Gary Corona (management)", "access": "2026-10-07读取发行人IR CDN完整转录中参与者表及pp.10–11有关问答；未听音频。", "retrieval_level": "issuer-hosted full transcript; attributable exchanges read", "summary": "Mitchell用积压订单、交期和两年累计增长检验收入指引，再检查涨价与通胀对利润率的净影响。" }, { "id": "EXP200-S92-B", "title": "Trane Technologies Q4 2025 Earnings", "date": "2026-01-29", "url": "https://s2.q4cdn.com/950394465/files/doc_financials/2025/q4/Trane-Technologies-Q4-2025-Earnings_Transcript.pdf", "authors": "Julian Mitchell (questioner); Christopher Kuehn and David Regnery (management)", "access": "2026-10-07读取发行人IR CDN完整转录中参与者表及pp.9–10问答；未听音频。", "retrieval_level": "issuer-hosted full transcript; attributable exchanges read", "summary": "Mitchell将订单增长映射到次年收入节奏，并以渠道库存和折价线索检验住宅业务判断；管理层回答交付周期和定价情况。", "additional_readings": [ { "accessed": "2026-10-08", "scope": "pp.10–13 Scott Davis/Andrew Kaplowitz; pp.20–22 Joseph Ritchie/Jeffrey Sprague; pp.24–25 Stephen Tusa；正文问答及p.2名录重读，未听录音。", "authors": "Named analysts asking questions; David Regnery / Christopher Kuehn answering", "summary": "pp.10–13 Scott Davis/Andrew Kaplowitz; pp.20–22 Joseph Ritchie/Jeffrey Sprague; pp.24–25 Stephen Tusa；正文问答及p.2名录重读，未听录音。", "access": "pp.10–13 Scott Davis/Andrew Kaplowitz; pp.20–22 Joseph Ritchie/Jeffrey Sprague; pp.24–25 Stephen Tusa；正文问答及p.2名录重读，未听录音。" } ] }, { "id": "O8BAS01", "title": "Quick View: Has AI really eaten software?", "date": "2026-02-04", "url": "https://www.janushenderson.com/en-gb/investor/article/quick-view-has-ai-really-eaten-software/", "authors": "Graeme Clark", "summary": "讨论AI对席位收入、核心系统与附加模块的不同影响；业务依赖不自动消除商业模式风险。", "access": "2026-10-08 新读取官方原文；正文 Key takeaways 至 concluding paragraph，网页段落125–142；未观看视频或独立核查厂商迁移速度。" }, { "id": "O8BAS02", "title": "Tech stocks and tariffs: 5 key considerations for investors", "date": "2025-04-09", "url": "https://www.janushenderson.com/en-gb/investor/article/tech-stocks-and-tariffs-5-key-considerations-for-investors/", "authors": "Alison Porter, Graeme Clark and Richard Clode", "summary": "团队区分直接关税影响与需求、信心和流动性的间接冲击。", "access": "2026-10-08 新读取官方原文；正文五项考虑与买入时机段落、作者区；历史政策判断仅按2025-04-09口径，不代表现行关税。" }, { "id": "O8BAS03", "title": "2025: The year AI comes of age", "date": "2025-08-21", "url": "https://www.janushenderson.com/en-us/offshore/article/2025-the-year-ai-comes-of-age/", "authors": "Denny Fish", "summary": "把推理算力需求、部署变现及采用者利润率连接，同时提示被替代企业风险。", "access": "2026-10-08 新读取官方原文；正文 If you build it、Enablers and early adopters、Schumpeter 章节；图表原始数据未复算。" }, { "id": "O8BAS04", "title": "Still early innings in the AI investment cycle", "date": "2025-12-02", "url": "https://www.janushenderson.com/en-us/institutional/article/still-early-innings-in-the-ai-investment-cycle/", "authors": "Denny Fish", "summary": "提出使能者、增强者、终端使用者三层，并将电力和施工时滞视作供给调节因素。", "access": "2026-10-08 新读取官方原文；官方完整视频文字稿段落99–112；未与音频逐字校对。满载与避免繁荣衰退是作者判断，非独立验证事实。" }, { "id": "O8BAS05", "title": "Investment Manager’s Report: Summary — FY2024", "date": "2024-07-16", "url": "https://www.pctannualhighlights.co.uk/2024/Investment-Managers-Report/", "authors": "Ben Rogoff and Alastair Unwin", "summary": "共同报告比较盈利、估值和资本再投入，讨论软件增长放缓与利润率改善。", "access": "2026-10-08 新读取官方原文；选读 Technology Outlook 下 Earnings Outlook、Valuation、Magnificent 7、Survival of the fittest 及页面署名日期；非完整年报审阅。" }, { "id": "O8BAS06", "title": "Investment Manager’s Report: Summary — FY2025", "date": null, "url": "https://www.pctannualhighlights.co.uk/2025/Investment-Managers-Report/", "authors": "Ben Rogoff and Alastair Unwin", "summary": "共同报告提高对大平台资本密集度及软件替代风险的关注，披露相应组合调整。", "access": "2026-10-08 新读取官方原文；选读 Mag-7 update、Disruption ahead、Cloud update、AI to the rescue、Technology/AI risks、Concentration risk；FY截至2025-04-30，页面有July 2025图表但无明确发布日期，不把财年末当发表日。" }, { "id": "O8BAS07", "title": "Meet The Manager — Dom Rizzo", "date": "2024-04", "url": "https://www.troweprice.com/financial-intermediary/uk/en/thinking/articles/2024/q2/meet-the-manager-dom-rizzo.html", "authors": "Dom Rizzo", "summary": "具名问答提出关键技术、结构增长、基本面改善与合理估值四项筛选。", "access": "2026-10-08 新读取官方原文；完整问答正文；页面仅标April 2024。" }, { "id": "O8BAS08", "title": "Behind the Markets Podcast: A Conversation with Dom Rizzo", "date": "2023-08-18", "url": "https://www.wisdomtree.com/us/insights/blog/behind-the-markets-podcast-a-conversation-with-dom-rizzo", "authors": "Christopher Gannatti, interviewing and summarizing Dom Rizzo", "summary": "采访主持人原始总结介绍Rizzo四项框架及硬件竞争中软件切换成本。", "access": "2026-10-08 新读取官方原文；主持人发布的完整文章，不是逐字稿；采访发生2023-08-11，未听音频。主张按主持人转述保留，非Rizzo原话。" }, { "id": "O8BAS09", "title": "Podcast: Are investors underestimating AI?", "date": "2025-07-11", "url": "https://www.franklintempleton.com/corporate/articles/2025/equity/are-investors-underestimating-ai", "authors": "Jonathan Curtis; host John Przygocki", "summary": "讨论AI生产率的公司利润桥、研究工具与人类投资判断边界。", "access": "2026-10-08 新读取官方完整文字稿；官方Franklin Equity Insights索引标示2025-07-11发布，文末观点截至2025-07-01；未与音频校对。", "date_evidence_url": "https://www.franklintempleton.co.uk/insights-collections/franklin-equity-insights" }, { "id": "O8BAS10", "title": "The AI productivity boom that no one is talking about", "date": "2025-07-24", "url": "https://www.franklintempleton.co.uk/articles/2025/equity/the-ai-productivity-boom-that-no-one-is-talking-about", "authors": "Jonathan Curtis", "summary": "具名文章将关注从AI供给商扩展到各行业采用者。", "access": "2026-10-08 新读取官方原文；新读取官网文章正文及结论；24 July 2025来自官方Franklin Equity Insights索引；所链PDF下载失败，未声称读完整白皮书。", "date_evidence_url": "https://www.franklintempleton.co.uk/insights-collections/franklin-equity-insights" }, { "id": "O8BB-TS24", "title": "Tom Slater: Manager Review — November 2024", "date": "2024-11", "url": "https://www.bailliegifford.com/en/uk/individual-investors/insights/ic-article/sm-articles-2024-q4-tom-slater-manager-review-10051274/", "authors": "Tom Slater", "summary": "在AI成本与资本开支可持续性之间建立约束，并区分技术信心和具体持仓。", "access": "2026-10-08：官方HTML正文读取；重点Artificial Intelligence及Capital Allocation；仅披露月份，未核验交易明细或图表。" }, { "id": "O8BB-TS25", "title": "Tom Slater: Manager Review — May 2025", "date": "2025-05", "url": "https://www.bailliegifford.com/en/belgium/professional-investor/insights/ic-article/sm-articles-2025-bg-tom-slater-manager-review-10055356/", "authors": "Tom Slater", "summary": "讨论AI普及下的供应商价值捕获、风险回报及企业适应力。", "access": "2026-10-08：官方HTML正文读取；重点The Rise of AI and the Shifting Shape of Value、Resilience as a Strategic Advantage；未重算基金业绩。" }, { "id": "O8BB-LB24", "title": "Lawrence Burns on artificial intelligence: manager review", "date": "2024-05", "url": "https://www.scottishmortgage.com/en/uk/individual-investors/insights/ic-article/sm-articles-2024-q2-lawrence-burns-manager-review-10047502", "authors": "Lawrence Burns", "summary": "用硬件、基础设施、应用三层定位受益与时滞，同时保留硬件需求消化期的可能。", "access": "2026-10-08：官方HTML正文读取；The Layers of Artificial Intelligence、Applications、Conclusion；技术参数与市场数字是作者当时陈述，未独立确认。" }, { "id": "O8BB-LB25", "title": "Lawrence Burns: Manager Review — May 2025", "date": "2025-05", "url": "https://www.bailliegifford.com/en/belgium/professional-investor/insights/ic-article/sm-articles-2025-bg-lawrence-burns-manager-review-10055358/", "authors": "Lawrence Burns", "summary": "讨论异常优秀企业的多次业务扩张、长期投入与适应能力，并强调新增业务价值的不确定性。", "access": "2026-10-08：官方HTML正文读取；Outliers Address Large Opportunities及Outliers are Adaptable；未重算收益图表。" }, { "id": "O8BB-DB24", "title": "Discovery Q2 investment update — July 2024", "date": "2024-07", "url": "https://media.bailliegifford.com/mws/5hwbby2a/20240712084931_baillie-gifford-film-discovery-q2-update-transcript-july-2024.pdf", "authors": "Douglas Brodie; interviewer Bill Chater", "summary": "区分经营进展、执行波折与增长空间收缩，讨论Raspberry Pi工业应用和边缘AI机会。", "access": "2026-10-08：官方PDF实质访谈第1–4页读取；第1页答话标签误写Dougald，封面与主持人明确Douglas；不把主持人的总结归于Brodie。" }, { "id": "O8BB-DB25", "title": "Discovery Q2 investment update — July 2025", "date": "2025-07", "url": "https://media.bailliegifford.com/mws/4umpoof3/20250709114359_discovery-q2-update-transcript-jul-2025.pdf", "authors": "Douglas Brodie", "summary": "区分技术驱动和政策驱动的增长，承认组合回撤，并跟踪Sweetgreen等公司的需求和竞争风险。", "access": "2026-10-08：官方PDF第1–2页投资评论完整读取；后续业绩与法规页未作收益审计。" }, { "id": "O8BB-BW22", "title": "ARK Venture Commentary, 12/9/2022", "date": "2022-12-15", "url": "https://www.ark-funds.com/articles/venture-fund/ark-venture-commentary-12-9-2022", "authors": "Brett Winton / ARK", "summary": "区分生成式AI流量热潮、持久优势与收费模式；推理成本影响广告模式可行性。", "access": "2026-10-08：官方HTML两个AI小节读取；页面发布日期Dec 15，标题12/9为原始期号，不当作发布日期；第三方App数据未重算。" }, { "id": "O8BB-BW23", "title": "ARK Venture Commentary, 7/11/2023", "date": "2023-07-11", "url": "https://www.ark-funds.com/articles/venture-fund/ark-venture-commentary-7-11-2023", "authors": "Brett Winton / ARK", "summary": "讨论数据平台与模型开发协同，同时指出供应商投资回流收入及对竞争的潜在约束。", "access": "2026-10-08：官方HTML三个研究小节完整读取；日期与署名核验；Crunchbase底层数据和具体投资条款未独立审计。" }, { "id": "O8BB-TK23", "title": "ARK’s Expected Value For Tesla In 2027: $2,000 Per Share", "date": "2023-04-21", "url": "https://www.ark-invest.com/articles/valuation-models/arks-tesla-price-target-2027", "authors": "Tasha Keeney, Sam Korus and Daniel Maguire", "summary": "公开模型把投产、资本效率与robotaxi推出时间连接至估值分布，并披露样本与模型局限。", "access": "2026-10-08：官方网页正文重点读取Model Inputs、Share Price Forecast、Risks and Limits；核验联合署名和日期；未运行GitHub模拟或重算图表，非2027结果审计。" }, { "id": "O8BB-TK24", "title": "Countdown To Cybercab, Tesla’s Multi-Trillion Dollar Robotaxi Opportunity", "date": "2024-10-09", "url": "https://www.ark-invest.com/articles/analyst-research/countdown-to-cybercab", "authors": "Tasha Keeney", "summary": "将兼容车辆、车队扩张、服务利用率与平台抽成连接，并区分地址市场和预测收入。", "access": "2026-10-08：官方网页实质正文读取，包括数据口径注记；未核验供应商安全性原始数据或图表；这是2024预测样本，不代表当前技术/监管状态。" }, { "id": "O8BCS01", "title": "邱国鹭：回归投资的本质，做时间的朋友", "date": "2017-03-10", "url": "https://www.gyasset.com/p/site/investTeamNewsArticle?id=136&type=5", "authors": "邱国鹭；高毅官网演讲实录", "summary": "从行业、公司和管理层分析可持续回报。", "access": "官方正文直接读取：三个维度、管理KPI与回款段；未审计案例收益。", "checked_at": "2026-10-08", "locator": "管理层维度：中层KPI与回款" }, { "id": "O8BCS02", "title": "邱国鹭：价值投资应该如何数月亮", "date": "2018-06-12", "url": "https://www.gyasset.com/p/site/investTeamNewsArticle?id=172&type=5", "authors": "邱国鹭；高毅官网转载高礼价值投资研究院", "summary": "集中度必须转化为议价权，且应先界定竞争边界。", "access": "官方短文正文与作者日期已读；非投资执行记录。", "checked_at": "2026-10-08", "locator": "行业格局及“定义天空”段" }, { "id": "O8BCS03", "title": "高毅资产邓晓峰：以光伏行业为例，分析产业演进和迭代", "date": "2022-08-02", "url": "https://www.gyasset.com/p/site/investTeamNewsArticle?id=289&type=6", "authors": "邓晓峰演讲；券商中国整理，高毅官网刊载", "summary": "技术进步可扩大产业，也会淘汰存量企业和资产。", "access": "官方刊载演讲整理；读取行业演进、瓶颈和成熟期段，原文明确部分内容删减。", "checked_at": "2026-10-08", "locator": "行业演进与迭代；成熟行业回报" }, { "id": "O8BCS04", "title": "高毅资产·毅路同行｜对话邓晓峰：复苏共识下的思考和分析", "date": "2023-02-25", "url": "https://www.gyasset.com/p/site/investTeamNewsArticle?id=308&type=6", "authors": "邓晓峰交流回答；高毅资产整理", "summary": "分别判断供需周期、竞争盈利与投资兑现时间。", "access": "官方交流整理：工业金属、电动车、逆向投资与时间风险问答已读；日期为网页刊载日。", "checked_at": "2026-10-08", "locator": "回顾投资策略：长期与时间风险" }, { "id": "O8BCS05", "title": "高毅资产冯柳再谈投资方法：寻找共识 依靠常识 接受自己的不完美", "date": "2019-06-17", "url": "https://www.gyasset.com/p/site/investTeamNewsArticle?id=220&type=9", "authors": "冯柳回答；中国基金报记者采访，高毅官网刊载", "summary": "以自知和风险定价设边界；高仓位组合调仓可能为新增机会筹资，稳定客户资金支持较长持有。", "access": "官方转载2019采访：选股流程、风险定价、高仓位调仓及长期客户资金段已读。", "checked_at": "2026-10-08", "locator": "2019中国基金报记者问答：持仓与高仓位资金调动；直销客户与长期锁定资金" }, { "id": "O8BCS06", "title": "最新专访：如果你读懂了冯柳，就知道“抄作业”没有用", "date": "2020-05-26", "url": "https://www.gyasset.com/p/site/investTeamNewsArticle?id=251&type=9", "authors": "冯柳回答；上海证券报记者赵明超，高毅官网刊载", "summary": "区分估值保护与估值回归；公开持仓受组合和流动性影响。", "access": "官方转载采访：成长/价值、加减仓机制、公开持仓限制问答已读；不据采访认证实际风控效果。", "checked_at": "2026-10-08", "locator": "公开持仓不完整、资金调动与抄作业问答；成长/价值股与估值保护问答" }, { "id": "O8BCS07", "title": "景顺长城新兴成长混合2025年第2季度报告", "date": "2025-07-21", "url": "https://www.igwfmc.com/upload/dqgg/20250721/2025-07-21%E6%99%AF%E9%A1%BA%E9%95%BF%E5%9F%8E%E6%96%B0%E5%85%B4%E6%88%90%E9%95%BF%E6%B7%B7%E5%90%88%E5%9E%8B%E8%AF%81%E5%88%B8%E6%8A%95%E8%B5%84%E5%9F%BA%E9%87%912025%E5%B9%B4%E7%AC%AC2%E5%AD%A3%E5%BA%A6%E6%8A%A5%E5%91%8A.pdf", "authors": "景顺长城基金管理人报告；本基金经理刘彦春，非个人独著认证", "summary": "讨论量价背离和内需恢复迟于预期，并关注企业质量。", "access": "官方12页PDF；印刷pp5–7的经理表与§4.4已读，未审计全报告。", "checked_at": "2026-10-08", "locator": "2025年第2季度报告印刷p5经理表、pp6–7 §4.4" }, { "id": "O8BCS08", "title": "景顺长城新兴成长混合2025年年度报告", "date": "2026-03-27", "url": "https://www.igwfmc.com/upload/dqgg/20260327/2026-03-27%E6%99%AF%E9%A1%BA%E9%95%BF%E5%9F%8E%E6%96%B0%E5%85%B4%E6%88%90%E9%95%BF%E6%B7%B7%E5%90%88%E5%9E%8B%E8%AF%81%E5%88%B8%E6%8A%95%E8%B5%84%E5%9F%BA%E9%87%912025%E5%B9%B4%E5%B9%B4%E5%BA%A6%E6%8A%A5%E5%91%8A.pdf", "authors": "景顺长城基金管理人报告；本基金经理刘彦春，非个人独著认证", "summary": "通过投资、库存与需求推导盈利恢复，仍需验证。", "access": "官方68页PDF；印刷p10经理表及pp13–14 §4.4–4.5已读；2026展望不作已兑现事实。", "checked_at": "2026-10-08", "locator": "2025年年度报告印刷p10经理表、pp13–14 §4.4–4.5" }, { "id": "O8BCS09", "title": "睿远成长价值混合2025年中期报告", "date": "2025-08-27", "url": "https://www.foresightfund.com/upload/pdf/1756255838958-ruiyuanchengzhangjiazhihunhexingzhengquantouzijijin2025nianzhongqibaogao.pdf", "authors": "睿远基金管理人报告；傅鹏博、朱璘共同管理，非傅鹏博独著", "summary": "把PCB和光模块增长联系到量产、产能、良率；关注价格上涨的需求条件。", "access": "官方70页PDF；印刷pp12–14 §4.4–4.5已读，封面送出日期已核；不是完整归因审计。", "checked_at": "2026-10-08", "locator": "中期报告§4.4.1：PCB与光模块" }, { "id": "O8BCS10", "title": "睿远成长价值混合2025年年度报告", "date": "2026-03-27", "url": "https://www.foresightfund.com/upload/pdf/1774572963264-ruiyuanchengzhangjiazhihunhexingzhengquantouzijijin2025nianniandubaogao.pdf", "authors": "睿远基金管理人报告；傅鹏博、朱璘共同管理，非傅鹏博独著", "summary": "扩大AI相关配置，同时明确资本开支回报尚无定论。", "access": "官方82页PDF；印刷pp11–12经理表和pp13–14 §4.4.1已读；集中度变化不自动等同主动买入。", "checked_at": "2026-10-08", "locator": "年度报告§4.4.1：集中度与资本回报疑问" }, { "id": "O8SAD1", "title": "AMD Q1 2026 Earnings Call", "date": "2026-05-05", "url": "https://d1io3yog0oux5.cloudfront.net/_e6fb18640ea7bfff80a48dc92268a25f/amd/db/841/9232/webcast_transcript/AMD_1Q_2026_Earnings.pdf", "access": "AMD官方Financial Results页链接的PDF；2026-10-08直接下载并读具名问答正文，未听音频；网页搜索解析不可用但原PDF可读取。", "authors": "AMD management and attributed analyst participants", "summary": "选读具名问题及相邻管理层回答；研究方法为编辑归纳。" }, { "id": "O8SAD2", "title": "AMD Q2 2026 Earnings Call", "date": "2026-08-04", "url": "https://d1io3yog0oux5.cloudfront.net/_e6fb18640ea7bfff80a48dc92268a25f/amd/db/841/9237/webcast_transcript/AMD2Q26EarningsCallTranscript.pdf", "access": "AMD官方Financial Results页链接的PDF；2026-10-08直接下载并读具名问答正文，未听音频；网页搜索解析不可用但原PDF可读取。", "authors": "AMD management and attributed analyst participants", "summary": "选读具名问题及相邻管理层回答；研究方法为编辑归纳。" }, { "id": "O8SAD3", "title": "AMD Q2 2025 Earnings Call", "date": "2025-08-05", "url": "https://d1io3yog0oux5.cloudfront.net/_e6fb18640ea7bfff80a48dc92268a25f/amd/db/841/9167/webcast_transcript/amd2q2025.pdf", "access": "AMD官方Financial Results页链接的PDF；2026-10-08直接下载并读具名问答正文，未听音频；网页搜索解析不可用但原PDF可读取。", "authors": "AMD management and attributed analyst participants", "summary": "选读具名问题及相邻管理层回答；研究方法为编辑归纳。" }, { "id": "O8SAD4", "title": "AMD Q2 2024 Earnings Call", "date": "2024-07-30", "url": "https://d1io3yog0oux5.cloudfront.net/_e6fb18640ea7bfff80a48dc92268a25f/amd/db/841/8899/webcast_transcript/AMD+2Q24call.pdf", "access": "AMD官方页链接PDF直接下载并读取Harsh Kumar p12；未听音频。答复正文标签误作Mitch Haws，后续称Lisa；只归为管理层，不强认具体答者。", "authors": "Issuer management and named analyst interlocutor", "summary": "仅选读具名公开问答，问题与管理层答复分开。", "additional_readings": [ { "date": "2026-10-08", "analyst": "Toshiya Hari", "locator": "pp.10–11", "summary": "检验MI300指引是需求还是供给限制，并问盈利变化。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://d1io3yog0oux5.cloudfront.net/_e6fb18640ea7bfff80a48dc92268a25f/amd/db/841/8899/webcast_transcript/AMD+2Q24call.pdf" } ] }, { "id": "O8SAD5", "title": "AMD Q3 2024 Earnings Call", "date": "2024-10-29", "url": "https://d1io3yog0oux5.cloudfront.net/_e6fb18640ea7bfff80a48dc92268a25f/amd/db/841/8900/webcast_transcript/AMD+3Q24call.pdf", "access": "AMD官方页链接PDF直接下载并读取Harsh Kumar pp15–16；未听音频。", "authors": "Issuer management and named analyst interlocutor", "summary": "仅选读具名公开问答，问题与管理层答复分开。", "additional_readings": [ { "date": "2026-10-08", "analyst": "Toshiya Hari", "locator": "pp.7–8", "summary": "追问GPU指引上调、客户扩展与毛利组合。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://d1io3yog0oux5.cloudfront.net/_e6fb18640ea7bfff80a48dc92268a25f/amd/db/841/8900/webcast_transcript/AMD+3Q24call.pdf" } ] }, { "id": "O8SAD6", "title": "Cirrus Logic Oppenheimer Technology, Internet & Communications Conference", "date": "2024-08-14", "url": "https://s21.q4cdn.com/307563939/files/doc_events/2024/Aug/14/corrected-transcript_-cirrus-logic-inc-crus-us-oppenheimer-technology-internet-communications-conference-14-august-2024-11_35-am-et.pdf", "access": "发行人CDN FactSet更正稿；2026-10-08读取Rick Schafer pp7–9,12–13，未听音频。", "authors": "Issuer management and named analyst interlocutor", "summary": "仅选读具名公开问答，问题与管理层答复分开。" }, { "id": "O8SAD7", "title": "MaxLinear Oppenheimer 25th Annual Technology Conference — issuer-filed transcript", "date": "2022-08-09", "url": "https://www.sec.gov/Archives/edgar/data/1288469/000110465922088223/tm2223040d1_425.htm", "access": "SEC Rule 425发行人备案正文；读取开头供应、交期和backlog问答，会议日期为2022-08-09。历史材料，不当作2026当前观点。", "authors": "Issuer management and named analyst interlocutor", "summary": "仅选读具名公开问答，问题与管理层答复分开。" }, { "id": "O8SB-MS3", "title": "Microsoft FY2025 Q3 earnings call — selected Q&A", "date": "2025-04-30", "url": "https://www.microsoft.com/en-us/investor/events/fy-2025/earnings-fy-2025-q3", "authors": "具名分析师提问；公司管理层回答，见各人物定位", "summary": "Keirstead问增长来源，Thill问迁移，Rangan问资本效率；管理层回应须独立检验。", "access": "2026-10-08取得Microsoft官方HTML全文并阅读具名问答。", "retrieval_level": "primary-full-text; selected Q&A read", "source_type": "official earnings transcript", "additional_readings": [ { "date": "2026-10-08", "analyst": "Kirk Materne", "locator": "Kirk Materne具名问答", "summary": "区分Azure非AI改善与此前优化周期。", "access": "实际读取发行人正文具名提问，未听录音", "url": "https://www.microsoft.com/en-us/investor/events/fy-2025/earnings-fy-2025-q3" } ] }, { "id": "O8SB-META1", "title": "Meta Q1 2025 earnings call — selected Q&A", "date": "2025-04-30", "url": "https://s21.q4cdn.com/399680738/files/doc_financials/2025/q1/Transcripts/META-Q1-2025-Earnings-Call-Transcript-1.pdf", "authors": "具名分析师提问；公司管理层回答，见各人物定位", "summary": "具名问答涉及用户行为、产品形态、基础设施成本及回报。", "access": "2026-10-08通过网页检索器读取官方PDF正文，印刷页9–13；本地直接下载超时，不声称本地副本校验。", "retrieval_level": "primary-full-text; selected Q&A read", "source_type": "official earnings transcript" }, { "id": "O8SB-META2", "title": "Meta Q2 2025 earnings call — selected Q&A", "date": "2025-07-30", "url": "https://s21.q4cdn.com/399680738/files/doc_financials/2025/q2/META-Q2-2025-Earnings-Call-Transcript.pdf", "authors": "具名分析师提问；公司管理层回答，见各人物定位", "summary": "具名问答涉及技术约束、资本投入、融资及内部使用回报。", "access": "2026-10-08通过网页检索器读取官方PDF正文，印刷页9–16；本地直接下载超时，不声称本地副本校验。", "retrieval_level": "primary-full-text; selected Q&A read", "source_type": "official earnings transcript" }, { "id": "O8SC-TTQ325", "title": "Trane Q3 2025 Earnings", "date": "2025-10-30", "url": "https://s2.q4cdn.com/950394465/files/doc_events/2025/Oct/30/Trane-Technologies-Q3-2025-Earnings-Transcript.pdf", "authors": "Issuer management and named sell-side questioners; transcript publisher retains authorship", "access": "2026-10-08读取发行人托管PDF的具名问答；逐人页码见档案/卡片。已下载PDF、提取正文核对；未听录音。", "retrieval_level": "issuer-hosted transcript; selected attributable exchanges read", "summary": "具名问题与管理层答复分开；只用于有定位的研究方法样本。" }, { "id": "O8SC-NVTQ225", "title": "nVent Q2 2025 Earnings Call", "date": "2025-08-01", "url": "https://s22.q4cdn.com/268397047/files/doc_financials/2025/q2/Q2-Corrected-Transcript.pdf", "authors": "Issuer management and named sell-side questioners; transcript publisher retains authorship", "access": "2026-10-08读取发行人托管PDF的具名问答；逐人页码见档案/卡片。通过web工具读取PDF正文，本地下载451；未听录音。", "retrieval_level": "issuer-hosted transcript; selected attributable exchanges read", "summary": "具名问题与管理层答复分开；只用于有定位的研究方法样本。" }, { "id": "O8SC-CIENQ125", "title": "Ciena Q1 2025 Financial Results", "date": "2025-03-11", "url": "https://s25.q4cdn.com/550667411/files/doc_financials/2025/q1/CIENA-Q1-2025-Financial-Results-Transcript.pdf", "authors": "Issuer management and named sell-side questioners; transcript publisher retains authorship", "access": "2026-10-08读取发行人托管PDF的具名问答；逐人页码见档案/卡片。已下载PDF、提取正文核对；未听录音。", "retrieval_level": "issuer-hosted transcript; selected attributable exchanges read", "summary": "具名问题与管理层答复分开；只用于有定位的研究方法样本。", "additional_readings": [ { "date": "2026-10-08", "summary": "新增读取Ruben Roy具名问答：Q1供应商替换、CPO机会/威胁；Q3 scale-across网络产品范围与宽带业务资源配置。", "access": "重新读取既有发行人PDF提取正文；具名问答单独归属，未听录音。" } ] }, { "id": "O8SC-CIENQ325", "title": "Ciena Q3 2025 Financial Results", "date": "2025-09-04", "url": "https://s25.q4cdn.com/550667411/files/doc_financials/2025/q3/Transcript-Cienas-Fiscal-Third-Quarter-2025-Financial-Results-Conference-Call.pdf", "authors": "Issuer management and named sell-side questioners; transcript publisher retains authorship", "access": "2026-10-08读取发行人托管PDF的具名问答；逐人页码见档案/卡片。已下载PDF、提取正文核对；未听录音。", "retrieval_level": "issuer-hosted transcript; selected attributable exchanges read", "summary": "具名问题与管理层答复分开；只用于有定位的研究方法样本。", "additional_readings": [ { "date": "2026-10-08", "summary": "新增读取Ruben Roy具名问答：Q1供应商替换、CPO机会/威胁；Q3 scale-across网络产品范围与宽带业务资源配置。", "access": "重新读取既有发行人PDF提取正文；具名问答单独归属，未听录音。" } ] }, { "id": "R2B1-EV24", "title": "The Magnificent 7 and positioning for a market broadening", "date": "2024-05", "authors": "Nick Evans", "url": "https://www.polarcapital.co.uk/static/documents/Global_Technology_Video_-_The_Magnificent_7_and_positioning_for_a_market_broadening_-_May_2024.pdf", "access": "2026-10-08 读取访谈PDF第1页实质正文。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "区分AI基础设施、受益者与采用者；看企业之间增长差异而非统一大盘标签。" }, { "id": "R2B1-EV26", "title": "Global Technology video transcript", "date": "2026-01", "authors": "Nick Evans", "url": "https://www.polarcapital.co.uk/static/documents/Global_Tech_Video_Transcript_-_January_2026.pdf", "access": "2026-10-08 读取PDF第1–2页访谈正文。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "以AI重新划分利润池，关注软件替代风险、基础设施需求与资本开支对自由现金流的影响。" }, { "id": "R2B1-ME26Q2", "title": "BlackRock Technology Opportunities Fund commentary", "date": "2026-06-30", "authors": "Tony Kim and Reid Menge；组合管理团队共同材料", "url": "https://www.blackrock.com/us/individual/literature/fund-commentary/oef-technology-opportunities-fund-qtd-commentary.pdf", "access": "2026-10-08 读取PDF第1–2页；采用文件内Q2 2026日期而非搜索索引日期。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "团队关注推理效率、代理工作负载和可度量资本回报，偏好已显现商业化及持久优势。", "date_basis": "报告期末；精确发布日期未核" }, { "id": "R2B1-ME26SP", "title": "The Bid: Space — The Final Investing Frontier", "date": "2026-09-25", "authors": "Reid Menge受访；BlackRock The Bid", "url": "https://www.blackrock.com/uk/solutions/podcasts/the-bid", "access": "2026-10-08 读取第272集官方完整文字稿中Menge发言；栏目滚动页面，标题用于定位。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "把成本下降、能力提升和需求结合，同时强调发射执行、太空环境和地面通电时间；太空算力判断属预测。" }, { "id": "R2B1-SO26", "title": "Fidelity International Outlook 2026", "date": "2025", "authors": "Hyun Ho Sohn；展望合集中的具名个人段落", "url": "https://www.fidelityinternational.com/static/master/media/pdf/outlook/Fidelity-International-Outlook-2026.pdf", "access": "2026-10-08 读取印刷第13页/PDF索引12；准备年份2025，确切发布日期未核。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "担心资本密集度、外部融资与半导体估值透支周期，要求区分基础设施增长和投资回报。", "date_basis": "展望准备年份；精确发布日期未核" }, { "id": "R2B1-SO26Q", "title": "Global technology: Overdone risks and diverse opportunities", "date": "2026-04-05", "authors": "Hyun Ho Sohn问答；Fidelity供稿，RSMR合作平台转载", "url": "https://www.rsmr.co.uk/rsmr-connected/industry-insights/fidelity/global-technology-overdone-risks-and-diverse-opportunities/", "access": "2026-10-08 读取全文具名问答；非RSMR独立研究，未比对原始录音。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "指出企业软件工作流、数据、安全与集成形成阻力；用管理层、专家和内部IT调研检验AI替代速度。" }, { "id": "R2B1-KE18", "title": "Technology’s Golden Era", "date": "2018-07", "authors": "Ken Allen问答；T. Rowe Price", "url": "https://www.troweprice.com/content/dam/iinvestor/images/PricePt_dwnld_TechGoldenEra07302018.pdf", "access": "2026-10-08 读取PDF第1–3页Ken Allen问答。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "重视持久业务、合理价格和中长期现金；看云规模优势同时保留本地部署需求。" }, { "id": "R2B1-KE17", "title": "Artificial Intelligence: A New Era of Opportunity", "date": "2017-12", "authors": "T. Rowe Price机构文章；分别引述Ken Allen、Josh Spencer、Paul Greene", "url": "https://www.troweprice.com/content/dam/tpd/Articles/PDFs/AI_Price_Perspective_NOV-17_GL_P4.pdf", "access": "2026-10-08 读取第2–3页Allen具名段落；文件名NOV但采用正文December 2017。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "Allen强调云分发以及数据、算力和人才组合；其他经理的判断不归给Allen。" }, { "id": "R2B1-TU22", "title": "Investing in Companies at the Start of the Runway", "date": "2022-06", "authors": "Alan Tu", "url": "https://www.troweprice.com/content/dam/trp-ecl/global/en/ipc/assets/us-retail-intermediary/2022/june/investing-in-companies-at-the-start-of-the-runway-id0004907/Investing-in-Companies-at-the-Start-of-the-Runway.pdf", "access": "2026-10-08 读取PDF第1–4页。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "从产品进入市场方式、客户扩张和竞争优势寻找长增长跑道，同时要求估值纪律。" }, { "id": "R2B1-TU21", "title": "Investing in the Infrastructure Behind the Digital Economy（日文官方译本）", "date": "2021-03", "authors": "Alan Tu, James Stillwagon, Tony Wang共同署名", "url": "https://www.troweprice.com/content/dam/tpd/global/ja/pdfs/%28J%29Investing%20in%20the%20Infrastructure%20Behind%20the%20Digital%20Economy_GL%20Std.pdf", "access": "2026-10-08 读取日文官方PDF第1–4页，保留共同署名；不把团队观点拆为Tu独著。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "团队从软件、云、半导体设备与安全切入数字化基础设施，讨论竞争壁垒、增长基数及估值。" }, { "id": "R2B1-WE26", "title": "Investing in companies harnessing AI to transform their business", "date": "2026-05-04", "authors": "Kyle Weaver", "url": "https://clearingcustody.fidelity.com/insights/topics/portfolio-manager-insights/investing-in-companies-harnessing-ai-to-transform-their-business", "access": "2026-10-08 读取署名文章正文。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "把AI机会扩展到采用者，用客服和数字开户案例连接运营变化；案例陈述不等于已证明因果收益。" }, { "id": "R2B1-WE26Q2", "title": "Fidelity Advisor Growth Opportunities Fund quarterly commentary", "date": "2026-06-30", "authors": "Kyle Weaver and Becky Baker；团队共同材料", "url": "https://institutional.fidelity.com/app/proxy/content?literatureURL=%2F9910025.PDF", "access": "2026-10-08 读取PDF第2页投资理念、第5页展望；确认Q2 2026。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "团队采用3–7年视角，结合现金增长、资本配置及不同增长类型，追踪真实AI使用与供给约束。", "date_basis": "报告期末；精确发布日期未核" }, { "id": "R2B1-KA26", "title": "Fidelity Blue Chip Growth Fund: Portfolio Manager Q&A", "date": "2026-07-31", "authors": "Sonu Kalra", "url": "https://institutional.fidelity.com/app/proxy/content?literatureURL=%2FRD_QAA_7822.PDF", "access": "2026-10-08 读取PDF第1–4页；滚动PDF实际日期July 31, 2026。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "以竞争优势、定价能力和管理层判断增长持久性，关注AI半导体机会并保留宏观及估值风险。" }, { "id": "R2B1-KA24", "title": "Exploring the AI investment opportunity across public and private markets", "date": "2024-05", "authors": "Sonu Kalra, Asher Anolic, Karin Fronczke, Annie Wang共同署名", "url": "https://clearingcustody.fidelity.com/app/proxy/content?literatureURL=%2F9916259.PDF", "access": "2026-10-08 读取公开PDF实质正文；不采用图表中无效4/31/2024脚注作为日期。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "团队跨公私市场观察AI价值链，区分流动性、采用速度及长期生产率兑现的不确定性。" }, { "id": "R2B1-LA26D", "title": "Our Partnership with Databricks", "date": "2026-07-20", "authors": "Thomas Laffont, Lucas Swisher, Jade Lai, Abhi Srinivas, Philena Sun", "url": "https://www.coatue.com/blog/perspective/our-partnership-with-databricks", "access": "2026-10-08 读取官网完整实质正文及利益披露。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "团队以分析、交易和代理应用扩张解释增长，强调数据治理与研发周期；Coatue持有经济利益。" }, { "id": "R2B1-LA26A", "title": "Our Partnership with Anthropic: Enterprise Intelligence has Liftoff", "date": "2026-02-12", "authors": "Thomas Laffont, Lucas Swisher, Jade Lai, Abhi Srinivas", "url": "https://www.coatue.com/blog/press/our-partnership-with-anthropic-enterprise-intelligence-has-liftoff", "access": "2026-10-08 读取官网完整实质正文及利益披露；未独立验证营收与工程效率宣传数字。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "团队用代码活动和工作流扩展支持企业采用论点；从开发能力延伸到平台价值，存在被投企业利益冲突。" }, { "id": "R2B1-AI06", "title": "Inside a hedge fund: An interview with the managing partner of Maverick Capital", "date": "2006-04-01", "authors": "Lee Ainslie受访；Richard Dobbs and Tim Koller采访", "url": "https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/inside-a-hedge-fund-an-interview-with-the-managing-partner-of-maverick-capital", "access": "2026-10-08 读取原发布者完整访谈正文。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "以管理层、持续增长和资本回报筛选，以可持续自由现金流/企业价值衡量价格，持仓按未来回报重审。" }, { "id": "R2B1-AI21", "title": "A Conversation with Lee Ainslie, Founder and Managing Partner of Maverick Capital", "date": "2021-05-16", "authors": "Lee Ainslie受访；Business Today原采访", "url": "https://journal.businesstoday.org/bt-online/2021/a-conversation-with-lee-ainslie-founder-and-managing-partner-of-maverick-capital", "access": "2026-10-08 读取原采访全文。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "说明基本面与因素风险互补；用统计化替代数据替代过度加权的零散观察，并保留资本保护目标。" }, { "id": "R2B1-DA20", "title": "Will Danoff: Succeeding at Scale — The Long View", "date": "2020-01-29", "authors": "Will Danoff受访；Morningstar Jeff Ptak and Christine Benz", "url": "https://the-long-view.simplecast.com/episodes/will-danoff/transcript", "access": "2026-10-08 读取Morningstar官方Simplecast托管逐字稿中管理层复盘、竞争优势、每股现金增长及持仓重估段落。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "强调隔半年至一年追踪管理层承诺；偏好高资本回报、现金增长和客户价值，并用新证据逐步建立信心。" }, { "id": "R2B1-DA26I", "title": "Will Danoff: Be Very Careful of Unprofitable Companies — The Long View", "date": "2026-06-23", "authors": "Will Danoff受访；Morningstar Amy Arnott and Christine Benz", "url": "https://the-long-view.simplecast.com/episodes/will-danoff-be-very-careful-of-unprofitable-companies-1CsuaI3d", "access": "2026-10-08 读取Morningstar节目官方Simplecast托管完整逐字稿，尤其unit economics、管理层核验、竞争优势及亏损公司段落；Morningstar主域对应页403，合法公开托管可读。", "retrieval_level": "原发布者公开正文；合著/转载限制见作者与读取说明", "summary": "从门店单位经济性检验扩张；将管理层许诺与后续执行对照，并要求竞争压力下的盈利记录而非短期收入火箭。" }, { "id": "R2B2-J1", "title": "Management Commentary: October 2025", "url": "https://www.jacobmutualfunds.com/blog/managementcommentary-october-2025", "date": "2025-10-15", "authors": "Ryan Jacob and Darren Chervitz；网站编辑Hayden Cook", "summary": "Internet Fund部分讨论Circle支付基础设施与Bed Bath & Beyond的tZERO资产及重组价值，不能把整个合署信归于Ryan一人。", "access": "2026-10-08已读官方正文相关段落；未审计底层数据。", "date_basis": "原文显示日期" }, { "id": "R2B2-J2", "title": "Management Commentary: January 2026", "url": "https://www.jacobmutualfunds.com/blog/managementcommentary-january-2026", "date": "2026-01-13", "authors": "Ryan Jacob and Darren Chervitz；网站编辑Hayden Cook", "summary": "Internet Fund新增Snowflake的逻辑是跨云数据层、客户黏性与AI能力；竞争与估值仍需检验。", "access": "2026-10-08已读官方正文相关段落；未审计底层数据。", "date_basis": "原文显示日期" }, { "id": "R2B2-L1", "title": "Firsthand Funds Semi-Annual Report June 30 2023", "url": "https://firsthandfunds.com/documents/reports/semiannual_report.pdf", "date": "2023-06-30", "authors": "Kevin Landis签署President’s Letter", "summary": "信件把未配置大市值股的相对拖累与Chegg被生成式AI替代的新客风险分开。", "access": "2026-10-08已读PDF第4–6页；日期为报告期末，送出日在正文未核定。", "date_basis": "报告期末（period end）；发表日期未核定（publication unknown）" }, { "id": "R2B2-L2", "title": "2015 Annual Shareholder Letter", "url": "https://firsthandtvf.com/index.php?fuseaction=press.shareholder_letter", "date": null, "authors": "Kevin Landis", "summary": "区分私企经营进度、IPO锁定、退出现金与基金NAV折价；成熟需要多年，上市不等于可兑现。", "access": "2026-10-08全文已读；2015年度信，正文涉及2016年事项，精确发表日未知。", "date_basis": "发表日期未知（publication unknown）" }, { "id": "R2B2-S1", "title": "Letter to Investors", "url": "https://www.seilernfunds.com/blog/letter-from-peter-seilern/", "date": "2024-09-26", "authors": "Peter Seilern", "summary": "退休交接信强调清晰规则、可复制流程、团队独立性与长期激励；不承诺每个周期跑赢。", "access": "2026-10-08已读官方正文相关段落；未审计底层数据。", "date_basis": "原文显示日期" }, { "id": "R2B2-S2", "title": "Investment Philosophy", "url": "https://www.seilernfunds.com/seilern-philosophy/", "date": null, "authors": "Seilern机构", "summary": "十项质量标准包含低资本强度、高投入资本回报和低净负债；机构以研究与集中组合实施。", "access": "2026-10-08已读官方正文相关段落；未审计底层数据。", "date_basis": "发表日期未知（publication unknown）" }, { "id": "R2B2-A16", "title": "Allianz Technology Trust Annual Financial Report 2016", "url": "https://www.allianztechnologytrust.com/-/media/allianzgi/eu/allianz-technology-trust/docs/annual-financial-report/annual-financial-report-allianz-technology-trust-30-11-2016.pdf", "date": "2016-11-30", "authors": "AllianzGI Technology team；Walter Price与Huachen Chen为共同负责人", "summary": "管理人报告将新技术高增长与成熟业务整合、回购分开，并以产品周期和子行业分散风险。", "access": "2026-10-08下载官方PDF并提取正文；PDF pp.11、14–16；日期为财年末，非出版日。", "date_basis": "报告期末（period end）；发表日期未核定（publication unknown）" }, { "id": "R2B2-A20", "title": "Allianz Technology Trust Annual Financial Report 2020", "url": "https://www.allianztechnologytrust.com/-/media/allianzgi/eu/allianz-technology-trust/docs/annual-financial-report/annual-financial-report-allianz-technology-trust-31-12-2020.pdf", "date": "2020-12-31", "authors": "AllianzGI Technology team；Walter Price与Huachen Chen为共同负责人", "summary": "管理人报告分拆疫情下远程采用、付费转换与低利率带来的估值扩张；绝对上涨与相对收益不同。", "access": "2026-10-08下载官方PDF；PDF pp.20、23、40–46相关章节；日期为财年末。", "date_basis": "报告期末（period end）；发表日期未核定（publication unknown）" }, { "id": "R2B2-G1", "title": "Equal weighted tech – an interview with Anthony Ginsberg", "url": "https://hanetf.com/news-insights/equal-weighted-tech-an-interview-with-anthony-ginsberg/", "date": "2025-07", "authors": "Anthony Ginsberg / HANetf访谈", "summary": "访谈以八主题各15股等权回应巨头集中；同时承认技术主题交叉与轮动难预测。", "access": "2026-10-08全文读取；月份由官方相关文章目录标注；未独立审计收益表。", "date_basis": "只核定月份；精确发表日未知，依据见access" }, { "id": "R2B2-G2", "title": "The New Age of Tech | Tech Megatrend Outlook 2026", "url": "https://hanetf.com/news-insights/the-new-age-of-tech-tech-megatrend-outlook-2026/", "date": "2026-01", "authors": "HANetf / HAN-GINS机构展望，正文无个人署名", "summary": "机构展望以现金融资和盈利支持AI投资，并称十主题等权；与2025八主题不同，定义变更需核对。", "access": "2026-10-08正文读取；月份见官方相关文章目录；不把预租率当实际利用率。", "date_basis": "只核定月份；精确发表日未知，依据见access" }, { "id": "R2B2-M1", "title": "Carmignac’s Mark Denham shares his investment recipe", "url": "https://carmignaceuropeanequities.citywirecontent.com/2020/05/carmignacs-aaa-rated-mark-denham-shares-his-investment-recipe/", "date": "2020-05", "authors": "Mark Denham；Carmignac制作的Citywire赞助访谈", "summary": "以资本回报、再投资和经营利润率稳定性作筛选，再做基本面与ESG研究，持有期3–5年。", "access": "2026-10-08访谈全文读取；精确日未知，月份来自发布路径；赞助材料，不采用评级为能力证据。", "date_basis": "只核定月份；精确发表日未知，依据见access" }, { "id": "R2B2-M2", "title": "Beyond the plate – Why we favour personal care over food companies", "url": "https://www.carmignac.com/en/articles/beyond-the-plate-why-we-favour-personal-care-over-food-companies-3227-11140?_format=pdf", "date": "2025-02-24", "authors": "Mark Denham", "summary": "将日常消费标签拆为食品与个人护理；从需求变化、有机增长与经营杠杆推导不同盈利及估值。", "access": "2026-10-08官方PDF pp.1–2读取；其中共识为原文当时引用，非本轮独立核验当前共识。", "date_basis": "原文显示日期" }, { "id": "R2B2-H1", "title": "The outlook for 2025 – Mark Hawtin section", "url": "https://www.liontrust.com/insights/blogs/2024/12/the-outlook-for-2025---multi-asset", "date": "2024-12", "authors": "Liontrust；选读Mark Hawtin具名段", "summary": "Hawtin关注指数集中和AI受益向使用者扩散；其他团队发言没有归给他。", "access": "2026-10-08官方HTML正文读取；月份来自路径，精确发表日未核定。", "date_basis": "只核定月份；精确发表日未知，依据见access" }, { "id": "R2B2-H2", "title": "DeepSeek – what just happened in AI?", "url": "https://professionalparaplanner.co.uk/deepseek-what-just-happened-in-ai/", "date": "2025-01-29", "authors": "Mark Hawtin；Professional Paraplanner刊载具名观点", "summary": "文章提出AI普及与基础设施股估值可分离，并说明此前削减CapEx相关敞口；成本与技术比较为作者当时主张，未独立核实。", "access": "2026-10-08全文读取；具名作者文章刊于媒体，非获取内部报告；不能将文中训练成本当全成本或把MoE称全新发明。", "date_basis": "原文显示日期" }, { "id": "R2B2-W2", "title": "Wedgewood Partners Second Quarter 2025 Client Letter", "url": "https://wedgewoodpartners.com/wp-content/uploads/2025/07/Wedgewood-Partners-Second-Quarter-2025-Client-Letter-A-Day-in-the-Stock-Market-Life.pdf", "date": "2025-07", "authors": "David A. Rolfe / Michael X. Quigley / Christopher T. Jersan", "summary": "Zoetis案例从宠物医疗渗透、产品延伸研发和资本回报构造投资逻辑，同时区分疫情提升与正常化增长。", "access": "2026-10-08读原始合署信公开PDF镜像pp.5–8及p.16署名；官方URL随后返回502，未核对镜像与官方字节一致性。", "mirror_url": "https://imonkey-files.s3-us-west-1.amazonaws.com/2Q25-Wedgewood-Partners-Client-Letter-A-Day-in-the-Stock-Market-Life.pdf", "date_basis": "信末署名月份；非报告期末；精确发表日未知" }, { "id": "R2B2-W4", "title": "Wedgewood Partners Fourth Quarter 2025 Client Letter", "url": "https://imonkey-files.s3-us-west-1.amazonaws.com/4Q25-Wedgewood%20Partners.pdf", "date": "2026-01", "authors": "David A. Rolfe / Michael X. Quigley / Christopher T. Jersan", "summary": "Amazon新仓论证分拆广告增长与零售生产率对利润贡献，再讨论AWS硬件、资本开支与估值；分部利润不能直接当核心零售效率。", "access": "2026-10-08读原始合署信第三方公开PDF镜像pp.6–8及p.24署名；未与官方字节比对。", "date_basis": "信末署名月份；非报告期末；精确发表日未知" }, { "id": "R2B3-B66C", "title": "万家全球成长一年持有期混合(QDII)2025年第1季度报告", "date": "2025-04-21", "url": "https://www.wjasset.com/upload/pdf/1744971636392-wjqqczyncyqhhxzqtzjj%28QDII%292025nd1jdbg.pdf", "authors": "万家基金管理人；黄兴亮为报告期列示基金经理，非个人独著", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：§4.1经理表、§4.5投资运作分析，印刷页5–8", "retrieval_level": "substantive primary material read", "summary": "管理人解释AI应用和自主科技机会，并按价格透支及经营压力调整持仓。", "locator": "§4.1经理表、§4.5投资运作分析，印刷页5–8" }, { "id": "R2B3-B66D", "title": "万家全球成长一年持有期混合(QDII)2026年第1季度报告", "date": "2026-04-21", "url": "https://www.wjasset.com/upload/pdf/1776681225527-wjqqczyncyqhhxzqtzjj%28QDII%292026nd1jdbg.pdf", "authors": "万家基金管理人；黄兴亮为报告期列示基金经理，非个人独著", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：§4.1经理表、§4.5投资运作分析，印刷页5–8", "retrieval_level": "substantive primary material read", "summary": "管理人讨论风险偏好和重资产交易，调减部分生物科技并增加AI数据中心相关资产。", "locator": "§4.1经理表、§4.5投资运作分析，印刷页5–8" }, { "id": "R2B3-B70A", "title": "李巍：在不确定的世界寻找成长", "date": "2020-07-06", "url": "https://www.gffunds.com.cn/gywm/4/202007/t20200706_358867.shtml", "authors": "李巍受访；朱昂提问；广发官方转载", "access": "2026-10-08读取；官方网页正文已通过网页检索读取（本地HTTP405）。读取现金流、PEG和管理层投资段落。", "retrieval_level": "substantive primary material read", "summary": "问答把经营现金流、资本开支、定价权和管理层资本配置连接起来，反思机械PEG。", "locator": "正文问答：现金流与PEG；广发官网转载点拾投资朱昂访谈" }, { "id": "R2B3-B70B", "title": "李巍：大变局时代，如何安放我们的投资框架", "date": "2022-09-13", "url": "https://edu.gffunds.com.cn/cfxy/dksfk/202209/t20220913_377864.shtml", "authors": "李巍", "access": "2026-10-08读取；官方网页正文已通过网页检索读取（本地HTTP405）。读取2022风险反思与行业分散不足段落。", "retrieval_level": "substantive primary material read", "summary": "第一人称反思风险虽被预见却未充分进入组合；行业分散不等于风格因子分散。", "locator": "正文：风险模型与2022年组合反思" }, { "id": "R2B3-B72A", "title": "唯一10年10倍的公募基金经理，这段话将影响你的下一个10年", "date": "2017-12-14", "url": "https://www.fullgoal.com.cn/contents/2017/12/28-45ba8dd00d1341f99f58d2d6ba243949.html", "authors": "朱少醒受访；富国官网转载", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：问答5、9、16、19、20；正文日期12月14日，URL路径12月28日", "retrieval_level": "substantive primary material read", "summary": "问答讨论自下而上、行业均衡、三年可见度和先研究后交易；宣传业绩未审计。", "locator": "问答5、9、16、19、20；正文日期12月14日，URL路径12月28日" }, { "id": "R2B3-B72B", "title": "朱少醒：淡化择时，精选个股", "date": "2012-07-09", "url": "https://www.fullgoal.com.cn/contents/2012/7/9-c43b07708a2f450c87ae2e840b87abae.html", "authors": "朱少醒受访；富国官网发布", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：正文投资理念及择时段落", "retrieval_level": "substantive primary material read", "summary": "谈价格波动与基本面差异，通过公司质量和成长寻找机会。", "locator": "正文投资理念及择时段落" }, { "id": "R2B3-B75B", "title": "易方达蓝筹精选混合型证券投资基金2023年年度报告", "date": "2024-03-29", "url": "https://cdn.efunds.com.cn/owch/data/bulletin/20240329/%E6%98%93%E6%96%B9%E8%BE%BE%E8%93%9D%E7%AD%B9%E7%B2%BE%E9%80%89%E6%B7%B7%E5%90%88%E5%9E%8B%E8%AF%81%E5%88%B8%E6%8A%95%E8%B5%84%E5%9F%BA%E9%87%912023%E5%B9%B4%E5%B9%B4%E5%BA%A6%E6%8A%A5%E5%91%8A.pdf", "authors": "易方达基金管理人；报告期列示张坤，非个人独著", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：§4.1经理表、§4.4–4.5，印刷页9–12；期间截止2023-12-31", "retrieval_level": "substantive primary material read", "summary": "管理人讨论资本配置、持续高增长的低基准概率，以及现金产生能力对AI投入的支撑。", "locator": "§4.1经理表、§4.4–4.5，印刷页9–12；期间截止2023-12-31" }, { "id": "R2B3-B75C", "title": "易方达蓝筹精选混合型证券投资基金2024年年度报告", "date": "2025-03-31", "url": "https://cdn.efunds.com.cn/owch/data/bulletin/20250331/%E6%98%93%E6%96%B9%E8%BE%BE%E8%93%9D%E7%AD%B9%E7%B2%BE%E9%80%89%E6%B7%B7%E5%90%88%E5%9E%8B%E8%AF%81%E5%88%B8%E6%8A%95%E8%B5%84%E5%9F%BA%E9%87%912024%E5%B9%B4%E5%B9%B4%E5%BA%A6%E6%8A%A5%E5%91%8A.pdf", "authors": "易方达基金管理人；报告期列示张坤，非个人独著", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：§4.1经理表、§4.5，印刷页9–13；期间截止2024-12-31", "retrieval_level": "substantive primary material read", "summary": "管理人质疑二元市场叙事，强调效率改善、自由现金流与股东回报。", "locator": "§4.1经理表、§4.5，印刷页9–13；期间截止2024-12-31" }, { "id": "R2B3-B77B", "title": "陈光明：创造长期价值的源泉是优秀公司", "date": "2019-03-28", "url": "https://www.foresightfund.com/contents/2019/3/28-eee82b56ed024cdc916c6bb0a4a17d1f.html", "authors": "陈光明演讲；中国基金报整理、睿远官网转载", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：全文演讲；演讲发生2019-03-21，官网刊载2019-03-28", "retrieval_level": "substantive primary material read", "summary": "演讲把长期回报拆为分红、业绩增长和估值，并强调公司质量、治理及长期资金制度。", "locator": "全文演讲；演讲发生2019-03-21，官网刊载2019-03-28" }, { "id": "R2B3-B77A", "title": "陈光明：价值投资的智慧与进化", "date": "2025-07-31", "url": "https://www.foresightfund.com/contents/2025/7/31-4fd9f076bd8341c5bd981de77e9eeab3.html", "authors": "陈光明", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：官方刊载的本人书序正文；不审计其中历史收益表述", "retrieval_level": "substantive primary material read", "summary": "署名前言讨论价格与价值、安全边际及从廉价资产走向再投资能力。", "locator": "官方刊载的本人书序正文；不审计其中历史收益表述" }, { "id": "R2B3-B82B", "title": "Wesco Financial Corporation 2009 shareholder letter", "date": "2010-02-26", "url": "https://www.berkshirehathaway.com/wesco/cm2009.pdf", "authors": "Charles T. Munger", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：Charles T. Munger署名信，印刷页1–7", "retrieval_level": "substantive primary material read", "summary": "讨论浮存金成本、承保不确定性及周期和结构压力的区别。", "locator": "Charles T. Munger署名信，印刷页1–7" }, { "id": "R2B3-B82A", "title": "Vice Chairman’s Thoughts — Past and Future（2014年度股东信附文）", "date": "2014", "url": "https://www.berkshirehathaway.com/letters/2014ltr.pdf", "authors": "Charles T. Munger", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：仅Munger署名附文印刷页39–43；2014为财年，2015年发布，不把Buffett正文归给Munger", "retrieval_level": "substantive primary material read", "summary": "讨论分权经营、集中资本配置、留存收益检验及保守负债。", "locator": "仅Munger署名附文印刷页39–43；2014为财年，2015年发布，不把Buffett正文归给Munger" }, { "id": "R2B3-B87A", "title": "Evolution of a Manager — Value Investor Insight", "date": "2014-03-31", "url": "https://www.gothamfunds.com/Download.aspx?ID=2e2c42b7-ba7a-44cc-872f-b5d5d857e7f7&Inline=1", "authors": "Joel Greenblatt受访；Value Investor Insight采访", "access": "2026-10-08读取；通过网页检索读取官方托管获授权转载PDF两页；本地HTTP403。", "retrieval_level": "substantive primary material read", "summary": "访谈比较集中与分散价值策略，解释多空组合总敞口与净敞口。", "locator": "官方托管授权转载，两页" }, { "id": "R2B3-B87B", "title": "A Value Investor Defends Value Investing (Despite Its Recent Track Record)", "date": "2019-10-06", "url": "https://www.gothamfunds.com/Download.aspx?ID=4745663f-08f5-43ec-ae0c-f1e5dba7a775&Inline=1", "authors": "Joel Greenblatt受访；Chuck Jaffe采访", "access": "2026-10-08读取；通过网页检索读取官方托管获授权转载PDF两页；本地HTTP403。", "retrieval_level": "substantive primary material read", "summary": "访谈以正常化现金流和资本回报讨论价值，而非把低账面倍数等同低估。", "locator": "官方托管WSJ授权转载，两页" }, { "id": "R2B3-B89A", "title": "Bill Miller 2Q 2021 Market Letter", "date": "2021-07-09", "url": "https://millervalue.com/bill-miller-2q-2021-market-letter/", "authors": "Bill Miller（非Bill Miller IV）", "access": "2026-10-08读取；通过网页检索读取本人市场信全文；本地HTTP403。", "retrieval_level": "substantive primary material read", "summary": "市场信区分事实、解释与概率，讨论公开风险和已计价预期；不是风险必被充分定价的证明。", "locator": "官网全文；本人署名2021-07-05，发布2021-07-09" }, { "id": "R2B3-B89B", "title": "Bill Miller’s Market Observations — March 2022", "date": "2022-03-22", "url": "https://millervalue.com/bill-millers-market-observations-march-2022/", "authors": "Bill Miller（非Bill Miller IV）", "access": "2026-10-08读取；通过网页检索读取本人市场观察全文；本地HTTP403。", "retrieval_level": "substantive primary material read", "summary": "讨论战争、利率和市场预期中的不确定性，提出价值轮动判断。", "locator": "官网全文；本人署名2022-03-20，发布2022-03-22" }, { "id": "R2B3-B99B", "title": "The Prospect of Value Investing in China", "date": "2015-10-23", "url": "https://cdn.prod.website-files.com/5ef3c7300432b40ed865991a/5ef3c7300432b46a7e659977_The%20Prospect%20of%20Value%20Investing%20in%20China%20English%20Translation.pdf", "authors": "李录演讲；官方英文译稿", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：官方网站链接英文翻译PDF，印刷页1–3、13–15；日期为演讲日", "retrieval_level": "substantive primary material read", "summary": "讲座讨论受托责任、激励、能力圈及对判断前提的证伪。", "locator": "官方网站链接英文翻译PDF，印刷页1–3、13–15；日期为演讲日" }, { "id": "R2B3-B99A", "title": "Global Value Investing in Our Era", "date": "2024-12-07", "url": "https://cdn.prod.website-files.com/5ef3c7300432b40ed865991a/67a4f75703627bd3a927077e_Global%20Value%20Investing%20in%20Our%20Era%20%282024-12-07%29.pdf", "authors": "李录演讲；官方英文稿", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：官方链接PDF，导言及第四部分印刷页15–17；日期为演讲日", "retrieval_level": "substantive primary material read", "summary": "讲座连接微观企业行动与宏观约束，区分名义资产和购买力财富。", "locator": "官方链接PDF，导言及第四部分印刷页15–17；日期为演讲日" }, { "id": "R2B3-B100A", "title": "Mohnish Pabrai’s Session at the Heilbrunn Center", "date": "2026-04-21", "url": "https://cdn.prod.website-files.com/691db7fda7023478951386c6/6a4b99597e7e01d646f4d746_20260528_mohnish_pabrais_session_at_heilbrunn_center_on_april_21_2026_v3.pdf", "authors": "Mohnish Pabrai演讲/问答逐字稿", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：官方链接逐字稿印刷页1–3；日期为活动日，文件名20260528为发布版本", "retrieval_level": "substantive primary material read", "summary": "讲座使用能力圈和有限决策次数解释选股纪律；高确信度不自动证明低风险。", "locator": "官方链接逐字稿印刷页1–3；日期为活动日，文件名20260528为发布版本" }, { "id": "R2B3-B100B", "title": "Mohnish Pabrai’s Session with University of Nebraska Omaha", "date": "2026-05-01", "url": "https://cdn.prod.website-files.com/691db7fda7023478951386c6/6a4b991bd3823e4b88190b51_20260525_mohnish_pabrais_session_with_university_of_nebraska_omaha_on_may_1_2026_v3.pdf", "authors": "Mohnish Pabrai演讲/问答逐字稿", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：官方链接逐字稿印刷页1–5；活动日2026-05-01；两讲主题重叠不算独立验证", "retrieval_level": "substantive primary material read", "summary": "讲座讨论错误常见、少数赢家贡献及不机械卖出赢家；克隆作为线索有核验边界。", "locator": "官方链接逐字稿印刷页1–5；活动日2026-05-01；两讲主题重叠不算独立验证" }, { "id": "R2B3-B91A", "title": "Talks at GS: Stanley Druckenmiller", "date": "2021-02-03", "url": "https://www.goldmansachs.com/insights/talks-at-gs/stanley-druckenmiller.html", "authors": "Stanley Druckenmiller受访；Goldman Sachs主办方摘录", "access": "2026-10-08读取；官方正文/PDF文本已取得；本次实际阅读范围：网页精选引语正文；活动日2021-01-21，发布2021-02-03；未观看完整视频", "retrieval_level": "substantive primary material read", "summary": "主办方精选本人引语讨论跨资产寻找机会、流动性带来的纠错能力及集中仓位。", "locator": "网页精选引语正文；活动日2021-01-21，发布2021-02-03；未观看完整视频" }, { "id": "R2S1-ARM425", "title": "Arm FYE2025 Q4 earnings call", "date": "2025-05-07", "url": "https://investors.arm.com/static-files/181d5019-29bd-4ba5-af29-45888e25c637", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-ARM326", "title": "Arm FYE2026 Q3 earnings call", "date": "2026-02-04", "url": "https://investors.arm.com/static-files/ba30dca7-d2dd-4d44-b4c5-9fd953fedaea", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-ARM426", "title": "Arm FYE2026 Q4 earnings call", "date": "2026-05-06", "url": "https://investors.arm.com/static-files/78526857-5997-46eb-9b65-0d3249d83711", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-MPWR126", "title": "MPS Q1 2026 earnings call", "date": "2026-04-30", "url": "https://media.monolithicpower.com/investor-relations/events/MPS_Q1_2026_Earnings_Call_Transcript.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-MPWR226", "title": "MPS Q2 2026 earnings call", "date": "2026-07-30", "url": "https://media.monolithicpower.com/investor-relations/events/Monolithic_Power_Systems_Q2_2026_Earnings_Webinar_Transcript.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-MPWR225", "title": "MPS Q2 2025 earnings webinar", "date": "2025-07-31", "url": "https://media.monolithicpower.com/investor-relations/events/Q2_2025_MPS_Earnings_Webinar_Transcript_1.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-CRUS22", "title": "Cirrus Logic KeyBanc Technology Leadership Forum 2022", "date": "2022-08-08", "url": "https://s21.q4cdn.com/307563939/files/doc_downloads/2022/08/CRUS-USQ_Transcript_2022-08-08.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-CRUS26", "title": "Cirrus Logic KeyBanc Technology Leadership Forum 2026", "date": "2026-08-11", "url": "https://s21.q4cdn.com/307563939/files/doc_events/2026/Aug/11/CRUS-US-CORRECTED-TRANSCRIPT-Cirrus-Logic-IncCRUSUS-KeyBanc-Technology-Leadership-Forum-1-ET-PM-12-08-2026-2026-08-12-0.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-TSM124", "title": "TSMC Q1 2024 earnings call", "date": "2024-04-18", "url": "https://investor.tsmc.com/chinese/encrypt/files/encrypt_file/reports/2024-04/34ff75e23e53246302ce3a8d90d0423c57c6b120/TSMC%201Q24%20Transcript.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-TSM422", "title": "TSMC Q4 2022 earnings call", "date": "2023-01-12", "url": "https://investor.tsmc.com/english/encrypt/files/encrypt_file/reports/2023-01/84f1ef6f66c6c90b05d2e581066a01a8f1c4f410/TSMC%204Q22%20Transcript.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-AIX423", "title": "AIXTRON FY2023 results call", "date": "2024-02-29", "url": "https://www.aixtron.com/investoren/events/telefonkonferenz/2023/Edited%20Transcript%20FY-23%20Results%20Call.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-AIX325", "title": "AIXTRON Q3 2025 earnings call", "date": "2025-10-30", "url": "https://www.aixtron.com/investoren/events/telefonkonferenz/2025/AIXTRON_SE_Q3_2025_transcript.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。", "additional_readings": [ { "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。", "access": "2026-10-08实际阅读具名问答正文；发行人托管正文/PDF已读取；未听录音。", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "batch": "2026-10-08" } ] }, { "id": "R2S1-STX126", "title": "Seagate FY2026 Q1 earnings call", "date": "2025-10-28", "url": "https://s24.q4cdn.com/101481333/files/doc_financials/2026/q1/CORRECTED-TRANSCRIPT_-Seagate-Technology-Holdings-Plc-STX-US-Q1-2026-Earnings-Call-28-October-2025-5_00-PM-ET.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-STX223", "title": "Seagate FY2023 Q2 earnings call", "date": "2023-01-25", "url": "https://s24.q4cdn.com/101481333/files/doc_financials/2023/q2/CORRECTED-TRANSCRIPT_-Seagate-Technology-Holdings-Plc%28STX-US%29%2C-Q2-2023-Earnings-Call%2C-25-January-2023-5_00-PM-ET.1.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-IONQ126", "title": "IonQ Q1 2026 earnings call", "date": "2026-05-06", "url": "https://s28.q4cdn.com/828571518/files/doc_financials/2026/q1/CORRECTED-TRANSCRIPT_-IonQ-Q1-2026-Earnings-Call-6-May-2026-4_30-PM-ET.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-ARGUS25", "title": "Argus Analysts Top Picks for 2025", "date": "2025-01-08", "url": "https://www.argusresearch.com/assets/doc/20250108_WebinarSlides.pdf", "authors": "Argus team; Jim Kelleher named on selected Apple/NVIDIA pages", "access": "2026-10-08下载官方PDF读取具名Apple/NVIDIA页；不是把整份团队材料归于一人。原文有CPU-based/Blackwood等表述错误，不作为技术事实引用。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-ARGUS26", "title": "Argus Analysts Top Picks for 2026", "date": "2026-01-07", "url": "https://www.argusresearch.com/assets/doc/20260107_WebinarSlides.pdf", "authors": "Argus team; Jim Kelleher named on selected Apple/NVIDIA pages", "access": "2026-10-08下载官方PDF读取具名Apple/NVIDIA页；不是把整份团队材料归于一人。原文有CPU-based/Blackwood等表述错误，不作为技术事实引用。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-RFM1", "title": "OpenAI – The Numbers Game", "date": "2026-04-10", "url": "https://www.radiofreemobile.com/openai-the-numbers-game/", "authors": "Richard Windsor", "access": "2026-10-08读取本人署名全文；所有商业数值是作者估计或引用，未另行认证为公司报告事实。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-RFM2", "title": "AMD Q1 26 – Money for Nothing", "date": "2026-05-07", "url": "https://www.radiofreemobile.com/amd-q1-26-money-for-nothing/", "authors": "Richard Windsor", "access": "2026-10-08读取本人署名全文；尤其行权价格推算是作者猜测，未认证条款或接受其结论。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-D2D1", "title": "D2D Cont'd – Everything is an AI Trade", "date": "2025-07-14", "url": "https://d2d.substack.com/p/d2d-contd-everything-is-an-ai-trade", "authors": "Digits to Dollars / Jay Goldberg; publication uses we", "access": "2026-10-08读取公开全文，按D2D署名研究归属，不推断团队成员私人观点。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-D2D2", "title": "D2D Cont'd: Fear or Faith?", "date": "2025-10-22", "url": "https://d2d.substack.com/p/d2d-contd-fear-or-faith", "authors": "Digits to Dollars / Jay Goldberg; publication uses we", "access": "2026-10-08读取公开全文；原文区分作者模型、公司预测及重复计算，均保留为待检验论证。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-ALTR410", "title": "Altera Q4 2010 earnings call", "date": "2011-01-25", "url": "https://www.sec.gov/Archives/edgar/data/768251/000076825111000005/conferencecalltranscript.htm", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取SEC存档Gus Richard/Piper Jaffray具名问答；历史方法样本。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-POWI305", "title": "Power Integrations Q3 2005 earnings call", "date": "2005-10-19", "url": "https://www.sec.gov/Archives/edgar/data/833640/000119312505207677/dex992.htm", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取SEC存档Gus Richard/First Albany具名问答；历史方法样本。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-ARM224", "title": "Arm FYE2024 Q2 earnings call", "date": "2023-11-08", "url": "https://investors.arm.com/static-files/04e0fd13-9243-49ba-927f-288765598926", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S1-MSI412", "title": "Motorola Solutions Q4 2012 earnings call", "date": "2013-01-23", "url": "https://www.motorolasolutions.com/content/dam/msi/investors/doc_events/2013/q4-2012-earnings-conference-call.pdf", "authors": "Issuer management and named analyst questioners", "access": "2026-10-08读取发行人托管PDF具名问答正文，未听录音。", "retrieval_level": "primary selected full-text passages", "summary": "仅使用具体定位的归属明确的研究样本；管理层回复与分析师问题分开。" }, { "id": "R2S2-ASML424", "title": "ASML Q4 2024 earnings Q&A — public transcript mirror", "date": "2025-01-29", "url": "https://www.roic.ai/quote/ASML/transcripts/2024-year/4-quarter", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；公开第三方原始问答转录镜像，未听录音；可能有转录错误。", "retrieval_level": "public full transcript mirror; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。", "additional_readings": [ { "date": "2026-10-08", "access": "已独立读取ROIC公开转录具名问题及管理层回答；未核对原录音，存在转录误差风险。", "summary": "Sandeep Deshpande两期：High-NA订单/量产验证及下一年订单覆盖；Sara Russo仅Q4：披露频度、订单时间和指导覆盖。" } ] }, { "id": "R2S2-ASML125", "title": "ASML Q1 2025 earnings Q&A — public transcript mirror", "date": "2025-04-16", "url": "https://www.roic.ai/quote/ASML/transcripts/2025-year/1-quarter", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；公开第三方原始问答转录镜像，未听录音；可能有转录错误。", "retrieval_level": "public full transcript mirror; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。", "additional_readings": [ { "date": "2026-10-08", "access": "已独立读取ROIC公开转录具名问题及管理层回答；未核对原录音，存在转录误差风险。", "summary": "Sandeep Deshpande两期：High-NA订单/量产验证及下一年订单覆盖；Sara Russo仅Q4：披露频度、订单时间和指导覆盖。" } ] }, { "id": "R2S2-ASML225", "title": "ASML Q2 2025 earnings Q&A — public transcript mirror", "date": "2025-07-16", "url": "https://www.roic.ai/quote/ASML/transcripts/2025-year/2-quarter", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；公开第三方原始问答转录镜像，未听录音；可能有转录错误。", "retrieval_level": "public full transcript mirror; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-LAM425", "title": "Lam Research FY2025 Q4 earnings Q&A — public transcript mirror", "date": "2025-07-30", "url": "https://www.roic.ai/quote/LRCX/transcripts/2025-year/4-quarter", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；公开第三方原始问答转录镜像，未听录音；可能有转录错误。", "retrieval_level": "public full transcript mirror; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-STM125", "title": "STMicroelectronics Q1 2025 earnings Q&A — public transcript mirror", "date": "2025-04-24", "url": "https://www.roic.ai/quote/STM/transcripts/2025-year/1-quarter", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；公开第三方原始问答转录镜像，未听录音；可能有转录错误。", "retrieval_level": "public full transcript mirror; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-NOW225", "title": "ServiceNow Q2 2025 earnings Q&A — public transcript mirror", "date": "2025-07-23", "url": "https://www.roic.ai/quote/NOW/transcripts/2025-year/2-quarter", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；公开第三方原始问答转录镜像，未听录音；可能有转录错误。", "retrieval_level": "public full transcript mirror; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-ADBE224", "title": "Adobe FY2024 Q2 earnings Q&A — public transcript mirror", "date": "2024-06-13", "url": "https://www.roic.ai/quote/ADBE/transcripts/2024-year/2-quarter", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；公开第三方原始问答转录镜像，未听录音；可能有转录错误。", "retrieval_level": "public full transcript mirror; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-SPLK123", "title": "Splunk FY2023 Q1 earnings Q&A — public transcript mirror", "date": "2022-05-25", "url": "https://www.roic.ai/quote/SPLK/transcripts/2023-year/1-quarter", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；公开第三方原始问答转录镜像，未听录音；可能有转录错误。", "retrieval_level": "public full transcript mirror; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-ADBE225", "title": "Adobe FY2025 Q2 earnings call", "date": "2025-06-12", "url": "https://www.adobe.com/cc-shared/assets/investor-relations/pdfs/adbe-q2fy25-transcript.pdf", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；发行人托管正文/PDF已读取；未听录音。", "retrieval_level": "issuer-hosted full transcript; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-ADBE422", "title": "Adobe FY2022 Q4 earnings call", "date": "2022-12-15", "url": "https://www.adobe.com/cc-shared/assets/investor-relations/pdfs/q4-fy2022-earnings-transcript.pdf", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；发行人托管正文/PDF已读取；未听录音。", "retrieval_level": "issuer-hosted full transcript; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-UI324", "title": "UiPath FY2024 Q3 earnings call", "date": "2023-11-30", "url": "https://d1io3yog0oux5.cloudfront.net/_ce622def36020bd12ae7bda661b8d25a/uipath/db/1195/13709/webcast_transcript/UiPath_-_3Q_2024_-_Earnings_Call_Transcript.pdf", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；发行人托管正文/PDF已读取；未听录音。", "retrieval_level": "issuer-hosted full transcript; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-MS125", "title": "Microsoft FY2025 Q1 earnings call", "date": "2024-10-30", "url": "https://www.microsoft.com/en-us/investor/events/fy-2025/earnings-fy-2025-q1", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；发行人托管正文/PDF已读取；未听录音。", "retrieval_level": "issuer-hosted full transcript; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-COUR123", "title": "Coursera Q1 2023 earnings call", "date": "2023-04-27", "url": "https://s27.q4cdn.com/928340662/files/doc_financials/2023/q1/5_Transcript_Q1-2023.pdf", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；发行人托管正文/PDF已读取；未听录音。", "retrieval_level": "issuer-hosted full transcript; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S2-TS419", "title": "TSMC Q4 2019 earnings call", "date": "2020-01-16", "url": "https://investor.tsmc.com/english/encrypt/files/encrypt_file/english/2019/Q4/TSMC%204Q19%20transcript.pdf", "authors": "Issuer management and named analyst questioners; transcript publisher retains authorship", "access": "2026-10-08实际阅读具名问答正文；发行人托管正文/PDF已读取；未听录音。", "retrieval_level": "issuer-hosted full transcript; attributable questions read", "summary": "用于具体提问样本；管理层回答与分析师提问分开，不认证个人预测、持续方法或业绩。" }, { "id": "R2S3-JBLQ124", "title": "Jabil Q1 FY2025 Earnings Call", "date": "2024-12-18", "url": "https://s27.q4cdn.com/276975351/files/doc_financials/2025/q1/JBL-USQ_Transcript_2024-12-18.pdf", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；发行人托管PDF；未用录音逐字复核。", "retrieval_level": "issuer-hosted-transcript-full-text", "summary": "已读具名问答：S78 追问数据中心新增收入的利润率及并购取向。 / 管理层：管理层称数据中心不稀释企业利润率，资本配置另述。；S79 区分现有美国产能承接与后续扩产能力。 / 管理层：Dastoor称余量有限，部分地点可能扩张。；S80 再问ATE/WFE节奏及硅光产品演进。 / 管理层：Dastoor称测试较强而硅光起点较小。" }, { "id": "R2S3-JBLQ424", "title": "Jabil Q4 FY2024 Earnings Call", "date": "2024-09-26", "url": "https://s27.q4cdn.com/276975351/files/doc_financials/2024/q4/JBL-USQ_Transcript_2024-09-26.pdf", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；发行人托管PDF；未用录音逐字复核。", "retrieval_level": "issuer-hosted-transcript-full-text", "summary": "已读具名问答：S75 以收入规模、产能利用与费用率检验6%利润率路径。 / 管理层：Dastoor用毛利和SG&A组合解释目标，时间仍不精确。；S78 分辨后半年EPS提升来自利用率、组合还是回购。 / 管理层：Dastoor强调季节性及回购安排，未给完整桥。；S79 追问汽车需求变化的短期负杠杆及FCF改善桥。 / 管理层：管理层指产能延期及营运资本/CapEx变化。；S80 拆解半导体设备增长并核对AI收入跨分部定义。 / 管理层：管理层区分测试、WFE与云/光子业务。" }, { "id": "R2S3-DELLQ125", "title": "Dell Q1 FY2025 Earnings Call", "date": "2024-05-30", "url": "https://investors.delltechnologies.com/static-files/0b83385d-e4ad-4c58-86d5-e6ed90e2992b", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；发行人托管PDF；未用录音逐字复核。", "retrieval_level": "issuer-hosted-transcript-full-text", "summary": "已读具名问答：S77 区分主动降价抢份额与竞争应对，并拆解毛利压力。 / 管理层：管理层称大型项目更具竞争，CFO把压力区分rate和mix。" }, { "id": "R2S3-DELLQ225", "title": "Dell Q2 FY2025 Earnings Call", "date": "2024-08-29", "url": "https://investors.delltechnologies.com/static-files/5baec214-ec57-45c9-8e4a-83e6ed9355f5", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；发行人托管PDF；未用录音逐字复核。", "retrieval_level": "issuer-hosted-transcript-full-text", "summary": "已读具名问答：S75 询问企业AI客户能带来多少服务、存储附加及毛利差异。 / 管理层：Clarke称企业利润率更高，但未给附加率量化桥。；S77 追问AI积压不增是否反映Blackwell或液冷供给时点。 / 管理层：Clarke强调客户可接货、部件供给及管线转订单。" }, { "id": "R2S3-FRSHQ325", "title": "Freshworks Q3 2025 Earnings Call", "date": "2025-11-05", "url": "https://s21.q4cdn.com/987526491/files/doc_financials/2025/q3/FRSH-Q3-25-Earnings-Transcript.pdf", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；发行人托管PDF；未用录音逐字复核。", "retrieval_level": "issuer-hosted-transcript-full-text", "summary": "已读具名问答：S65 询问独立ESM销售是否另设团队、如何定价。 / 管理层：Woodside称沿用现有销售，市场扩展仍待验证。" }, { "id": "R2S3-NAVANQ426", "title": "Navan Q4 FY2026 Earnings Call", "date": "2026-03-25", "url": "https://investors.navan.com/static-files/16602ac2-7215-443d-a8f8-011ac356b23d", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；发行人托管PDF；未用录音逐字复核。", "retrieval_level": "issuer-hosted-transcript-full-text", "summary": "已读具名问答：S65 询问餐厅预订经济性及对指引贡献。 / 管理层：Sindicich称Navan Edge尚非指引重要贡献。" }, { "id": "R2S3-MSIQ325", "title": "Motorola Solutions Q3 2025 Earnings Call", "date": "2025-10-30", "url": "https://www.motorolasolutions.com/content/dam/msi/investors/doc_financials/2025/q3/msi_q3_2025_earnings_call_transcript_10_30_25.pdf", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；发行人托管PDF；未用录音逐字复核。", "retrieval_level": "issuer-hosted-transcript-full-text", "summary": "已读具名问答：S71 询问SVX上市后的竞争者定价反应。 / 管理层：Molloy以集成设备和服务TCO回应；未量化份额。" }, { "id": "R2S3-MSICONF26", "title": "Motorola Solutions at Piper Sandler Growth Frontiers", "date": "2026-09-16", "url": "https://www.motorolasolutions.com/content/dam/msi/investors/doc_events/2026/final_transcript_msi_at_piper_sandler_growth_frontiers_conference_09-16-26.pdf", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；发行人托管PDF；未用录音逐字复核。", "retrieval_level": "issuer-hosted-transcript-full-text", "summary": "已读具名问答：S71 追问AI策略及适合救援人员工作流的原因。 / 管理层：Saptharishi解释把AI嵌入既有核心工作流。" }, { "id": "R2S3-AMZNQ424", "title": "Amazon Q4 2024 Earnings Call — public transcript mirror", "date": "2025-02-06", "url": "https://www.fool.com/earnings/call-transcripts/2025/02/06/amazoncom-amzn-q4-2024-earnings-call-transcript/", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S66 追问资本开支年化规模与AWS增长是否受供给限制。 / 管理层：Jassy将投资与先建后变现现金周期联系。", "corroboration_urls": [ "https://ir.aboutamazon.com/events/event-details/2025/Q4-2024-Amazoncom-Inc-Earnings-Conference-Call/default.aspx" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-EQIXQ225", "title": "Equinix Q2 2025 Earnings Call — public transcript mirror", "date": "2025-07-30", "url": "https://www.roic.ai/quote/EQIX/transcripts/2025-year/2-quarter", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。 Roic页面标题July31；本记录采用官方IR实际电话会日期July30。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S84 问追加资本支出能否提前交付。 / 管理层：CEO强调能源及供应链约束。；S85 问稳定期能否缩短及预租降低风险。 / 管理层：CEO称预售、大客户和建设分期影响爬坡。；S88 询问互联新增锯齿波动及持续性。 / 管理层：CEO指云和AI网络需求及产品使用。；S90 追问资本化利息变化及历史处理。 / 管理层：CFO纠正此前已资本化，项目周期推高金额。", "corroboration_urls": [ "https://investor.equinix.com/news-events/press-releases/detail/1076/equinix-reports-second-quarter-2025-results" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-EQIXQ325", "title": "Equinix Q3 2025 Earnings Call — public transcript mirror", "date": "2025-10-29", "url": "https://www.fool.com/earnings/call-transcripts/2025/10/29/equinix-eqix-q3-2025-earnings-call-transcript/", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S84 追问预售增长与销售窗口延长关系。 / 管理层：CEO称窗口由3–6个月延至12个月。；S85 问供给紧张定价与次年AFFO融资。 / 管理层：CFO区分借款成本和资本化利息。；S88 问新云入口相对传统云的战略价值。 / 管理层：CEO以企业访问和生态吸引力回应。；S90 问互联ARPU与电力密度、地区复制性。 / 管理层：CEO强调美洲客户组合。；S82 追问预售与年化签约正常运行水平。 / 管理层：CEO强调波动，CFO区分收入时点。；S83 问新增土地分配零售与xScale。 / 管理层：CEO表示按地点和用途价值分配。", "corroboration_urls": [ "https://investor.equinix.com/news-events/press-releases/detail/1086/equinix-reports-strong-third-quarter-2025-results" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-ETNQ125", "title": "Eaton Q1 2025 Earnings Call — public transcript mirror", "date": "2025-05-02", "url": "https://www.roic.ai/quote/ETN/transcripts/2025-year/1-quarter", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S95 问高基数下订单与电力需求。 / 管理层：CEO指大单多年交付造成波动。", "corroboration_urls": [ "https://www.eaton.com/us/en-us/company/news-insights/news-releases/2025/eaton-to-announce-first-quarter-2025-earnings-on-may-2--2025.html" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-ETNQ225", "title": "Eaton Q2 2025 Earnings Call — public transcript mirror", "date": "2025-08-05", "url": "https://www.roic.ai/quote/ETN/transcripts/2025-year/2-quarter", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S95 问增长是否说明份额提升及数据中心产品布局。 / 管理层：CEO用公司与市场增速比较；是管理层推断。", "corroboration_urls": [ "https://www.eaton.com/us/en-us/company/news-insights/news-releases/2025/eaton-to-announce-second-quarter-2025-earnings-on-august-5--2025.html" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-AMEQ424", "title": "AMETEK Q4 2024 Earnings Call — public transcript mirror", "date": "2025-02-04", "url": "https://www.roic.ai/quote/AME/transcripts/2024-year/4-quarter", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S98 问Kern并购竞争定位及组合退出。 / 管理层：Zapico指精密加工细分互补性。", "corroboration_urls": [ "https://www.ametek.com/newsroom/news/investor/2025/january/ametek-announces-fourth-quarter-2024-earnings-call-and-webcasted-investor-conference-call-information" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-AMEQ225", "title": "AMETEK Q2 2025 Earnings Call — public transcript mirror", "date": "2025-07-31", "url": "https://www.roic.ai/quote/AME/transcripts/2025-year/2-quarter", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S98 问航天并购优先级及自动化周期。 / 管理层：Zapico称订单到收入存在3–6月滞后。", "corroboration_urls": [ "https://www.ametek.com/newsroom/news/investor/2025/july/ametek-announces-second-quarter-2025-earnings-call-and-webcasted-investor-conference-call-information" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-PANWQ126", "title": "Palo Alto Networks Q1 FY2026 — public transcript mirror", "date": "2025-11-19", "url": "https://www.roic.ai/quote/PANW/transcripts/2026-year/1-quarter", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S73 问并购过渡期利润稀释与协同时点。 / 管理层：CFO给出剔除一次性费用后的FCF区间。", "corroboration_urls": [ "https://investors.paloaltonetworks.com/news-releases/news-release-details/palo-alto-networks-reports-fiscal-first-quarter-2026-financial" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-DLRQ225", "title": "Digital Realty Q2 2025 Earnings Call — public transcript mirror", "date": "2025-07-24", "url": "https://www.roic.ai/quote/DLR/transcripts/2025-year/2-quarter", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S82 问公用事业预付款对成本和竞争影响。 / 管理层：Power称筛选真实开发方；未量化成本。；S83 问小容量增长来自市场还是份额及销售改变。 / 管理层：管理层归因全球平台和客户覆盖。；S87 问开发CapEx规模持续多久。 / 管理层：CFO和Power区分总投资与公司份额资金。", "corroboration_urls": [ "https://investor.digitalrealty.com/static-files/7689bc64-e5cc-45ae-aa51-f5f9eac71cf0" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-EQIXQ125", "title": "Equinix Q1 2025 Earnings Call — public transcript mirror", "date": "2025-04-30", "url": "https://www.roic.ai/quote/EQIX/transcripts/2025-year/1-quarter", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S86 问供给约束下提价和产能匹配。 / 管理层：CEO指续约价与加快项目交付。", "corroboration_urls": [ "https://investor.equinix.com/news-events/press-releases/detail/1068/equinix-reports-first-quarter-2025-results" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-EQIXQ324", "title": "Equinix Q3 2024 Earnings Call — public transcript mirror", "date": "2024-10-30", "url": "https://www.fool.com/earnings/call-transcripts/2024/10/30/equinix-eqix-q3-2024-earnings-call-transcript/", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S86 问xScale收益贡献何时进入运行水平。 / 管理层：CFO区分一次性费用和经常性收益爬坡。", "corroboration_urls": [ "https://investor.equinix.com/news-events/press-releases/detail/1055/equinix-reports-third-quarter-2024-results" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-EQIXQ423", "title": "Equinix Q4 2023 Earnings Call — public transcript mirror", "date": "2024-02-14", "url": "https://www.roic.ai/quote/EQIX/transcripts/2023-year/4-quarter", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S87 问竞争流失和价格变化。 / 管理层：Meyers指大客户与容量限制，价格仍稳。", "corroboration_urls": [ "https://investor.equinix.com/news-events/press-releases/detail/1030/equinix-reports-fourth-quarter-and-full-year-2023-results" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-EQIXQ122", "title": "Equinix Q1 2022 Earnings Call — public transcript mirror", "date": "2022-04-27", "url": "https://www.roic.ai/quote/EQIX/transcripts/2022-year/1-quarter", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S89 问长期利润率目标与可控成本杠杆。 / 管理层：Meyers区分电价和内部效率项目。", "corroboration_urls": [ "https://investor.equinix.com/news-events/press-releases/detail/2/equinix-reports-first-quarter-2022-results" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "R2S3-EQIXBARC26", "title": "Equinix at Barclays Financial Services Conference — public speaker transcript", "date": "2026-09-14", "url": "https://earningsapi.io/transcripts/equinix-inc_eqix_earnings_call_transcript_2026-09-14", "authors": "Management and individually named questioners; transcript publisher retains authorship", "access": "2026-10-08实际读取具名问答正文；第三方公开全文转录，未用录音逐字复核，官方IR核对活动日期。 页面自动摘要错误标作Q3业绩，未采用摘要或其数字；仅读取Speaker Segments具名实质问答。", "retrieval_level": "public-full-transcript-mirror", "summary": "已读具名问答：S89 追问延迟敏感负载占需求的角色。 / 管理层：Shaw区分不同负载及园区位置。", "corroboration_urls": [ "https://investor.equinix.com/news-events/press-releases/detail/1124/media-alert-equinix-to-speak-at-upcoming-investor" ], "corroboration_note": "官方IR仅核对活动及日期，不算另一份个人方法材料。" }, { "id": "O9G1-B04-S1", "title": "AI: We are only scraping the surface", "date": "2023-08-14", "url": "https://www.polarcapital.co.uk/static/documents/Polar_Capital_Artificial_Intelligence_Fund_-_August_2023.pdf", "authors": "Xuesong Zhao", "summary": "把独有数据及非科技采用者纳入AI受益研究，考察技术如何改变成本和商业模式。", "access": "2026-10-09 原发布者PDF第1、4–5页，末页正文明确本人署名与日期。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B04-S2", "title": "Artificial Intelligence Fund video transcript", "date": "2026-05", "url": "https://www.polarcapital.co.uk/static/documents/Polar_Capital_Artificial_Intelligence_Video_Transcript_-_May_2026.pdf", "authors": "Xuesong Zhao", "summary": "软件功能可被复制时重估原有数据优势；区分盈利修正与估值重估。", "access": "2026-10-09 官方PDF下载转文本，实读第1–2页；日期来自文件名，正文无具体录制日。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B10-S1", "title": "I couldn’t sleep with full index weights to Magnificent Seven", "date": "2024-10-21", "url": "https://www.theaic.co.uk/aic/news/industry-news/mike-seidenberg-i-couldnt-sleep-with-full-index-weights-to-magnificent-seven", "authors": "Mike Seidenberg", "summary": "承认低配巨头可能拖累相对表现，但把股东承受的集中风险纳入组合选择。", "access": "2026-10-09 AIC公开转载Citywire原访谈摘段；已读本人完整回答，仅该问答，不冒充一小时全片。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B10-S2", "title": "371. Finding real value in tech investing", "date": "2025-10-16", "url": "https://www.fundcalibre.com/episode-371-finding-real-value-in-tech-investing/", "authors": "Mike Seidenberg", "summary": "在高增长、合理估值增长和低估值可选性间配置；网络安全需求持续不保证供应商不会过时。", "access": "2026-10-09 原采访机构公开完整机器转录；实读mega-cap、AI、cyber和valuation段；2025-10-09录制，10-16发布；未核音频，转录有错误。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B18-S1", "title": "Cloud computing: still in its early stages and spreading globally", "date": "2017-09", "url": "https://www.troweprice.com/content/dam/ide/articles/pdfs/258759_PP_Cloud_Computing.pdf", "authors": "Josh Spencer", "summary": "将客户成本、扩容弹性与SaaS稳态利润率结合，同时保留估值约束。", "access": "2026-10-09 官方PDF第1–3页实质正文，本人具名；PDF封面2017-09，非搜索抓取年份。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B18-S2", "title": "Semiconductor stocks: buying when the chips are down", "date": "2018-11", "url": "https://www.troweprice.com/content/dam/tpd/Articles/PDFs/semiconductor-stocks-buying-when-the-chips-are-down.pdf", "authors": "Josh Spencer", "summary": "用供给集中度、扩产资本强度和终端内容增长检验周期底部利润，而非只外推峰值。", "access": "2026-10-09 官方PDF第1–3页，本人具名，2018-11；历史判断未作事后成功认证。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B25-S1", "title": "Fidelity Growth Company Fund PM Q&A — May 2018", "date": "2018-05-31", "url": "https://www.readkong.com/page/fidelity-growth-company-fund-key-takeaways-5851471", "authors": "Steve Wymer", "summary": "看三至五年增长机会和商业模式延展性，同时因组合构建调整高确信持仓。", "access": "2026-10-09 Fidelity原问答公开第三方全文镜像，第2–3页正文与经理姓名可见；未取得2018官方存档逐字校验。日期是报告截至日。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B25-S2", "title": "Fidelity Growth Company Fund PM Q&A — May 2026", "date": "2026-05-31", "url": "https://www.fidelity.com/bin-public/060_www_fidelity_com/documents/SHDOCS/FDGRX/hosts/sh_comm_pmqa.000025.RETAIL_pdf.pdf", "authors": "Steve Wymer", "summary": "把持续AI采用与企业投资回报及基建约束相连；增长持仓仍需管理集中度。", "access": "2026-10-09 官方滚动PDF第2–4页，实际头部为2026-05-31；截至日不等于发布日；未复制来源未独立核验的业绩或IPO数字。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B30-S1", "title": "Graham & Doddsville Issue XXV — Alex Sacerdote", "date": "2015", "url": "https://business.columbia.edu/sites/default/files-efs/imce-uploads/Graham%20%26%20Doddsville_Issue%2025_0.pdf", "authors": "Alex Sacerdote", "summary": "把采用拐点与竞争优势、未被价格反映的盈利能力联合考察；市场增长不保证每家厂商获利。", "access": "2026-10-09 Columbia原始刊物，2015年秋刊，第6–12页访谈；重点读第7–9页S曲线和退出案例。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B30-S2", "title": "Invest Like the Best EP477 — Investing on the S-Curve", "date": "2026-06-09", "url": "https://podscripts.co/podcasts/invest-like-the-best-with-patrick-oshaughnessy/alex-sacerdote-how-to-invest-through-technology-cycles-invest-like-the-best-ep477", "authors": "Alex Sacerdote", "summary": "区分采用速度与市场上限，并在渗透路径偏离时修正假设；企业整合难度也影响采用速度。", "access": "2026-10-09 公开第三方机器逐字稿，实读19:23–29:40；官方Colossus节目页核对身份、节目号和日期，官方完整稿需登录，未访问。未核音频；明显转录错误不作为事实。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B13-S1", "title": "TSMC Q2 2004 Earnings Conference Call — Pramanick Q&A", "date": "2004-07-29", "url": "https://investor.tsmc.com/chinese/encrypt/files/encrypt_file/english/2004/Q2/04Q2%20Conf%20Call-Transcript.pdf", "authors": "Shekhar Pramanick", "summary": "卖方时期追问出货指引是否已考虑库存风险，并追问收入增长能否转成毛利率改善。", "access": "2026-10-09 TSMC IR官方承载的Thomson完整逐字稿，下载后读第7–8页本人具名提问；2004-07-29为会议日；当时Schwab Soundview。管理层回答不归给提问人。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G1-B13-S2", "title": "Investing in the metaverse: 13 stocks to buy for 2022 — attributed interview portions", "date": "2022-01-20", "url": "https://fortune.com/2022/01/20/metaverse-investing-13-stocks-to-buy-in-2022/", "authors": "Shekhar Pramanick", "summary": "考察芯片需求与公司盈利。", "access": "2026-10-09 Fortune原采访正文The infrastructure bet节，仅使用Pramanick具名段；其他受访者观点排除。本人为Columbia Seligman买方经理。", "retrieval_level": "公开正文已读；归属与读取边界见access" }, { "id": "O9G2-M21", "title": "Steve Mandel — Investing Behind Change", "date": "2021-07-20", "authors": "Steve Mandel，访谈主持 Patrick O’Shaughnessy", "url": "https://podscripts.co/podcasts/invest-like-the-best-with-patrick-oshaughnessy/steve-mandel-investing-behind-change-invest-like-the-best-ep-235", "summary": "从可复制的单位经济与管理质量理解业务安全边际；由具体公司发现多年变迁，而非从主题标签倒推股票。", "access": "2026-10-09 公开完整转录已读00:09–00:22投资方法段；节目原页核日期与人物；自动转录存在姓名错字。", "retrieval_level": "公开原始材料转录/信件镜像；非官方校订，未核对原音频或官方文件字节", "original_url": "https://colossus.com/episode/mandel-investing-behind-change/" }, { "id": "O9G2-M25", "title": "Lessons from Steve Mandel", "date": "2025-09-15", "authors": "Steve Mandel；主持Rick Buhrman、Paul Buser；Kevin Gee转录", "url": "https://aletteraday.substack.com/p/letter-298-steve-mandel-paul-buser", "summary": "以自由现金流收益率、再投资回报与融资依赖共同看估值；高回报扩张导致的低FCF不能直接视为劣质，AI赢家也可能因买价过高给投资者低回报。", "access": "2026-10-09 公开完整转录已读Valuation Frameworks and Non-Linear Thinking等章节；镜像发布日期2025-10-21，节目日期由Colossus原页核为2025-09-15。", "retrieval_level": "公开原始材料转录/信件镜像；非官方校订，未核对原音频或官方文件字节", "original_url": "https://colossus.com/episode/lessons-from-steve-mandel/", "mirror_publication_date": "2025-10-21" }, { "id": "O9G2-H09", "title": "Viking 4Q 2008 investor letter", "date": "2009-01-09", "authors": "O. Andreas Halvorsen（信件主文）；附件另有Dave Ott邮件", "url": "https://pdfcoffee.com/viking-onshore-4q-2008-letter-pdf-free.html", "summary": "把多空选股价差、净敞口与总敞口分开；降低总杠杆可能牺牲账面收益，却保留融资安全与危机中继续研究交易的能力。", "access": "2026-10-09 公开信件转录正文已读页1–3，抬头明确From: O. Andreas Halvorsen、January 9, 2009；附件Dave Ott邮件不是Halvorsen个人原话；未认证上传者身份或官方字节一致。", "retrieval_level": "公开原始材料转录/信件镜像；非官方校订，未核对原音频或官方文件字节", "date_basis": "信件抬头日期，报告期为2008Q4" }, { "id": "O9G2-H10", "title": "Andreas Halvorsen 2Q10 Investor Letter", "date": "2010-07-30", "authors": "O. Andreas Halvorsen（转载信件主文）", "url": "https://www.gurufocus.com/news/101623/andreas-halvorsen-2q10-investor-letter?mobile=true", "summary": "检视高确信度持仓、分散度和多空激励；多头与空头的流动性及损失不对称，不能按同样集中规则处理。", "access": "2026-10-09 公开长篇信件主文转载已读开头From: O. Andreas Halvorsen及集中度、做空激励、总敞口段；原信无可确认具体写信日，date用镜像公布日，报告期2010Q2；未取得原附件及官方原件字节核验。", "retrieval_level": "公开原始材料转录/信件镜像；非官方校订，未核对原音频或官方文件字节", "date_basis": "GuruFocus镜像发布时间2010-07-30；原信确切写作日未知" }, { "id": "O9G2-D25", "title": "Dan Sundheim of D1 Capital on the art of public market investing", "date": "2025-10-22", "authors": "Dan Sundheim；Cheeky Pint访谈", "url": "https://cheekypint.transistor.fm/15/transcript", "summary": "三年预期回报依赖退出倍数；股价提前兑现会降低未来IRR，但过早卖出长期优秀企业同样有机会成本。", "access": "2026-10-09 节目官方完整逐字稿，已读00:07:02–00:11:44的IRR、退出估值及Netflix投资段。", "retrieval_level": "原节目公开完整逐字稿正文" }, { "id": "O9G2-D26", "title": "Dan Sundheim — The Art of Public and Private Market Investing", "date": "2026-02-24", "authors": "Dan Sundheim；主持Patrick O’Shaughnessy", "url": "https://podscripts.co/podcasts/invest-like-the-best-with-patrick-oshaughnessy/dan-sundheim-the-art-of-public-and-private-market-investing-invest-like-the-best-ep460", "summary": "把LLM训练视为前置固定投资，并检验采用速度、后续资本供给与回报；Netflix式成本摊薄与Spotify式个性化黏性是类比，模型商品化及延迟采用可能破坏回报。", "access": "2026-10-09 公开完整转录已读约00:12–00:20资本回报、Netflix/Spotify类比与业务聚焦段；动态广告造成时间戳与官方show notes略有差别，未逐秒核音频。", "retrieval_level": "公开原始材料转录/信件镜像；非官方校订，未核对原音频或官方文件字节", "original_url": "https://colossus.com/episode/the-art-of-public-private-market-investing/" }, { "id": "O9G2-SH22", "title": "Where VC Flows From Here — Scott Shleifer and Hani Enaya", "date": "2022-10-26", "authors": "Scott Shleifer，主持Hani Enaya；Kevin Gee转录", "url": "https://aletteraday.substack.com/p/letter-307-scott-shleifer-and-hani", "original_url": "https://fii-institute.org/wp-content/uploads/2022/07/FII6-Program-2022.10.26.1356KSA.docx-1.pdf", "mirror_publication_date": "2025-11-25", "date_basis": "FII官方议程Day2，2022-10-26；镜像发表于2025-11-25", "summary": "本人讨论低渗透率市场的增长空间、高增量利润及融资依赖；属于单次方法材料，不能由此认证稳定策略或把多年后利润用于上市时点估值。", "access": "2026-10-09 已读取完整公开转录的渗透率、成本及融资段；官方议程第34页核对嘉宾、主持及日期。", "retrieval_level": "第三方公开完整转录；作者自述转录错误自负，未核原音频" }, { "id": "O9G2-SH23", "title": "Tiger Global India investor-call remarks — original reporting selected passages", "date": "2023-02-01", "event_date": "2023-01-31", "date_basis": "原刊2023-02-01；文中前日Tuesday通话为2023-01-31", "authors": "Scott Shleifer（具名引语）；Manish Singh（原报道记者）", "url": "https://techcrunch.com/2023/02/01/tiger-global-india-returns-way-below-average-but-firm-remains-bullish-says-scott-shleifer/", "summary": "增长需转为利润。", "access": "2026-10-09 原刊正文已读，仅取本人直接引语段；未取得完整通话或录音，媒体叙述不作为本人论证。", "retrieval_level": "原报道所载本人实质论证选段；非完整采访或官方逐字稿", "scope": "只支持增长与利润验证这一窄方法维度；未认证所引基金业绩或投资金额。" }, { "id": "O9G3-N1", "title": "CNN First Move: Dan Niles on Tesla earnings and risk", "url": "https://transcripts.cnn.com/show/fmjc/date/2024-10-23/segment/01", "date": "2024-10-23", "summary": "Niles从汽车毛利、交付指引过渡至估值与小仓位风险控制。", "access": "2026-10-09：CNN官方逐字稿本人段落已读；rush transcript，未核音频。", "locator": "NILES具名段，约18:25前后；网页L203–246", "date_basis": "节目播出日期" }, { "id": "O9G3-N2", "title": "CNN Outfront: Dan Niles on trade policy", "url": "https://transcripts.cnn.com/show/ebo/date/2025-04-07/segment/01", "date": "2025-04-07", "summary": "Niles讨论政策持续时间、市场准入及科技企业的跨境风险。", "access": "2026-10-09：CNN官方逐字稿本人段落已读；不把Richard Quest或Dan Ives段归给Niles。", "locator": "NILES具名段L103–119，19:20前后", "date_basis": "节目播出日期" }, { "id": "O9G3-J1", "title": "The Eric Jackson Interview: Overlooked Opportunities And Hunting The Next 100X", "url": "https://www.thenew.money/article/eric-jackson-overlooked-opportunities-and-hunting-the-next-100x", "date": "2026-01-06", "summary": "Jackson以生存能力、资产与执行的区别检验困境平台的上行空间。", "access": "2026-10-09：原采访发布方书面Q&A正文已读；本人持仓倡议，未审计收益和目标。", "locator": "首筛、不对称、Carvana及Opendoor四项问答L56–103", "date_basis": "页面显示发表日期" }, { "id": "O9G3-J2", "title": "Opendoor Q3 2025 earnings call: Eric Jackson video question", "url": "https://au.investing.com/news/transcripts/earnings-call-transcript-opendoor-q3-2025-reveals-challenges-93CH-4110021", "date": "2025-11-07", "summary": "Jackson预录问题追问人员规模及iBuying之外的附加服务收入路径。", "access": "2026-10-09：公开完整逐字稿镜像，已读Jackson提问和紧随回复；存在自动转写错误，未核音频；只归属提问，不归属管理层承诺。", "locator": "Full transcript内Eric Jackson video submission段L165–171", "date_basis": "澳洲镜像显示发表日07/11/2025；不是独立核实的通话日期" }, { "id": "O9G3-M1", "title": "AI capex will be pivotal figure in tech, chip earnings", "url": "https://finance.yahoo.com/video/ai-capex-pivotal-figure-tech-162421261.html", "date": "2024-08-01", "summary": "Meeks区分AI资本开支支持的业务判断与财报期暂停增仓的时点决策。", "access": "2026-10-09：搜索服务返回Yahoo原始页面的带时间戳采访正文，已读00:00–09:14；直连429，未核音频，非whole-page直读。", "locator": "Paul Meeks 01:35、04:01、06:16、08:20；检索精确标题", "date_basis": "页面显示发表日期" }, { "id": "O9G3-M2", "title": "Why this tech investor is on a buying strike", "url": "https://finance.yahoo.com/video/why-tech-investor-buying-strike-100022413.html", "date": "2025-04-01", "summary": "Meeks将消费与企业需求分开，要求随后财报确认AI支出，并区分基建投入与应用变现。", "access": "2026-10-09：搜索服务返回Yahoo原始页面带时间戳采访正文，已读00:00–08:20；直连429，未核音频。", "locator": "Paul Meeks 00:36、01:50、05:25、07:01；Lisa Su 03:53插播不归Meeks", "date_basis": "页面显示发表日期" }, { "id": "O9G3-E1", "title": "Macquarie growth chief: disruptive themes and 13 holdings", "url": "https://www.businessinsider.com/stocks-to-buy-outperform-innovation-disruption-macquarie-growth-small-caps-2022-4", "date": "2022-04-28", "summary": "Ely把更低成本、更高效率的替代路径作为长期需求迁移线索。", "access": "2026-10-09：原采访发布方公开返回正文已读；编辑报道含本人具名引语，非本人独著或完整录音转录。", "locator": "投资流程及数字银行段L181–202；个股列表不复述", "date_basis": "页面显示发表日期" }, { "id": "O9G3-E2", "title": "Growth-investment chief explains his three-part investing strategy", "url": "https://www.businessinsider.com/stocks-to-buy-massive-growth-couple-years-growth-investment-chief-2022-6", "date": "2022-06-27", "summary": "Ely强调集中、创新替代、领先公司，并把投资期限与承受风险联系。", "access": "2026-10-09：原采访发布方公开返回正文已读；只提炼流程段，未验证文中表现和股票预测。", "locator": "流程及期限段L182–204", "date_basis": "页面显示发表日期" }, { "id": "O9G3-A1", "title": "Rethinking Risk: Uncertainty and Investment Returns", "url": "https://www.bailliegifford.com/en/ireland/professional-investor/insights/ic-article/2021-q3-rethinking-risk-author-interview-ins-ar-0856/?printversion=true", "date": "2021-08", "summary": "Anderson质疑风险可被波动率完全度量，提出少数公司的上行偏斜及长期技术机会问题。", "access": "2026-10-09：Baillie Gifford官方双人对话已读；只归属JA明确发言，John Kay回答不转归Anderson。", "locator": "JA发言L104、111、114、124", "date_basis": "原文August2021制作批准，日不详" }, { "id": "O9G3-A2", "title": "Collecting thoughts", "url": "https://www.bailliegifford.com/en/uk/individual-investors/insights/ic-article/2022-q1-trust-44-influencers-collecting-thoughts-in-an-evolving-world-10008352/", "date": "2022-04", "summary": "Anderson署名文章讨论少数公司主导回报、规模收益递增及低估Microsoft的反思。", "access": "2026-10-09：Baillie Gifford官方署名正文已读；历史Scottish Mortgage时期，不表述为当前Lingotto持仓。", "locator": "Company and return concentration及结尾L119–139", "date_basis": "原文April2022／Spring2022，日不详" }, { "id": "O9G4-B57-A", "title": "LTGG Q1 investment update", "date": "2023-04", "url": "https://media.bailliegifford.com/mws/xefj51ce/ltgg-q1-investment-update-transcript.pdf", "authors": "Michael Pye受访；Stewart Hogg主持", "locator": "PDF第1–2页，MP回答财务韧性和适应能力", "summary": "Pye把现金储备与逆周期研发及人才投入的能力相联系。", "access": "2026-10-09读取；官网PDF逐字稿文本已读；文件仅列月份，非具体日；实际阅读范围：PDF第1–2页，MP回答财务韧性和适应能力", "retrieval_level": "substantive primary material read", "date_basis": "文件明确月份；具体日未知" }, { "id": "O9G4-B57-B", "title": "Long Term Global Growth Q4 investment update", "date": "2024-01", "url": "https://media.bailliegifford.com/mws/elujt4z5/baillie-gifford-film-ltgg-q4-update-transcript-jan-2024.pdf", "authors": "Michael Pye受访；Carla Bannatyne主持", "locator": "PDF第1–3页图像，MP回答持仓增长、自由现金流及规模效应", "summary": "Pye用增量收入成本与现金生成能力解释规模化，提醒价格与经营变化可以不同步。", "access": "2026-10-09读取；官网PDF截图直接阅读第1–3页；文本提取失败；文件仅列月份；实际阅读范围：PDF第1–3页图像，MP回答持仓增长、自由现金流及规模效应", "retrieval_level": "substantive primary material read", "date_basis": "文件明确月份；具体日未知" }, { "id": "O9G4-B61-A", "title": "专访冯明远：每月看5家公司", "date": "2020-02-14", "url": "https://www.jiemian.com/article/3905765.html", "authors": "界面新闻黄慧玲采访；冯明远回答", "locator": "投资与研究两节：规模、卖出条件、亏损原因、产业演进", "summary": "冯明远区分研发投入和竞争力衰退造成的亏损，并讨论规模增长对纠错和流动性的约束。", "access": "2026-10-09读取；原刊全文问答已读，交叉复核页面日期；实际阅读范围：投资与研究两节，规模、卖出条件、亏损原因、产业演进", "retrieval_level": "substantive primary material read", "date_basis": "原刊页面直接列2020/02/14 08:00（交叉复核定位）；界面官方新浪同文旁证", "date_corrob_url": "https://finance.sina.cn/2020-02-14/detail-iimxxstf1268996.d.html?vt=4" }, { "id": "O9G4-B61-B", "title": "科技为矛周期为盾：选出跑得又快又远的公司", "date": "2020-10-15", "url": "https://www.chnfund.com/article/AR2020101510331068621443", "authors": "中国基金报记者张燕北采访；冯明远具名陈述", "locator": "寻找好公司一节；五因素、产业链研究和分散", "summary": "冯明远把企业、行业、财务及动态估值结合，解释技术迭代下分散持股的用途；李淑彦周期评论不归于冯。", "access": "2026-10-09读取；公开原始材料正文已读；实际阅读范围：寻找好公司一节；五因素、产业链研究和分散", "retrieval_level": "substantive primary material read" }, { "id": "O9G4-B62-A", "title": "广发双擎升级混合2025年第3季度报告", "date": "2025-10-28", "url": "https://www.gffunds.com.cn/jjgg/dqgg/202510/P020251028324901714324.pdf", "authors": "广发基金管理人；刘格菘为报告期经理，非个人独著", "locator": "PDF第6页经理表；第7–8页§4.4", "summary": "管理人用壁垒、可落地需求和供给约束筛选，估值上升后更严格检查供需。", "access": "2026-10-09读取；官方PDF文字及第7–8页截图已读；实际阅读范围：PDF第6页经理表；第7–8页§4.4", "retrieval_level": "substantive primary material read" }, { "id": "O9G4-B62-B", "title": "广发双擎升级混合2025年年度报告", "date": "2026-03-30", "url": "https://www.9fzt.com/detail/fund_005911_2_10708415.html", "authors": "广发基金管理人；刘格菘为报告期经理，非个人独著", "locator": "原报告第11–12页经理表、第14页§4.4和§4.5", "summary": "管理人将AI需求与受限供给、价格弹性及企业利润相接，商业闭环是当时预期而非已实现事实。", "access": "2026-10-09读取；九方智投公开原报告全文镜像；相关段落直接阅读；官网PDF检索命中但打开失败，未做字节校验；实际阅读范围：原报告第11–12页经理表、第14页§4.4和§4.5", "retrieval_level": "substantive primary material read", "original_url": "https://www.gffunds.com.cn/jjgg/dqgg/202603/P020260330313961874534.pdf" }, { "id": "O9G4-B64-A", "title": "金鹰科技创新股票2024年年度报告", "date": "2025-03-31", "url": "https://pdf.dfcfw.com/pdf/H2_AN202503311649465138_1.pdf", "authors": "金鹰基金管理人；陈颖为基金经理，张展华、何欢列经理助理", "locator": "印刷第13页经理表、第15–16页§4.4和§4.5", "summary": "管理人提出终端AI渗透向企业应用传导的假设，并在成长性和估值之间筛选。", "access": "2026-10-09读取；东方财富公开原报告PDF镜像文本已读；截图超时，未与官网副本逐字节校验；实际阅读范围：印刷第13页经理表、第15–16页§4.4和§4.5", "retrieval_level": "substantive primary material read" }, { "id": "O9G4-B64-B", "title": "金鹰科技创新股票2025年年度报告", "date": "2026-03-31", "url": "https://pdf.dfcfw.com/pdf/H2_AN202603311820899764_1.pdf", "authors": "金鹰基金管理人；陈颖为报告期经理，非个人独著", "locator": "印刷第13页经理表、第16–17页§4.4和§4.5", "summary": "管理人把模型、产业数据与应用场景结合视为机会，同时保留成长和低估值两个筛选方向。", "access": "2026-10-09读取；东方财富公开原报告PDF镜像文本已读；截图超时，未与官网副本逐字节校验；实际阅读范围：印刷第13页经理表、第16–17页§4.4和§4.5", "retrieval_level": "substantive primary material read" }, { "id": "O9G4-B65-A", "title": "为什么科技投资进入了基本面投资时代", "date": "2021-12-28", "url": "https://edu.chinaamc.com/c/2021-12-28/513383.shtml", "authors": "朱昂采访周克平；华夏基金投教网站刊载", "locator": "投资框架问答：四类资产、现金流、DCF、组合管理", "summary": "周克平按现金流和企业演化区分项目、产品、平台和生态，并强调前瞻自由现金流。", "access": "2026-10-09读取；官方刊载访谈正文已读；日期取URL刊载路径，实际访谈发生日未核；实际阅读范围：投资框架问答，四类资产、现金流、DCF、组合管理", "retrieval_level": "substantive primary material read", "date_basis": "官网URL刊载路径2021-12-28；本次提取未见独立日期栏，实际访谈日未核" }, { "id": "O9G4-B65-B", "title": "九大头部基金经理最新发声", "date": "2022-06-11", "url": "https://www.chnfund.com/article/AR2022061116242760113822", "authors": "中国基金报记者采访；仅归属标记周克平的回答", "locator": "周克平回答科创板结构因素、制度流动性、盈利估值三问", "summary": "周克平区分周期扰动和结构增长，并用前置投入、后置利润解释估值判断；其他八位回答排除。", "access": "2026-10-09读取；公开原始材料正文已读；实际阅读范围：周克平回答科创板结构因素、制度流动性、盈利估值三问", "retrieval_level": "substantive primary material read" }, { "id": "O9G4-B67-A", "title": "财通多策略福鑫2024年年度报告", "date": "2025-03-29", "url": "https://static.cninfo.com.cn/finalpage/2025-03-29/1222955470.PDF", "authors": "财通基金管理人；金梓才为报告期经理，非个人独著", "locator": "第10页经理表、第12–13页§4.4和§4.5", "summary": "管理人把海外与国内算力投资节奏作比较，同时以租金优势解释线下折扣业态机会。", "access": "2026-10-09读取；巨潮公开披露PDF文本直接阅读；截图超时；实际阅读范围：第10页经理表、第12–13页§4.4和§4.5", "retrieval_level": "substantive primary material read" }, { "id": "O9G4-B67-B", "title": "财通多策略福鑫2025年年度报告", "date": "2026-03-31", "url": "https://www.hongdianfund.com/FundAnnouncement/Detail/501046/0000000000001pk5ia", "authors": "财通基金管理人；金梓才为报告期经理，非个人独著", "locator": "§4.1.2经理表，§4.4.1与§4.5（原报告第10、12–13页）", "summary": "管理人将量价变化拆成产品材料升级与供不应求，讨论PCB产业链供给和竞争约束。", "access": "2026-10-09读取；虹点基金销售平台公开原报告全文；网页HTML直接取得并读取相关章节；交易所原件打开失败，未做字节校验；实际阅读范围：§4.1.2经理表，§4.4.1与§4.5（原报告第10、12–13页）", "retrieval_level": "substantive primary material read", "original_url": "https://www.sse.com.cn/disclosure/fund/announcement/c/new/2026-03-31/501046_20260331_MN9Z.pdf" }, { "id": "O9G5-B68A", "title": "汇丰晋信低碳先锋2022年年度报告", "date": "2023-03-17", "url": "https://static.cninfo.com.cn/finalpage/2023-03-17/1216142063.PDF", "summary": "报告复盘新能源估值切换未兑现及错过调整窗口，区分终端增长和产能过剩导致的增量不增利。", "locator": "§4.1.2经理表；§4.4.1及§4.5，印刷页14–18", "access": "2026-10-09读取；巨潮公开原始报告PDF正文", "authors": "汇丰晋信基金管理人；陆彬为报告期基金经理", "retrieval_level": "substantive primary material read" }, { "id": "O9G5-B68B", "title": "汇丰晋信低碳先锋2025年第2季度报告", "date": "2025-07-21", "url": "https://pdf.dfcfw.com/pdf/H2_AN202507211712818307_1.pdf", "summary": "报告区分锂电竞争格局、光伏控产价格及风电需求，使用产能利用率与盈利优势判断出清。", "locator": "§4.1，页6–7；§4.4，页8", "access": "2026-10-09读取；东方财富托管的管理人原始PDF；非基金官网，未逐字与官网版本比对", "authors": "汇丰晋信基金管理人；陆彬为报告期基金经理", "retrieval_level": "substantive primary material read" }, { "id": "O9G5-B69A", "title": "长城久嘉创新成长2024年年度报告", "date": "2025-03-29", "url": "https://pdf.dfcfw.com/pdf/H2_AN202503291648950276_1.pdf", "summary": "报告说明从国产算力到AI终端、智能驾驶及应用布局的变化，将行业趋势与预期落地分开。", "locator": "§4.1.2及§4.4.1，页11–13", "access": "2026-10-09读取；东方财富托管的管理人原始PDF；非基金官网，未逐字比对官网", "authors": "长城基金管理人；尤国梁为报告期基金经理", "retrieval_level": "substantive primary material read" }, { "id": "O9G5-B69B", "title": "长城久嘉创新成长2025年中期报告", "date": "2025-08-29", "url": "https://pdf.dfcfw.com/pdf/H2_AN202508291736331254_1.pdf", "summary": "报告解释开源模型推动应用布局，并将关税敏感性、军工业绩与事件催化共同用于配置判断。", "locator": "§4.1.2页10；§4.4.1页11", "access": "2026-10-09读取；东方财富托管的管理人原始PDF；非基金官网，未逐字比对官网", "authors": "长城基金管理人；尤国梁为报告期基金经理", "retrieval_level": "substantive primary material read" }, { "id": "O9G5-B73A", "title": "兴全合润2025年第1季度报告", "date": "2025-04-22", "url": "https://www.xqfunds.com/data/20250421174551herun.pdf", "summary": "报告将技术变革的非线性增长、细分景气反转与风险收益条件结合，讨论AI终端及资本开支传导。", "locator": "§4.1页4–5；§4.4页6", "access": "2026-10-09读取；基金官网原始PDF正文；日期取封面送出日期而非URL", "authors": "兴证全球基金管理人；谢治宇为报告期基金经理", "retrieval_level": "substantive primary material read" }, { "id": "O9G5-B73B", "title": "兴全合润2025年第3季度报告", "date": "2025-10-28", "url": "https://www.xqfunds.com/data/20251027155643herun.pdf", "summary": "报告对AI资本开支热潮保留谨慎，要求检验终端渗透、模型效率与硬件迭代对长期价值的影响。", "locator": "§4.1页5–6；§4.4页7", "access": "2026-10-09读取；基金官网原始PDF正文；日期取封面送出日期而非URL", "authors": "兴证全球基金管理人；谢治宇为报告期基金经理", "retrieval_level": "substantive primary material read" }, { "id": "O9G5-B74A", "title": "嘉实基金归凯：成长需要想象力", "date": "2020-07-07", "url": "https://www.jsfund.cn/main/a/20200707/431636.shtml", "summary": "归凯将商业模式、竞争优势及产业趋势拆开评估，并提醒成长行业仍有库存周期和回报率约束。", "locator": "三个维度精选好公司；深度研究创造持续回报", "access": "2026-10-09读取；嘉实官网转载采访正文；仅归属直接引述，排除宣传绩效描述", "authors": "归凯采访；嘉实官网发布", "retrieval_level": "substantive primary material read" }, { "id": "O9G5-B74B", "title": "嘉实泰和2025年第2季度报告", "date": "2025-07-21", "url": "https://static.cninfo.com.cn/finalpage/2025-07-21/1224231330.PDF", "summary": "管理人区分宏观修复与企业成本价格压力，并披露增持芯片、消费电子和面板、减持医疗器械的操作。", "locator": "§4.1页4–5；§4.4页5–6", "access": "2026-10-09读取；巨潮公开管理人原始PDF正文；实质运作评论，不只产品投资范围", "authors": "嘉实基金管理人；归凯为报告期基金经理", "retrieval_level": "substantive primary material read" }, { "id": "O9G5-B83A", "title": "Common Stocks and Uncommon Profits：Clues from the Past 授权节选", "date": "1996", "url": "https://www.abebooks.co.uk/9780471119289/Common-Stocks-Uncommon-Profits-Writings-0471119288/plp", "summary": "Fisher强调研发需转化为商业利润，失败项目和长时间滞后令当期研发金额难以直接衡量价值。", "locator": "Excerpt / Chapter One / Clues from the Past，研发实验室及商业化段落", "access": "2026-10-09读取；AbeBooks呈现出版社授权原著节选，页面明确Reprinted by permission；未读全书；不把Ken Fisher序言归于Philip", "authors": "Philip A. Fisher；Wiley授权节选", "retrieval_level": "substantive primary material read", "date_basis": "1996为所读版本版权/出版年；原著1958年，不是新发表观点" }, { "id": "O9G5-B83B", "title": "Paths to Wealth Through Common Stocks：Chapter One 授权节选", "date": "2007", "url": "https://www.amazon.com/Paths-Wealth-Through-Common-Stocks/dp/0470139498", "summary": "Fisher将管理能力及商业研发进步与市场缓慢重估联系起来，要求区分价值变化和价格已计入程度。", "locator": "Editorial Reviews / Excerpt / Chapter One，Adjusting to Key Influences of the 1960s开篇", "access": "2026-10-09读取；Amazon展示Wiley授权原著节选，明确Reprinted by permission；不是读全书或读评论；Wiley直接样章PDF403", "authors": "Philip A. Fisher；Wiley授权节选", "retrieval_level": "substantive primary material read", "date_basis": "2007为所读再版版权年；原著1960年，不是当时新观点" }, { "id": "O9G5-B84A", "title": "PBS FRONTLINE：Interview with Peter Lynch", "date": "1997-01-14", "url": "https://www.pbs.org/wgbh/pages/frontline/shows/betting/pros/lynch.html", "summary": "Lynch强调公司研究和保留基本面改善的赢家，容许部分判断错误；市场下跌不能替代业务分析。", "locator": "Was that your secret?；Magellan早期投资与Chrysler问答", "access": "2026-10-09读取；重读2026-10-08已保存PBS完整采访文本；10月9日原网址403，以公开RBCPA转载交叉核对；不是本日重新抓取官网成功", "authors": "Peter Lynch；PBS采访", "retrieval_level": "substantive primary material read", "date_basis": "关联节目播出日，PBS tapes页已核；采访录制日和网页上传日未标明", "mirror_url": "https://www.rbcpa.com/commentary-archive/pbs-interview-with-peter-lynch/", "date_source": "https://www.pbs.org/wgbh/pages/frontline/shows/betting/etal/tapes.html" }, { "id": "O9G5-B84B", "title": "Peter Lynch National Press Club 演讲主体转录", "date": "1994-10-07", "url": "https://brewbooks.blog/2021/02/22/transcript-of-peter-lynch-8-october-1994-lecture-to-the-national-press-club/", "summary": "演讲将熟悉业务与资产负债表研究结合，以行业库存、占用率等可观察指标替代宏观择时，反对仅因跌价而买入。", "locator": "转录时间戳11:20–16:52、23:17–28:48", "access": "2026-10-09读取；公开第三方个人转录，已读演讲主体；不含后续Q&A，未逐秒核音频；原题10/8与NPC官方回顾10/7冲突，采用主办方日期", "authors": "Peter Lynch演讲；John Brew转录", "retrieval_level": "substantive primary material read", "date_basis": "演讲日按NPC官方2013-12-06历史回顾；转录发布2021-02-22", "date_source": "https://www.press.org/newsroom/past-npc-president-klein-remembers-nelson-mandelas-visit-club" }, { "id": "O9G6-ASML416", "title": "ASML Q4 2016 earnings call", "date": "2017-01-18", "url": "https://www.roic.ai/quote/ASML:UW/transcripts/2016-year/4-quarter", "authors": "Issuer management and named analyst questioners", "access": "公开原始访谈/电话会转录镜像；已读具名实质段落，未听原录音或与官方文件逐字节核对；可能存在转录错误。", "retrieval_level": "primary-source substantive passages via public mirror", "summary": "Meunier追问收购价格分摊对毛利率的影响；管理层明确不能把该调整直接称现金毛利率。", "locator": "Francois Meunier Q&A; lines 605–636" }, { "id": "O9G6-ASML117", "title": "ASML Q1 2017 earnings call", "date": "2017-04-19", "url": "https://www.roic.ai/quote/ASML:LI/transcripts/2017-year/1-quarter", "authors": "Issuer management and named analyst questioners", "access": "公开原始访谈/电话会转录镜像；已读具名实质段落，未听原录音或与官方文件逐字节核对；可能存在转录错误。", "retrieval_level": "primary-source substantive passages via public mirror", "summary": "Meunier拆分毛利率同比提升来自服务/升级产品组合还是工具自身利润率。", "locator": "Francois Meunier Q&A; lines 888–907" }, { "id": "O9G6-AXT124", "title": "AXT Q1 2024 earnings call", "date": "2024-05-02", "url": "https://www.roic.ai/quote/AXTI/transcripts/2024-year/1-quarter", "authors": "Issuer management and named analyst questioners", "access": "公开原始访谈/电话会转录镜像；已读具名实质段落，未听原录音或与官方文件逐字节核对；可能存在转录错误。", "retrieval_level": "primary-source substantive passages via public mirror", "summary": "Bryson纠正管理层把传统磷化铟去库存问题答成AI需求，追问库存压制收入的量和恢复时点。", "locator": "Matthew Bryson Q&A; lines 652–751" }, { "id": "O9G6-DBD226", "title": "Diebold Nixdorf Q2 2026 earnings call", "date": "2026-08-07", "url": "https://www.fool.com/earnings/call-transcripts/2026/08/07/diebold-nixdorf-dbd-q2-2026-earnings-call-transcript/", "authors": "Issuer management and named analyst questioners", "access": "公开原始访谈/电话会转录镜像；已读具名实质段落，未听原录音或与官方文件逐字节核对；可能存在转录错误。", "retrieval_level": "primary-source substantive passages via public mirror", "summary": "Bryson追问银行设备延期具体原因，再问高毛利来自季度特殊项目还是产品组合改善。", "locator": "Matthew Bryson Q&A; lines 311–320" }, { "id": "O9G6-ISIL16", "title": "Renesas acquisition of Intersil conference call", "date": "2016-09-13", "url": "https://www.sec.gov/Archives/edgar/data/1096325/000119312516708642/d257613ddefa14a.htm", "authors": "Issuer management and named analyst questioners", "access": "已读发行人或监管机构托管具名原始正文；未听电话会录音。", "retrieval_level": "primary selected full-text passages", "summary": "Acree挑战收购方此前市场份额下滑与转型能力，追问跨地域管理团队保留。", "locator": "Cody Acree Q&A, transcript pp.4–5" }, { "id": "O9G6-SMTC424", "title": "Semtech Q4 FY2024 earnings call", "date": "2024-03-28", "url": "https://investors.semtech.com/media/document/32bb4187-40aa-4071-906a-bba6e05f78b4/assets/SMTC-USQ_Transcript_2024-03-28.pdf?disposition=inline", "authors": "Issuer management and named analyst questioners", "access": "已读发行人或监管机构托管具名原始正文；未听电话会录音。", "retrieval_level": "primary selected full-text passages", "summary": "Acree追问铜与光的架构替代及Semtech是否因某一方案获益，管理层回答需看具体应用和组合。", "locator": "Cody Acree Q&A, p.8" }, { "id": "O9G6-KLGS25", "title": "Seth Klarman — Goldman Sachs Talks", "date": "2025-05-27", "url": "https://www.bishoprockcap.com/assets/library/seth-klarman/July%202025%20Seth%20Klarman%20on%20finding%20value%20and%20maintaining%20discipline%20(transcript)", "authors": "Seth Klarman and named interviewer", "access": "公开原始访谈/电话会转录镜像；已读具名实质段落，未听原录音或与官方文件逐字节核对；可能存在转录错误。 此镜像标有自动转录置信度，未验证每个词；仅作有限样本。", "retrieval_level": "primary-source substantive passages via public mirror", "summary": "Klarman将折价与价值兑现催化剂并列，要求预先设想投资失败的原因。", "locator": "02:13–04:45", "date_basis": "录制日，Goldman Sachs官方节目页核实；发布日2025-07-23", "official_url": "https://www.goldmansachs.com/insights/talks-at-gs/seth-klarman" }, { "id": "O9G6-KLCO25", "title": "Seth Klarman — Value Investing with Legends", "date": "2025-08-01", "url": "https://www.bishoprockcap.com/assets/library/seth-klarman/August%202025%20Seth%20Klarman%20-%20Contrarian%20Investing,%20Discipline,%20and%20Building%20Baupost%20(transcript)", "authors": "Seth Klarman and named interviewer", "access": "公开原始访谈/电话会转录镜像；已读具名实质段落，未听原录音或与官方文件逐字节核对；可能存在转录错误。 此镜像标有自动转录置信度，未验证每个词；仅作有限样本。", "retrieval_level": "primary-source substantive passages via public mirror", "summary": "Klarman回忆拆分交易：先扣清算现金，再计算所获分拆权益的隐含成本；这是本人回忆而非审计后的历史回报。", "locator": "04:22–06:47", "date_basis": "节目发布日；Apple Podcasts节目发行记录核实", "official_url": "https://podcasts.apple.com/za/podcast/seth-klarman-contrarian-investing-discipline-and/id1459993408?i=1000720195700" }, { "id": "O9G6-TEP10", "title": "CNBC Squawk Box — David Tepper excerpts", "date": "2010-09-24", "url": "https://www.bishoprockcap.com/assets/library/david-tepper/2010,%20September%20CNBC%20Transcript%20CNBC's%20'Squawk%20Box'%20Sits%20Down%20with%20Appaloosa%20Management%20President%20&%20Founder%20David%20Tepper.pdf", "authors": "David Tepper / CNBC", "access": "公开原始访谈/电话会转录镜像；已读具名实质段落，未听原录音或与官方文件逐字节核对；可能存在转录错误。 2010为节目方选摘，2018为unofficial transcript；不冒充原站全文字节核验。", "retrieval_level": "primary-source substantive passages via public mirror", "summary": "Tepper分两条宏观路径：自主复苏或经济不振引发QE；同时说明长期持续依赖QE并非好事。", "locator": "pp.1,3；CNBC发布的unofficial transcript excerpts之公开镜像" }, { "id": "O9G6-TEP18", "title": "CNBC interview — David Tepper with Scott Wapner", "date": "2018-09-13", "url": "https://www.bishoprockcap.com/assets/library/david-tepper/2018,%20September%20CNBC%20Exclusive%20CNBC%20Transcript%20Appaloosa%20Management's%20David%20Tepper%20Speaks%20with%20CNBC's%20Scott%20Wapner%20(live.pdf", "authors": "David Tepper / CNBC", "access": "公开原始访谈/电话会转录镜像；已读具名实质段落，未听原录音或与官方文件逐字节核对；可能存在转录错误。 2010为节目方选摘，2018为unofficial transcript；不冒充原站全文字节核验。 已定位CNBC原页面地址但原站检索访问受限，本次实读使用公开镜像。", "retrieval_level": "primary-source substantive passages via public mirror", "summary": "Tepper把关税、汇率、美元盈利与估值联动；说明用个股多头和市场空头表达约25%的S&P等效敞口。", "locator": "pp.4–6；CNBC unofficial transcript公开镜像", "original_url": "https://www.cnbc.com/2018/09/13/cnbc-exclusive-cnbc-transcript-appaloosa-managements-david-tepper-speaks-with-cnbcs-scott-wapner-today.html" }, { "id": "O9G6-HOHN22", "title": "TCI letter to Alphabet CEO", "date": "2022-11-15", "url": "https://www.tcifund.com/files/corporateengageement/alphabet/15th%20November%202022.pdf", "authors": "Christopher Hohn, TCI", "access": "已视觉读取官方4页原信，核对日期、抬头及签名。", "retrieval_level": "primary PDF visual full-text read", "summary": "Hohn要求约束成本及Other Bets损失，设置利润率目标，以现金流支持回购；信中估值与预测仅为当时主张。", "locator": "pp.1–4", "audit": "root agent directly inspected all PDF pages and signature/date; group author integrated per-page verification." }, { "id": "O9G6-HOHN23", "title": "TCI follow-up letter to Alphabet CEO", "date": "2023-01-20", "url": "https://www.tcifund.com/files/corporateengageement/alphabet/20th%20January%202023.pdf", "authors": "Christopher Hohn, TCI", "access": "已视觉读取官方1页原信，核对日期、抬头及签名。", "retrieval_level": "primary PDF visual full-text read", "summary": "Hohn认可12000减员但认为不足，建议约150000目标人数并降低薪酬/SBC；不是实际已实现的削减结果。", "locator": "p.1", "audit": "root agent directly inspected all PDF pages and signature/date; group author integrated per-page verification." }, { "id": "FM-LA-ESSAY24", "title": "Situational Awareness: IIIa. Racing to the Trillion-Dollar Cluster", "date": "2024-06", "url": "https://situational-awareness.ai/racing-to-the-trillion-dollar-cluster/", "authors": "Leopold Aschenbrenner", "access": "公开作者原文；已读 Training compute、Power requirements、Overall compute 与 AI revenue 段。", "retrieval_level": "primary HTML substantive sections read", "summary": "2024年将训练算力趋势转为整座集群的资本与电力需求，区别一次训练租金和建设成本。", "locator": "Training compute；Power requirements；Overall compute；AI revenue", "audit": "2026-10-09读取；这是2024年前瞻情景，未核验其后预测实现率。" }, { "id": "FM-LA-DW24", "title": "Dwarkesh Podcast: Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history", "date": "2024-06-04", "url": "https://www.dwarkesh.com/p/leopold-aschenbrenner", "authors": "Leopold Aschenbrenner（受访者）；Dwarkesh Patel（主持人）", "access": "原始节目公开文字稿；已读00:01–00:20、04:07:58–04:10:58（含主题收入敞口论述）及04:25:58反思段。", "retrieval_level": "primary interview transcript substantive sections read", "summary": "访谈将AI产业方向与投资时序分开：主题敞口须足够重要，还要承受收入兑现延迟；并主张随证据更新。", "locator": "00:01:20–00:07:11；04:07:58、04:10:58；04:25:58", "audit": "与同日论文是两份不同形式的原始材料，但围绕同一论点，不能视为独立事实验证或跨时期稳定性证明。" }, { "id": "FM-LA-IDENTITY", "title": "Situational Awareness LP — About", "date": "undated; retrieved 2026-10-09", "url": "https://situationalawarenesslp.com/about", "authors": "Situational Awareness LP", "access": "基金管理机构官方简介全文公开。", "retrieval_level": "primary corporate identity page read", "summary": "官方确认Leopold Aschenbrenner于2024年创办Situational Awareness；本页仅用于身份核验。", "locator": "About，第1–2段", "audit": "无发布日期；检索日不是任职起始日。无AUM、回报或当前持仓断言。" }, { "id": "FM-RJ-EURO", "title": "Serendipity brings power of the Rolodex to early-stage investment", "date": "2022（检索索引年份；页面未显示具体日期）", "url": "https://www.euromoney.com/article/29xkk5q0z0q4naqqgo3k0/capital-markets/serendipity-brings-power-of-the-rolodex-to-early-stage-investment/", "authors": "Euromoney；Rob Jesudason（受访者）", "access": "原始采访公开全文已读；以本人直接发言为方法依据。", "retrieval_level": "primary interview substantive sections read", "summary": "用永久资本减少赎回对持有期限的干扰，并以行业经验筛查真实客户需求。", "locator": "永久资本结构段；Solving problems；Commitment", "audit": "2026-10-09读取；正文讨论2021年结果，索引约2022年发表，未核验精确日期。历史投资范围不当作当前完整策略；不引用业绩数字。" }, { "id": "FM-RJ-POD26", "title": "Rob Jesudason (Serendipity Capital): Why quantum investors must triangulate across rival modalities", "date": "2026-08-17", "url": "https://podcast.yboger.com/2026/08/17/rob-jesudason-ceo-and-founder-of-serendipity-capital/", "authors": "Rob Jesudason（受访者）；Yuval Boger（主持人）", "access": "原始节目公开完整文字稿已读；节目注明末尾数分钟因录音问题缺失。", "retrieval_level": "primary interview transcript substantive sections read", "summary": "跨技术路线交叉求证，区分物理原理可行和工程批量交付；强调资本期限与技术成熟度匹配。", "locator": "Transcript：永久资本；retail investor评估问题；managerial execution and engineering scale", "audit": "主持人为QuEra首席商务官，受访者为相关领域投资者，存在商业利益关系；主持人提问及研究者反例不归为Rob观点。技术、持仓与规模自述未独立审计。" }, { "id": "FM-RJ-IDENTITY", "title": "Serendipity Capital — Team", "date": "undated; retrieved 2026-10-09", "url": "https://www.serendipitycapital.com/team", "authors": "Serendipity Capital", "access": "机构官方团队页全文公开。", "retrieval_level": "primary corporate identity page read", "summary": "确认Rob Jesudason为联合创始人、Partner、CEO & CIO；仅用于身份核验。", "locator": "FOUNDERS — Rob Jesudason", "audit": "检索日不是任职起始日；不与其他同名基金合并。" } ], "profiles": [ { "id": "wood-ark", "name": "Cathie Wood / ARK", "role": "买方基金管理；此处主要归纳ARK团队公开研究", "scope": "技术创新 → 成本/需求 → 公司价值捕获 → 组合与估值", "focus": [ "成本曲线与弹性", "技术融合", "公司与股东", "信念与证伪" ], "boundary": "不是Cathie Wood私有思想的完整复刻。机构流程无日期，未满足个人跨两份 dated 材料的稳定方法认证。", "timeline": [ { "date": "undated", "text": "公开流程连接主题筛选、公司模型与组合监测。来源日期未知，不能定位方法形成时间。", "source_ids": [ "D01" ] }, { "date": "2026-09-10", "text": "公开摘要延伸到投资周期及组合情景；这是材料范围变化，不证明个人观点转向。", "source_ids": [ "D02" ] }, { "date": "undated", "text": "机构成本曲线与需求弹性框架；五年模型纳入销量、成本、稀释和估值。", "source_ids": [ "D01" ] }, { "date": "2026-09-10", "text": "2026摘要把AI成本下降与token增长联系；800/135与“近3倍”表述不一致，应重算而非沿用。", "source_ids": [ "D02" ] } ], "gaps": [ "完整报告和底层模型未全审阅", "个人与团队贡献不能混同", "没有基金业绩/预测命中率审计", "两份机构材料；无日期流程页不构成两次具名个人研究，保留既有wood-ark。" ], "comparison": "更侧重从技术机制到机会与估值；这是本次样本的编辑比较，不是永久人格。", "side": "buy" }, { "id": "meeker-bond", "name": "Mary Meeker / BOND", "role": "历史卖方研究 → 创投/成长投资；保留BOND共著归属", "scope": "扩散数据 → 使用/商业化 → 平台/专业化 → 资本与价值捕获", "focus": [ "采用曲线与区域类比", "平台与专业化", "增长与利润张力", "数据口径审查" ], "boundary": "两份 dated 团队材料可做纵向比较，不足以覆盖全部职业思想；未将团队观点全部归为本人。", "timeline": [ { "date": "2009-12-16", "text": "移动互联网发布说明强调技术组合、区域经验与扩散。未审阅完整旧研报。", "source_ids": [ "D03" ] }, { "date": "2025-05-30", "text": "AI共著报告样本增加成本、资本回收及竞争视角。不是证明旧观点被放弃。", "source_ids": [ "D04" ] } ], "gaps": [ "2009完整研报未审阅", "2025仅核查指定文本与部分脚注", "没有逐预测结果或投资业绩审计" ], "comparison": "更侧重跨时段趋势证据与商业化路径；只代表选定材料。", "side": "buy" }, { "id": "marks-oaktree", "name": "Howard Marks / Oaktree", "side": "buy", "role": "买方信用投资与资产管理；不是科技行业覆盖分析师", "scope": "经营价值 → 支付价格 → 市场心理 → 下行与风险预算", "focus": [ "价格与价值", "周期心理", "风险不对称", "安全边际" ], "boundary": "公开材料样本与编辑模型，不是完整个人思想。未认证所有方法为反复模式；没有研报全文、完整仓位或预测命中率审计。", "timeline": [ { "date": "2025-01-07", "text": "讨论泡沫心理与购买价格，不作精确顶部预测。", "source_ids": [ "D05" ] }, { "date": "2025-08-13", "text": "系统讨论盈利能力、价值与价格，不能推定为AI观点反转。", "source_ids": [ "D06" ] } ], "gaps": [ "科技公司经营预测不属于此处材料的强项", "估值范围需要公司级现金流补充" ], "comparison": "比较研究问题与必要条件，不按名气或股价表现排名。", "assessment": { "strength": "把好技术与好投资分开，避免支付任何价格。", "weakness": "心理诊断不提供崩盘日期，也不能替代收入/现金模型。", "next_evidence": "补充独立有日期的研报/访谈或具名提问；冻结原判断和反例后再评估结果。" } }, { "id": "druckenmiller-duquesne", "name": "Stanley Druckenmiller / Duquesne", "side": "buy", "role": "买方宏观投资者；当前节目身份为Duquesne Family Office负责人", "scope": "变化信号 → 分阶段验证 → 仓位与盈亏 → 信号更新", "focus": [ "非线性变化", "分阶段信念", "仓位规模", "信号拥挤失效" ], "boundary": "已有2026年官方逐字稿与2021年主办方实质精选引语支持有限的跨材料方法比较；2021年全文视频未读取，不认证完整方法、当前仓位或预测准确率。", "timeline": [ { "date": "2021-02-03", "text": "主办方精选本人引语讨论跨资产寻找机会、流动性带来的纠错能力及集中仓位。仅主办方精选引语，非完整访谈。", "source_ids": [ "R2B3-B91A" ] }, { "date": "2026-02-27", "text": "同一访谈回顾AI加仓、出售遗憾及旧信号失效；回忆不是多份历史证据。", "source_ids": [ "D07" ] } ], "gaps": [ "2021年仅阅读主办方精选引语，尚缺完整逐字稿", "未读取完整组合或验证回忆中的交易时间", "跨材料比较仍不能认证永久稳定模式或投资业绩" ], "comparison": "比较研究问题与必要条件，不按名气或股价表现排名。", "assessment": { "strength": "把资产选择、集中度和可快速纠错的流动性共同考虑。", "weakness": "主办方摘录存在选择性；历史回忆和高集中度不能证明可复制。", "next_evidence": "取得2021年完整逐字稿，并对真实建仓、减仓时间和流动性压力独立核查。" } }, { "id": "loeb-thirdpoint", "name": "Dan Loeb / Third Point", "side": "buy", "role": "买方事件驱动基金经理；保留基金团队分析归属", "scope": "技术瓶颈 → 国际价值捕获 → 催化剂/估值 → 融资与敞口", "focus": [ "HBM差异化", "控股折价", "催化剂", "资本与融资" ], "boundary": "公开材料样本与编辑模型，不是完整个人思想。未认证所有方法为反复模式；没有研报全文、完整仓位或预测命中率审计。", "timeline": [ { "date": "2025-04-30", "text": "政策不确定下报告降低敞口并偏向催化剂。", "source_ids": [ "D08" ] }, { "date": "2025-10-31", "text": "讨论国际AI机会与融资风险；不是断言所有减仓已完全逆转。", "source_ids": [ "D09" ] } ], "gaps": [ "投资信具有持仓与营销利益", "未验证HBM预测与未来折价收敛", "没有完整净敞口/对冲账本" ], "comparison": "比较研究问题与必要条件，不按名气或股价表现排名。", "assessment": { "strength": "连接产品机制、估值与融资，允许股票和信用有不同结论。", "weakness": "通过周期利润与控股折价仍可能掩盖治理和资本风险。", "next_evidence": "补充独立有日期的研报/访谈或具名提问；冻结原判断和反例后再评估结果。" } }, { "id": "moore-ms", "name": "Joseph Moore / Morgan Stanley", "side": "sell", "role": "卖方半导体分析师；机构按2025–2026具名材料记录", "scope": "市场可达性 → 可交付需求 → 合同保障 → 现金/收入", "focus": [ "政策可达市场", "需求存量与流量", "合同押金", "可实现收入", "供需差与可交付收入" ], "boundary": "公开材料的方法样本；新增两期具名Q&A，不把管理层回答当作分析师立场，未认证稳定反复模式。", "timeline": [ { "date": "2025-08-27", "text": "追问中国机会及许可必要条件。", "source_ids": [ "D10" ] }, { "date": "2025-11-19", "text": "追问累计平台需求与已出货金额的桥。", "source_ids": [ "D11" ] }, { "date": "2026-05-20", "text": "Q1追问分部竞争差异", "source_ids": [ "E7S01" ] }, { "date": "2026-06-24", "text": "追问押金与RPO、预付款的区别。", "source_ids": [ "D12" ] }, { "date": "2026-08-26", "text": "Q2追问需求与供给增长差额。", "source_ids": [ "S3D02" ] } ], "gaps": [ "没有读取许可卖方研报全文", "本样本三种问题不能证明同一种稳定方法", "镜像未音频验证" ], "comparison": "比较研究问题与必要条件，不按名气或股价表现排名。", "assessment": { "strength": "把宏大叙事拆成市场准入、交付与会计条件。", "weakness": "提问揭示不确定性，不证明他本人持有某个价格目标模型。", "next_evidence": "补充独立有日期的研报/访谈或具名提问；冻结原判断和反例后再评估结果。" }, "current_role_as_of": "2026-10-07", "role_evidence": "https://ir.amd.com/stock-data/analyst-coverage", "current_role": "AMD covering analyst, Morgan Stanley (issuer list accessed 2026-10-07; undated list)" }, { "id": "arcuri-ubs", "name": "Timothy Arcuri / UBS", "side": "sell", "role": "卖方半导体分析师；按具名材料时点保留UBS身份", "scope": "行业增长 → 公司兑现 → 最短板 → 合同分期", "focus": [ "TAM到公司", "多重瓶颈", "合同地板", "分期模型", "专用产品与开放模型的价值归属" ], "boundary": "公开材料的方法样本；新增两期具名Q&A，不把管理层回答当作分析师立场，未认证稳定反复模式。", "timeline": [ { "date": "2025-08-27", "text": "追问行业增长与公司收入增长的关系。", "source_ids": [ "D10" ] }, { "date": "2025-11-19", "text": "追问电力、融资、内存与晶圆瓶颈。", "source_ids": [ "D11" ] }, { "date": "2026-05-20", "text": "Q1问专用推理产品", "source_ids": [ "E7S01" ] }, { "date": "2026-06-24", "text": "追问合同地板及RPO分期，不简单年均化。", "source_ids": [ "D12" ] }, { "date": "2026-08-26", "text": "Q2问开放模型与前沿公司的关系。", "source_ids": [ "S3D02" ] } ], "gaps": [ "完整研究模型与客户调研不可得", "不将CAGR与合同最低额当作公司指引" ], "comparison": "比较研究问题与必要条件，不按名气或股价表现排名。", "assessment": { "strength": "从可兑现约束检验增长，要求季度兑现路径。", "weakness": "供应上限不等于确定销量，客户需求和竞争仍须补齐。", "next_evidence": "补充独立有日期的研报/访谈或具名提问；冻结原判断和反例后再评估结果。" }, "current_role_as_of": "2026-10-07", "role_evidence": "https://ir.amd.com/stock-data/analyst-coverage", "current_role": "AMD covering analyst, UBS (issuer list accessed 2026-10-07; undated list)" }, { "id": "muse-cantor", "name": "C.J. Muse / Cantor Fitzgerald", "side": "sell", "role": "卖方半导体分析师；按2025–2026材料身份记录", "scope": "晶圆到机架 → 通电/融资 → 需求消化 → 流动性", "focus": [ "部署时滞", "供需闭合", "融资与ROI", "可返还资金" ], "boundary": "公开材料样本与编辑模型，不是完整个人思想。未认证所有方法为反复模式；没有研报全文、完整仓位或预测命中率审计。", "timeline": [ { "date": "2025-08-27", "text": "连接生产周期与电力、冷却和未来增长。", "source_ids": [ "D10" ] }, { "date": "2025-11-19", "text": "追问供给追上需求的窗口及投资回报背景。", "source_ids": [ "D11" ] }, { "date": "2026-06-24", "text": "追问返还押金是否改变可用现金与资本回报。", "source_ids": [ "D12" ] } ], "gaps": [ "研报全文与持仓不可得", "管理层没有提供所有瓶颈的量化排名" ], "comparison": "比较研究问题与必要条件，不按名气或股价表现排名。", "assessment": { "strength": "覆盖物理部署与资金约束，不把卖出芯片等同于盈利运行。", "weakness": "供需窗口需要量化，不可用技术性能倍数替代客户ROI。", "next_evidence": "补充独立有日期的研报/访谈或具名提问；冻结原判断和反例后再评估结果。" } }, { "id": "rasgon-bernstein", "name": "Stacy Rasgon / Bernstein", "side": "sell", "role": "卖方美国半导体分析师；按2025具名名录身份记录", "scope": "总增长 → 产品组合 → 成本/毛利 → 费用与盈利", "focus": [ "组合拆解", "输入成本", "成本与定价", "运营杠杆", "分部与价格的增长桥" ], "boundary": "公开材料的方法样本；新增两期具名Q&A，不把管理层回答当作分析师立场，未认证稳定反复模式。", "timeline": [ { "date": "2025-08-27", "text": "将下一季度增量拆为平台和网络。", "source_ids": [ "D10" ] }, { "date": "2025-11-19", "text": "追问毛利维持的成本/定价手段及费用增长。", "source_ids": [ "D11" ] }, { "date": "2026-05-20", "text": "Q1澄清新云归类", "source_ids": [ "E7S01" ] }, { "date": "2026-08-26", "text": "Q2追问产品与价格贡献。", "source_ids": [ "S3D02" ] } ], "gaps": [ "未取得完整研报及其底层分产品估计", "两次提问只够方法样本，不认证完整思想" ], "comparison": "比较研究问题与必要条件，不按名气或股价表现排名。", "assessment": { "strength": "要求数字桥，防止增量和成本重复计算。", "weakness": "管理层定性回答仍不能识别每产品销量、ASP和毛利。", "next_evidence": "补充独立有日期的研报/访谈或具名提问；冻结原判断和反例后再评估结果。" }, "current_role_as_of": "2026-10-07", "role_evidence": "https://ir.amd.com/stock-data/analyst-coverage", "current_role": "AMD covering analyst, Bernstein Research (issuer list accessed 2026-10-07; undated list)" }, { "id": "newman-bernstein", "name": "Mark Newman / Bernstein", "side": "sell", "role": "卖方美国IT硬件分析师；2025重返Bernstein，非仍只覆盖全球存储", "scope": "客户类别 → 工作负载 → 数量/配置/价格 → 订单兑现", "focus": [ "CPU与AI分类", "客户结构", "量价配置拆分", "订单与交付" ], "boundary": "公开材料样本与编辑模型，不是完整个人思想。未认证所有方法为反复模式；没有研报全文、完整仓位或预测命中率审计。", "timeline": [ { "date": "2026-09-01", "text": "追问传统/AI服务器内企业与云客户构成。", "source_ids": [ "D14" ] }, { "date": "2026-09-02", "text": "追问单位、CPU核心、ASP及订单兑现约束。", "source_ids": [ "D13" ] } ], "gaps": [ "两份材料仅有两次具名交换，未达反复方法认证", "历史New Memory Paradigm完整报告未审阅", "不能将当前IT硬件覆盖误写成仅存储覆盖" ], "comparison": "比较研究问题与必要条件，不按名气或股价表现排名。", "assessment": { "strength": "对CPU/agentic叙事特别重视分母、类别和物理单位。", "weakness": "CPU核心数、服务器台数和收入可反向变化，需要实际配置。", "next_evidence": "补充独立有日期的研报/访谈或具名提问；冻结原判断和反例后再评估结果。" }, "role_evidence": "https://www.bernsteinresearch.com/about/" }, { "id": "smith-fundsmith", "name": "Terry Smith / Fundsmith", "side": "buy", "role": "材料时点2025-01及2026-07：署名CEO、Fundsmith LLP，买方股票基金管理；最近已读取角色证据为2026-07，未单独认证2026-10-06任职。", "scope": "企业质量与再投资 → 现金实现/估值 → 持有期限与基金资金流", "focus": [ "增量资本回报", "现金转换", "客户集中与资本品周期", "开放式基金生存约束" ], "boundary": "限已读取公开材料；署名与团队归属分开。没有完整交易、许可研报或结果审计；不把两份材料升级为永久个人思想。", "timeline": [ { "date": "2025-01", "text": "2024年信强调低换手与耐心，同时把AI资本投入的回报列为开放问题。", "source_ids": [ "B1D01" ] }, { "date": "2026-07", "text": "明确修订Do nothing过程，更重视基本面和价格动量；不是改为无估值约束追涨。", "source_ids": [ "B1D02" ] } ], "gaps": [ "未核查两时点间所有材料，不能定位每项改变的开始日", "尚需独立区分资金流、真实基本面恶化与估值驱动交易", "最新预测的3–5年窗口未结束，不可声称命中" ], "comparison": "编辑比较：相对Nomad的长期结构视角，此样本更直面开放式基金负债端。两者都必须验证资本回报，持有期限需受结构约束；这是样本比较而非优劣排名。", "assessment": { "strength": "把质量、支付价格和执行条件连起来，能暴露好公司与可持有投资的差别。", "weakness": "对动量/被动资金的解释仍需外部因果检验，不能由亏损或赎回推出市场必然失灵。", "next_evidence": "冻结资金流、执行成本、基本面指标和换手判据，核查后续报告能否改善投资者净结果。" } }, { "id": "sleep-nomad", "name": "Nick Sleep / Nomad（历史角色）", "side": "buy", "role": "2008年末/2009年半年原信：Nicholas Sleep署名，和Qais Zakaria共同管理，Sleep, Zakaria and Company为投资顾问；官方IGY现页面提供慈善身份，未认证当前基金管理职务或持仓。", "scope": "规模节约如何回馈客户 → 增长与竞争优势 → 长期估值、资本余量", "focus": [ "规模回馈反馈环", "长期资本配置", "价格/价值分离", "资金与认知余量" ], "boundary": "限已读取公开材料；署名与团队归属分开。没有完整交易、许可研报或结果审计；不把两份材料升级为永久个人思想。 两份独立期间原信来自同一批准选集，发表日未标；共同管理思想不独占归于Sleep。", "timeline": [ { "date": "2008-12-31", "text": "期间年度信面对危机，讨论余量与价格/价值，并解释规模回馈机制。", "source_ids": [ "B1D03" ] }, { "date": "2009-06-30", "text": "期间半年信延伸该机制至估值期限和持有条件，属于拓展而非已证明的观点反转。", "source_ids": [ "B1D04" ] } ], "gaps": [ "仅两封历史信，尚无完整思想演变序列", "集中持有是否适合特定委托、流动性和负债结构需独立证明", "历史回报表未独立审计，存活企业案例存在选择偏差" ], "comparison": "编辑比较：给云算力价格下降提供成本/留存/利用率反馈假说；不能从零售案例直接推出GPU云经济。与Smith比较需先匹配基金结构、期限和现金需求。", "assessment": { "strength": "把低利润率拆成主动让利与结构低效两种解释，要求找到再投资的反馈闭环。", "weakness": "文化和长期复利叙事难量化，过长终值期可能遮蔽技术替代、杠杆与永久损失。", "next_evidence": "对照失败的规模扩张者，逐期检验客户留存、单位成本、资本强度与增量回报，而非只看胜者。" } }, { "id": "ackman-pershing", "name": "Bill Ackman / Pershing Square", "side": "buy", "role": "买方；2026-02-11官方团队表列CEO / Portfolio Manager，2026中期信以CEO联署；截至本次截止日不把旧持仓当当前完整组合。", "scope": "业务现金流与护城河 → 管理和资本配置 → EPS与估值 → 封闭基金NAV/折价", "focus": [ "盈利复利与倍数风险分开", "as-is价值和优化价值分开", "长期资金与流动性", "股东回报与基金折价" ], "boundary": "三份独立日期原始材料：Ackman个人署名年信、Ackman/Israel联合署名中期信，以及官方团队演示。可分析公开方法和团队边界，不认证完整永久哲学；均非本周新发表观点。", "timeline": [ { "date": "2026-02-11", "text": "团队演示将核心选股约束、融资和资本配置同列，强调基础价值不依赖未来并购。", "source_ids": [ "B2D01" ] }, { "date": "2026-02-18", "text": "个人署名年信讨论价格发现和长期经营价值；与团队演示互证同一研究领域，不等同全员个人归因。", "source_ids": [ "B2D06" ] }, { "date": "2026-08-13", "text": "联合信进一步区分盈利增长与倍数波动。相同约束不同材料有延续，不证明立场反转；其长期增长判断尚有未来验证窗口。", "source_ids": [ "B2D02" ] } ], "gaps": [ "未验证2026-10-06完整实时持仓及当前估值", "联合署名与团队材料不能精确拆分个人研究贡献", "没有收益归因或预测命中率独立审计", "尚无三次跨两场可归因问答，不授予稳定个人方法认证" ], "comparison": "相较本次Greenlight样本，更适合训练高质量业务复利、优化价值与持有期限；这是本次选材的比较。", "source_ids": [ "B2D01", "B2D02", "B2D06" ], "current_role_status": "2026-08报告仍列CEO；本次未另做10月当日人事调查。" }, { "id": "einhorn-greenlight", "name": "David Einhorn / Greenlight Capital", "side": "buy", "role": "买方长短仓；2024 Sohn官方议程确认为President；2024演示个人署名，其后季度信为机构落款。", "scope": "现金回收与周期估值 → 市场结构假说 → 产业链现金/融资核验 → 论点与风险分离", "focus": [ "现金分配而非仅等重估", "周期低谷与正常化盈利", "行业收入口径与重复加总", "资金来源和资本开支寿命" ], "boundary": "两份不同日期本人署名演示经公开逐页镜像审阅；另两份不同日期机构原始季度信明确团队归因。满足跨材料公开样本，仍不足认证个人永久方法或业绩。官网文件解析受限，镜像未逐字官方校验；不是本周新观点。", "timeline": [ { "date": "2024-04-03", "text": "个人Sohn演示使用现金回收与周期目标组织化工公司论证。", "source_ids": [ "B2D03" ] }, { "date": "2024-04-24", "text": "机构信将分红、回购的可持续性作为外部重估之外的回报来源。", "source_ids": [ "B2D04" ] }, { "date": "2024-10-23", "text": "个人Peloton演示把订阅与硬件分开，将同行成本/SBC比较接入利润模型；与化工演示体现跨行业量化，但不能因两个案例宣称全领域稳定方法。", "source_ids": [ "B2D07" ] }, { "date": "2025-10-14", "text": "机构信增加AI融资及产业链收入口径检验；材料领域扩展，不是同一变量观点反转证据。", "source_ids": [ "B2D05" ] } ], "gaps": [ "两份本人演示，但没有跨两场三次可归因问答，不授予稳定个人问答方法认证", "当前职务和2026完整组合未另行确认", "市场结构及AI技术机制表述是作者判断，不是经本次审计的科学事实", "未复现信中第三方研究总量，也未验证具体投资结果" ], "comparison": "本次样本较强调靠公司现金分配回收价值、周期正常化和资金链核验；不能以此判定其永远反技术或永远低估值。", "source_ids": [ "B2D03", "B2D04", "B2D05", "B2D07" ], "current_role_status": "可确认2024出版时职务；截止2026-10-06未以历史角色代替当前履历核查。" }, { "id": "mauboussin-counterpoint", "name": "Michael Mauboussin / Counterpoint Global", "side": "buy", "role": "资产管理机构的买方投资研究；本样本为与Dan Callahan共同署名的方法研究，不是基金交易记录或卖方行业覆盖", "current_role": { "as_of": "2026-10-06", "title": "Head of Consilient Research, Counterpoint Global, Morgan Stanley Investment Management", "url": "https://www.morganstanley.com/im/en-us/individual-investor/about-us/people-and-teams/investment-professionals/michael-mauboussin.html", "corroboration_url": "https://www.michaelmauboussin.com/about", "access": "官方MSIM生平搜索文本与本人About页核验；不把当前职称倒推为每篇历史材料的职称。" }, "scope": "价格内含预期 → 增量资本回报与再投资 → 现金流 → 历史参照类与证据更新", "focus": [ "预期与证券回报", "ROIC与ROIIC", "无形资本和现金一致性", "基准概率与参照类" ], "boundary": "三份不同日期的共同署名公开材料；方法地图是编辑归纳，不能归为个人私有观点、实际持仓或业绩。当前职位与每篇来源角色分开，机构不单独计为一位分析师。", "timeline": [ { "date": "2021-04-14", "text": "MEROI连接价格隐含预期与投资测量。", "source_ids": [ "B3D1" ] }, { "date": "2023-06-06", "text": "ROIC与现金流、预期修订和竞争持续性结合；材料范围扩展，不证明观点逆转。", "source_ids": [ "B3D2" ] }, { "date": "2026-05-13", "text": "回应AI增长参照类的疑问，保留历史约束与动态更新。", "source_ids": [ "B3D3" ] } ], "gaps": [ "没有本人或基金当时交易、持仓与预测命中率审计", "当前PDF存在后续版权/合规页脚，正文日期按封面；未确认全部版本差异", "历史参照类底层数据、因果关系和公司当前财务未独立重算" ], "comparison": "相较单点盈利预测，更强调支付价格、投资需求和概率边界；这种方法不替代行业产能、客户研究或组合授权。", "assessment": { "strength": "把高质量公司与高回报证券分开，并要求增长、资本回报和现金相互约束。", "weakness": "模型对资本化寿命、未来ROIIC和参照类选择敏感；输入假设不能靠作者声望验证。", "next_evidence": "以一个陌生行业冻结价格隐含模型，再用下一份有日期的经营披露检验先验和现金需求。" } }, { "id": "arya-bofa", "name": "Vivek Arya / Bank of America", "side": "sell", "role": "卖方半导体分析师；材料发生时NVIDIA名录为BofA Securities，ADI主持人也如此；当前AMAT名单为Bank of America", "role_evidence": "https://ir.amd.com/stock-data/analyst-coverage", "current_role_as_of": "2026-10-07", "scope": "电力/客户预算 → 供应商可捕获份额 → 收入增长；再检验周期与价格的来源", "focus": [ "每GW收入与系统边界", "客户融资及预算约束", "总增长的周期/价格拆分", "增量利润桥", "增量市场与生态融资分开" ], "boundary": "公开材料的方法样本；新增两期具名Q&A，不把管理层回答当作分析师立场，未认证稳定反复模式。", "timeline": [ { "date": "2025-11-19", "text": "NVIDIA：追问每GW内容价值及长期扩建资金来源。", "source_ids": [ "D11" ] }, { "date": "2026-02-25", "text": "NVIDIA：把预算不增长设为条件，追问公司仍能否增长。", "source_ids": [ "S1D01" ] }, { "date": "2026-05-20", "text": "Q1追问独立CPU与替代", "source_ids": [ "E7S01" ] }, { "date": "2026-08-19", "text": "ADI：将同比加速分为长期、周期和价格贡献，询问次年及经营杠杆。", "source_ids": [ "S1D03" ] }, { "date": "2026-08-26", "text": "Q2追问生态支持及竞争性自研。", "source_ids": [ "S3D02" ] } ], "gaps": [ "没有取得BofA许可研报或完整底层估计", "本次融资/预算同机制仅两次交换，未达三次跨两call认证门槛", "ADI名录机构别名与主持人口径不同，未将其误当买方身份", "未审计其真实预测结果、目标价、收益或2026-10-06最新投资立场" ], "comparison": "本次样本重视大市场到公司收入的分母、份额和预算闭合；这是编辑比较。", "assessment": { "strength": "在技术机会之外追问收入金额及谁付得起，适合F9、F10、F18。", "weakness": "管理层叙事不能替代客户现金流和实际部署；不同GW定义也不能直接比较。", "next_evidence": "补有日期同机制第三次独立Q&A或授权完整报告，冻结模型后用客户预算与部署验证。" }, "evolution": "观察到问题从每GW与融资扩展到静态预算的增长边界；这是问题约束变化，不是证实由看多转看空。", "counterexample": "预算不增长时，通过其他设备替换、份额与新客户仍可增长；内容价值上升也可能挤压客户回报。", "current_role": "AMD covering analyst, Bank of America Global Research (issuer list accessed 2026-10-07; undated list)" }, { "id": "sur-jpm", "name": "Harlan Sur / JPMorgan", "side": "sell", "role": "卖方半导体/设备分析师；Lam、NVIDIA和ADI有日期名录均为JPMorgan；当前AMAT名单仍列JP Morgan", "role_evidence": "https://ir.appliedmaterials.com/stock-information/analyst-coverage", "current_role_as_of": "2026-10-06", "scope": "政策/架构变化 → 地理供应和网络内容 → 成本/毛利的可复核驱动", "focus": [ "制造调配时滞", "网络附着与内容价值", "量价组合与固定成本吸收", "披露不足下的条件模型" ], "boundary": "三份官方托管公开问答给出三个方法样本；关税、网络和毛利不同机制，不认证稳定个人模式。两段追问不能充作两个独立call。", "timeline": [ { "date": "2025-04-23", "text": "Lam：问关税下美国制造能力以及前期库存后的成本风险。", "source_ids": [ "S1D02" ] }, { "date": "2026-02-25", "text": "NVIDIA：问网络占比、Spectrum系列订单和未来收入运行率。", "source_ids": [ "S1D01" ] }, { "date": "2026-08-19", "text": "ADI：尝试由利润指引推回毛利，问组合/量/价格贡献。", "source_ids": [ "S1D03" ] } ], "gaps": [ "没有JPMorgan许可研究全文及其估计模型", "三个议题不构成同机制三次认证", "Lam管理层不披露完整地理制造及材料成本，不能伪精确预测", "未用第三方收益排名证明研究质量或推断当前评级" ], "comparison": "本次样本把宏观与平台叙事落到具体供应节点和利润驱动，仍需独立核验。", "assessment": { "strength": "能把制造地点、网络架构及价格组合转成可追问经营变量。", "weakness": "量、组合、价格和固定成本吸收相互作用；只凭盈利指引不能唯一反推出毛利来源。", "next_evidence": "逐产品/厂区产能、网络附着率与ASP、价格实施时点及成本吸收桥。" }, "evolution": "跨期比较显示研究对象变化，不能把关税到网络到毛利描述成投资观点反转。", "counterexample": "更高网络占比来自GPU延迟而非网络增长；利用率改善也可能抵不过输入成本上涨。" }, { "id": "rolland-susquehanna", "name": "Christopher Rolland / Susquehanna", "side": "sell", "role": "卖方半导体分析师；2024–2025原稿本人身份为Susquehanna，2026现职尚未以雇主原页独立确认", "scope": "季度收入差异 → 量/内容/时点分解 → 客户多元化兑现", "focus": [ "季节性口径", "结构增长与提前订单", "内容与单位", "设计到收入" ], "boundary": "两份有日期本人Q&A方法样本，不是完整研究报告；未认证同机制三次跨两场的反复方法。代理同事问题不自动归为本人。", "timeline": [ { "date": "2024-11-04", "text": "本人追问异常季节性与内容、销量、提前出货的解释。", "source_ids": [ "S2D-B01" ] }, { "date": "2025-08-05", "text": "本人追问超指引来源及PC/汽车多元化里程碑；问题范围扩展不等于观点反转。", "source_ids": [ "S2D-B02" ] } ], "gaps": [ "未取得许可研报、目标价模型或投资回报审计", "2026当前雇主角色未独立确认", "Cirrus 2025-11-04和2026-02-03为Dylan Ollivier代问，排除个人方法认证" ], "comparison": "本样本侧重季度变化的可识别性，不能由一次收入强势直接提高长期需求。", "assessment": { "strength": "先检查分母、产品内容与时点，再讨论可持续增量。", "weakness": "管理层不能量化所有提前订单，仍需情景与渠道验证。", "next_evidence": "下一次具名本人Q&A或原始署名研究补充，同变量冻结判断后核对后续终端销量。" } }, { "id": "curtis-jefferies", "name": "Blayne Curtis / Jefferies", "side": "sell", "role": "卖方半导体分析师；2025-10/2026-01 Lam原稿以及2026-08-19 ADI原稿名录均为Jefferies；2026-10现职未单独验证。", "scope": "总投资指引 → 区域/终端结构 → 建设/升级约束 → 收入节奏", "focus": [ "投资时间分布", "终端与区域去重", "棕地升级与绿地扩产", "可交付上限" ], "boundary": "两份有日期公开Q&A样本，非许可研报审阅；不同问题不构成完整稳定思想认证。", "timeline": [ { "date": "2025-10-22", "text": "追問后半年权重是否各终端共同驱动；管理层仅给WFE总体解释。", "source_ids": [ "S2D-B03" ] }, { "date": "2026-01-28", "text": "追问Reliant区域和NAND升级的节奏/空间约束。", "source_ids": [ "S2D-B04" ] } ], "gaps": [ "总WFE回答没有给分部季度数字", "未审阅完整Jefferies产能/目标价模型", "扫描OCR保留转录风险，2026-10现职未独立确认" ], "comparison": "把增长拆回区域与部署约束，比仅按年度总额等分更适合建季度模型。", "assessment": { "strength": "识别总量不变下的季度、区域、产品组合差异。", "weakness": "问句本身不能证明需求兑现；需要客户安装计划和许可边界。", "next_evidence": "公开客户预算、项目里程碑和具名新Q&A；不要求私人客户清单。" }, "role_evidence_source_ids": [ "S2D-B03", "S2D-B04", "S1D03" ] }, { "id": "rakers-wf", "name": "Aaron Rakers / Wells Fargo", "side": "sell", "role": "卖方科技分析师；2025–2026来源均署Wells Fargo Securities。最近直接核查的角色日期为2026-08-26，不推定此后从未变化。", "role_as_of": "2026-08-26", "role_evidence_source_ids": [ "S3D02" ], "scope": "网络/CPU/加速器架构 → 工作负载与可部署容量 → 价值捕获和经济约束", "focus": [ "系统层机会与价值捕获", "异构推理与CPU独立需求", "GPU与自研ASIC替代", "电力约束和每GW价值密度" ], "boundary": "四份有日期的发行人托管具名提问样本，不是私人思想、完整研报或覆盖全貌。四种机制各有一问，未认证任何机制为反复模式；三张以上合成卡不增加真实提问数。", "timeline": [ { "date": "2025-08-27", "text": "追问Spectrum-XGS定位和新增网络机会；不是本人给出的TAM测算。", "source_ids": [ "D10" ] }, { "date": "2025-11-19", "text": "追问Anthropic合作背景下AI ASIC与GPU竞争判断是否变化；问题不等于接受管理层的优势论。", "source_ids": [ "D11" ] }, { "date": "2026-02-25", "text": "追问Vera独立CPU在异构推理中的作用；只反映新增研究问题，不能证明观点反转。", "source_ids": [ "S1D01" ] }, { "date": "2026-08-26", "text": "追问每GW美元密度及未来部署供给能否非线性扩张；不把管理层方向回答视为本人预测。", "source_ids": [ "S3D02" ] } ], "gaps": [ "仅核查选定NVIDIA问题，未审阅许可卖方研报", "没有相同机制三次跨两电话会的稳定模式认证", "没有预测收益或目标价命中审计", "管理层每GW回报和技术优势仍需独立工作负载证据" ], "comparison": "选定样本提供系统架构向商业约束的提问入口；这是编辑比较，不是分析师能力或排名。", "assessment": { "strength": "问题跨网络、CPU、ASIC与电力瓶颈，能逼近产品组合和部署约束。", "weakness": "宏观TAM或架构优势答案仍缺客户级验证和净收入、利润、现金桥。", "next_evidence": "以2026-11-17 NVIDIA下一次业绩会为检查点；收集实际交付、接受、利润和客户异构部署证据；缺披露即保留未知。" } }, { "id": "rogoff-polar", "name": "Ben Rogoff / Polar Capital", "side": "buy", "role": "买方科技基金经理；2026材料与Alastair Unwin共同署名", "scope": "AI架构变化 → 价值迁移 → 原有龙头终值与相对估值", "focus": [ "技术进展与股价分离", "价值链迁移", "资本密集度", "软件终值" ], "boundary": "两份实质方法样本；共同署名观点不作个人独有方法认证，2026网页发布日期未明确。", "timeline": [ { "date": "2024-09-25", "text": "将市场波动与模型进展分开，维持AI主题。", "source_ids": [ "B100-BR1" ] }, { "date": "FY2026", "text": "与Unwin共同报告强调模型、芯片和基础设施对既有软件利润池的挑战。", "source_ids": [ "B100-BR2" ] } ], "gaps": [ "2026发布日期未明确", "未审计长期预测命中率", "历史类比须独立验证" ], "comparison": "适合检验旧有护城河能否穿越新架构。", "method_status": "method-sample; not certified recurring pattern" }, { "id": "kim-blackrock", "name": "Tony Kim / BlackRock", "side": "buy", "role": "买方；Global Technology团队负责人", "scope": "AI三层价值捕获与物理供应链", "focus": [ "基础设施/模型/应用", "资本开支与功率", "供应约束", "竞争结构变化" ], "boundary": "两份官方访谈方法样本；2026材料缺明确发布日期，未经跨周期稳定方法认证。", "timeline": [ { "date": "2025-09-11", "text": "分三层审视AI机会和资本进入壁垒。", "source_ids": [ "B100-TK1" ] }, { "date": "2026; publication date unknown", "text": "强调基础设施稀缺与模型/应用之间价值迁移。", "source_ids": [ "B100-TK2" ] } ], "gaps": [ "资本开支预测非已实现需求", "功率与计算能力不能混用", "技术能力倍数缺统一基准" ], "comparison": "把投资叙事压回电力、供给及价值捕获的位置。", "method_status": "method-sample; not certified recurring pattern" }, { "id": "clode-janus", "name": "Richard Clode / Janus Henderson", "side": "buy", "role": "买方；Global Technology Leaders基金经理", "scope": "技术成长 → 预期差 → 估值及现金流纪律", "focus": [ "PE/FCF收益率", "未计价增长", "产能与需求", "微型炒作周期" ], "boundary": "一份个人Q&A、一份三人共同文章，适合方法样本；未认证个人稳定思维模式。", "timeline": [ { "date": "2025-08-13", "text": "沿用盈利及现金流估值，并比较当代与互联网泡沫时的盈利基础。", "source_ids": [ "B100-RC1" ] }, { "date": "2025-12-04", "text": "团队强调算力供需约束并监测循环投资和亏损科技股炒作。", "source_ids": [ "B100-RC2" ] } ], "gaps": [ "共同文章贡献未拆到个人", "未复算图表数据", "资金循环对终端需求影响仍须核验" ], "comparison": "适合把增长故事与买入价格、现金流接起来。", "method_status": "method-sample; not certified recurring pattern" }, { "id": "wick-seligman", "name": "Paul Wick / Columbia Threadneedle", "side": "buy", "role": "买方；Seligman科技团队负责人（来源时点）", "scope": "半导体供需弹性与平台用途变化", "focus": [ "代工利用率", "产能时滞", "知识工作工具", "供应链约束" ], "boundary": "2021与2023圆桌中可明确归属Wick的发言；不将同场其他人观点归于他，不推定2026完整职位。", "timeline": [ { "date": "2021-06", "text": "从代工集中及高利用率解释供给缺乏余量。", "source_ids": [ "B100-PW1" ] }, { "date": "2023-09-27", "text": "将AI定位为知识工作工具，区别于以分发为重要作用的互联网/移动技术。", "source_ids": [ "B100-PW2" ] } ], "gaps": [ "需补2026任职及最新独立观点", "两主题不足证明统一稳定方法", "未审计预测绩效" ], "comparison": "适合检验结构增长和短期产能紧缺是否被重复定价。", "method_status": "method-sample; not certified recurring pattern" }, { "id": "benjamin-fidelity", "name": "Adam Benjamin / Fidelity（来源时点）", "side": "buy", "role": "买方科技研究与基金管理；已于2026-07-15前卸任FSELX经理", "scope": "AI与非AI分化 → 库存周期 → 预期差与投资复盘", "focus": [ "IT预算挤出", "非AI库存周期", "技术护城河", "承认判断过早" ], "boundary": "使用2026年1—2月研究材料；SEC7月文件纠正职位。不能依据旧网页称其当前FSELX经理。", "timeline": [ { "date": "2026-01-07", "text": "讨论AI资本投入与传统IT预算竞争。", "source_ids": [ "B100-AB1" ] }, { "date": "2026-02-28", "text": "承认2024/2025非AI复苏判断过早，重估库存及需求。", "source_ids": [ "B100-AB2" ] }, { "date": "2026-07-15", "text": "SEC文件显示FSELX由Sonu Kalra接任。", "source_ids": [ "B100-ABROLE" ] } ], "gaps": [ "当前其他任职待核实", "滚动PDF须保存时点", "未独立审计基金回报或预测绩效" ], "comparison": "适合把行业判断与组合实际代价、后续修订连接。", "method_status": "method-sample; not certified recurring pattern" }, { "id": "baker-atreides", "name": "Gavin Baker / Atreides", "side": "buy", "role": "买方Managing Partner and CIO", "scope": "路线图假设 → 失败概率 → 利润率与利润额分离", "focus": [ "路线图风险", "假设复盘", "AI毛利", "资本回报" ], "boundary": "两篇实质一手方法样本；尚未认证跨时期重复方法。", "timeline": [ { "date": "2020-08-14", "text": "自述Intel投资失误：检验10nm时没有充分检验7nm假设；清单不能弥补关键变量遗漏。", "source_ids": [ "B100-GB1" ] }, { "date": "2025-10-30", "text": "认为AI计算成本改变软件经济，利润率下降须与利润额和资本需求分开判断。", "source_ids": [ "B100-GB2" ] } ], "gaps": [ "未审计长期预测命中率及投资业绩", "公开材料存在选择性披露", "跨来源稳定模式未认证" ], "comparison": "路线图假设 → 失败概率 → 利润率与利润额分离", "method_status": "method-sample; not certified recurring pattern" }, { "id": "gerstner-altimeter", "name": "Brad Gerstner / Altimeter", "side": "buy", "role": "买方Altimeter创始人；2022署名Founder and CEO", "scope": "资本纪律 → 现金流桥 → 概率与证券条款", "focus": [ "现金流", "资本配置", "概率估值", "公私市场" ], "boundary": "两篇实质一手方法样本；尚未认证跨时期重复方法。", "timeline": [ { "date": "2022-10-24", "text": "建议削减人员成本并限制资本开支和元宇宙支出，保留AI投资信念。", "source_ids": [ "B100-BG1" ] }, { "date": "2023-12-05", "text": "将技术乐观与交易价格分离；要求估值最终转为现金流，并检验回报概率和证券上行限制。", "source_ids": [ "B100-BG2" ] } ], "gaps": [ "未审计长期预测命中率及投资业绩", "公开材料存在选择性披露", "跨来源稳定模式未认证" ], "comparison": "资本纪律 → 现金流桥 → 概率与证券条款", "method_status": "method-sample; not certified recurring pattern" }, { "id": "hendi-susanto", "name": "Hendi Susanto", "side": "buy", "role": "买方基金经理／科技研究；按材料日期归属", "scope": "半导体与设备、自动化、AI投资链", "focus": [ "周期与长期需求分离", "客户集中与议价权", "技术换代的利润传导" ], "boundary": "方法样本；2025材料为共同署名，不构成个人稳定方法认证，也不代表当前投资建议。", "timeline": [ { "date": "2024-08-12", "text": "署名评论同时讨论AI长期机会和短期库存、估值约束。", "source_ids": [ "BX-HS-20240812" ] }, { "date": "2025-04-22", "text": "与Ryuta Makino共同调研日本企业，拓展至客户集中和技术换代。", "source_ids": [ "BX-HS-20250422" ] } ], "gaps": [ "共同研究中的个人贡献无法拆分", "未审计持仓、预测结果或收益", "管理层陈述仍需发行人原文交叉核对" ], "comparison": "编辑比较：擅长把技术受益链与周期、议价和估值分开讨论。" }, { "id": "zheng-weishan", "name": "郑巍山", "side": "buy", "role": "买方基金经理；此处研究2021—2022年银河基金公开材料", "scope": "半导体细分景气、国产渗透与产品竞争力", "focus": [ "细分行业口径", "成本转嫁", "数据验证逻辑" ], "boundary": "历史方法样本，不代表当前岗位、持仓或观点。两篇短文不足以评价择时能力。", "timeline": [ { "date": "2021-09-10", "text": "讨论半导体回调和产品结构、国产化的关系。", "source_ids": [ "BX-ZWS-20210910" ] }, { "date": "2022-06-09", "text": "用一季报指标讨论行业分化与代工成本传导。", "source_ids": [ "BX-ZWS-20220609" ] } ], "gaps": [ "缺最新季度报告与组合归因", "未核验底层券商行业统计样本", "当前任职需另行确认" ], "comparison": "编辑比较：以细分产品和成本传导检验宏观科技主题。" }, { "id": "buffett-berkshire", "name": "Warren Buffett", "side": "buy", "role": "买方资本配置者；2023/2024投资信时期为Berkshire CEO，当前头衔未核验", "scope": "经营盈利质量 → 增量资本回报 → 管理者激励与纠错", "focus": [ "经营与估值波动", "增量资本配置", "管理者激励" ], "boundary": "方法论样本，非AI行业专职分析师。两期投资信不等于完整思想复刻；未颁发稳定方法认证。", "timeline": [ { "date": "2024-02-24", "text": "以经营盈利和增量资本回报讨论企业价值。", "source_ids": [ "B71D01" ] }, { "date": "2025-02-22", "text": "讨论投资与识人错误以及及时修正；新增资本应影响管理激励考核。", "source_ids": [ "B71D02" ] } ], "gaps": [ "2026当前职务未单独核验", "没有逐项投资结果或预测准确率审计" ], "comparison": "本次样本偏重经营价值与资本回报；以下AI迁移练习由编辑合成。" }, { "id": "nygren-oakmark", "name": "Bill Nygren", "side": "buy", "role": "买方基金经理；Harris | Oakmark Partner and Co-CIO-U.S.（官网核验）", "scope": "非共识判断 → 估值影响 → AI盈利持续期与安全边际", "focus": [ "预期差", "周期正常化", "安全边际", "永久损失" ], "boundary": "公开方法论与AI硬件估值观点样本，不是半导体专职覆盖。未作稳定方法认证。", "timeline": [ { "date": "2022-Q4", "text": "明确研究员任务是发现重要共识错误、计算估值影响并持续监测。", "source_ids": [ "B71D03" ] }, { "date": "2026-06-30", "text": "将同一估值问题应用于AI硬件：当前利润强不消除持续期不确定性。", "source_ids": [ "B71D04" ] } ], "gaps": [ "没有底层持仓估值模型", "未审计AI不持仓决策的最终投资结果" ], "comparison": "更直接挑战AI硬件的盈利持续期；不能把未持仓编辑成确定性看空。" }, { "id": "philippe-laffont", "name": "Philippe Laffont / Coatue", "side": "buy", "role": "Coatue创始人；CTEK Primary Portfolio Manager（2026-05-01招募书）", "scope": "长期技术主题 → 跨产业链机会 → 模型更新与风险预算", "focus": [ "长期主题", "GPU需求传导", "现金与风险暴露", "公共/私人资产口径" ], "boundary": "一份个人访谈加一份机构招募书的方法样本，不是两份个人材料认证。机构流程与本人观点分别归属；没有复制私有投资体系。", "timeline": [ { "date": "2025-06-12", "text": "个人访谈连接长期主题、GPU周边产业与现金风险控制。", "source_ids": [ "EXP200-B28-A" ] }, { "date": "2026-05-01", "text": "机构文件披露数据支持、持续监测和风险工具，并明确组合责任。", "source_ids": [ "EXP200-B28-B" ] } ], "gaps": [ "还缺第二份个人完整材料验证稳定方法", "私有模型及历史交易回报未审计", "私人资产估值更新可能滞后" ], "comparison": "本样本强调主题扩张与资本存续同时约束；长期叙事不自动证明安全边际。" }, { "id": "lipacis-evercore", "name": "Mark Lipacis / Evercore ISI", "side": "sell", "role": "现任Evercore半导体与半导体设备团队负责人；方法样本发生于Jefferies任职期", "role_evidence": "https://www.evercore.com/team/mark-lipacis/", "current_role_as_of": "2026-10-07", "scope": "渠道报表 → 库存与消耗 → 周期判断", "focus": [ "渠道库存与真实消耗对账", "库存周数", "出货与消耗" ], "boundary": "2011与2023历史问题样本，不能直接代表2026投资立场或已认证稳定方法。", "timeline": [ { "date": "2011-12-08", "text": "在Jefferies任职期追问TI渠道库存周数与此前基线。", "source_ids": [ "E7S04" ] }, { "date": "2023-06-13", "text": "在Jefferies任职期对照ADI渠道叙事与分销商库存报表。", "source_ids": [ "E7S03" ] } ], "gaps": [ "近期具名同机制第三次材料待补", "没有授权研报底层估计或投资结果审计" ], "comparison": "本次样本注重库存口径与外部报表交叉验证。", "current_role": "Evercore ISI senior managing director, head of Semiconductor and Semiconductor Capital Equipment Team (official biography accessed 2026-10-07)" }, { "id": "charlie-chan-ms", "nomination_id": "S36", "name": "Charlie Chan", "role": "卖方；Morgan Stanley（2025年官方财报问答身份）", "side": "sell", "scope": "先进制程、汇率/定价、系统创新与AI市场边界", "focus": [ "汇率到毛利的传导", "系统与制程互补", "市场限制与增长基数" ], "boundary": "方法样本：已读两份不同日期的官方问答；提问只证明其检验的问题，不构成投资评级、明确预测或个人稳定方法认证。", "timeline": [ { "date": "2025-07-17", "text": "询问汇率冲击能否由定价吸收，以及中国市场变化对AI增长假设的影响。", "source_ids": [ "S36SRC1" ] }, { "date": "2025-10-16", "text": "追问系统创新与先进制程需求的兼容，并回访中国市场约束。", "source_ids": [ "S36SRC2" ] } ], "gaps": [ "缺少本人完整证券研究报告和估值模型", "未审计预测结果、持仓或投资收益", "2025问答不保证2026具体任职及覆盖范围" ], "comparison": "样本侧重把技术与地理约束转成财务假设；属于编辑归纳。", "evidence_status": "method-sample" }, { "id": "keith-weiss-ms", "nomination_id": "S51", "name": "Keith Weiss", "role": "卖方；Morgan Stanley美国软件研究负责人（2025年本人官方播客自述）", "side": "sell", "scope": "AI商业采用、云增长归因、客户集中与商业模式", "focus": [ "研发到收入的阶段区别", "增长执行归因", "客户转竞争者风险", "席位到用量计费" ], "boundary": "方法样本：两次官方问答及一次本人官方播客；管理层回答不归为Weiss论断，不声称已认证长期个人方法。", "timeline": [ { "date": "2025-01-29", "text": "追问Azure执行偏差及未来增长修复。", "source_ids": [ "S51SRC1" ] }, { "date": "2025-03-03", "text": "本人讨论商业采用阶段与自动化推动计费方式变化。", "source_ids": [ "S51SRC3" ] }, { "date": "2025-07-30", "text": "询问AI大客户自建基础设施及成为竞争者的风险。", "source_ids": [ "S51SRC2" ] } ], "gaps": [ "尚缺完整研究报告底层数据", "未核验2025商业采用预测的事后结果", "不能用管理层解释代替独立验证" ], "comparison": "样本把收入增长、客户经济与平台关系连在一起；编辑比较，不是永久人格。", "evidence_status": "method-sample" }, { "id": "mark-moerdler-bernstein", "nomination_id": "S52", "name": "Mark Moerdler", "role": "卖方；Bernstein Research（2025年公司官方问答身份）", "side": "sell", "scope": "软件AI变现、席位与用量、增量利润", "focus": [ "计费单位变化", "收入替代与增量", "应用差异", "利润率约束" ], "boundary": "方法样本：两份官方问答；Moerdler提出席位/用量与盈利问题，管理层乐观回应不等于他给出的预测或投资建议。", "timeline": [ { "date": "2025-07-30", "text": "询问不同SaaS应用的AI变现方式及长期利润率。", "source_ids": [ "S51SRC2" ] }, { "date": "2025-09-11", "text": "追问Creative Cloud用量能否补偿潜在席位压力及重要性何时显现。", "source_ids": [ "S52SRC2" ] } ], "gaps": [ "缺少本人完整模型与评级论证", "只有问题样本，不能认证稳定投资方法", "尚未跟踪席位、用量及AI成本的同队列数据" ], "comparison": "样本侧重商业化单元与收入替代；编辑归纳，不外推为个人所有观点。", "evidence_status": "method-sample" }, { "id": "michael-elias", "name": "Michael Elias", "side": "sell", "role": "TD Cowen通讯基础设施研究；当前官方简介为Managing Director", "scope": "可交付容量 → 合同与定价 → 开发收益率与资本成本", "focus": [ "真实通电容量", "需求去重", "收益率差", "训练与推理选址" ], "boundary": "基于两次公开访谈的方法样本；问答不是本人买卖建议。QTS转录的Ellis拼写按官方主持人元数据识别，未由音频复核。", "timeline": [ { "date": "2025-05-29", "text": "询问成本、融资与收益率差，以及供给瓶颈。", "source_ids": [ "EXP200-S81-A" ] }, { "date": "2025-08-12", "text": "追问电力申请重复与项目可交付性。", "source_ids": [ "EXP200-S81-B" ] } ], "gaps": [ "未审计评级、目标价和投资回报", "未读取付费底层模型", "访谈为管理层自述，需要项目级独立验证" ], "comparison": "本次样本把需求叙事转成可交付项目与收益率差；不以MW申请量替代实际需求。" }, { "id": "julian-mitchell", "name": "Julian Mitchell", "side": "sell", "role": "Barclays工业研究分析师；Eaton官方覆盖表确认机构", "scope": "订单/积压 → 发货节奏 → 有机增长 → 价格成本与利润率", "focus": [ "交付节奏", "指引基数", "价格成本", "库存与订单" ], "boundary": "两次发行人问答的方法样本；问题显示检验方向，管理层回答不是Mitchell的预测。", "timeline": [ { "date": "2026-01-29", "text": "以订单和交期反推收入节奏，并追问库存与折价。", "source_ids": [ "EXP200-S92-B" ] }, { "date": "2026-05-01", "text": "检查两年增长基数与积压转收入，再检验价格成本抵消。", "source_ids": [ "EXP200-S92-A" ] } ], "gaps": [ "没有付费研究报告和目标价模型", "未审计预测结果", "管理层口径仍需财报和后续兑现核对" ], "comparison": "本次样本强调从前置订单到当期收入和利润的时间桥。" }, { "id": "unwin-polar", "name": "Alastair Unwin / Polar Capital（共同署名样本）", "side": "buy", "role": "买方科技基金经理；来源时为Partner、Deputy Fund Manager", "scope": "增长和资本密集度 → 利润池持续性 → 估值与组合约束", "focus": [ "资本密集度", "软件替代", "基本面与估值", "集中度" ], "boundary": "两份年度报告共同署名；不把团队报告拆成Unwin个人独有方法。FY2025摘要无明确发表日。", "timeline": [ { "date": "2024-07-16", "text": "共同报告对比增长、利润率和再投资，保留大平台差异。", "source_ids": [ "O8BAS05" ] }, { "date": "FY2025", "text": "共同报告明确提升替代风险权重；不是个人观点的独立转向认证。", "source_ids": [ "O8BAS06" ] } ], "gaps": [ "共同作者贡献未拆分", "FY2025发表日期未核验", "未审计组合绩效或预测结果" ], "comparison": "适合检验收入增长是否覆盖新增资本负担。", "method_status": "method-sample; not certified recurring pattern", "source_evidence": [ { "source_id": "O8BAS05", "claim": "共同把大平台增长、利润率、再投入和估值放在一起评价。", "locator": "Technology Outlook / Magnificent 7；Survival of the fittest", "read_scope": "网页上述章节及署名日期；非完整年报", "evidence_state": "newly-read primary" }, { "source_id": "O8BAS06", "claim": "共同提高对资本密集度、软件替代和大平台仓位的警惕。", "locator": "Mag-7 update；Disruption ahead；Concentration risk", "read_scope": "选读这些章节及Cloud update、Technology/AI risks；无发表日", "evidence_state": "newly-read primary" } ] }, { "id": "porter-janus", "name": "Alison Porter / Janus Henderson（团队样本）", "side": "buy", "role": "买方Global Technology Leaders基金经理；2025文章共同作者", "scope": "政策冲击 → 需求与流动性 → 估值纪律", "focus": [ "二阶需求", "周期与结构", "流动性", "估值" ], "boundary": "两份团队共同署名材料，不代表Porter个人独有观点。", "timeline": [ { "date": "2025-04-09", "text": "共同区分直接政策暴露和需求传导。", "source_ids": [ "O8BAS02" ] }, { "date": "2025-12-04", "text": "共同关注AI容量制约及循环投资风险。", "source_ids": [ "B100-RC2" ] } ], "gaps": [ "个人与团队贡献无法逐句拆分", "政策为历史时点", "未核验机构底层估值模型" ], "comparison": "团队政策分析可用于防止只计算进口税率。", "method_status": "method-sample; not certified recurring pattern", "source_evidence": [ { "source_id": "O8BAS02", "claim": "团队区分政策事件和经济风险；关注间接需求及流动性。", "locator": "五项考虑1–5；Is now a buying opportunity?", "read_scope": "文章完整分析正文与作者区", "evidence_state": "newly-read primary" }, { "source_id": "B100-RC2", "claim": "团队提示电力施工约束和投资循环。", "locator": "AI is transformative；Why do we think skepticism about AI valuations is healthy?", "read_scope": "上述两节及日期署名；图表未复算", "evidence_state": "re-read primary, existing source ID" } ] }, { "id": "clark-janus", "name": "Graeme Clark / Janus Henderson", "side": "buy", "role": "买方Global Technology Leaders基金经理；2026文章具名作者", "scope": "AI替代叙事 → 工作流依赖 → 可收费收入", "focus": [ "企业软件", "付费席位", "数据与工作流", "反向证据" ], "boundary": "一份个人文章与一份团队材料；席位变现是待检验假设，不能当成保证。", "timeline": [ { "date": "2025-04-09", "text": "共同关注经济放缓的间接需求影响。", "source_ids": [ "O8BAS02" ] }, { "date": "2026-02-04", "text": "个人文章区分席位、核心系统和附加功能被替代的机制。", "source_ids": [ "O8BAS01" ] } ], "gaps": [ "AI代理实际收费模式需逐公司核验", "核心系统难替代不等于旧收入安全", "市场下跌的因果未独立验证" ], "comparison": "与Polar的软件替代忧虑形成同变量的相反假设。", "method_status": "method-sample; not certified recurring pattern", "source_evidence": [ { "source_id": "O8BAS01", "claim": "个人文章区分AI席位、核心系统迁移和模块收入，认为企业上下文仍重要。", "locator": "What are our views on the three main concerns；What does this mean", "read_scope": "完整正文，不验证文中第三方产品性能", "evidence_state": "newly-read primary" }, { "source_id": "O8BAS02", "claim": "共同讨论间接需求风险。", "locator": "Considerations 2–3", "read_scope": "完整分析正文及共同作者区", "evidence_state": "newly-read primary" } ] }, { "id": "narang-columbia", "name": "Rahul Narang / Columbia Threadneedle", "side": "buy", "role": "买方科技组合经理；2021、2023圆桌来源时角色，现职未重核", "scope": "终端行为变化 → 需求结构 → 技术采用", "focus": [ "需求错判", "库存与终端", "平台转换", "商业化" ], "boundary": "两场具名圆桌可归属到Narang自己的答复；不得挪用Wick或Devgan回答。", "timeline": [ { "date": "2021-06", "text": "解释疫情中家庭科技需求超出原有预期。", "source_ids": [ "B100-PW1" ] }, { "date": "2023-09-27", "text": "讨论AI平台转换与企业用途；宏观增益是展望。", "source_ids": [ "B100-PW2" ] } ], "gaps": [ "2026现职未核验", "宏观增益缺公司份额桥", "没有预测绩效审计" ], "comparison": "先查真实终端行为，再把主题映射到产品和现金流。", "method_status": "method-sample; not certified recurring pattern", "source_evidence": [ { "source_id": "B100-PW1", "claim": "Narang自己回答指出疫情需求错判与科技家庭消费韧性。", "locator": "PDF printed page2 COVID answer；page3 long-term trends answer", "read_scope": "PDF页1–3文字；其他说话者未归其名下", "evidence_state": "re-read primary, existing source ID" }, { "source_id": "B100-PW2", "claim": "Narang自己回答讨论AI平台转换、企业用途与采用速度。", "locator": "How does advent compare；areas of opportunity；How fast", "read_scope": "官方Q&A正文相关Narang回答；不采用Wick关于控制权的表述", "evidence_state": "re-read primary, existing source ID" } ] }, { "id": "fish-janus", "name": "Denny Fish / Janus Henderson", "side": "buy", "role": "买方Global Technology and Innovation组合经理；2025具名材料", "scope": "AI供给建设 → 部署与使用 → 采用者利润率", "focus": [ "推理需求", "供给时滞", "客户变现", "利润改善" ], "boundary": "两份有日期的个人材料支持方法样本；自然供给约束不能证明不存在过度投资。", "timeline": [ { "date": "2025-08-21", "text": "讨论推理需求和采用者经营利润的联系。", "source_ids": [ "O8BAS03" ] }, { "date": "2025-12-02", "text": "用三类受益者及施工、电力时滞解释产业扩散。", "source_ids": [ "O8BAS04" ] } ], "gaps": [ "满载不等于高价付费使用", "需求和盈利判断未独立审计", "长期AGI叙事不是确定现金流" ], "comparison": "把供给约束和资本回报分开测试。", "method_status": "method-sample; not certified recurring pattern", "source_evidence": [ { "source_id": "O8BAS03", "claim": "Fish将推理计算、部署变现和采用者利润改善相连。", "locator": "If you build it；Enablers and early adopters", "read_scope": "正文全部分析段，图表底数未复算", "evidence_state": "newly-read primary" }, { "source_id": "O8BAS04", "claim": "Fish以三层受益者与建设、电力时滞解释AI扩散。", "locator": "官方transcript第2–10段", "read_scope": "完整嵌入文字稿，未听音频", "evidence_state": "newly-read primary" } ] }, { "id": "rizzo-trowe", "name": "Dom Rizzo / T. Rowe Price", "side": "buy", "role": "买方Global Technology Equity策略经理；2023–2024来源角色", "scope": "关键技术与结构增长 → 基本面改善 → 合理买入价格", "focus": [ "关键技术", "市场份额", "现金转换", "价格纪律" ], "boundary": "一份官方问答与一份采访主持人原始总结；未听采访音频，不能把主持人措辞当本人逐字引语。", "timeline": [ { "date": "2023-08-18", "text": "主持人总结四项框架与软硬件切换成本。", "source_ids": [ "O8BAS08" ] }, { "date": "2024-04", "text": "官方具名问答说明四项筛选及周期中的灵活性。", "source_ids": [ "O8BAS07" ] } ], "gaps": [ "第二材料为主持人原始总结、非完整访谈逐字稿", "2026现职未重新核验", "不认证跨所有时期的固定方法" ], "comparison": "把技术不可替代性与合理价格同时作为必要检验。", "method_status": "method-sample; not certified recurring pattern", "source_evidence": [ { "source_id": "O8BAS07", "claim": "四项选择框架，并承认很难四项同时完美。", "locator": "Please explain your stock selection process", "read_scope": "完整官方问答", "evidence_state": "newly-read primary" }, { "source_id": "O8BAS08", "claim": "采访主持人总结框架，并讨论GPU硬件与CUDA切换壁垒。", "locator": "Framework for Company Evaluation；Semiconductors", "read_scope": "主持人原始文章全文；无音频核对", "evidence_state": "newly-read first-party interviewer account; not verbatim transcript" } ] }, { "id": "curtis-franklin", "name": "Jonathan Curtis / Franklin Equity", "side": "buy", "role": "买方组合经理；2025访谈时称CIO，当前标题与历史角色分开", "scope": "采用者生产率 → 可兑现公司收益 → 已计价程度", "focus": [ "AI采用者", "生产率现金化", "研究工具", "价值捕获" ], "boundary": "官网访谈与官网短文支持方法样本；短文所链完整白皮书未取得。", "timeline": [ { "date": "2025-07-11", "text": "7月11日发布的官方访谈（观点截至7月1日）讨论效率增益与研究工作的人工判断。", "source_ids": [ "O8BAS09" ] }, { "date": "2025-07-24", "text": "短文把研究范围延伸到非科技采用者。", "source_ids": [ "O8BAS10" ] } ], "gaps": [ "完整白皮书未取得", "宏观生产率并非可直接销售TAM", "未核验其收益和定价判断" ], "comparison": "检验技术进步、经济价值和供应商收入之间的漏损。", "method_status": "method-sample; not certified recurring pattern", "source_evidence": [ { "source_id": "O8BAS09", "claim": "区分AI销售收入、采用者效率与金融研究工具的作用。", "locator": "Opening answers；financial services；quantify knowledge-worker impact", "read_scope": "完整官方文字稿；文末as-of July1日期", "evidence_state": "newly-read primary transcript" }, { "source_id": "O8BAS10", "claim": "将AI研究扩展到采用技术的非科技企业。", "locator": "Key takeaways；Why AI’s biggest investment opportunity is not in tech；Our conclusion", "read_scope": "官网短文全部正文；所链完整白皮书未取得", "evidence_state": "newly-read primary short article" } ] }, { "id": "O8BB-tom-slater", "name": "Tom Slater", "role": "买方；Scottish Mortgage基金经理（2024–2025来源时）", "scope": "AI采用 → 利润池 → 风险回报与资本分配", "focus": [ "技术进步与股东回报", "AI价格及成本", "企业适应能力" ], "boundary": "两篇经理报告支持历史方法样本；减持原因是作者披露，非完整交易决策或绩效归因。", "timeline": [ { "date": "2024-11", "text": "对AI变革保持信心，但指出成本与资本开支持续性限制。", "source_ids": [ "O8BB-TS24" ] }, { "date": "2025-05", "text": "进一步讨论普及与利润空间、估值不对称性，关注AI使用方；不能据此宣称成功预测芯片股价格。", "source_ids": [ "O8BB-TS25" ] } ], "gaps": [ "缺具体持仓成本、交易日期及完整估值模型", "供应商利润压缩与采用扩张必须分别验证", "未审计预测和投资业绩" ], "comparison": "样本侧重把同一技术趋势拆成供应商与使用方的不同回报路径。", "side": "buy" }, { "id": "O8BB-lawrence-burns", "name": "Lawrence Burns", "role": "买方；Scottish Mortgage副基金经理（2024–2025来源时）", "scope": "AI价值链分层 → 采用时滞 → 多业务成长与适应力", "focus": [ "硬件/基础设施/应用", "需求消化期", "第二增长业务", "经营适应力" ], "boundary": "跨两篇报告的样本，不是永久投资人格；示例企业为作者选择，存在幸存者偏差。", "timeline": [ { "date": "2024-05", "text": "分层定位AI机会，同时提示硬件周期及应用采用延迟。", "source_ids": [ "O8BB-LB24" ] }, { "date": "2025-05", "text": "从技术层次扩展到企业多次业务增长与组织适应，要求把可能价值与确定价值分开。", "source_ids": [ "O8BB-LB25" ] } ], "gaps": [ "技术参数如GPT-4规模为来源估计，不作为已验证事实", "缺未成功公司对照组及多业务概率校准", "未审计真实交易或组合收益" ], "comparison": "样本侧重机会分层与业务扩张的条件，不由技术规模直接推定估值。", "side": "buy" }, { "id": "O8BB-douglas-brodie", "name": "Douglas Brodie", "role": "买方；Baillie Gifford Discovery基金经理（2024–2025来源时）", "scope": "创新进展 → 经营兑现 → 暂时波折或结构性失效", "focus": [ "技术与商业化", "客户迁移", "执行风险", "政策与需求分解" ], "boundary": "访谈和更新仅支持历史方法样本；宏观对股价的解释仍是作者判断。", "timeline": [ { "date": "2024-07", "text": "将新产品兑现与增长空间变化区分；介绍工业应用机会及Sweetgreen新增持仓逻辑。", "source_ids": [ "O8BB-DB24" ] }, { "date": "2025-07", "text": "Sweetgreen出现需求与竞争挑战；保留这项不利证据，不把此前看好改写成一直谨慎。", "source_ids": [ "O8BB-DB25" ] } ], "gaps": [ "缺公司级模型和实时持仓", "未核验原始临床数据或公司运营指标", "来源将宏观与股价关联，因果未独立识别" ], "comparison": "样本把产品进展、组织执行和最终需求分别检验，保留前后不利变化。", "side": "buy" }, { "id": "O8BB-brett-winton", "name": "Brett Winton / ARK", "role": "买方研究；ARK具名公开评论（2022–2023来源时）", "scope": "AI技术采用 → 付费单位经济 → 融资回流与竞争", "focus": [ "免费热度与收入", "推理成本", "资金来源", "供应商融资" ], "boundary": "署名为Winton、观点用ARK机构口径；不映射为Cathie Wood个人信念，不推断实际投资决策。", "timeline": [ { "date": "2022-12-15", "text": "把AI应用流行与持久优势分开，并关注付费和推理成本。", "source_ids": [ "O8BB-BW22" ] }, { "date": "2023-07-11", "text": "加入融资回流供应商收入及竞争约束的检验；不等同于断言所有AI投资循环都不可持续。", "source_ids": [ "O8BB-BW23" ] } ], "gaps": [ "第三方收入年化和融资分类未复算", "合同限制、补贴与外部客户收入须逐案检查", "无预测准确率或基金收益审计" ], "comparison": "样本连接商业模式与资金闭环，技术乐观与融资风险可以同时成立。", "side": "buy" }, { "id": "O8BB-tasha-keeney", "name": "Tasha Keeney / ARK", "role": "买方研究；ARK自动驾驶研究与联合估值模型作者（2023–2024来源时）", "scope": "技术资格 → 付费利用率 → 单位经济 → 估值分布", "focus": [ "推出日期敏感性", "车队资格与利用率", "抽成与平台利润", "可比安全样本" ], "boundary": "2023模型为Keeney、Korus、Maguire联合研究；2024为Keeney署名。历史预测及作者假设，不是当前安全性证明。", "timeline": [ { "date": "2023-04-21", "text": "联合模型公开模拟输入、推出日期及分位数；模拟规模不证明概率校准。", "source_ids": [ "O8BB-TK23" ] }, { "date": "2024-10-09", "text": "进一步讨论兼容车辆、服务规模和付费利用率；监督驾驶与无人驾驶数据在原文注记中明确不同。", "source_ids": [ "O8BB-TK24" ] } ], "gaps": [ "未执行公开模拟或审计输入相关性", "未核验安全原始样本与监管资格", "2027及2029估值窗口未作结果审计", "2023与2024预测须保持原始时间与条件，不能用相邻成功替代" ], "comparison": "样本的价值在于公开关键假设和时点敏感性，同时需要独立挑战基率与分母。", "side": "buy" }, { "id": "qiu-gaoyi", "name": "邱国鹭", "side": "buy", "role": "高毅资产董事长（2017/2018材料时期）", "scope": "竞争格局 → 定价权 → 可持续回报", "focus": [ "竞争边界", "规模可复制性", "管理激励与现金" ], "boundary": "仅为所列时期公开材料的方法样本，不认证永久个人思想或投资业绩。", "timeline": [ { "date": "2017-03-10", "text": "从行业、公司及管理机制检验盈利质量。", "source_ids": [ "O8BCS01" ] }, { "date": "2018-06-12", "text": "进一步强调市场定义与集中度背后的定价权。", "source_ids": [ "O8BCS02" ] } ], "gaps": [ "未审计真实交易与回报", "样本未覆盖2018年之后的观点演变" ], "comparison": "编辑比较：竞争格局 → 定价权 → 可持续回报；不能据此排名投资能力。" }, { "id": "deng-gaoyi", "name": "邓晓峰", "side": "buy", "role": "高毅资产合伙人、首席投资官（2022/2023材料时期）", "scope": "产业进步 → 企业存活与盈利 → 兑现时间", "focus": [ "技术替代", "资本周期", "时间风险" ], "boundary": "仅为所列时期公开材料的方法样本，不认证永久个人思想或投资业绩。", "timeline": [ { "date": "2022-08-02", "text": "光伏案例区分产业进步与企业生存。", "source_ids": [ "O8BCS03" ] }, { "date": "2023-02-25", "text": "在电动车竞争和工业金属供需中应用阶段分析。", "source_ids": [ "O8BCS04" ] } ], "gaps": [ "2023宏观判断未进行结果审计", "演讲由机构/媒体整理，有删节" ], "comparison": "编辑比较：产业进步 → 企业存活与盈利 → 兑现时间；不能据此排名投资能力。" }, { "id": "feng-gaoyi", "name": "冯柳", "side": "buy", "role": "高毅资产董事总经理（2019材料时期）；2026官网列合伙人、资深基金经理", "scope": "风险定价 → 条件与赔率 → 组合约束", "focus": [ "估值保护", "流动性与持仓披露", "认知边界" ], "boundary": "仅为所列时期公开材料的方法样本，不认证永久个人思想或投资业绩。", "timeline": [ { "date": "2019-06-17", "text": "阐述风险定价及资金结构的约束。", "source_ids": [ "O8BCS05" ] }, { "date": "2020-05-26", "text": "澄清估值保护不等于等待重估、交易不等于观点。", "source_ids": [ "O8BCS06" ] } ], "gaps": [ "不复制本人交易规则为普适建议", "未验证完整持仓及交易效果" ], "comparison": "编辑比较：风险定价 → 条件与赔率 → 组合约束；不能据此排名投资能力。" }, { "id": "liu-igw", "name": "刘彦春", "side": "buy", "role": "景顺长城新兴成长基金经理（2025报告时期）；材料属管理人报告", "scope": "宏观量价 → 企业盈利 → 质量与估值", "focus": [ "名义需求", "复苏条件", "展望证伪" ], "boundary": "仅为所列时期公开材料的方法样本，不认证永久个人思想或投资业绩。管理人报告不等于个人独著。", "timeline": [ { "date": "2025-07-21", "text": "承认内需与物价恢复慢于预期。", "source_ids": [ "O8BCS07" ] }, { "date": "2026-03-27", "text": "提出库存、产能与资产负债表改善的恢复路径。", "source_ids": [ "O8BCS08" ] } ], "gaps": [ "未核验2026年10月现任管理职务", "复苏时间与因果仍须后续检验" ], "comparison": "编辑比较：宏观量价 → 企业盈利 → 质量与估值；不能据此排名投资能力。" }, { "id": "fu-foresight", "name": "傅鹏博 / 睿远成长价值", "side": "buy", "role": "傅鹏博、朱璘共同管理（2025报告时期）", "scope": "产业量产与良率 → 组合配置 → 资本回报验证", "focus": [ "AI产业链", "量产良率", "集中风险" ], "boundary": "仅为所列时期公开材料的方法样本，不认证永久个人思想或投资业绩。管理人报告不等于个人独著。", "timeline": [ { "date": "2025-08-27", "text": "从产品量产及制造改善解释产业链表现。", "source_ids": [ "O8BCS09" ] }, { "date": "2026-03-27", "text": "AI配置提高，同时保留对海外资本开支回报的疑问。", "source_ids": [ "O8BCS10" ] } ], "gaps": [ "机构及共同管理观点不能归为傅鹏博独著", "未检验实际交易级归因或长期预测结果" ], "comparison": "编辑比较：产业量产与良率 → 组合配置 → 资本回报验证；不能据此排名投资能力。" }, { "id": "omalley-barclays", "name": "Tom O’Malley / Barclays", "side": "sell", "role": "卖方分析师；机构按所读电话会当时名录记录，2026-10现职未在本批重新认证。", "scope": "CPU与GPU增长 → 时点和供给约束 → 组合毛利", "focus": [ "CPU/GPU收入拆分", "季度兑现节奏", "毛利组合" ], "boundary": "两场具名公开问答的方法样本；不是许可研报或私人思想，也未认证同机制三次跨两场的反复模式。", "timeline": [ { "date": "2026-05-05", "text": "问CPU/GPU季度节奏与供给、电力限制；管理层澄清AI收入因中国业务环比下降，给出Helios爬坡节奏。", "source_ids": [ "O8SAD1" ] }, { "date": "2026-08-04", "text": "问CPU/GPU收入交叉时点与利润组合；管理层强调两者增长及不同毛利贡献，没有给确定交叉日期。", "source_ids": [ "O8SAD2" ] } ], "gaps": [ "未审阅许可研究报告或完整预测模型", "当前任职和收益/预测命中率未在本批审计", "管理层答案需要独立客户与经营数据验证" ], "comparison": "仅比较选定问题的方法与证据边界，不按声望排名。" }, { "id": "buchalter-td-cowen", "name": "Joshua Buchalter / TD Cowen", "side": "sell", "role": "卖方分析师；机构按所读电话会当时名录记录，2026-10现职未在本批重新认证。", "scope": "市场扩张 → 数量/内容/价格 → 需求与良率证据", "focus": [ "CPU市场量价", "核心数与ASP", "需求与良率分离" ], "boundary": "两场具名公开问答的方法样本；不是许可研报或私人思想，也未认证同机制三次跨两场的反复模式。", "timeline": [ { "date": "2026-05-05", "text": "追问CPU市场预测变化、竞争份额和MI450客户范围；管理层给出工作负载、客户预测及推理部署解释。", "source_ids": [ "O8SAD1" ] }, { "date": "2026-08-04", "text": "问2027 CPU增长的量价及核心数代理，并区分Helios超计划是需求还是良率；管理层澄清指需求量，良率改善为后续预期。", "source_ids": [ "O8SAD2" ] } ], "gaps": [ "未审阅许可研究报告或完整预测模型", "当前任职和收益/预测命中率未在本批审计", "管理层答案需要独立客户与经营数据验证" ], "comparison": "仅比较选定问题的方法与证据边界，不按声望排名。" }, { "id": "malik-citi", "name": "Atif Malik / Citi", "side": "sell", "role": "卖方分析师；机构按所读电话会当时名录记录，2026-10现职未在本批重新认证。", "scope": "推理架构 → 软件与合作模式 → 净经济价值", "focus": [ "推理软件护城河", "分离式推理", "费用经营杠杆" ], "boundary": "两场具名公开问答的方法样本；不是许可研报或私人思想，也未认证同机制三次跨两场的反复模式。", "timeline": [ { "date": "2026-02-25", "text": "问推理投资增加后CUDA的重要性；管理层给出软件栈及系统性能论证，非分析师背书。", "source_ids": [ "S1D01" ] }, { "date": "2026-08-04", "text": "问Cerebras合作的快速推理收入前景与费用增速；管理层给出产品可用时间和费用慢于收入增长的方向，未量化收入。", "source_ids": [ "O8SAD2" ] } ], "gaps": [ "未审阅许可研究报告或完整预测模型", "当前任职和收益/预测命中率未在本批审计", "管理层答案需要独立客户与经营数据验证" ], "comparison": "仅比较选定问题的方法与证据边界，不按声望排名。" }, { "id": "reitzes-melius", "name": "Ben Reitzes / Melius Research", "side": "sell", "role": "卖方分析师；机构按所读电话会当时名录记录，2026-10现职未在本批重新认证。", "scope": "收入运行率 → 许可与兑现 → 技术价值和利润可持续性", "focus": [ "收入增量归因", "许可到交付时滞", "性能价值与毛利" ], "boundary": "两场具名公开问答的方法样本；不是许可研报或私人思想，也未认证同机制三次跨两场的反复模式。", "timeline": [ { "date": "2025-08-05", "text": "追问季度增量是否主要GPU及中国许可恢复后的爬坡；管理层确认MI355为主要驱动，但提示许可后仍需时间。", "source_ids": [ "O8SAD3" ] }, { "date": "2026-02-25", "text": "追问高毛利可持续性与内存创新；管理层将答案连接代际客户价值，没有提供保证毛利的成本桥。", "source_ids": [ "S1D01" ] } ], "gaps": [ "未审阅许可研究报告或完整预测模型", "当前任职和收益/预测命中率未在本批审计", "管理层答案需要独立客户与经营数据验证" ], "comparison": "仅比较选定问题的方法与证据边界，不按声望排名。" }, { "id": "kumar-piper", "name": "Harsh Kumar / Piper Sandler", "side": "sell", "role": "按来源当时机构署名记录的卖方分析师；本批不更新现职认证。", "scope": "单芯片能力 → 系统集成与客户采用 → 利润爬坡", "focus": [ "机架级差异", "客户采用与信任", "收入增长与毛利" ], "boundary": "具名公开问答方法样本；明确保留历史日期，不把管理层论证当分析师本人预测。", "timeline": [ { "date": "2024-07-30", "text": "问Instinct机架级差距和UALink的补齐时点。 管理层谈互连、网络和系统投入，没有给差距关闭的确定日期。", "source_ids": [ "O8SAD4" ] }, { "date": "2024-10-29", "text": "问客户采用障碍及Instinct毛利达到公司水平的条件。 Lisa Su强调软件熟悉度和多代客户信任，Jean Hu给出规模改善方向，未给到达日期。", "source_ids": [ "O8SAD5" ] } ], "gaps": [ "历史两场公开问答不证明2026当前投资观点", "未审阅许可研报或完整模型，未审计预测业绩", "跨议题样本，未认证同机制的稳定反复模式" ], "comparison": "比较问题如何识别约束，不作能力或业绩排名。" }, { "id": "schafer-oppenheimer", "name": "Rick Schafer / Oppenheimer", "side": "sell", "role": "按来源当时机构署名记录的卖方分析师；本批不更新现职认证。", "scope": "供给与产品组合 → 可服务市场 → 工艺和利润约束", "focus": [ "供给可替代性", "TAM与SAM", "工艺迁移经济性" ], "boundary": "具名公开问答方法样本；明确保留历史日期，不把管理层论证当分析师本人预测。", "timeline": [ { "date": "2022-08-09", "text": "问消费市场放缓释放的晶圆与基板供应能多快帮助MaxLinear，及订单可见度。 管理层称供应改善仍取决于节点与厂商，较长交期要求相应backlog；这不是已兑现收入。", "source_ids": [ "O8SAD7" ] }, { "date": "2024-08-14", "text": "问PC可服务机会及Arm/x86影响，再问22nm转移对毛利和ASP的作用。 管理层区分可服务高端市场与总量，指出新节点成本、良率和模拟电路缩放限制，不提供通用提价规则。", "source_ids": [ "O8SAD6" ] } ], "gaps": [ "历史两场公开问答不证明2026当前投资观点", "未审阅许可研报或完整模型，未审计预测业绩", "跨议题样本，未认证同机制的稳定反复模式" ], "comparison": "比较问题如何识别约束，不作能力或业绩排名。" }, { "id": "bruce-lu-gs", "nomination_id": "S38", "name": "Bruce Lu", "side": "sell", "role": "来源时卖方；Goldman Sachs（2025问答）", "scope": "制程迁移、产能上限与token需求的收入传导", "focus": [ "新节点量价拆分", "节点间产能转换", "token与收入增速边界" ], "boundary": "两次具名公开问答的方法样本，非完整个人投资思想或反复方法认证；管理层回答不归为分析师观点。", "timeline": [ { "date": "2025-07-17", "text": "问新节点爬坡与旧节点供需。", "source_ids": [ "S36SRC1" ], "locator": "印刷页9–10，Bruce Lu" }, { "date": "2025-10-16", "text": "追问token增长如何传到代工收入。", "source_ids": [ "S36SRC2" ], "locator": "印刷页11–13，Bruce Lu" } ], "gaps": [ "仅研究两次公开问答样本；未取得完整证券研报、估值模型或底层渠道数据。", "未审计预测结果或投资绩效；2025来源角色不等于2026现职核验。" ], "comparison": "将提问转为可检验模型是编辑解释；不构成分析师能力排名。", "evidence_status": "method-sample" }, { "id": "gokul-hariharan-jpm", "nomination_id": "S39", "name": "Gokul Hariharan", "side": "sell", "role": "来源时卖方；J.P. Morgan（2025问答）", "scope": "需求信号、先进封装瓶颈与资本强度", "focus": [ "供需缺口", "指引反推", "CapEx收入比较" ], "boundary": "两次具名公开问答的方法样本，非完整个人投资思想或反复方法认证；管理层回答不归为分析师观点。", "timeline": [ { "date": "2025-07-17", "text": "问CoWoS供需与指引隐含的季度变化。", "source_ids": [ "S36SRC1" ], "locator": "印刷页6–8，Gokul Hariharan" }, { "date": "2025-10-16", "text": "追问AI增长与资本开支、封装扩产的关系。", "source_ids": [ "S36SRC2" ], "locator": "印刷页6–7，Gokul Hariharan" } ], "gaps": [ "仅研究两次公开问答样本；未取得完整证券研报、估值模型或底层渠道数据。", "未审计预测结果或投资绩效；2025来源角色不等于2026现职核验。" ], "comparison": "将提问转为可检验模型是编辑解释；不构成分析师能力排名。", "evidence_status": "method-sample" }, { "id": "karl-keirstead-ubs", "nomination_id": "S53", "name": "Karl Keirstead", "side": "sell", "role": "来源时卖方；UBS（2025问答）", "scope": "云增长归因与伙伴资本责任", "focus": [ "训练与推理投资", "需求与供给提前", "增长质量" ], "boundary": "两次具名公开问答的方法样本，非完整个人投资思想或反复方法认证；管理层回答不归为分析师观点。", "timeline": [ { "date": "2025-01-29", "text": "问OpenAI关系变化的资本开支含义。", "source_ids": [ "S51SRC1" ], "locator": "HTML Q&A：KARL KEIRSTEAD, UBS" }, { "date": "2025-04-30", "text": "问AI增长贡献来源；Hood纠正超预期归因。", "source_ids": [ "O8SB-MS3" ], "locator": "HTML Q&A：KARL KEIRSTEAD及AMY HOOD回复" } ], "gaps": [ "仅研究两次公开问答样本；未取得完整证券研报、估值模型或底层渠道数据。", "未审计预测结果或投资绩效；2025来源角色不等于2026现职核验。" ], "comparison": "将提问转为可检验模型是编辑解释；不构成分析师能力排名。", "evidence_status": "method-sample" }, { "id": "brent-thill-jefferies", "nomination_id": "S56", "name": "Brent Thill", "side": "sell", "role": "来源时卖方；Jefferies（2025问答）", "scope": "效率下降价与云工作负载扩散", "focus": [ "推理成本曲线", "云迁移", "需求弹性" ], "boundary": "两次具名公开问答的方法样本，非完整个人投资思想或反复方法认证；管理层回答不归为分析师观点。", "timeline": [ { "date": "2025-01-29", "text": "问更低成本能否推动AI规模化。", "source_ids": [ "S51SRC1" ], "locator": "HTML Q&A：BRENT THILL, Jefferies" }, { "date": "2025-04-30", "text": "问云迁移加速的具体驱动。", "source_ids": [ "O8SB-MS3" ], "locator": "HTML Q&A：BRENT THILL及SATYA NADELLA回复" } ], "gaps": [ "仅研究两次公开问答样本；未取得完整证券研报、估值模型或底层渠道数据。", "未审计预测结果或投资绩效；2025来源角色不等于2026现职核验。" ], "comparison": "将提问转为可检验模型是编辑解释；不构成分析师能力排名。", "evidence_status": "method-sample" }, { "id": "kash-rangan-historical", "nomination_id": "S57", "name": "Kash Rangan", "side": "sell", "role": "历史卖方；Former Goldman Sachs（2025问答）", "scope": "收入增长与资本效率的跨期检验", "focus": [ "CapEx增速与水平", "存量资产效率", "合同到现金" ], "boundary": "两次具名公开问答的方法样本，非完整个人投资思想或反复方法认证；管理层回答不归为分析师观点。", "timeline": [ { "date": "2025-04-30", "text": "问云收入加速能否伴随CapEx增速下降。", "source_ids": [ "O8SB-MS3" ], "locator": "HTML Q&A：KASH RANGAN, Goldman Sachs" }, { "date": "2025-07-30", "text": "追问资本支出与收入曲线的后续关系。", "source_ids": [ "S51SRC2" ], "locator": "HTML Q&A：KASH RANGAN及AMY HOOD回复" } ], "gaps": [ "仅研究两次公开问答样本；未取得完整证券研报、估值模型或底层渠道数据。", "未审计预测结果或投资绩效；2025来源角色不等于2026现职核验。", "保留退休/历史卖方标签；本批没有重新确认退休日期。" ], "comparison": "将提问转为可检验模型是编辑解释；不构成分析师能力排名。", "evidence_status": "method-sample" }, { "id": "brian-nowak-ms", "nomination_id": "S67", "name": "Brian Nowak", "side": "sell", "role": "来源时卖方；Morgan Stanley（2025问答）", "scope": "技术瓶颈与用户行为验证", "focus": [ "模型能力", "重复使用", "推荐系统回报" ], "boundary": "两次具名公开问答的方法样本，非完整个人投资思想或反复方法认证；管理层回答不归为分析师观点。", "timeline": [ { "date": "2025-04-30", "text": "问模型进展与用户重复使用行为。", "source_ids": [ "O8SB-META1" ], "locator": "印刷页9–11，Brian Nowak" }, { "date": "2025-07-30", "text": "问未来技术约束及核心平台参与度。", "source_ids": [ "O8SB-META2" ], "locator": "印刷页12–13，Brian Nowak" } ], "gaps": [ "仅研究两次公开问答样本；未取得完整证券研报、估值模型或底层渠道数据。", "未审计预测结果或投资绩效；2025来源角色不等于2026现职核验。" ], "comparison": "将提问转为可检验模型是编辑解释；不构成分析师能力排名。", "evidence_status": "method-sample" }, { "id": "eric-sheridan-gs", "nomination_id": "S68", "name": "Eric Sheridan", "side": "sell", "role": "来源时卖方；Goldman Sachs（2025问答）", "scope": "产品分发选择与AI投入的财务传导", "focus": [ "独立应用定位", "人才与计算投资", "折旧时点" ], "boundary": "两次具名公开问答的方法样本，非完整个人投资思想或反复方法认证；管理层回答不归为分析师观点。", "timeline": [ { "date": "2025-04-30", "text": "问独立应用的效用、频次与竞争位置。", "source_ids": [ "O8SB-META1" ], "locator": "印刷页11，Eric Sheridan" }, { "date": "2025-07-30", "text": "问技术学习如何改变人才与算力投入。", "source_ids": [ "O8SB-META2" ], "locator": "印刷页9–12，Eric Sheridan" } ], "gaps": [ "仅研究两次公开问答样本；未取得完整证券研报、估值模型或底层渠道数据。", "未审计预测结果或投资绩效；2025来源角色不等于2026现职核验。" ], "comparison": "将提问转为可检验模型是编辑解释；不构成分析师能力排名。", "evidence_status": "method-sample" }, { "id": "doug-anmuth-jpm", "nomination_id": "S69", "name": "Doug Anmuth", "side": "sell", "role": "来源时卖方；J.P. Morgan（2025问答）", "scope": "基础设施支出拆分与合作融资边界", "focus": [ "成本与交付", "自建与合作", "资金义务" ], "boundary": "两次具名公开问答的方法样本，非完整个人投资思想或反复方法认证；管理层回答不归为分析师观点。", "timeline": [ { "date": "2025-04-30", "text": "问硬件成本、建造投入及伙伴分担。", "source_ids": [ "O8SB-META1" ], "locator": "印刷页12–13，Douglas Anmuth" }, { "date": "2025-07-30", "text": "再问外部融资及开源策略。", "source_ids": [ "O8SB-META2" ], "locator": "印刷页13–14，Douglas Anmuth" } ], "gaps": [ "仅研究两次公开问答样本；未取得完整证券研报、估值模型或底层渠道数据。", "未审计预测结果或投资绩效；2025来源角色不等于2026现职核验。" ], "comparison": "将提问转为可检验模型是编辑解释；不构成分析师能力排名。", "evidence_status": "method-sample" }, { "id": "justin-post-bofa", "nomination_id": "S70", "name": "Justin Post", "side": "sell", "role": "来源时卖方；Bank of America / Merrill Lynch（2025问答）", "scope": "内部算力使用与投资回报的口径", "focus": [ "内部收益", "外售机会成本", "分阶段投入" ], "boundary": "两次具名公开问答的方法样本，非完整个人投资思想或反复方法认证；管理层回答不归为分析师观点。", "timeline": [ { "date": "2025-04-30", "text": "问没有企业云收入时的资本回报。", "source_ids": [ "O8SB-META1" ], "locator": "印刷页12，Justin Post" }, { "date": "2025-07-30", "text": "追问内部与外部使用及ROI。", "source_ids": [ "O8SB-META2" ], "locator": "印刷页14–16，Justin Post" } ], "gaps": [ "仅研究两次公开问答样本；未取得完整证券研报、估值模型或底层渠道数据。", "未审计预测结果或投资绩效；2025来源角色不等于2026现职核验。" ], "comparison": "将提问转为可检验模型是编辑解释；不构成分析师能力排名。", "evidence_status": "method-sample" }, { "id": "meta-marshall", "name": "Meta Marshall", "side": "sell", "role": "Morgan Stanley分析师（下列电话会时身份；不是本轮现职认证）", "scope": "客户季节性；毛利率改善来源", "focus": [ "客户季节性", "毛利率改善来源" ], "boundary": "跨两份实质材料的公开提问样本；不认证永久个人方法，管理层答复不是分析师预测。", "timeline": [ { "date": "2025-03-11", "text": "检查云客户季节性与关税假设。", "source_ids": [ "O8SC-CIENQ125" ] }, { "date": "2025-09-04", "text": "区分产品组合、规模降本与关税对毛利的贡献。", "source_ids": [ "O8SC-CIENQ325" ] } ], "gaps": [ "未取得个人付费研究与完整估值模型", "未检验预测结果或投资业绩", "选读历史电话会，不代表全部覆盖或当前判断" ], "comparison": "客户季节性；毛利率改善来源" }, { "id": "simon-leopold", "name": "Simon Leopold", "side": "sell", "role": "Raymond James分析师（下列电话会时身份；不是本轮现职认证）", "scope": "供应链迁移时滞；DCI用途和增量内容", "focus": [ "供应链迁移时滞", "DCI用途和增量内容" ], "boundary": "跨两份实质材料的公开提问样本；不认证永久个人方法，管理层答复不是分析师预测。", "timeline": [ { "date": "2025-03-11", "text": "问制造地调整、关税敞口及供应链迁移。", "source_ids": [ "O8SC-CIENQ125" ] }, { "date": "2025-09-04", "text": "区分新型训练互连与传统DCI，追问路由增长来源。", "source_ids": [ "O8SC-CIENQ325" ] } ], "gaps": [ "未取得个人付费研究与完整估值模型", "未检验预测结果或投资业绩", "选读历史电话会，不代表全部覆盖或当前判断" ], "comparison": "供应链迁移时滞；DCI用途和增量内容" }, { "id": "amit-daryanani", "name": "Amit Daryanani", "side": "sell", "role": "Evercore分析师（下列电话会时身份；不是本轮现职认证）", "scope": "非经常性毛利；分客户增长一致性", "focus": [ "非经常性毛利", "分客户增长一致性" ], "boundary": "跨两份实质材料的公开提问样本；不认证永久个人方法，管理层答复不是分析师预测。", "timeline": [ { "date": "2025-03-11", "text": "追查一次性毛利收益及订单强而短期收入温和的原因。", "source_ids": [ "O8SC-CIENQ125" ] }, { "date": "2025-09-04", "text": "检验云加速与总体指引是否隐含其他业务减速。", "source_ids": [ "O8SC-CIENQ325" ] } ], "gaps": [ "未取得个人付费研究与完整估值模型", "未检验预测结果或投资业绩", "选读历史电话会，不代表全部覆盖或当前判断" ], "comparison": "非经常性毛利；分客户增长一致性" }, { "id": "andy-kaplowitz", "name": "Andy Kaplowitz", "side": "sell", "role": "Citigroup分析师（下列电话会时身份；不是本轮现职认证）", "scope": "增量利润率；液冷设计与制冷内容", "focus": [ "增量利润率", "液冷设计与制冷内容" ], "boundary": "跨两份实质材料的公开提问样本；不认证永久个人方法，管理层答复不是分析师预测。", "timeline": [ { "date": "2025-10-30", "text": "检查订单加速、收入节奏与渠道库存。", "source_ids": [ "O8SC-TTQ325" ] }, { "date": "2026-01-29", "text": "追问有机增量利润率、价格成本及未来制冷架构。", "source_ids": [ "EXP200-S92-B" ] } ], "gaps": [ "未取得个人付费研究与完整估值模型", "未检验预测结果或投资业绩", "选读历史电话会，不代表全部覆盖或当前判断" ], "comparison": "增量利润率；液冷设计与制冷内容" }, { "id": "scott-davis", "name": "Scott Davis", "side": "sell", "role": "Melius Research分析师（下列电话会时身份；不是本轮现职认证）", "scope": "订单兑现窗口；增长广度", "focus": [ "订单兑现窗口", "增长广度" ], "boundary": "跨两份实质材料的公开提问样本；不认证永久个人方法，管理层答复不是分析师预测。", "timeline": [ { "date": "2025-10-30", "text": "问订单交期是否延至后年，并追问参考设计新增内容。", "source_ids": [ "O8SC-TTQ325" ] }, { "date": "2026-01-29", "text": "问订单增长是扩散到多个行业还是集中在数据中心。", "source_ids": [ "EXP200-S92-B" ] } ], "gaps": [ "未取得个人付费研究与完整估值模型", "未检验预测结果或投资业绩", "选读历史电话会，不代表全部覆盖或当前判断" ], "comparison": "订单兑现窗口；增长广度" }, { "id": "jeff-sprague", "name": "Jeff Sprague", "side": "sell", "role": "Vertical Research Partners分析师（下列电话会时身份；不是本轮现职认证）", "scope": "施工就绪与发货；成本转嫁时滞", "focus": [ "施工就绪与发货", "成本转嫁时滞" ], "boundary": "跨两份实质材料的公开提问样本；不认证永久个人方法，管理层答复不是分析师预测。", "timeline": [ { "date": "2025-10-30", "text": "问延期是项目噪声还是施工系统瓶颈。", "source_ids": [ "O8SC-TTQ325" ] }, { "date": "2026-01-29", "text": "问关税跨年影响、金属成本和调价速度。", "source_ids": [ "EXP200-S92-B" ] } ], "gaps": [ "未取得个人付费研究与完整估值模型", "未检验预测结果或投资业绩", "选读历史电话会，不代表全部覆盖或当前判断" ], "comparison": "施工就绪与发货；成本转嫁时滞" }, { "id": "joe-ritchie", "name": "Joe Ritchie", "side": "sell", "role": "Goldman Sachs分析师（下列电话会时身份；不是本轮现职认证）", "scope": "模块化施工；区域项目规模", "focus": [ "模块化施工", "区域项目规模" ], "boundary": "跨两份实质材料的公开提问样本；不认证永久个人方法，管理层答复不是分析师预测。", "timeline": [ { "date": "2025-10-30", "text": "问模块化项目变化与不同供应链交期。", "source_ids": [ "O8SC-TTQ325" ] }, { "date": "2026-01-29", "text": "问欧洲数据中心机会及区域配置。", "source_ids": [ "EXP200-S92-B" ] } ], "gaps": [ "未取得个人付费研究与完整估值模型", "未检验预测结果或投资业绩", "选读历史电话会，不代表全部覆盖或当前判断" ], "comparison": "模块化施工；区域项目规模" }, { "id": "stephen-tusa", "name": "Stephen Tusa", "side": "sell", "role": "JPMorgan分析师（下列电话会时身份；不是本轮现职认证）", "scope": "统计口径追问；订单与收入时点", "focus": [ "统计口径追问", "订单与收入时点" ], "boundary": "跨两份实质材料的公开提问样本；不认证永久个人方法，管理层答复不是分析师预测。", "timeline": [ { "date": "2025-10-30", "text": "追问积压同比基数与渠道进销差异。", "source_ids": [ "O8SC-TTQ325" ] }, { "date": "2026-01-29", "text": "要求子项与总体订单增长对账，并区分订单和收入。", "source_ids": [ "EXP200-S92-B" ] } ], "gaps": [ "未取得个人付费研究与完整估值模型", "未检验预测结果或投资业绩", "选读历史电话会，不代表全部覆盖或当前判断" ], "comparison": "统计口径追问；订单与收入时点" }, { "id": "nicole-deblase", "name": "Nicole DeBlase", "side": "sell", "role": "Deutsche Bank分析师（下列电话会时身份；不是本轮现职认证）", "scope": "核心业务与AI增量；服务附着", "focus": [ "核心业务与AI增量", "服务附着" ], "boundary": "跨两份实质材料的公开提问样本；不认证永久个人方法，管理层答复不是分析师预测。", "timeline": [ { "date": "2025-08-01", "text": "追问非数据中心订单与液冷服务能力。", "source_ids": [ "O8SC-NVTQ225" ] }, { "date": "2025-10-30", "text": "检查欧洲订单驱动及住宅价格、组合、销量。", "source_ids": [ "O8SC-TTQ325" ] } ], "gaps": [ "未取得个人付费研究与完整估值模型", "未检验预测结果或投资业绩", "选读历史电话会，不代表全部覆盖或当前判断" ], "comparison": "核心业务与AI增量；服务附着" }, { "id": "deane-dray", "name": "Deane Dray", "side": "sell", "role": "RBC Capital Markets分析师（下列电话会时身份；不是本轮现职认证）", "scope": "客户自研与价值捕获；地区异质性", "focus": [ "客户自研与价值捕获", "地区异质性" ], "boundary": "跨两份实质材料的公开提问样本；不认证永久个人方法，管理层答复不是分析师预测。", "timeline": [ { "date": "2025-08-01", "text": "追问积压转化及客户自研液冷是否绕过供应商。", "source_ids": [ "O8SC-NVTQ225" ] }, { "date": "2025-10-30", "text": "质疑地区项目规模概括，另问液冷合作。", "source_ids": [ "O8SC-TTQ325" ] } ], "gaps": [ "未取得个人付费研究与完整估值模型", "未检验预测结果或投资业绩", "选读历史电话会，不代表全部覆盖或当前判断" ], "comparison": "客户自研与价值捕获；地区异质性" }, { "id": "nick-evans", "name": "Nick Evans", "side": "buy", "role": "Polar Capital Partner and Fund Manager（2024/2026访谈时）", "scope": "AI利润池迁移；软件收入与现金成本", "focus": [ "AI利润池迁移", "软件收入与现金成本" ], "boundary": "两份公开实质材料的方法样本；合著只认证参与团队材料，不认证个人稳定哲学或投资业绩。", "timeline": [ { "date": "2024-05", "text": "区分AI基础设施、受益者与采用者；看企业之间增长差异而非统一大盘标签。", "source_ids": [ "R2B1-EV24" ] }, { "date": "2026-01", "text": "以AI重新划分利润池，关注软件替代风险、基础设施需求与资本开支对自由现金流的影响。", "source_ids": [ "R2B1-EV26" ] } ], "gaps": [ "未取得完整持仓、交易或模型；未审计预测结果及风险调整业绩", "来源身份与现职分列；历史职务不外推至2026年10月" ], "comparison": "AI利润池迁移；软件收入与现金成本" }, { "id": "reid-menge", "name": "Reid Menge", "side": "buy", "role": "BlackRock基金经理；Q2 2026共同管理评论和2026访谈时身份", "scope": "技术成本与采用；时间价值及工程风险", "focus": [ "技术成本与采用", "时间价值及工程风险" ], "boundary": "两份公开实质材料的方法样本；合著只认证参与团队材料，不认证个人稳定哲学或投资业绩。", "timeline": [ { "date": "2026-06-30", "text": "团队关注推理效率、代理工作负载和可度量资本回报，偏好已显现商业化及持久优势。", "source_ids": [ "R2B1-ME26Q2" ] }, { "date": "2026-09-25", "text": "把成本下降、能力提升和需求结合，同时强调发射执行、太空环境和地面通电时间；太空算力判断属预测。", "source_ids": [ "R2B1-ME26SP" ] } ], "gaps": [ "未取得完整持仓、交易或模型；未审计预测结果及风险调整业绩", "来源身份与现职分列；历史职务不外推至2026年10月" ], "comparison": "技术成本与采用；时间价值及工程风险" }, { "id": "hyun-ho-sohn", "name": "Hyun Ho Sohn", "side": "buy", "role": "Fidelity International Portfolio Manager（2025展望及2026问答时）", "scope": "企业采用摩擦；AI资本回报", "focus": [ "企业采用摩擦", "AI资本回报" ], "boundary": "两份公开实质材料的方法样本；合著只认证参与团队材料，不认证个人稳定哲学或投资业绩。", "timeline": [ { "date": "2025", "text": "担心资本密集度、外部融资与半导体估值透支周期，要求区分基础设施增长和投资回报。", "source_ids": [ "R2B1-SO26" ] }, { "date": "2026-04-05", "text": "指出企业软件工作流、数据、安全与集成形成阻力；用管理层、专家和内部IT调研检验AI替代速度。", "source_ids": [ "R2B1-SO26Q" ] } ], "gaps": [ "未取得完整持仓、交易或模型；未审计预测结果及风险调整业绩", "来源身份与现职分列；历史职务不外推至2026年10月" ], "comparison": "企业采用摩擦；AI资本回报" }, { "id": "ken-allen", "name": "Ken Allen", "side": "buy", "role": "T. Rowe Price Science & Technology经理（2017/2018材料时）；当前large-cap value analyst身份沿用已核传记", "scope": "持续现金流与估值；云规模优势", "focus": [ "持续现金流与估值", "云规模优势" ], "boundary": "两份公开实质材料的方法样本；合著只认证参与团队材料，不认证个人稳定哲学或投资业绩。", "timeline": [ { "date": "2018-07", "text": "重视持久业务、合理价格和中长期现金；看云规模优势同时保留本地部署需求。", "source_ids": [ "R2B1-KE18" ] }, { "date": "2017-12", "text": "Allen强调云分发以及数据、算力和人才组合；其他经理的判断不归给Allen。", "source_ids": [ "R2B1-KE17" ] } ], "gaps": [ "未取得完整持仓、交易或模型；未审计预测结果及风险调整业绩", "来源身份与现职分列；历史职务不外推至2026年10月" ], "comparison": "持续现金流与估值；云规模优势" }, { "id": "alan-tu", "name": "Alan Tu", "side": "buy", "role": "T. Rowe Price Global Technology Fund经理（2021/2022材料时）；不认证现职", "scope": "长增长跑道；产品分发与基础设施壁垒", "focus": [ "长增长跑道", "产品分发与基础设施壁垒" ], "boundary": "两份公开实质材料的方法样本；合著只认证参与团队材料，不认证个人稳定哲学或投资业绩。", "timeline": [ { "date": "2022-06", "text": "从产品进入市场方式、客户扩张和竞争优势寻找长增长跑道，同时要求估值纪律。", "source_ids": [ "R2B1-TU22" ] }, { "date": "2021-03", "text": "团队从软件、云、半导体设备与安全切入数字化基础设施，讨论竞争壁垒、增长基数及估值。", "source_ids": [ "R2B1-TU21" ] } ], "gaps": [ "未取得完整持仓、交易或模型；未审计预测结果及风险调整业绩", "来源身份与现职分列；历史职务不外推至2026年10月" ], "comparison": "长增长跑道；产品分发与基础设施壁垒" }, { "id": "kyle-weaver", "name": "Kyle Weaver", "side": "buy", "role": "Fidelity Growth Opportunities共同基金经理（2026材料时）", "scope": "AI采用者；工时与现金兑现", "focus": [ "AI采用者", "工时与现金兑现" ], "boundary": "两份公开实质材料的方法样本；合著只认证参与团队材料，不认证个人稳定哲学或投资业绩。", "timeline": [ { "date": "2026-05-04", "text": "把AI机会扩展到采用者，用客服和数字开户案例连接运营变化；案例陈述不等于已证明因果收益。", "source_ids": [ "R2B1-WE26" ] }, { "date": "2026-06-30", "text": "团队采用3–7年视角，结合现金增长、资本配置及不同增长类型，追踪真实AI使用与供给约束。", "source_ids": [ "R2B1-WE26Q2" ] } ], "gaps": [ "未取得完整持仓、交易或模型；未审计预测结果及风险调整业绩", "来源身份与现职分列；历史职务不外推至2026年10月" ], "comparison": "AI采用者；工时与现金兑现" }, { "id": "sonu-kalra", "name": "Sonu Kalra", "side": "buy", "role": "Fidelity Blue Chip Growth经理（2026 Q&A时）；2024材料为四人合著", "scope": "增长持久性；公私市场流动性", "focus": [ "增长持久性", "公私市场流动性" ], "boundary": "两份公开实质材料的方法样本；合著只认证参与团队材料，不认证个人稳定哲学或投资业绩。", "timeline": [ { "date": "2026-07-31", "text": "以竞争优势、定价能力和管理层判断增长持久性，关注AI半导体机会并保留宏观及估值风险。", "source_ids": [ "R2B1-KA26" ] }, { "date": "2024-05", "text": "团队跨公私市场观察AI价值链，区分流动性、采用速度及长期生产率兑现的不确定性。", "source_ids": [ "R2B1-KA24" ] } ], "gaps": [ "未取得完整持仓、交易或模型；未审计预测结果及风险调整业绩", "来源身份与现职分列；历史职务不外推至2026年10月" ], "comparison": "增长持久性；公私市场流动性" }, { "id": "thomas-laffont", "name": "Thomas Laffont", "side": "buy", "role": "Coatue Co-Founder and CIO of Privates（已核官网身份）；两文均具名团队共著", "scope": "企业工作流扩展；使用信号与商业回报", "focus": [ "企业工作流扩展", "使用信号与商业回报" ], "boundary": "两份公开实质材料的方法样本；合著只认证参与团队材料，不认证个人稳定哲学或投资业绩。", "timeline": [ { "date": "2026-07-20", "text": "团队以分析、交易和代理应用扩张解释增长，强调数据治理与研发周期；Coatue持有经济利益。", "source_ids": [ "R2B1-LA26D" ] }, { "date": "2026-02-12", "text": "团队用代码活动和工作流扩展支持企业采用论点；从开发能力延伸到平台价值，存在被投企业利益冲突。", "source_ids": [ "R2B1-LA26A" ] } ], "gaps": [ "未取得完整持仓、交易或模型；未审计预测结果及风险调整业绩", "来源身份与现职分列；历史职务不外推至2026年10月" ], "comparison": "企业工作流扩展；使用信号与商业回报" }, { "id": "lee-ainslie", "name": "Lee Ainslie", "side": "buy", "role": "Maverick Capital managing partner（2006及2021采访时）", "scope": "可持续现金价值；替代数据与因素风险", "focus": [ "可持续现金价值", "替代数据与因素风险" ], "boundary": "两份公开实质材料的方法样本；合著只认证参与团队材料，不认证个人稳定哲学或投资业绩。", "timeline": [ { "date": "2006-04-01", "text": "以管理层、持续增长和资本回报筛选，以可持续自由现金流/企业价值衡量价格，持仓按未来回报重审。", "source_ids": [ "R2B1-AI06" ] }, { "date": "2021-05-16", "text": "说明基本面与因素风险互补；用统计化替代数据替代过度加权的零散观察，并保留资本保护目标。", "source_ids": [ "R2B1-AI21" ] } ], "gaps": [ "未取得完整持仓、交易或模型；未审计预测结果及风险调整业绩", "来源身份与现职分列；历史职务不外推至2026年10月" ], "comparison": "可持续现金价值；替代数据与因素风险" }, { "id": "will-danoff", "name": "Will Danoff", "side": "buy", "role": "Fidelity Contrafund经理（2026年6月访谈时）；宣布年底退休不等于已离任", "scope": "单位经济性；管理层执行与持久盈利", "focus": [ "单位经济性", "管理层执行与持久盈利" ], "boundary": "两份公开实质材料的方法样本；合著只认证参与团队材料，不认证个人稳定哲学或投资业绩。", "timeline": [ { "date": "2020-01-29", "text": "强调隔半年至一年追踪管理层承诺；偏好高资本回报、现金增长和客户价值，并用新证据逐步建立信心。", "source_ids": [ "R2B1-DA20" ] }, { "date": "2026-06-23", "text": "从门店单位经济性检验扩张；将管理层许诺与后续执行对照，并要求竞争压力下的盈利记录而非短期收入火箭。", "source_ids": [ "R2B1-DA26I" ] } ], "gaps": [ "未取得完整持仓、交易或模型；未审计预测结果及风险调整业绩", "来源身份与现职分列；历史职务不外推至2026年10月" ], "comparison": "单位经济性；管理层执行与持久盈利" }, { "id": "ryan-jacob", "name": "Ryan Jacob", "side": "buy", "role": "Chief Portfolio Manager（按材料时期；非现职认证）", "scope": "平台黏性如何转成股东回报？", "focus": [ "平台黏性如何转成股东回报？" ], "boundary": "两份材料的有限方法样本；不认证永久个人方法或投资业绩。合署信只读取Ryan所管Internet Fund段，Discovery Fund观点不单独归于Ryan。", "timeline": [ { "date": "2025-10-15", "text": "Internet Fund部分讨论Circle支付基础设施与Bed Bath & Beyond的tZERO资产及重组价值，不能把整个合署信归于Ryan一人。", "source_ids": [ "R2B2-J1" ] }, { "date": "2026-01-13", "text": "Internet Fund新增Snowflake的逻辑是跨云数据层、客户黏性与AI能力；竞争与估值仍需检验。", "source_ids": [ "R2B2-J2" ] } ], "gaps": [ "合署信只读取Ryan所管Internet Fund段，Discovery Fund观点不单独归于Ryan。", "未取得完整交易、估值及投资结果审计" ], "comparison": "平台黏性如何转成股东回报？" }, { "id": "kevin-landis", "name": "Kevin Landis", "side": "buy", "role": "Founder and investment manager（按材料时期；非现职认证）", "scope": "IPO与账面价值怎样变成可用现金？", "focus": [ "IPO与账面价值怎样变成可用现金？" ], "boundary": "两份材料的有限方法样本；不认证永久个人方法或投资业绩。历史信件和不同基金产品不能拼成当前组合；本卡侧重可迁移流动性约束。", "timeline": [ { "date": "2023-06-30", "text": "报告期末对应材料（非发表日期）：信件把未配置大市值股的相对拖累与Chegg被生成式AI替代的新客风险分开。", "source_ids": [ "R2B2-L1" ], "date_basis": "period end; publication unknown" }, { "date": null, "text": "区分私企经营进度、IPO锁定、退出现金与基金NAV折价；成熟需要多年，上市不等于可兑现。", "source_ids": [ "R2B2-L2" ] } ], "gaps": [ "历史信件和不同基金产品不能拼成当前组合；本卡侧重可迁移流动性约束。", "未取得完整交易、估值及投资结果审计" ], "comparison": "IPO与账面价值怎样变成可用现金？" }, { "id": "peter-seilern", "name": "Peter Seilern", "side": "buy", "role": "Founder; retirement announced 2024-09-26（按材料时期；非现职认证）", "scope": "高ROIC与可持续再投资", "focus": [ "高ROIC与可持续再投资" ], "boundary": "两份材料的有限方法样本；不认证永久个人方法或投资业绩。一份个人交接信加无日期机构流程页；不视为两份个人独著或跨期方法认证，按退休前历史身份。", "timeline": [ { "date": "2024-09-26", "text": "退休交接信强调清晰规则、可复制流程、团队独立性与长期激励；不承诺每个周期跑赢。", "source_ids": [ "R2B2-S1" ] }, { "date": null, "text": "十项质量标准包含低资本强度、高投入资本回报和低净负债；机构以研究与集中组合实施。", "source_ids": [ "R2B2-S2" ] } ], "gaps": [ "一份个人交接信加无日期机构流程页；不视为两份个人独著或跨期方法认证，按退休前历史身份。", "未取得完整交易、估值及投资结果审计" ], "comparison": "高ROIC与可持续再投资" }, { "id": "walter-price", "name": "Walter Price", "side": "buy", "role": "Managing Director and Senior Portfolio Manager (2016)（按材料时期；非现职认证）", "scope": "采用增长与估值扩张分拆", "focus": [ "采用增长与估值扩张分拆" ], "boundary": "两份材料的有限方法样本；不认证永久个人方法或投资业绩。两份为共同负责团队的机构报告；不能分配为Walter独立观点，不验证现职。", "timeline": [ { "date": "2016-11-30", "text": "报告期末对应材料（非发表日期）：管理人报告将新技术高增长与成熟业务整合、回购分开，并以产品周期和子行业分散风险。", "source_ids": [ "R2B2-A16" ], "date_basis": "period end; publication unknown" }, { "date": "2020-12-31", "text": "报告期末对应材料（非发表日期）：管理人报告分拆疫情下远程采用、付费转换与低利率带来的估值扩张；绝对上涨与相对收益不同。", "source_ids": [ "R2B2-A20" ], "date_basis": "period end; publication unknown" } ], "gaps": [ "两份为共同负责团队的机构报告；不能分配为Walter独立观点，不验证现职。", "未取得完整交易、估值及投资结果审计" ], "comparison": "采用增长与估值扩张分拆" }, { "id": "huachen-chen", "name": "Huachen Chen", "side": "buy", "role": "Senior Portfolio Manager (2016)（按材料时期；非现职认证）", "scope": "科技生命周期分散是否真能降低风险？", "focus": [ "科技生命周期分散是否真能降低风险？" ], "boundary": "两份材料的有限方法样本；不认证永久个人方法或投资业绩。机构团队资料，Huachen为共同负责人，不将共同文本包装成其个人独有方法；历史样本。", "timeline": [ { "date": "2016-11-30", "text": "报告期末对应材料（非发表日期）：管理人报告将新技术高增长与成熟业务整合、回购分开，并以产品周期和子行业分散风险。", "source_ids": [ "R2B2-A16" ], "date_basis": "period end; publication unknown" }, { "date": "2020-12-31", "text": "报告期末对应材料（非发表日期）：管理人报告分拆疫情下远程采用、付费转换与低利率带来的估值扩张；绝对上涨与相对收益不同。", "source_ids": [ "R2B2-A20" ], "date_basis": "period end; publication unknown" } ], "gaps": [ "机构团队资料，Huachen为共同负责人，不将共同文本包装成其个人独有方法；历史样本。", "未取得完整交易、估值及投资结果审计" ], "comparison": "科技生命周期分散是否真能降低风险？" }, { "id": "anthony-ginsberg", "name": "Anthony Ginsberg", "side": "buy", "role": "CEO（按材料时期；非现职认证）", "scope": "等权主题背后的共同客户风险", "focus": [ "等权主题背后的共同客户风险" ], "boundary": "两份材料的有限方法样本；不认证永久个人方法或投资业绩。一份个人访谈加一份机构无署名展望；八到十主题变化未核定生效时间，不自动称当前规则。", "timeline": [ { "date": "2025-07", "text": "访谈以八主题各15股等权回应巨头集中；同时承认技术主题交叉与轮动难预测。", "source_ids": [ "R2B2-G1" ] }, { "date": "2026-01", "text": "机构展望以现金融资和盈利支持AI投资，并称十主题等权；与2025八主题不同，定义变更需核对。", "source_ids": [ "R2B2-G2" ] } ], "gaps": [ "一份个人访谈加一份机构无署名展望；八到十主题变化未核定生效时间，不自动称当前规则。", "未取得完整交易、估值及投资结果审计" ], "comparison": "等权主题背后的共同客户风险" }, { "id": "mark-denham", "name": "Mark Denham", "side": "buy", "role": "Head of Equities and Fund Manager（按材料时期；非现职认证）", "scope": "防御性标签不能代替需求与利润模型", "focus": [ "防御性标签不能代替需求与利润模型" ], "boundary": "两份材料的有限方法样本；不认证永久个人方法或投资业绩。跨年具名方法样本，不外推实际未来收益；个人护理不是天然无周期，原文需求预测尚待验证。", "timeline": [ { "date": "2020-05", "text": "以资本回报、再投资和经营利润率稳定性作筛选，再做基本面与ESG研究，持有期3–5年。", "source_ids": [ "R2B2-M1" ] }, { "date": "2025-02-24", "text": "将日常消费标签拆为食品与个人护理；从需求变化、有机增长与经营杠杆推导不同盈利及估值。", "source_ids": [ "R2B2-M2" ] } ], "gaps": [ "跨年具名方法样本，不外推实际未来收益；个人护理不是天然无周期，原文需求预测尚待验证。", "未取得完整交易、估值及投资结果审计" ], "comparison": "防御性标签不能代替需求与利润模型" }, { "id": "mark-hawtin", "name": "Mark Hawtin", "side": "buy", "role": "Head of Global Equities (2024 appointment)（按材料时期；非现职认证）", "scope": "正确的AI趋势为何仍可能买贵？", "focus": [ "正确的AI趋势为何仍可能买贵？" ], "boundary": "两份材料的有限方法样本；不认证永久个人方法或投资业绩。两份具名样本；技术成本原文跨模型口径未独立核实，不采其点估计；源时期角色。", "timeline": [ { "date": "2024-12", "text": "Hawtin关注指数集中和AI受益向使用者扩散；其他团队发言没有归给他。", "source_ids": [ "R2B2-H1" ] }, { "date": "2025-01-29", "text": "文章提出AI普及与基础设施股估值可分离，并说明此前削减CapEx相关敞口；成本与技术比较为作者当时主张，未独立核实。", "source_ids": [ "R2B2-H2" ] } ], "gaps": [ "两份具名样本；技术成本原文跨模型口径未独立核实，不采其点估计；源时期角色。", "未取得完整交易、估值及投资结果审计" ], "comparison": "正确的AI趋势为何仍可能买贵？" }, { "id": "david-rolfe", "name": "David Rolfe", "side": "buy", "role": "Chief Investment Officer（按材料时期；非现职认证）", "scope": "利润增长来自核心效率还是高毛利副业？", "focus": [ "利润增长来自核心效率还是高毛利副业？" ], "boundary": "两份材料的有限方法样本；不认证永久个人方法或投资业绩。两份均为三人合署客户信，不称Rolfe独著；两份均以公开原文PDF镜像读取，未核验官方字节。", "timeline": [ { "date": "2025-07", "text": "Zoetis案例从宠物医疗渗透、产品延伸研发和资本回报构造投资逻辑，同时区分疫情提升与正常化增长。", "source_ids": [ "R2B2-W2" ] }, { "date": "2026-01", "text": "Amazon新仓论证分拆广告增长与零售生产率对利润贡献，再讨论AWS硬件、资本开支与估值；分部利润不能直接当核心零售效率。", "source_ids": [ "R2B2-W4" ] } ], "gaps": [ "两份均为三人合署客户信，不称Rolfe独著；两份均以公开原文PDF镜像读取，未核验官方字节。", "未取得完整交易、估值及投资结果审计" ], "comparison": "利润增长来自核心效率还是高毛利副业？" }, { "id": "r2b3-b66", "name": "黄兴亮", "side": "buy", "role": "万家基金；资料时期身份，未借历史材料认证2026年现职", "scope": "公开投资论证 → 假设拆解 → 边界与反例 → 合成迁移", "focus": [ "科技主题与价格纪律", "风险偏好", "经营验证" ], "boundary": "两份实质公开材料支持有限方法样本，不认证永久个人方法、投资能力或预测准确率。基金管理人报告归于机构，经理表证明任职，不证明本人独著或独立完成决策。", "timeline": [ { "date": "2025-04-21", "text": "管理人解释AI应用和自主科技机会，并按价格透支及经营压力调整持仓。", "source_ids": [ "R2B3-B66C" ] }, { "date": "2026-04-21", "text": "管理人讨论风险偏好和重资产交易，调减部分生物科技并增加AI数据中心相关资产。", "source_ids": [ "R2B3-B66D" ] } ], "gaps": [ "未核查完整历史组合和真实交易决策", "没有预测命中率或投资业绩审计", "当前职务须用独立现职来源更新" ], "comparison": "编辑比较重心为科技主题与价格纪律、风险偏好、经营验证；不按名气排名。" }, { "id": "r2b3-b70", "name": "李巍", "side": "buy", "role": "广发基金；资料时期身份，未借历史材料认证2026年现职", "scope": "公开投资论证 → 假设拆解 → 边界与反例 → 合成迁移", "focus": [ "现金流质量", "资本配置", "共同风险因子" ], "boundary": "两份实质公开材料支持有限方法样本，不认证永久个人方法、投资能力或预测准确率。", "timeline": [ { "date": "2020-07-06", "text": "问答把经营现金流、资本开支、定价权和管理层资本配置连接起来，反思机械PEG。", "source_ids": [ "R2B3-B70A" ] }, { "date": "2022-09-13", "text": "第一人称反思风险虽被预见却未充分进入组合；行业分散不等于风格因子分散。", "source_ids": [ "R2B3-B70B" ] } ], "gaps": [ "未核查完整历史组合和真实交易决策", "没有预测命中率或投资业绩审计", "当前职务须用独立现职来源更新" ], "comparison": "编辑比较重心为现金流质量、资本配置、共同风险因子；不按名气排名。" }, { "id": "r2b3-b72", "name": "朱少醒", "side": "buy", "role": "富国基金；资料时期身份，未借历史材料认证2026年现职", "scope": "公开投资论证 → 假设拆解 → 边界与反例 → 合成迁移", "focus": [ "自下而上", "持有期限", "机会成本" ], "boundary": "两份实质公开材料支持有限方法样本，不认证永久个人方法、投资能力或预测准确率。", "timeline": [ { "date": "2017-12-14", "text": "问答讨论自下而上、行业均衡、三年可见度和先研究后交易；宣传业绩未审计。", "source_ids": [ "R2B3-B72A" ] }, { "date": "2012-07-09", "text": "谈价格波动与基本面差异，通过公司质量和成长寻找机会。", "source_ids": [ "R2B3-B72B" ] } ], "gaps": [ "未核查完整历史组合和真实交易决策", "没有预测命中率或投资业绩审计", "当前职务须用独立现职来源更新" ], "comparison": "编辑比较重心为自下而上、持有期限、机会成本；不按名气排名。" }, { "id": "r2b3-b75", "name": "张坤", "side": "buy", "role": "易方达基金；资料时期身份，未借历史材料认证2026年现职", "scope": "公开投资论证 → 假设拆解 → 边界与反例 → 合成迁移", "focus": [ "自由现金流", "再投资效率", "资本回报" ], "boundary": "两份实质公开材料支持有限方法样本，不认证永久个人方法、投资能力或预测准确率。基金管理人报告归于机构，经理表证明任职，不证明本人独著或独立完成决策。", "timeline": [ { "date": "2024-03-29", "text": "管理人讨论资本配置、持续高增长的低基准概率，以及现金产生能力对AI投入的支撑。", "source_ids": [ "R2B3-B75B" ] }, { "date": "2025-03-31", "text": "管理人质疑二元市场叙事，强调效率改善、自由现金流与股东回报。", "source_ids": [ "R2B3-B75C" ] } ], "gaps": [ "未核查完整历史组合和真实交易决策", "没有预测命中率或投资业绩审计", "当前职务须用独立现职来源更新" ], "comparison": "编辑比较重心为自由现金流、再投资效率、资本回报；不按名气排名。" }, { "id": "r2b3-b77", "name": "陈光明", "side": "buy", "role": "睿远基金；资料时期身份，未借历史材料认证2026年现职", "scope": "公开投资论证 → 假设拆解 → 边界与反例 → 合成迁移", "focus": [ "价值与价格", "安全边际", "优秀公司再投资" ], "boundary": "两份实质公开材料支持有限方法样本，不认证永久个人方法、投资能力或预测准确率。", "timeline": [ { "date": "2019-03-28", "text": "演讲把长期回报拆为分红、业绩增长和估值，并强调公司质量、治理及长期资金制度。", "source_ids": [ "R2B3-B77B" ] }, { "date": "2025-07-31", "text": "署名前言讨论价格与价值、安全边际及从廉价资产走向再投资能力。", "source_ids": [ "R2B3-B77A" ] } ], "gaps": [ "未核查完整历史组合和真实交易决策", "没有预测命中率或投资业绩审计", "当前职务须用独立现职来源更新" ], "comparison": "编辑比较重心为价值与价格、安全边际、优秀公司再投资；不按名气排名。" }, { "id": "r2b3-b82", "name": "Charlie Munger", "side": "buy", "role": "Wesco / Berkshire；历史副主席、投资者研究对象，非现任职务", "scope": "公开投资论证 → 假设拆解 → 边界与反例 → 合成迁移", "focus": [ "浮存金成本", "资本配置", "结构与周期" ], "boundary": "两份实质公开材料支持有限方法样本，不认证永久个人方法、投资能力或预测准确率。历史人物研究，不表示现任职务。", "timeline": [ { "date": "2010-02-26", "text": "讨论浮存金成本、承保不确定性及周期和结构压力的区别。", "source_ids": [ "R2B3-B82B" ] }, { "date": "2014", "text": "讨论分权经营、集中资本配置、留存收益检验及保守负债。", "source_ids": [ "R2B3-B82A" ] } ], "gaps": [ "未核查完整历史组合和真实交易决策", "没有预测命中率或投资业绩审计" ], "comparison": "编辑比较重心为浮存金成本、资本配置、结构与周期；不按名气排名。" }, { "id": "r2b3-b87", "name": "Joel Greenblatt", "side": "buy", "role": "Gotham Asset Management；资料时期身份，未借历史材料认证2026年现职", "scope": "公开投资论证 → 假设拆解 → 边界与反例 → 合成迁移", "focus": [ "内在价值", "正常化现金流", "多空敞口" ], "boundary": "两份实质公开材料支持有限方法样本，不认证永久个人方法、投资能力或预测准确率。", "timeline": [ { "date": "2014-03-31", "text": "访谈比较集中与分散价值策略，解释多空组合总敞口与净敞口。", "source_ids": [ "R2B3-B87A" ] }, { "date": "2019-10-06", "text": "访谈以正常化现金流和资本回报讨论价值，而非把低账面倍数等同低估。", "source_ids": [ "R2B3-B87B" ] } ], "gaps": [ "未核查完整历史组合和真实交易决策", "没有预测命中率或投资业绩审计", "当前职务须用独立现职来源更新" ], "comparison": "编辑比较重心为内在价值、正常化现金流、多空敞口；不按名气排名。" }, { "id": "r2b3-b89", "name": "Bill Miller", "side": "buy", "role": "Miller Value Partners；资料时期身份，未借历史材料认证2026年现职", "scope": "公开投资论证 → 假设拆解 → 边界与反例 → 合成迁移", "focus": [ "事实与解释", "概率与预期", "反证" ], "boundary": "两份实质公开材料支持有限方法样本，不认证永久个人方法、投资能力或预测准确率。", "timeline": [ { "date": "2021-07-09", "text": "市场信区分事实、解释与概率，讨论公开风险和已计价预期；不是风险必被充分定价的证明。", "source_ids": [ "R2B3-B89A" ] }, { "date": "2022-03-22", "text": "讨论战争、利率和市场预期中的不确定性，提出价值轮动判断。", "source_ids": [ "R2B3-B89B" ] } ], "gaps": [ "未核查完整历史组合和真实交易决策", "没有预测命中率或投资业绩审计", "当前职务须用独立现职来源更新" ], "comparison": "编辑比较重心为事实与解释、概率与预期、反证；不按名气排名。" }, { "id": "r2b3-b99", "name": "李录", "side": "buy", "role": "Himalaya Capital；资料时期身份，未借历史材料认证2026年现职", "scope": "公开投资论证 → 假设拆解 → 边界与反例 → 合成迁移", "focus": [ "能力圈", "受托责任", "真实购买力" ], "boundary": "两份实质公开材料支持有限方法样本，不认证永久个人方法、投资能力或预测准确率。", "timeline": [ { "date": "2015-10-23", "text": "讲座讨论受托责任、激励、能力圈及对判断前提的证伪。", "source_ids": [ "R2B3-B99B" ] }, { "date": "2024-12-07", "text": "讲座连接微观企业行动与宏观约束，区分名义资产和购买力财富。", "source_ids": [ "R2B3-B99A" ] } ], "gaps": [ "未核查完整历史组合和真实交易决策", "没有预测命中率或投资业绩审计", "当前职务须用独立现职来源更新" ], "comparison": "编辑比较重心为能力圈、受托责任、真实购买力；不按名气排名。" }, { "id": "r2b3-b100", "name": "Mohnish Pabrai", "side": "buy", "role": "Pabrai Investment Funds；资料时期身份，未借历史材料认证2026年现职", "scope": "公开投资论证 → 假设拆解 → 边界与反例 → 合成迁移", "focus": [ "有限决策", "集中与赢家", "克隆后的独立验证" ], "boundary": "两份实质公开材料支持有限方法样本，不认证永久个人方法、投资能力或预测准确率。", "timeline": [ { "date": "2026-04-21", "text": "讲座使用能力圈和有限决策次数解释选股纪律；高确信度不自动证明低风险。", "source_ids": [ "R2B3-B100A" ] }, { "date": "2026-05-01", "text": "讲座讨论错误常见、少数赢家贡献及不机械卖出赢家；克隆作为线索有核验边界。", "source_ids": [ "R2B3-B100B" ] } ], "gaps": [ "未核查完整历史组合和真实交易决策", "没有预测命中率或投资业绩审计", "当前职务须用独立现职来源更新", "两场讲座主题有重叠，不能当独立实证复制" ], "comparison": "编辑比较重心为有限决策、集中与赢家、克隆后的独立验证；不按名气排名。" }, { "id": "r2s1-vijay-rakesh", "name": "Vijay Rakesh", "side": "sell", "role": "Mizuho；下列2026-05-06电话会时身份", "scope": "收入构成与新工作负载的商业化", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。CPU需求增加不等于版税费率或授权合同同比增加。", "timeline": [ { "date": "2026-05-06", "text": "追问agentic CPU需求如何传导至数据中心版税、再单独询问授权收入增长。", "source_ids": [ "R2S1-ARM426" ] }, { "date": "2026-05-06", "text": "询问硬件和平台服务收入组合及后续增速；管理层按客户起点回答而未承诺固定占比。", "source_ids": [ "R2S1-IONQ126" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "总收入=版税+授权；分别增长后相加。" }, { "id": "r2s1-john-vinh", "name": "John Vinh", "side": "sell", "role": "KeyBanc；2022/2026会议时身份", "scope": "单机内容、装配率与量的拆分", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。新平台功能可选和标准延期会阻止装配率提升。", "timeline": [ { "date": "2022-08-08", "text": "以季节性差异追问供需，并拆分下一代内容增长来自装配率、单价或新增芯片。", "source_ids": [ "R2S1-CRUS22" ] }, { "date": "2026-08-11", "text": "追问手机内容增长、PC标准迁移延迟及AI PC真正形成需求的条件。", "source_ids": [ "R2S1-CRUS26" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "收入=终端量×装配率×单机内容价格。" }, { "id": "r2s1-srini-pajjuri", "name": "Srini Pajjuri", "side": "sell", "role": "2025 Raymond James；2026 RBC；按来源时期区分", "scope": "端市场拆解与BOM挤压", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。高端与低端手机不可混为同一暴露。", "timeline": [ { "date": "2025-05-07", "text": "追问版税增长的端市场构成及云与网络占比；管理层答复不等于他的预测。", "source_ids": [ "R2S1-ARM425" ] }, { "date": "2026-02-04", "text": "追问内存成本是否已影响手机销量，以及BOM压力会否拖慢CSS/v9采用。", "source_ids": [ "R2S1-ARM326" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "版税=终端量×采用率×单位版税。" }, { "id": "r2s1-ruben-roy", "name": "Ruben Roy", "side": "sell", "role": "Stifel；2025 Ciena电话会时身份", "scope": "技术替代与可售产品边界", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。距离、能耗与时延约束决定产品，市场总额不能全分配给某一方案。", "timeline": [ { "date": "2025-03-11", "text": "询问华为替换周期及CPO对Ciena是机会、威胁还是中性。", "source_ids": [ "O8SC-CIENQ125" ] }, { "date": "2025-09-04", "text": "询问scale-across是否超出Coherent Lite范围，以及宽带业务是出售还是减少投资。", "source_ids": [ "O8SC-CIENQ325" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "可实现收入=项目数×适配比例×每项目可售内容。" }, { "id": "r2s1-tristan-gerra", "name": "Tristan Gerra", "side": "sell", "role": "Baird；2023/2025 Seagate电话会时身份", "scope": "替代品价格与长期合同", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。长期协议可能仅约定量而不锁价。", "timeline": [ { "date": "2023-01-25", "text": "询问NAND降价会否传导HDD价格或引起份额替代。", "source_ids": [ "R2S1-STX223" ] }, { "date": "2025-10-28", "text": "追问HAMR长期协议期限、价格条款和收入占比，及密度升级后交期。", "source_ids": [ "R2S1-STX126" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "收入=出货容量×每TB价格；合同覆盖率与价格锁定率分别核对。" }, { "id": "r2s1-chris-caso", "name": "Chris Caso", "side": "sell", "role": "Wolfe Research；2026 MPS电话会时身份", "scope": "CPU电源内容与组合集中度", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。横跨x86/Arm不能证明两边内容或份额相同。", "timeline": [ { "date": "2026-04-30", "text": "询问商用GPU与ASIC解决方案增量，并检验汽车下半年增长可见性。", "source_ids": [ "R2S1-MPWR126" ] }, { "date": "2026-07-30", "text": "询问CPU电源份额、agentic需求传导与高速增长如何影响业务多元化。", "source_ids": [ "R2S1-MPWR226" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "收入=CPU数×单CPU内容×份额；再与其他业务合并。" }, { "id": "r2s1-william-stein", "name": "William Stein", "side": "sell", "role": "Truist；2025/2026 MPS电话会时身份", "scope": "交期、产能与定价的相互检验", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。产能金额随产品组合变化，不能当作物理利用率精确值。", "timeline": [ { "date": "2025-07-31", "text": "以收入相对名义产能的差距追问短交期和订货模式。", "source_ids": [ "R2S1-MPWR225" ] }, { "date": "2026-07-30", "text": "直接追问价格对收入增长的贡献，继而检查机器人和楼宇自动化落地。", "source_ids": [ "R2S1-MPWR226" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "利用率=出货量/可用产能；收入=量×价。" }, { "id": "r2s1-richard-windsor", "name": "Richard Windsor", "side": "sell", "role": "Radio Free Mobile 创办人；署名公开研究", "scope": "容量货币化与股东经济成本", "focus": [ "Technology", "AI", "Platform economics" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。采购额、产能资本支出和股权稀释不能重复计算；权证公允价值不等于零行权价。", "timeline": [ { "date": "2026-04-10", "text": "将收入预测除以GW检验单位容量货币化，再加入研发和销售成本；作者估计仍需核验。", "source_ids": [ "R2S1-RFM1" ] }, { "date": "2026-05-07", "text": "将销售增长与客户权证稀释放在一起讨论；原文推测的行权门槛不作为事实。", "source_ids": [ "R2S1-RFM2" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "经营盈余=GW×每GW收入×经营利润率−固定成本；每股盈余=盈余/稀释后股数。" }, { "id": "r2s1-jim-kelleher", "name": "Jim Kelleher", "side": "sell", "role": "Argus Research；具名页面作者", "scope": "平台产品广度与增长驱动核验", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。客户名称多不等于经济风险独立；具名两页仅为公开样本。", "timeline": [ { "date": "2025-01-08", "text": "以设备更新、服务和AI平台产品集合组织投资论点；不将目标价当估值模型。", "source_ids": [ "R2S1-ARGUS25" ] }, { "date": "2026-01-07", "text": "将客户类型扩展到企业、主权与neocloud并考察非数据中心业务。", "source_ids": [ "R2S1-ARGUS26" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "最终需求暴露按资金来源归并；收入×毛利率映射利润。" }, { "id": "r2s1-jay-goldberg", "name": "Jay Goldberg", "side": "sell", "role": "D2D / Seaport；文章发表时期", "scope": "资本开支集中度与防御性投资假说", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。新增AI利润与保住的既有利润不能算作同一现金流两次。", "timeline": [ { "date": "2025-07-14", "text": "将跨行业增长还原为少数资本开支买家，并讨论开放ISA不等于免费设计。", "source_ids": [ "R2S1-D2D1" ] }, { "date": "2025-10-22", "text": "提出保护既有云利润的替代解释，同时质疑份额变化、融资能力和收入重复计算。", "source_ids": [ "R2S1-D2D2" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "防守价值=避免损失的利润×持续年数（简化，未折现）；与额外投入比较。" }, { "id": "r2s1-andrew-gardiner", "name": "Andrew Gardiner", "side": "sell", "role": "Citi（2025/2026所选call时）；现已自述转Arm IR", "scope": "版税结构与订单可兑现性", "focus": [ "European semiconductors", "Semiconductor equipment" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。高费率产品收入占比高于数量占比，不可直接混用。", "timeline": [ { "date": "2025-05-07", "text": "追问v9占比提升来自哪些端市场及CSS费率。", "source_ids": [ "R2S1-ARM425" ] }, { "date": "2026-05-06", "text": "追问新CPU需求增加来自原有客户加量还是新增客户，并检查供应交付约束。", "source_ids": [ "R2S1-ARM426" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "某类收入占比=该类量×费率/总版税。" }, { "id": "r2s1-didier-scemama", "name": "Didier Scemama", "side": "sell", "role": "BofA；AIXTRON下列call时身份", "scope": "订单周期与设备再利用", "focus": [ "European semiconductors", "Semiconductor equipment" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。总收入预测与占比必须同期；中古设备影响订单而非必然影响当期收入。", "timeline": [ { "date": "2024-02-29", "text": "询问客户出售旧MicroLED设备会否形成竞争及是否能改作其他工艺，另问研发费用弹性。", "source_ids": [ "R2S1-AIX423" ] }, { "date": "2025-10-30", "text": "将产品占比变动换算金额，追问光电子需求能否跨过单轮AI资本开支周期。", "source_ids": [ "R2S1-AIX325" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "分部收入=公司总收入×分部占比。" }, { "id": "r2s1-randy-abrams", "name": "Randy Abrams", "side": "sell", "role": "2023 Credit Suisse；2024 UBS（所选call时）", "scope": "制程投资口径与收入时滞", "focus": "半导体、设备与科技供应链", "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。跨地区比较须统一工艺节点、阶段、基础设施范围与产能单位。", "timeline": [ { "date": "2023-01-12", "text": "比较美国和台湾每单位晶圆产能投资，主动询问口径差异与新节点成本加速。", "source_ids": [ "R2S1-TSM422" ] }, { "date": "2024-04-18", "text": "追问N2量产到有意义收入的时差、tape-out及资本开支节奏。", "source_ids": [ "R2S1-TSM124" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "已装产能×利用率×良率×单位收入；量产到收入有周期延迟。" }, { "id": "r2s1-gus-richard", "name": "Gus Richard", "side": "sell", "role": "2005 First Albany；2011 Piper Jaffray；仅历史材料，不代表Northland现观点", "scope": "剔除一次性因素与实际产品里程碑", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。旧公开样本不能证明当前AI观点；消除一次性项不能消除终端需求不确定性。", "timeline": [ { "date": "2005-10-19", "text": "先剔除准备金转回与项目延迟再比较季节增长，接着区分降本留存和客户让利。", "source_ids": [ "R2S1-POWI305" ] }, { "date": "2011-01-25", "text": "把28nm送样计划拆成仍在晶圆厂与客户已经拿到产品，核对实际进度。", "source_ids": [ "R2S1-ALTR410" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。", "所选材料含较早历史样本，不能外推当前立场。" ], "comparison": "同口径增长=(本期经常收入/上期经常收入)-1；计划送样不等于完成送样。" }, { "id": "r2s1-pierre-ferragu", "name": "Pierre Ferragu", "side": "sell", "role": "2013 Bernstein；2023 New Street Research（材料当时）", "scope": "费用桥与产品开发收入错期", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。管理层对费用原因的解释不是分析师既有预测；GAAP与non-GAAP须同口径。", "timeline": [ { "date": "2013-01-23", "text": "由收入和营业利润率指导反推费用增长，并询问新品上市后的采用节奏。", "source_ids": [ "R2S1-MSI412" ] }, { "date": "2023-11-08", "text": "追问费用增加对应研发、客户工程或业务拓展，并区分共识误建模与新增支出。", "source_ids": [ "R2S1-ARM224" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。", "所选材料含较早历史样本，不能外推当前立场。" ], "comparison": "营业利润=收入×毛利率−经营费用；开发支出先于收入。" }, { "id": "r2s1-sandeep-deshpande", "name": "Sandeep Deshpande", "side": "sell", "role": "JPMorgan；2025所选电话会时", "scope": "订单可见性与技术导入阶段", "focus": [ "European semiconductors", "Semiconductor equipment" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。公开镜像转录尚未逐字核对官方录音；积压订单并非不可延期的现金。", "timeline": [ { "date": "2025-01-29", "text": "追问已下单工具与尚未确认量产导入时点的矛盾，检验延期风险。", "source_ids": [ "R2S2-ASML424" ] }, { "date": "2025-04-16", "text": "把下一年订单覆盖与上一年同期比较，并追问第三阶段High-NA订单时点。", "source_ids": [ "R2S2-ASML125" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "当期可实现收入=积压订单×当期可交付比例×验收比例。" }, { "id": "r2s1-sara-russo", "name": "Sara Russo", "side": "sell", "role": "Bernstein；2023/2025所选材料当时", "scope": "续约群组与订单收入可见性", "focus": [ "European semiconductors" ], "boundary": "仅限下列日期的公开具名材料，不据此认定稳定个人风格或当前职务。ASML为公开转录镜像，未听原录音；机构简介登录墙，不能用历史call认证现职。", "timeline": [ { "date": "2023-11-08", "text": "问同一续约客户IP采用与花费变化，再问订阅方案是否改善版税预测可见性。", "source_ids": [ "R2S1-ARM224" ] }, { "date": "2025-01-29", "text": "追问减少订单披露的理由，并将EUV订单与年度指导覆盖和次年交付分开。", "source_ids": [ "R2S2-ASML424" ] } ], "gaps": [ "公开具名两份材料仅支持方法样本；尚无私人完整模型、预测复盘及长期业绩审计。" ], "comparison": "群组净收入留存=老客户本期收入/老客户基期收入；订单须按交付期拆分。" }, { "id": "mehdi-hosseini", "name": "Mehdi Hosseini", "side": "sell", "role": "SIG / Susquehanna分析师（材料时身份）", "scope": "资本强度与交付弹性", "focus": [ "资本强度", "交付弹性" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-01-29", "text": "追问设备吞吐、产能弹性与客户承诺。", "source_ids": [ "R2S2-ASML424" ], "locator": "Hosseini具名问答" }, { "date": "2025-07-17", "text": "追问N2流片增长是否需要更高资本强度。", "source_ids": [ "S36SRC1" ], "locator": "pp.17–18" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对" ], "comparison": "资本强度与交付弹性" }, { "id": "krish-sankar", "name": "Krish Sankar", "side": "sell", "role": "TD Cowen分析师（材料时身份）", "scope": "订单可见度与GW到晶圆桥", "focus": [ "订单可见度", "GW到晶圆桥" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-04-16", "text": "追问次年增长需要的订单及区域构成。", "source_ids": [ "R2S2-ASML125" ], "locator": "Sankar具名问答" }, { "date": "2025-10-16", "text": "询问1GW AI容量如何映射晶圆需求及GPU/ASIC组合。", "source_ids": [ "S36SRC2" ], "locator": "pp.15–16" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对" ], "comparison": "订单可见度与GW到晶圆桥" }, { "id": "toshiya-hari", "name": "Toshiya Hari", "side": "sell", "role": "Goldman Sachs历史半导体分析师；已转NVIDIA企业IR，不计当前卖方", "scope": "供应约束与产品组合利润", "focus": [ "供应约束", "产品组合利润" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2024-07-30", "text": "检验MI300指引是需求还是供给限制，并问盈利变化。", "source_ids": [ "O8SAD4" ], "locator": "pp.10–11" }, { "date": "2024-10-29", "text": "追问GPU指引上调、客户扩展与毛利组合。", "source_ids": [ "O8SAD5" ], "locator": "pp.7–8" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "保留历史研究，不计当前卖方在岗人数；现职沿用已有一手转职证据" ], "comparison": "供应约束与产品组合利润" }, { "id": "joe-quatrochi", "name": "Joe Quatrochi", "side": "sell", "role": "Wells Fargo分析师（材料时身份）", "scope": "交期、客户空间与收入时点", "focus": [ "交期、客户空间", "收入时点" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-01-29", "text": "追问收入指引上端与预建库存、交期。", "source_ids": [ "R2S2-ASML424" ], "locator": "Quatrochi具名问答" }, { "date": "2025-04-16", "text": "检验关税是否促使提前交付及旧机库存可售性。", "source_ids": [ "R2S2-ASML125" ], "locator": "Quatrochi具名问答" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对" ], "comparison": "交期、客户空间与收入时点" }, { "id": "brian-chin", "name": "Brian Chin", "side": "sell", "role": "Stifel分析师（材料时身份）", "scope": "NAND升级与新增产能并行", "focus": [ "NAND升级", "新增产能并行" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-07-30", "text": "追问刻蚀工具跨代适用及服务业务拆分。", "source_ids": [ "R2S2-LAM425" ], "locator": "Brian Edward Chin具名问答" }, { "date": "2025-10-22", "text": "要求拆分AI支出到WFE构成，并检验升级与新建是否并行。", "source_ids": [ "S2D-B03" ], "locator": "Brian Chin具名问答；本轮读取Roic同场镜像" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对" ], "comparison": "NAND升级与新增产能并行" }, { "id": "brett-simpson", "name": "Brett Simpson", "side": "sell", "role": "Arete历史分析师；已转Qualcomm企业IR，不计当前卖方", "scope": "地区增长与成本传导", "focus": [ "地区增长", "成本传导" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2020-01-16", "text": "拆分中国增长来源及5G/HPC贡献。", "source_ids": [ "R2S2-TS419" ], "locator": "p.11" }, { "date": "2025-07-17", "text": "追问结构性成本、涨价与AI效率能否抵消毛利压力。", "source_ids": [ "S36SRC1" ], "locator": "pp.11–12" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "保留历史研究，不计当前卖方在岗人数；现职沿用已有一手转职证据" ], "comparison": "地区增长与成本传导" }, { "id": "janardan-menon", "name": "Janardan Menon", "side": "sell", "role": "Jefferies分析师（材料时身份）", "scope": "同业差异与产品组合", "focus": [ "同业差异", "产品组合" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-04-24", "text": "追问与TI工业复苏差异、渠道和毛利组合。", "source_ids": [ "R2S2-STM125" ], "locator": "Menon具名问答" }, { "date": "2025-10-30", "text": "拆分Opto/GaN/SiC走势，问晶圆尺寸转换。", "source_ids": [ "R2S1-AIX325" ], "locator": "pp.6–7" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对" ], "comparison": "同业差异与产品组合" }, { "id": "stephane-houri", "name": "Stéphane Houri", "side": "sell", "role": "ODDO BHF分析师（材料时身份）", "scope": "订单波动与年度桥", "focus": [ "订单波动", "年度桥" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-01-29", "text": "检验服务业务单季年化与全年指引差异。", "source_ids": [ "R2S2-ASML424" ], "locator": "Houri具名问答" }, { "date": "2025-04-16", "text": "追问订单波动是否来自关税，以及指引订单覆盖。", "source_ids": [ "R2S2-ASML125" ], "locator": "Houri具名问答" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对" ], "comparison": "订单波动与年度桥" }, { "id": "francois-xavier-bouvignies", "name": "François-Xavier Bouvignies", "side": "sell", "role": "UBS分析师（材料时身份）", "scope": "技术成熟度与采用经济性", "focus": [ "技术成熟度", "采用经济性" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-01-29", "text": "检验2nm扩产是否前置及AI能否抵消手机弱势。", "source_ids": [ "R2S2-ASML424" ], "locator": "Bouvignies具名问答" }, { "date": "2025-04-16", "text": "问High-NA定价能否促进采用及单次曝光替代节奏。", "source_ids": [ "R2S2-ASML125" ], "locator": "Bouvignies具名问答" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对" ], "comparison": "技术成熟度与采用经济性" }, { "id": "alexander-duval", "name": "Alexander Duval", "side": "sell", "role": "Goldman Sachs分析师（材料时身份）", "scope": "技术证据与设备强度", "focus": [ "技术证据", "设备强度" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-04-16", "text": "区分EUV研发消息与可扩展可靠产品，问推理需求。", "source_ids": [ "R2S2-ASML125" ], "locator": "Duval具名问答" }, { "date": "2025-07-16", "text": "追问DRAM光刻强度和通用平台时间线。", "source_ids": [ "R2S2-ASML225" ], "locator": "Duval具名问答" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对" ], "comparison": "技术证据与设备强度" }, { "id": "brad-zelnick", "name": "Brad Zelnick", "side": "sell", "role": "Deutsche Bank分析师（材料时身份）", "scope": "AI功能价值与收费捕获", "focus": [ "AI功能价值", "收费捕获" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-01-29", "text": "询问推理成本下降后产品包装与覆盖客户。", "source_ids": [ "S51SRC1" ], "locator": "Brad Zelnick具名问答" }, { "date": "2025-09-11", "text": "问AEP agents给客户创造的价值如何被Adobe捕获。", "source_ids": [ "S52SRC2" ], "locator": "pp.18–19" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验" ], "comparison": "AI功能价值与收费捕获" }, { "id": "kirk-materne", "name": "Kirk Materne", "side": "sell", "role": "Evercore ISI分析师（材料时身份）", "scope": "非AI增长与提价兑现", "focus": [ "非AI增长", "提价兑现" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-04-30", "text": "区分Azure非AI改善与此前优化周期。", "source_ids": [ "O8SB-MS3" ], "locator": "Kirk Materne具名问答" }, { "date": "2025-06-12", "text": "询问Creative Cloud Pro改变的客户反应及下半年收入。", "source_ids": [ "R2S2-ADBE225" ], "locator": "pp.13–14" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验" ], "comparison": "非AI增长与提价兑现" }, { "id": "raimo-lenschow", "name": "Raimo Lenschow", "side": "sell", "role": "Barclays分析师（材料时身份）", "scope": "产品扩展与客户工作流", "focus": [ "产品扩展", "客户工作流" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2023-11-30", "text": "询问测试自动化与SAP转型的需求。", "source_ids": [ "R2S2-UI324" ], "locator": "p.13" }, { "date": "2025-07-30", "text": "检验Copilot如何扩展到数据与agent工作流。", "source_ids": [ "S51SRC2" ], "locator": "Raimo Lenschow具名问答" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验" ], "comparison": "产品扩展与客户工作流" }, { "id": "alex-zukin", "name": "Alex Zukin", "side": "sell", "role": "Wolfe Research分析师（材料时身份）", "scope": "竞争边界与AI收入拆解", "focus": [ "竞争边界", "AI收入拆解" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-06-12", "text": "问Express竞争和Meta广告自动化的竞合。", "source_ids": [ "R2S2-ADBE225" ], "locator": "pp.15–16" }, { "date": "2025-09-11", "text": "拆分AI-first采用与传统价量对ARR改善的贡献。", "source_ids": [ "S52SRC2" ], "locator": "pp.15–16" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验" ], "comparison": "竞争边界与AI收入拆解" }, { "id": "derrick-wood", "name": "Derrick Wood", "side": "sell", "role": "Cowen / TD Cowen分析师（材料时身份）；原始记录亦作James Derrick Wood", "scope": "季节性与提前续约", "focus": [ "季节性", "提前续约" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2022-12-15", "text": "追问Creative/Document ARR季节性差异。", "source_ids": [ "R2S2-ADBE422" ], "locator": "p.12" }, { "date": "2025-07-23", "text": "问收入超指引是否由自托管比例、线性及业务构成驱动。", "source_ids": [ "R2S2-NOW225" ], "locator": "operator Derrick Wood / speaker James Wood问答" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对" ], "comparison": "季节性与提前续约" }, { "id": "rishi-jaluria", "name": "Rishi Jaluria", "side": "sell", "role": "RBC Capital Markets分析师（材料时身份）", "scope": "AI内部效率与商业部署", "focus": [ "AI内部效率", "商业部署" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2023-04-27", "text": "问GenAI如何改善内部工程和产品效率。", "source_ids": [ "R2S2-COUR123" ], "locator": "p.10" }, { "date": "2024-10-30", "text": "询问Copilot数字口径及企业/消费战略。", "source_ids": [ "R2S2-MS125" ], "locator": "Rishi Jaluria具名问答" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验" ], "comparison": "AI内部效率与商业部署" }, { "id": "tyler-radke", "name": "Tyler Radke", "side": "sell", "role": "Citi分析师（材料时身份）", "scope": "使用、价格与真实量", "focus": [ "使用、价格", "真实量" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2025-06-12", "text": "问Firefly生成量与货币化是否同步。", "source_ids": [ "R2S2-ADBE225" ], "locator": "pp.14–15" }, { "date": "2025-09-11", "text": "要求拆分提价后的真实业务量。", "source_ids": [ "S52SRC2" ], "locator": "p.14" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验" ], "comparison": "使用、价格与真实量" }, { "id": "gregg-moskowitz", "name": "Gregg Moskowitz", "side": "sell", "role": "Mizuho分析师（材料时身份）", "scope": "用户增长与销售组织", "focus": [ "用户增长", "销售组织" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2024-06-13", "text": "追问付费用户增长及数字渠道获客。", "source_ids": [ "R2S2-ADBE224" ], "locator": "Moskowitz具名问答" }, { "date": "2025-07-23", "text": "检验不同地区销售领导变化后的执行能力。", "source_ids": [ "R2S2-NOW225" ], "locator": "Moskowitz具名问答" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对" ], "comparison": "用户增长与销售组织" }, { "id": "michael-turits", "name": "Michael Turits", "side": "sell", "role": "KeyBanc历史软件研究分析师；已转软件投行，不计当前卖方", "scope": "定价价值与部署速度", "focus": [ "定价价值", "部署速度" ], "boundary": "跨两份材料的公开提问样本；管理层答复不是分析师预测，以下练习由编辑合成。", "timeline": [ { "date": "2022-05-25", "text": "检验计费方式价值与宏观延迟对消费的影响。", "source_ids": [ "R2S2-SPLK123" ], "locator": "Michael Turits具名问答" }, { "date": "2023-11-30", "text": "追问部署提速及固定汇率NRR仍下降。", "source_ids": [ "R2S2-UI324" ], "locator": "pp.17–18" } ], "gaps": [ "两份选读材料仅支持公开提问样本，未认证持续个人方法或预测业绩", "未取得个人付费报告、完整估值模型及投资回报核验", "至少一项为公开第三方问答转录，未与录音逐字校对", "保留历史研究，不计当前卖方在岗人数；现职沿用已有一手转职证据" ], "comparison": "定价价值与部署速度" }, { "id": "r-scott-berg", "name": "R. Scott Berg", "side": "sell", "role": "Needham & Company分析师（所引材料时身份）", "scope": "新产品分销成本；功能到收入贡献", "focus": [ "新产品分销成本", "功能到收入贡献" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-11-05", "text": "询问独立ESM销售是否另设团队、如何定价。", "source_ids": [ "R2S3-FRSHQ325" ] }, { "date": "2026-03-25", "text": "询问餐厅预订经济性及对指引贡献。", "source_ids": [ "R2S3-NAVANQ426" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "新产品分销成本；功能到收入贡献" }, { "id": "mark-mahaney", "name": "Mark Mahaney", "side": "sell", "role": "Evercore ISI分析师（所引材料时身份）", "scope": "资本开支与供给约束；亏损改善机制", "focus": [ "资本开支与供给约束", "亏损改善机制" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-02-06", "text": "追问资本开支年化规模与AWS增长是否受供给限制。", "source_ids": [ "R2S3-AMZNQ424" ] }, { "date": "2025-04-30", "text": "追問Reality Labs亏损何时及如何下降。", "source_ids": [ "O8SB-META1" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "资本开支与供给约束；亏损改善机制" }, { "id": "james-fish", "name": "James Fish", "side": "sell", "role": "Piper Sandler分析师（所引材料时身份）", "scope": "技术差异到竞争反应；工作流内AI采用", "focus": [ "技术差异到竞争反应", "工作流内AI采用" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-10-30", "text": "询问SVX上市后的竞争者定价反应。", "source_ids": [ "R2S3-MSIQ325" ] }, { "date": "2026-09-16", "text": "追问AI策略及适合救援人员工作流的原因。", "source_ids": [ "R2S3-MSICONF26" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "技术差异到竞争反应；工作流内AI采用" }, { "id": "samik-chatterjee", "name": "Samik Chatterjee", "side": "sell", "role": "JPMorgan分析师（所引材料时身份）", "scope": "企业附加服务；收入规模到利润率", "focus": [ "企业附加服务", "收入规模到利润率" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2024-08-29", "text": "询问企业AI客户能带来多少服务、存储附加及毛利差异。", "source_ids": [ "R2S3-DELLQ225" ] }, { "date": "2024-09-26", "text": "以收入规模、产能利用与费用率检验6%利润率路径。", "source_ids": [ "R2S3-JBLQ424" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "企业附加服务；收入规模到利润率" }, { "id": "wamsi-mohan", "name": "Wamsi Mohan", "side": "sell", "role": "Bank of America分析师（所引材料时身份）", "scope": "价格竞争与组合；订单到出货约束", "focus": [ "价格竞争与组合", "订单到出货约束" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2024-05-30", "text": "区分主动降价抢份额与竞争应对，并拆解毛利压力。", "source_ids": [ "R2S3-DELLQ125" ] }, { "date": "2024-08-29", "text": "追问AI积压不增是否反映Blackwell或液冷供给时点。", "source_ids": [ "R2S3-DELLQ225" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "价格竞争与组合；订单到出货约束" }, { "id": "david-vogt", "name": "David Vogt", "side": "sell", "role": "UBS分析师（所引材料时身份）", "scope": "利润增长归因；回购与经营杠杆", "focus": [ "利润增长归因", "回购与经营杠杆" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2024-09-26", "text": "分辨后半年EPS提升来自利用率、组合还是回购。", "source_ids": [ "R2S3-JBLQ424" ] }, { "date": "2024-12-18", "text": "追问数据中心新增收入的利润率及并购取向。", "source_ids": [ "R2S3-JBLQ124" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "利润增长归因；回购与经营杠杆" }, { "id": "steven-fox", "name": "Steven Fox", "side": "sell", "role": "Fox Advisors分析师（所引材料时身份）", "scope": "产能时滞；现金流与经营负杠杆", "focus": [ "产能时滞", "现金流与经营负杠杆" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2024-09-26", "text": "追问汽车需求变化的短期负杠杆及FCF改善桥。", "source_ids": [ "R2S3-JBLQ424" ] }, { "date": "2024-12-18", "text": "区分现有美国产能承接与后续扩产能力。", "source_ids": [ "R2S3-JBLQ124" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "产能时滞；现金流与经营负杠杆" }, { "id": "george-wang", "name": "George Wang", "side": "sell", "role": "Barclays分析师（所引材料时身份）", "scope": "ATE与WFE拆分；AI收入边界", "focus": [ "ATE与WFE拆分", "AI收入边界" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2024-09-26", "text": "拆解半导体设备增长并核对AI收入跨分部定义。", "source_ids": [ "R2S3-JBLQ424" ] }, { "date": "2024-12-18", "text": "再问ATE/WFE节奏及硅光产品演进。", "source_ids": [ "R2S3-JBLQ124" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "ATE与WFE拆分；AI收入边界" }, { "id": "ari-klein", "name": "Ari Klein", "side": "sell", "role": "BMO Capital Markets分析师（所引材料时身份）", "scope": "预售口径；产能交付时点", "focus": [ "预售口径", "产能交付时点" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-07-30", "text": "问追加资本支出能否提前交付。", "source_ids": [ "R2S3-EQIXQ225" ] }, { "date": "2025-10-29", "text": "追问预售增长与销售窗口延长关系。", "source_ids": [ "R2S3-EQIXQ325" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "预售口径；产能交付时点" }, { "id": "eric-luebchow", "name": "Eric Luebchow", "side": "sell", "role": "Wells Fargo分析师（所引材料时身份）", "scope": "投产到稳定期；融资对现金指标影响", "focus": [ "投产到稳定期", "融资对现金指标影响" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-07-30", "text": "问稳定期能否缩短及预租降低风险。", "source_ids": [ "R2S3-EQIXQ225" ] }, { "date": "2025-10-29", "text": "问供给紧张定价与次年AFFO融资。", "source_ids": [ "R2S3-EQIXQ325" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "投产到稳定期；融资对现金指标影响" }, { "id": "nick-del-deo", "name": "Nick Del Deo", "side": "sell", "role": "MoffettNathanson分析师（所引材料时身份）", "scope": "互联数量波动；生态价值与兑现", "focus": [ "互联数量波动", "生态价值与兑现" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-07-30", "text": "询问互联新增锯齿波动及持续性。", "source_ids": [ "R2S3-EQIXQ225" ] }, { "date": "2025-10-29", "text": "问新云入口相对传统云的战略价值。", "source_ids": [ "R2S3-EQIXQ325" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "互联数量波动；生态价值与兑现" }, { "id": "frank-louthan-iv", "name": "Frank Louthan IV", "side": "sell", "role": "Raymond James分析师（所引材料时身份）", "scope": "会计确认口径；互联价格与组合", "focus": [ "会计确认口径", "互联价格与组合" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-07-30", "text": "追问资本化利息变化及历史处理。", "source_ids": [ "R2S3-EQIXQ225" ] }, { "date": "2025-10-29", "text": "问互联ARPU与电力密度、地区复制性。", "source_ids": [ "R2S3-EQIXQ325" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "会计确认口径；互联价格与组合" }, { "id": "andrew-obin", "name": "Andrew Obin", "side": "sell", "role": "BofA Global Research分析师（所引材料时身份）", "scope": "订单周期；增长到份额推断", "focus": [ "订单周期", "增长到份额推断" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-05-02", "text": "问高基数下订单与电力需求。", "source_ids": [ "R2S3-ETNQ125" ] }, { "date": "2025-08-05", "text": "问增长是否说明份额提升及数据中心产品布局。", "source_ids": [ "R2S3-ETNQ225" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "订单周期；增长到份额推断" }, { "id": "christopher-glynn", "name": "Christopher Glynn", "side": "sell", "role": "Oppenheimer分析师（所引材料时身份）", "scope": "细分市场并购；订单到收入时滞", "focus": [ "细分市场并购", "订单到收入时滞" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-02-04", "text": "问Kern并购竞争定位及组合退出。", "source_ids": [ "R2S3-AMEQ424" ] }, { "date": "2025-07-31", "text": "问航天并购优先级及自动化周期。", "source_ids": [ "R2S3-AMEQ225" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "细分市场并购；订单到收入时滞" }, { "id": "tal-liani", "name": "Tal Liani", "side": "sell", "role": "BofA Securities分析师（所引材料时身份）", "scope": "技术应用边界；增长与并购过渡期", "focus": [ "技术应用边界", "增长与并购过渡期" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-03-11", "text": "追问相干光应用区间及服务增速。", "source_ids": [ "O8SC-CIENQ125" ] }, { "date": "2025-11-19", "text": "问并购过渡期利润稀释与协同时点。", "source_ids": [ "R2S3-PANWQ126" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "技术应用边界；增长与并购过渡期" }, { "id": "david-guarino", "name": "David Guarino", "side": "sell", "role": "Green Street Advisors分析师（所引材料时身份）", "scope": "签约口径；电力承诺现金成本", "focus": [ "签约口径", "电力承诺现金成本" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-07-24", "text": "问公用事业预付款对成本和竞争影响。", "source_ids": [ "R2S3-DLRQ225" ] }, { "date": "2025-10-29", "text": "追问预售与年化签约正常运行水平。", "source_ids": [ "R2S3-EQIXQ325" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "签约口径；电力承诺现金成本" }, { "id": "jonathan-petersen", "name": "Jonathan Petersen", "side": "sell", "role": "Jefferies分析师（所引材料时身份）", "scope": "市场与份额；土地用途选择", "focus": [ "市场与份额", "土地用途选择" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2025-07-24", "text": "问小容量增长来自市场还是份额及销售改变。", "source_ids": [ "R2S3-DLRQ225" ] }, { "date": "2025-10-29", "text": "问新增土地分配零售与xScale。", "source_ids": [ "R2S3-EQIXQ325" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "市场与份额；土地用途选择" }, { "id": "timothy-horan", "name": "Timothy Horan", "side": "sell", "role": "Oppenheimer & Co.分析师（所引材料时身份）", "scope": "经常性收益时间；供需与定价", "focus": [ "经常性收益时间", "供需与定价" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2024-10-30", "text": "问xScale收益贡献何时进入运行水平。", "source_ids": [ "R2S3-EQIXQ324" ] }, { "date": "2025-04-30", "text": "问供给约束下提价和产能匹配。", "source_ids": [ "R2S3-EQIXQ125" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "经常性收益时间；供需与定价" }, { "id": "richard-choe", "name": "Richard Choe", "side": "sell", "role": "JP Morgan分析师（所引材料时身份）", "scope": "资本开支归属；竞争与容量", "focus": [ "资本开支归属", "竞争与容量" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2024-02-14", "text": "问竞争流失和价格变化。", "source_ids": [ "R2S3-EQIXQ423" ] }, { "date": "2025-07-24", "text": "问开发CapEx规模持续多久。", "source_ids": [ "R2S3-DLRQ225" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "资本开支归属；竞争与容量" }, { "id": "brendan-lynch", "name": "Brendan Lynch", "side": "sell", "role": "Barclays分析师（所引材料时身份）", "scope": "成本可控性；负载位置与延迟", "focus": [ "成本可控性", "负载位置与延迟" ], "boundary": "两次公开问答选读支持方法样本，不认证永久个人风格；管理层答复不是分析师预测。", "timeline": [ { "date": "2022-04-27", "text": "问长期利润率目标与可控成本杠杆。", "source_ids": [ "R2S3-EQIXQ122" ] }, { "date": "2026-09-14", "text": "追问延迟敏感负载占需求的角色。", "source_ids": [ "R2S3-EQIXBARC26" ] } ], "gaps": [ "两次公开问答选读支持方法样本，不认证永久个人风格", "未审计个人预测准确度、投资业绩或付费模型", "当前任职未以本轮历史问答独立确认" ], "comparison": "成本可控性；负载位置与延迟" }, { "id": "xuesong-zhao", "name": "Xuesong Zhao", "side": "buy", "role": "Polar Capital AI基金Lead Fund Manager（2023文章、2026年5月访谈）", "scope": "数据护城河可证伪；AI采用者的利润兑现", "focus": [ "数据护城河可证伪", "AI采用者的利润兑现" ], "boundary": "仅提炼作者的可变假设；2023文中GPT-4参数等未经核验技术断言不作为事实采用。 两份公开材料不等于完整个人哲学或稳定方法认证。", "timeline": [ { "date": "2023-08-14", "text": "把独有数据及非科技采用者纳入AI受益研究，考察技术如何改变成本和商业模式。", "source_ids": [ "O9G1-B04-S1" ] }, { "date": "2026-05", "text": "软件功能可被复制时重估原有数据优势；区分盈利修正与估值重估。", "source_ids": [ "O9G1-B04-S2" ] } ], "gaps": [ "未审计预测准确率或风险调整业绩；两份材料仅支持有限方法样本", "来源时期身份不自动外推至2026年10月现职" ], "comparison": "数据护城河可证伪；AI采用者的利润兑现" }, { "id": "mike-seidenberg", "name": "Mike Seidenberg", "side": "buy", "role": "Allianz Technology Trust Lead Manager（2024/2025访谈时）", "scope": "基准集中风险；风格分散与科技选股", "focus": [ "基准集中风险", "风格分散与科技选股" ], "boundary": "机器转录未作音频复核；组合情景仅为教学，不是经理风险预测。 两份公开材料不等于完整个人哲学或稳定方法认证。", "timeline": [ { "date": "2024-10-21", "text": "承认低配巨头可能拖累相对表现，但把股东承受的集中风险纳入组合选择。", "source_ids": [ "O9G1-B10-S1" ] }, { "date": "2025-10-16", "text": "在高增长、合理估值增长和低估值可选性间配置；网络安全需求持续不保证供应商不会过时。", "source_ids": [ "O9G1-B10-S2" ] } ], "gaps": [ "未审计预测准确率或风险调整业绩；两份材料仅支持有限方法样本", "来源时期身份不自动外推至2026年10月现职" ], "comparison": "基准集中风险；风格分散与科技选股" }, { "id": "josh-spencer", "name": "Josh Spencer", "side": "buy", "role": "T. Rowe Price Global Technology Equity Strategy经理（2017/2018）", "scope": "云成本与稳态利润率；半导体周期和长期需求", "focus": [ "云成本与稳态利润率", "半导体周期和长期需求" ], "boundary": "历史技术基金方法样本，不把2017/2018职务冒充现职；原文看多不等于事后验证。 两份公开材料不等于完整个人哲学或稳定方法认证。", "timeline": [ { "date": "2017-09", "text": "将客户成本、扩容弹性与SaaS稳态利润率结合，同时保留估值约束。", "source_ids": [ "O9G1-B18-S1" ] }, { "date": "2018-11", "text": "用供给集中度、扩产资本强度和终端内容增长检验周期底部利润，而非只外推峰值。", "source_ids": [ "O9G1-B18-S2" ] } ], "gaps": [ "未审计预测准确率或风险调整业绩；两份材料仅支持有限方法样本", "来源时期身份不自动外推至2026年10月现职" ], "comparison": "云成本与稳态利润率；半导体周期和长期需求" }, { "id": "steve-wymer", "name": "Steve Wymer", "side": "buy", "role": "Fidelity Growth Company Fund经理（2018/2026问答时；正文名Steven Wymer）", "scope": "增长持续期；增长兑现与集中风险", "focus": [ "增长持续期", "增长兑现与集中风险" ], "boundary": "2018为未逐字核验的公开镜像，2026为滚动官方文档；日期是报告期间终点。 两份公开材料不等于完整个人哲学或稳定方法认证。", "timeline": [ { "date": "2018-05-31", "text": "看三至五年增长机会和商业模式延展性，同时因组合构建调整高确信持仓。", "source_ids": [ "O9G1-B25-S1" ] }, { "date": "2026-05-31", "text": "把持续AI采用与企业投资回报及基建约束相连；增长持仓仍需管理集中度。", "source_ids": [ "O9G1-B25-S2" ] } ], "gaps": [ "未审计预测准确率或风险调整业绩；两份材料仅支持有限方法样本", "来源时期身份不自动外推至2026年10月现职" ], "comparison": "增长持续期；增长兑现与集中风险" }, { "id": "alex-sacerdote", "name": "Alex Sacerdote", "side": "buy", "role": "Whale Rock Capital创始人/投资经理（2015/2026访谈）", "scope": "采用曲线斜率和上限；竞争优势与未反映的盈利", "focus": [ "采用曲线斜率和上限", "竞争优势与未反映的盈利" ], "boundary": "2026第三方机器稿未核音频；只取清楚的框架段，不将转录中的具体估值、公司历史或投资业绩当作已核事实。 两份公开材料不等于完整个人哲学或稳定方法认证。", "timeline": [ { "date": "2015", "text": "把采用拐点与竞争优势、未被价格反映的盈利能力联合考察；市场增长不保证每家厂商获利。", "source_ids": [ "O9G1-B30-S1" ] }, { "date": "2026-06-09", "text": "区分采用速度与市场上限，并在渗透路径偏离时修正假设；企业整合难度也影响采用速度。", "source_ids": [ "O9G1-B30-S2" ] } ], "gaps": [ "未审计预测准确率或风险调整业绩；两份材料仅支持有限方法样本", "来源时期身份不自动外推至2026年10月现职" ], "comparison": "采用曲线斜率和上限；竞争优势与未反映的盈利" }, { "id": "shekhar-pramanick", "name": "Shekhar Pramanick", "side": "buy", "role": "2004 Schwab Soundview卖方分析师；2022 Columbia Seligman买方经理；2025年末STK Technology Team Member", "scope": "出货与终端需求；收入到利润的约束", "focus": [ "出货与终端需求", "收入到利润的约束" ], "boundary": "跨角色的有限方法样本：一份2004卖方问答、一份2022买方采访；不声称两份均为买方或论证跨时期稳定。 两份公开材料不等于完整个人哲学或稳定方法认证。", "timeline": [ { "date": "2004-07-29", "text": "【2004卖方｜Schwab Soundview】卖方时期追问出货指引是否已考虑库存风险，并追问收入增长能否转成毛利率改善。", "source_ids": [ "O9G1-B13-S1" ] }, { "date": "2022-01-20", "text": "【2022买方｜Columbia Seligman】考察芯片需求与公司盈利。", "source_ids": [ "O9G1-B13-S2" ] } ], "gaps": [ "未审计预测准确率或风险调整业绩；两份材料仅支持有限方法样本", "来源时期身份不自动外推至2026年10月现职" ], "comparison": "出货与终端需求；收入到利润的约束" }, { "id": "stephen-mandel", "name": "Stephen Mandel", "side": "buy", "role": "Lone Pine创始人（2021/2025节目时）；官网2027前瞻职位不作2026现职", "scope": "单位经济可复制性；再投资回报与自由现金流；估值约束", "focus": [ "单位经济可复制性", "再投资回报与自由现金流", "估值约束" ], "boundary": "本人具名公开论证的方法样本；信件中的机构过程保留团队属性；题目全部编辑合成。", "timeline": [ { "date": "2021-07-20", "text": "从可复制的单位经济与管理质量理解业务安全边际；由具体公司发现多年变迁，而非从主题标签倒推股票。", "source_ids": [ "O9G2-M21" ] }, { "date": "2025-09-15", "text": "以自由现金流收益率、再投资回报与融资依赖共同看估值；高回报扩张导致的低FCF不能直接视为劣质，AI赢家也可能因买价过高给投资者低回报。", "source_ids": [ "O9G2-M25" ] } ], "gaps": [ "两份公开材料仅支持有限方法样本，未认证完整或稳定个人投资哲学", "镜像转录未逐字比对原音频/官方原件；未审计预测结果和投资业绩" ], "comparison": "单位经济可复制性；再投资回报与自由现金流；估值约束" }, { "id": "andreas-halvorsen", "name": "Andreas Halvorsen", "side": "buy", "role": "Viking Global投资负责人、信件作者（2009/2010）", "scope": "多空价差与风险敞口；流动性及融资风险；集中度激励", "focus": [ "多空价差与风险敞口", "流动性及融资风险", "集中度激励" ], "boundary": "本人具名公开论证的方法样本；信件中的机构过程保留团队属性；题目全部编辑合成。", "timeline": [ { "date": "2009-01-09", "text": "把多空选股价差、净敞口与总敞口分开；降低总杠杆可能牺牲账面收益，却保留融资安全与危机中继续研究交易的能力。（时间口径：信件抬头日期，报告期为2008Q4）", "source_ids": [ "O9G2-H09" ] }, { "date": "2010-07-30", "text": "检视高确信度持仓、分散度和多空激励；多头与空头的流动性及损失不对称，不能按同样集中规则处理。（时间口径：GuruFocus镜像发布时间2010-07-30；原信确切写作日未知）", "source_ids": [ "O9G2-H10" ] } ], "gaps": [ "两份公开材料仅支持有限方法样本，未认证完整或稳定个人投资哲学", "镜像转录未逐字比对原音频/官方原件；未审计预测结果和投资业绩" ], "comparison": "多空价差与风险敞口；流动性及融资风险；集中度激励" }, { "id": "dan-sundheim", "name": "Dan Sundheim", "side": "buy", "role": "D1 Capital创始人、CIO（2025/2026访谈时）", "scope": "前置资本投入与采用时滞；退出倍数及回报；AI产品差异化", "focus": [ "前置资本投入与采用时滞", "退出倍数及回报", "AI产品差异化" ], "boundary": "本人具名公开论证的方法样本；信件中的机构过程保留团队属性；题目全部编辑合成。", "timeline": [ { "date": "2025-10-22", "text": "三年预期回报依赖退出倍数；股价提前兑现会降低未来IRR，但过早卖出长期优秀企业同样有机会成本。", "source_ids": [ "O9G2-D25" ] }, { "date": "2026-02-24", "text": "把LLM训练视为前置固定投资，并检验采用速度、后续资本供给与回报；Netflix式成本摊薄与Spotify式个性化黏性是类比，模型商品化及延迟采用可能破坏回报。", "source_ids": [ "O9G2-D26" ] } ], "gaps": [ "两份公开材料仅支持有限方法样本，未认证完整或稳定个人投资哲学", "镜像转录未逐字比对原音频/官方原件；未审计预测结果和投资业绩" ], "comparison": "前置资本投入与采用时滞；退出倍数及回报；AI产品差异化" }, { "id": "scott-shleifer", "name": "Scott Shleifer", "side": "buy", "role": "Tiger Global Partner（2022 FII与2023通话来源期）", "scope": "渗透率与增量利润；资金续航约束", "focus": [ "渗透率与市场上限", "增量利润与资金续航" ], "boundary": "一个完整公开转录与一个原报道具名实质引语选段；仅作窄方法样本，不等同两份完整个人研究。", "timeline": [ { "date": "2022-10-26", "text": "本人讨论低渗透率市场的增长空间、高增量利润及融资依赖；属于单次方法材料，不能由此认证稳定策略或把多年后利润用于上市时点估值。", "source_ids": [ "O9G2-SH22" ] }, { "date": "2023-02-01", "text": "增长需转为利润。", "source_ids": [ "O9G2-SH23" ] } ], "gaps": [ "第二来源仅为原报道所载本人论证选段，未取得完整通话/音频；第一来源转录未核音频", "当前职务及顾问转型未由日期明确原始材料确认；不认证稳定个人哲学或基金业绩" ], "comparison": "把采用增长和股东可得现金分开，检验固定成本杠杆及融资需求。" }, { "id": "dan-niles", "name": "Dan Niles", "side": "buy", "role": "Niles Investment Management创始人／PM（两次采访时期）", "scope": "盈利改善如何与估值和仓位上限联动？", "focus": [ "盈利改善如何与估值和仓位上限联动？" ], "boundary": "两份历史公开材料的有限方法样本；不认证完整个人体系或投资业绩。两次不同情景的公开发言；政策判断是当时观点，不是已证因果。", "timeline": [ { "date": "2024-10-23", "text": "Niles从汽车毛利、交付指引过渡至估值与小仓位风险控制。", "source_ids": [ "O9G3-N1" ] }, { "date": "2025-04-07", "text": "Niles讨论政策持续时间、市场准入及科技企业的跨境风险。", "source_ids": [ "O9G3-N2" ] } ], "gaps": [ "两次不同情景的公开发言；政策判断是当时观点，不是已证因果。", "未完成完整模型、决策记录和结果审计" ], "comparison": "盈利改善如何与估值和仓位上限联动？" }, { "id": "eric-jackson", "name": "Eric Jackson", "side": "buy", "role": "EMJ Capital创始人／PM（材料时期）", "scope": "困境平台的价值能否熬过现金期限？", "focus": [ "困境平台的价值能否熬过现金期限？" ], "boundary": "两份历史公开材料的有限方法样本；不认证完整个人体系或投资业绩。公开持仓倡议与预录问题；管理层服务收入回答不归给Jackson；没有审核其百倍收益说法。", "timeline": [ { "date": "2025-11-07", "text": "Jackson预录问题追问人员规模及iBuying之外的附加服务收入路径。", "source_ids": [ "O9G3-J2" ] }, { "date": "2026-01-06", "text": "Jackson以生存能力、资产与执行的区别检验困境平台的上行空间。", "source_ids": [ "O9G3-J1" ] } ], "gaps": [ "公开持仓倡议与预录问题；管理层服务收入回答不归给Jackson；没有审核其百倍收益说法。", "未完成完整模型、决策记录和结果审计" ], "comparison": "困境平台的价值能否熬过现金期限？" }, { "id": "paul-meeks", "name": "J. Paul Meeks", "side": "buy", "role": "Harvest Portfolio Management Co-CIO／CIO（2024–2025历史买方）；现任Freedom卖方研究", "scope": "AI资本开支如何转成供应商可兑现收入？", "focus": [ "AI资本开支如何转成供应商可兑现收入？" ], "boundary": "两份历史公开材料的有限方法样本；不认证完整个人体系或投资业绩。两次采访属历史买方时期；现任卖方另记。搜索返回带时间戳正文，直连429；Lisa Su插播和主持人提问不归Meeks。", "timeline": [ { "date": "2024-08-01", "text": "Meeks区分AI资本开支支持的业务判断与财报期暂停增仓的时点决策。", "source_ids": [ "O9G3-M1" ] }, { "date": "2025-04-01", "text": "Meeks将消费与企业需求分开，要求随后财报确认AI支出，并区分基建投入与应用变现。", "source_ids": [ "O9G3-M2" ] } ], "gaps": [ "两次采访属历史买方时期；现任卖方另记。搜索返回带时间戳正文，直连429；Lisa Su插播和主持人提问不归Meeks。", "未完成完整模型、决策记录和结果审计" ], "comparison": "AI资本开支如何转成供应商可兑现收入？" }, { "id": "alex-ely", "name": "Alex Ely", "side": "buy", "role": "Macquarie美国成长股票CIO（2022采访时期，非现职认证）", "scope": "长期替代趋势是否足够补偿估值风险？", "focus": [ "长期替代趋势是否足够补偿估值风险？" ], "boundary": "两份历史公开材料的有限方法样本；不认证完整个人体系或投资业绩。原发布方两篇采访报道；Ely个人明确引语和记者转述分开；不复制个股清单或未核收益，非现任确认。", "timeline": [ { "date": "2022-04-28", "text": "第一份采访方法样本，摘要见来源记录。", "source_ids": [ "O9G3-E1" ] }, { "date": "2022-06-27", "text": "第二份独立采访方法样本，摘要见来源记录。", "source_ids": [ "O9G3-E2" ] } ], "gaps": [ "原发布方两篇采访报道；Ely个人明确引语和记者转述分开；不复制个股清单或未核收益，非现任确认。", "未完成完整模型、决策记录和结果审计" ], "comparison": "长期替代趋势是否足够补偿估值风险？" }, { "id": "james-anderson", "name": "James Anderson", "side": "buy", "role": "Baillie Gifford合伙人／Scottish Mortgage经理（2021–2022材料时期）", "scope": "少数大赢家与无法量化的不确定性如何共存？", "focus": [ "少数大赢家与无法量化的不确定性如何共存？" ], "boundary": "两份历史公开材料的有限方法样本；不认证完整个人体系或投资业绩。2021仅JA发言归Anderson，John Kay的回答保留其本人归属；日期只知月份；2022自我反思不等于完整收益审计。", "timeline": [ { "date": "2021-08", "text": "Anderson质疑风险可被波动率完全度量，提出少数公司的上行偏斜及长期技术机会问题。", "source_ids": [ "O9G3-A1" ] }, { "date": "2022-04", "text": "Anderson署名文章讨论少数公司主导回报、规模收益递增及低估Microsoft的反思。", "source_ids": [ "O9G3-A2" ] } ], "gaps": [ "2021仅JA发言归Anderson，John Kay的回答保留其本人归属；日期只知月份；2022自我反思不等于完整收益审计。", "未完成完整模型、决策记录和结果审计" ], "comparison": "少数大赢家与无法量化的不确定性如何共存？" }, { "id": "o9g4-b57", "name": "Michael Pye", "side": "buy", "role": "Baillie Gifford LTGG投资经理（2023–2024材料时期）；2026年退出团队，不列现任", "scope": "公开论证 → 机制拆解 → 反例与合成迁移", "focus": [ "财务韧性", "逆周期再投资", "增长质量" ], "boundary": "两份材料支持有限公开方法样本，不认证完整个人体系、持续有效性、真实执行或业绩。仅归属受访者明确署名回答，不把主持人或其他受访者观点合并。", "timeline": [ { "date": "2023-04", "text": "Pye把现金储备与逆周期研发及人才投入的能力相联系。", "source_ids": [ "O9G4-B57-A" ] }, { "date": "2024-01", "text": "Pye用增量收入成本与现金生成能力解释规模化，提醒价格与经营变化可以不同步。", "source_ids": [ "O9G4-B57-B" ] } ], "gaps": [ "未取得完整投资决策、仓位与结果审计", "需更多时期和反面证据验证方法边界" ], "comparison": "编辑比较重心：财务韧性、逆周期再投资、增长质量；非能力排序。" }, { "id": "o9g4-b61", "name": "冯明远", "side": "buy", "role": "信达澳亚基金；资料时期买方角色，现职另见候选身份记录", "scope": "公开论证 → 机制拆解 → 反例与合成迁移", "focus": [ "亏损归因", "产品竞争力", "规模与纠错" ], "boundary": "两份材料支持有限公开方法样本，不认证完整个人体系、持续有效性、真实执行或业绩。仅归属受访者明确署名回答，不把主持人或其他受访者观点合并。", "timeline": [ { "date": "2020-02-14", "text": "冯明远区分研发投入和竞争力衰退造成的亏损，并讨论规模增长对纠错和流动性的约束。", "source_ids": [ "O9G4-B61-A" ] }, { "date": "2020-10-15", "text": "冯明远把企业、行业、财务及动态估值结合，解释技术迭代下分散持股的用途；李淑彦周期评论不归于冯。", "source_ids": [ "O9G4-B61-B" ] } ], "gaps": [ "未取得完整投资决策、仓位与结果审计", "需更多时期和反面证据验证方法边界" ], "comparison": "编辑比较重心：亏损归因、产品竞争力、规模与纠错；非能力排序。" }, { "id": "o9g4-b62", "name": "刘格菘", "side": "buy", "role": "广发基金；资料时期买方角色，现职另见候选身份记录", "scope": "公开论证 → 机制拆解 → 反例与合成迁移", "focus": [ "供给约束", "价格利润弹性", "需求落地" ], "boundary": "两份材料支持有限公开方法样本，不认证完整个人体系、持续有效性、真实执行或业绩。基金管理人报告为机构观点；经理任职表仅建立责任时期联系，非个人独著证明。", "timeline": [ { "date": "2025-10-28", "text": "管理人用壁垒、可落地需求和供给约束筛选，估值上升后更严格检查供需。", "source_ids": [ "O9G4-B62-A" ] }, { "date": "2026-03-30", "text": "管理人将AI需求与受限供给、价格弹性及企业利润相接，商业闭环是当时预期而非已实现事实。", "source_ids": [ "O9G4-B62-B" ] } ], "gaps": [ "未取得完整投资决策、仓位与结果审计", "需更多时期和反面证据验证方法边界" ], "comparison": "编辑比较重心：供给约束、价格利润弹性、需求落地；非能力排序。" }, { "id": "o9g4-b64", "name": "陈颖", "side": "buy", "role": "金鹰基金；资料时期买方角色，现职另见候选身份记录", "scope": "公开论证 → 机制拆解 → 反例与合成迁移", "focus": [ "AI终端与应用", "渗透到付费", "成长与估值" ], "boundary": "两份材料支持有限公开方法样本，不认证完整个人体系、持续有效性、真实执行或业绩。基金管理人报告为机构观点；经理任职表仅建立责任时期联系，非个人独著证明。", "timeline": [ { "date": "2025-03-31", "text": "管理人提出终端AI渗透向企业应用传导的假设，并在成长性和估值之间筛选。", "source_ids": [ "O9G4-B64-A" ] }, { "date": "2026-03-31", "text": "管理人把模型、产业数据与应用场景结合视为机会，同时保留成长和低估值两个筛选方向。", "source_ids": [ "O9G4-B64-B" ] } ], "gaps": [ "未取得完整投资决策、仓位与结果审计", "需更多时期和反面证据验证方法边界" ], "comparison": "编辑比较重心：AI终端与应用、渗透到付费、成长与估值；非能力排序。" }, { "id": "o9g4-b65", "name": "周克平", "side": "buy", "role": "华夏基金经理（2021–2022材料时期）；2025-03-06已卸任创新未来，后续职务未核验", "scope": "公开论证 → 机制拆解 → 反例与合成迁移", "focus": [ "前瞻自由现金流", "前置投入", "周期与结构" ], "boundary": "两份材料支持有限公开方法样本，不认证完整个人体系、持续有效性、真实执行或业绩。仅归属受访者明确署名回答，不把主持人或其他受访者观点合并。", "timeline": [ { "date": "2021-12-28", "text": "周克平按现金流和企业演化区分项目、产品、平台和生态，并强调前瞻自由现金流。", "source_ids": [ "O9G4-B65-A" ] }, { "date": "2022-06-11", "text": "周克平区分周期扰动和结构增长，并用前置投入、后置利润解释估值判断；其他八位回答排除。", "source_ids": [ "O9G4-B65-B" ] } ], "gaps": [ "未取得完整投资决策、仓位与结果审计", "需更多时期和反面证据验证方法边界" ], "comparison": "编辑比较重心：前瞻自由现金流、前置投入、周期与结构；非能力排序。" }, { "id": "o9g4-b67", "name": "金梓才", "side": "buy", "role": "财通基金；资料时期买方角色，现职另见候选身份记录", "scope": "公开论证 → 机制拆解 → 反例与合成迁移", "focus": [ "产品升级与短缺涨价", "产业链利润", "竞争供给" ], "boundary": "两份材料支持有限公开方法样本，不认证完整个人体系、持续有效性、真实执行或业绩。基金管理人报告为机构观点；经理任职表仅建立责任时期联系，非个人独著证明。", "timeline": [ { "date": "2025-03-29", "text": "管理人把海外与国内算力投资节奏作比较，同时以租金优势解释线下折扣业态机会。", "source_ids": [ "O9G4-B67-A" ] }, { "date": "2026-03-31", "text": "管理人将量价变化拆成产品材料升级与供不应求，讨论PCB产业链供给和竞争约束。", "source_ids": [ "O9G4-B67-B" ] } ], "gaps": [ "未取得完整投资决策、仓位与结果审计", "需更多时期和反面证据验证方法边界" ], "comparison": "编辑比较重心：产品升级与短缺涨价、产业链利润、竞争供给；非能力排序。" }, { "id": "o9g5-b68", "name": "陆彬", "side": "buy", "role": "汇丰晋信；资料时期角色，未由历史资料核验2026年现职", "scope": "公开材料 → 方法样本 → 反例 → 合成迁移", "focus": [ "供给出清", "量价利润桥", "估值切换风险" ], "boundary": "两份公开材料构成有限方法样本，未认证个人永久方法、预测能力或投资表现。基金报告作者为管理人；经理表只证明报告期任职，不证明本人独著或独立完成所有决策。", "timeline": [ { "date": "2023-03-17", "text": "报告复盘新能源估值切换未兑现及错过调整窗口，区分终端增长和产能过剩导致的增量不增利。", "source_ids": [ "O9G5-B68A" ] }, { "date": "2025-07-21", "text": "报告区分锂电竞争格局、光伏控产价格及风电需求，使用产能利用率与盈利优势判断出清。", "source_ids": [ "O9G5-B68B" ] } ], "gaps": [ "未审计真实完整决策与结果", "方法样本不代表个人全部观点" ], "comparison": "编辑比较维度：供给出清、量价利润桥、估值切换风险；不按名气或未审计回报排名。" }, { "id": "o9g5-b69", "name": "尤国梁", "side": "buy", "role": "长城基金；资料时期角色，未由历史资料核验2026年现职", "scope": "公开材料 → 方法样本 → 反例 → 合成迁移", "focus": [ "产业链轮动", "应用付费转化", "催化与经营验证" ], "boundary": "两份公开材料构成有限方法样本，未认证个人永久方法、预测能力或投资表现。基金报告作者为管理人；经理表只证明报告期任职，不证明本人独著或独立完成所有决策。", "timeline": [ { "date": "2025-03-29", "text": "报告说明从国产算力到AI终端、智能驾驶及应用布局的变化，将行业趋势与预期落地分开。", "source_ids": [ "O9G5-B69A" ] }, { "date": "2025-08-29", "text": "报告解释开源模型推动应用布局，并将关税敏感性、军工业绩与事件催化共同用于配置判断。", "source_ids": [ "O9G5-B69B" ] } ], "gaps": [ "未审计真实完整决策与结果", "方法样本不代表个人全部观点" ], "comparison": "编辑比较维度：产业链轮动、应用付费转化、催化与经营验证；不按名气或未审计回报排名。" }, { "id": "o9g5-b73", "name": "谢治宇", "side": "buy", "role": "兴证全球基金；资料时期角色，未由历史资料核验2026年现职", "scope": "公开材料 → 方法样本 → 反例 → 合成迁移", "focus": [ "AI终端渗透", "效率与资本开支", "长期估值假设" ], "boundary": "两份公开材料构成有限方法样本，未认证个人永久方法、预测能力或投资表现。基金报告作者为管理人；经理表只证明报告期任职，不证明本人独著或独立完成所有决策。", "timeline": [ { "date": "2025-04-22", "text": "报告将技术变革的非线性增长、细分景气反转与风险收益条件结合，讨论AI终端及资本开支传导。", "source_ids": [ "O9G5-B73A" ] }, { "date": "2025-10-28", "text": "报告对AI资本开支热潮保留谨慎，要求检验终端渗透、模型效率与硬件迭代对长期价值的影响。", "source_ids": [ "O9G5-B73B" ] } ], "gaps": [ "未审计真实完整决策与结果", "方法样本不代表个人全部观点" ], "comparison": "编辑比较维度：AI终端渗透、效率与资本开支、长期估值假设；不按名气或未审计回报排名。" }, { "id": "o9g5-b74", "name": "归凯", "side": "buy", "role": "嘉实基金；资料时期角色，未由历史资料核验2026年现职", "scope": "公开材料 → 方法样本 → 反例 → 合成迁移", "focus": [ "商业质量", "周期与估值", "战略战术配置" ], "boundary": "两份公开材料构成有限方法样本，未认证个人永久方法、预测能力或投资表现。基金报告作者为管理人；经理表只证明报告期任职，不证明本人独著或独立完成所有决策。", "timeline": [ { "date": "2020-07-07", "text": "归凯将商业模式、竞争优势及产业趋势拆开评估，并提醒成长行业仍有库存周期和回报率约束。", "source_ids": [ "O9G5-B74A" ] }, { "date": "2025-07-21", "text": "管理人区分宏观修复与企业成本价格压力，并披露增持芯片、消费电子和面板、减持医疗器械的操作。", "source_ids": [ "O9G5-B74B" ] } ], "gaps": [ "未审计真实完整决策与结果", "方法样本不代表个人全部观点" ], "comparison": "编辑比较维度：商业质量、周期与估值、战略战术配置；不按名气或未审计回报排名。" }, { "id": "o9g5-b83", "name": "Philip Fisher", "side": "buy", "role": "历史投资作者；资料时期角色，未由历史资料核验2026年现职", "scope": "公开材料 → 方法样本 → 反例 → 合成迁移", "focus": [ "研发商业化", "时间滞后", "价值与重估" ], "boundary": "两份公开材料构成有限方法样本，未认证个人永久方法、预测能力或投资表现。历史方法研究，不代表现任覆盖或当下投资建议。两份为不同原著授权节选，未读全书；只分析所读段落，不将再版年当成观点形成时间。", "timeline": [ { "date": "1996", "text": "Fisher强调研发需转化为商业利润，失败项目和长时间滞后令当期研发金额难以直接衡量价值。（1996为所读版本版权/出版年；原著1958年，不是新发表观点）", "source_ids": [ "O9G5-B83A" ] }, { "date": "2007", "text": "Fisher将管理能力及商业研发进步与市场缓慢重估联系起来，要求区分价值变化和价格已计入程度。（2007为所读再版版权年；原著1960年，不是当时新观点）", "source_ids": [ "O9G5-B83B" ] } ], "gaps": [ "未审计真实完整决策与结果", "方法样本不代表个人全部观点" ], "comparison": "编辑比较维度：研发商业化、时间滞后、价值与重估；不按名气或未审计回报排名。" }, { "id": "o9g5-b84", "name": "Peter Lynch", "side": "buy", "role": "Fidelity Investments；资料时期角色，未由历史资料核验2026年现职", "scope": "公开材料 → 方法样本 → 反例 → 合成迁移", "focus": [ "业务理解", "行业事实", "资产负债表" ], "boundary": "两份公开材料构成有限方法样本，未认证个人永久方法、预测能力或投资表现。历史方法研究，不代表现任覆盖或当下投资建议。演讲为第三方转录主体，录音与Q&A未核，日期已按主办方更正。", "timeline": [ { "date": "1997-01-14", "text": "Lynch强调公司研究和保留基本面改善的赢家，容许部分判断错误；市场下跌不能替代业务分析。（关联节目播出日，PBS tapes页已核；采访录制日和网页上传日未标明）", "source_ids": [ "O9G5-B84A" ] }, { "date": "1994-10-07", "text": "演讲将熟悉业务与资产负债表研究结合，以行业库存、占用率等可观察指标替代宏观择时，反对仅因跌价而买入。（演讲日按NPC官方2013-12-06历史回顾；转录发布2021-02-22）", "source_ids": [ "O9G5-B84B" ] } ], "gaps": [ "未审计真实完整决策与结果", "方法样本不代表个人全部观点" ], "comparison": "编辑比较维度：业务理解、行业事实、资产负债表；不按名气或未审计回报排名。" }, { "id": "o9g6-francois-meunier", "name": "François Meunier", "side": "sell", "role": "Morgan Stanley分析师（2017来源时）；现职Norman K资管已另核", "scope": "毛利率桥接：组合与收购会计分开", "focus": [ "European semiconductors" ], "boundary": "加回PPA摊销后的毛利率不是现金流指标。；仅按材料时期归因，不代表当前持仓或稳定个人风格。", "timeline": [ { "date": "2017-01-18", "text": "Meunier追问收购价格分摊对毛利率的影响；管理层明确不能把该调整直接称现金毛利率。", "source_ids": [ "O9G6-ASML416" ] }, { "date": "2017-04-19", "text": "Meunier拆分毛利率同比提升来自服务/升级产品组合还是工具自身利润率。", "source_ids": [ "O9G6-ASML117" ] } ], "gaps": [ "两份公开材料仅支持有限方法样本；尚缺私人完整模型、预测复盘及跨时期方法检验。" ], "comparison": "毛利率=各业务收入权重×业务毛利率之和；PPA调整另列。" }, { "id": "o9g6-matt-bryson", "name": "Matt Bryson", "side": "sell", "role": "Wedbush分析师（2024及2026来源时）", "scope": "把去库存与交付延期从需求中拆开", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "去库存结束和项目延期逆转均不自动等于终端需求加速。；仅按材料时期归因，不代表当前持仓或稳定个人风格。", "timeline": [ { "date": "2024-05-02", "text": "Bryson纠正管理层把传统磷化铟去库存问题答成AI需求，追问库存压制收入的量和恢复时点。", "source_ids": [ "O9G6-AXT124" ] }, { "date": "2026-08-07", "text": "Bryson追问银行设备延期具体原因，再问高毛利来自季度特殊项目还是产品组合改善。", "source_ids": [ "O9G6-DBD226" ] } ], "gaps": [ "两份公开材料仅支持有限方法样本；尚缺私人完整模型、预测复盘及跨时期方法检验。" ], "comparison": "发货=终端消耗+渠道库存变化；订单转收入另受交付时点影响。" }, { "id": "o9g6-cody-acree", "name": "Cody Acree", "side": "sell", "role": "Drexel Hamilton（2016）、Benchmark（2024）分析师；现职StoneX另核", "scope": "技术替代必须落到供应商份额与内容量", "focus": [ "Semiconductors", "AI infrastructure" ], "boundary": "技术路线获胜不保证特定供应商收入增长；并购互补也不保证执行。；仅按材料时期归因，不代表当前持仓或稳定个人风格。", "timeline": [ { "date": "2016-09-13", "text": "Acree挑战收购方此前市场份额下滑与转型能力，追问跨地域管理团队保留。", "source_ids": [ "O9G6-ISIL16" ] }, { "date": "2024-03-28", "text": "Acree追问铜与光的架构替代及Semtech是否因某一方案获益，管理层回答需看具体应用和组合。", "source_ids": [ "O9G6-SMTC424" ] } ], "gaps": [ "两份公开材料仅支持有限方法样本；尚缺私人完整模型、预测复盘及跨时期方法检验。" ], "comparison": "供应商收入=各方案系统量×单系统内容量×份额；并购后还需验证组织执行。" }, { "id": "o9g6-seth-klarman", "name": "Seth Klarman", "side": "buy", "role": "Baupost CEO及组合经理（2025访谈时）", "scope": "拆分价值：便宜之外还需兑现路径", "focus": [ "投资方法论；非直接 AI 基建覆盖候选" ], "boundary": "自动转录公开镜像，未复听；历史案例是访谈回忆。折价不消除清算延迟、负债与现金流缩水风险。；仅按材料时期归因，不代表当前持仓或稳定个人风格。", "timeline": [ { "date": "2025-05-27", "text": "Klarman将折价与价值兑现催化剂并列，要求预先设想投资失败的原因。", "source_ids": [ "O9G6-KLGS25" ] }, { "date": "2025-08-01", "text": "Klarman回忆拆分交易：先扣清算现金，再计算所获分拆权益的隐含成本；这是本人回忆而非审计后的历史回报。", "source_ids": [ "O9G6-KLCO25" ] } ], "gaps": [ "两份公开材料仅支持有限方法样本；尚缺私人完整模型、预测复盘及跨时期方法检验。" ], "comparison": "剩余权益隐含成本=(买价−清算分配)/获配权益数量；再做分配下行情景。" }, { "id": "o9g6-david-tepper", "name": "David Tepper", "side": "buy", "role": "Appaloosa管理人（2010/2018访谈时）；不据历史材料更新现职", "scope": "政策情景与净敞口分别计算", "focus": [ "投资方法论；非直接 AI 基建覆盖候选" ], "boundary": "宏观路径不穷尽尾部风险，历史政策反应不保证重复；净敞口小不等于个股基差风险小。；仅按材料时期归因，不代表当前持仓或稳定个人风格。", "timeline": [ { "date": "2010-09-24", "text": "Tepper分两条宏观路径：自主复苏或经济不振引发QE；同时说明长期持续依赖QE并非好事。", "source_ids": [ "O9G6-TEP10" ] }, { "date": "2018-09-13", "text": "Tepper把关税、汇率、美元盈利与估值联动；说明用个股多头和市场空头表达约25%的S&P等效敞口。", "source_ids": [ "O9G6-TEP18" ] } ], "gaps": [ "两份公开材料仅支持有限方法样本；尚缺私人完整模型、预测复盘及跨时期方法检验。" ], "comparison": "情景收益=概率×各情景收益之和；市场等效净敞口=多头beta敞口−空头beta敞口。" }, { "id": "o9g6-chris-hohn", "name": "Chris Hohn", "side": "buy", "role": "TCI Founder and Managing Director（2022–2023签名时）", "scope": "成本削减须计入过渡成本与收入反作用", "focus": [ "投资方法论；非直接 AI 基建覆盖候选" ], "boundary": "信中median薪酬不能直接作全员平均全负担成本；SBC经济成本与现金支出不同。；仅按材料时期归因，不代表当前持仓或稳定个人风格。", "timeline": [ { "date": "2022-11-15", "text": "Hohn要求约束成本及Other Bets损失，设置利润率目标，以现金流支持回购；信中估值与预测仅为当时主张。", "source_ids": [ "O9G6-HOHN22" ] }, { "date": "2023-01-20", "text": "Hohn认可12000减员但认为不足，建议约150000目标人数并降低薪酬/SBC；不是实际已实现的削减结果。", "source_ids": [ "O9G6-HOHN23" ] } ], "gaps": [ "两份公开材料仅支持有限方法样本；尚缺私人完整模型、预测复盘及跨时期方法检验。" ], "comparison": "经常性成本节约=人数减少×平均全负担成本；首年净收益再扣遣散费及业务损失。" }, { "id": "leopold-aschenbrenner", "name": "Leopold Aschenbrenner", "side": "buy", "role": "Situational Awareness LP 创办人（官方确认于2024年创办）", "scope": "AI规模外推、物理交付约束与投资时序", "focus": [ "AI 基建", "算力与电力供给", "主题收入敞口", "兑现时序与资本生存" ], "boundary": "2024论文与同期访谈是前瞻方法样本；不是2026年已实现结果，也不证明基金回报。主持人质疑与研究者挑战不归为本人原话。", "timeline": [ { "date": "2024-06", "text": "以规模趋势估算集群需求，并将整座设施成本、电力和建设周期列入分析。", "source_ids": [ "FM-LA-ESSAY24" ] }, { "date": "2024-06-04", "text": "讨论主题收入占比、投资先后顺序与判断延迟时的承受能力；同场承认世界观需要更新。", "source_ids": [ "FM-LA-DW24" ] }, { "date": "2026-10-09（身份核验日）", "text": "官方简介确认2024年创办关系；不据此补写规模、仓位或业绩。", "source_ids": [ "FM-LA-IDENTITY" ] } ], "gaps": [ "尚缺公开完整组合风险规则与跨时期方法复盘。", "未审计2024年的AGI、电力、收入或集群时间表是否实现。", "两份方法材料时间接近且相互关联，不能证明长期一致性。" ], "comparison": "把技术需求、按时交付、收入捕获和股东回报分成不同待验证环节。" }, { "id": "rob-jesudason", "name": "Rob Jesudason", "side": "buy", "role": "Serendipity Capital 联合创始人、Partner、CEO & CIO", "scope": "永久资本、关键技术尽调与工程商业化", "focus": [ "永久资本", "量子技术", "跨路线交叉验证", "工程规模化" ], "boundary": "本人将机构描述为永久资本公司，不等同传统可赎回基金。两次访谈是方法样本；技术优胜判断和投资自述未独立验证，主持人的QuEra商业身份需保留。", "timeline": [ { "date": "2022（索引年份）", "text": "讨论资本期限与行业客户需求验证。", "source_ids": [ "FM-RJ-EURO" ] }, { "date": "2026-08-17", "text": "主张跨量子技术路线交叉求证，并检验规模交付能力。", "source_ids": [ "FM-RJ-POD26" ] }, { "date": "2026-10-09（身份核验日）", "text": "官方团队页确认创办人与CEO/CIO角色。", "source_ids": [ "FM-RJ-IDENTITY" ] } ], "gaps": [ "尚缺完整公开投资决策记录和失败案例复盘。", "永久资本不免除估值、融资与流动性风险。", "访谈观点不证明任何量子技术路线已胜出。" ], "comparison": "同时检验技术证据、工程交付和资本期限；产业前景与具体证券回报须分开判断。" } ], "cards": [ { "id": "CW01", "profile": "wood-ark", "title": "成本下降，需求与支出会怎样变化？", "source_ids": [ "D01" ], "locator": "Sizing the Opportunity", "source_paraphrase": "研究成本曲线与需求弹性。", "attribution": "团队/材料层面的转述；模型、边界、挑战均为我们的编辑归纳，不冒充本人。", "model": "合成模型：Q∝P^(−ε)，支出=P×Q；成本≠售价。", "boundary": "在可比质量、价格传导与弹性成立时，降价推动支出增加需要ε>1。", "counterexample": "新增调用主要免费，或效率提升减少每任务资源。", "test": "同质量付费成功任务、实现价格、资源/任务、贡献与采购。", "question": "合成：售价下降50%，付费成功任务增70%；哪些结论可以推出，哪些不可以？", "constraint": "新增负载先由原有闲置容量承接；修改采购模型。", "transfer": "迁移到SSD承接DRAM：容量节省与新增CPU/I/O成本怎样分开？", "tags": [ "F6", "F10", "F14", "T2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "CW02", "profile": "wood-ark", "title": "大市场怎样落到公司与股东？", "source_ids": [ "D01", "D02" ], "locator": "Stock Selection and Valuation; FAQs on investable opportunities", "source_paraphrase": "把技术机会进一步转为公司和估值研究。", "attribution": "团队/材料层面的转述；模型、边界、挑战均为我们的编辑归纳，不冒充本人。", "model": "收入=可服务市场×渗透×内容×份额；EPS与现金独立闭合。", "boundary": "TAM只是上界，需容量、资格、竞争、融资与稀释。", "counterexample": "市场扩大但份额流失，利润由买方获取。", "test": "同口径市场分层、单位利润、竞争响应、股数与现金融资。", "question": "合成：市场增50%，份额20%→12%，收入一定增吗？", "constraint": "增发使股数增加20%；从收入继续推到每股回报，列出缺口。", "transfer": "迁移HBM到光模块，重建内容、端口数与资格变量。", "tags": [ "F4", "F6", "F12", "T4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "CW03", "profile": "wood-ark", "title": "跨技术融合：累加还是相互依赖？", "source_ids": [ "D01", "D02" ], "locator": "Cross-theme collaboration; technology convergence section", "source_paraphrase": "强调跨技术主题相互作用。", "attribution": "团队/材料层面的转述；模型、边界、挑战均为我们的编辑归纳，不冒充本人。", "model": "端到端成功概率/产出由依赖链决定；比较独立、互补与替代。", "boundary": "只有明确接口、瓶颈与去重边界后才能量化。", "counterexample": "两个大市场其实是同一笔客户预算；一项延迟拖累全部。", "test": "依赖里程碑、边界BOM、预算重叠、强相关风险。", "question": "AI、机器人、能源三项改善各20%，能否推导系统改善60%？", "constraint": "能源供应延迟一年；重建可用产出而非相加主题。", "transfer": "迁移机架中的CPU、HBM、网络、冷却互补性。", "tags": [ "F6", "F7", "F9", "T3" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "CW04", "profile": "wood-ark", "title": "长周期信念怎样接受证伪？", "source_ids": [ "D01", "D02" ], "locator": "Portfolio and Risk Management; efficient-frontier discussion", "source_paraphrase": "公开说明监测投资论点，并讨论组合情景。", "attribution": "团队/材料层面的转述；模型、边界、挑战均为我们的编辑归纳，不冒充本人。", "model": "区分业务论点、估值、时间、流动性与组合共同因子。", "boundary": "模拟组合不证明未来分散效果；公开信不暴露完整决策。", "counterexample": "多家公司主题不同，但共享利率、融资、同一客户风险。", "test": "事前条件、现金跑道、集中度、因子暴露与修改日志。", "question": "两个主题基金相关性过去很低，能否证明融资冲击时仍分散？", "constraint": "两者客户都靠融资扩容；提出压力场景与止错证据。", "transfer": "迁移周期性内存供应商与数据中心经营者的共同融资风险。", "tags": [ "F11", "F12", "F14", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "MM01", "profile": "meeker-bond", "title": "区域先行者能否作为全球模板？", "source_ids": [ "D03", "D04" ], "locator": "2009 key takeaways; 2025 overview/usage sections", "source_paraphrase": "用采用与使用经验观察技术扩散。", "attribution": "团队/材料层面的转述；模型、边界、挑战均为我们的编辑归纳，不冒充本人。", "model": "分解人群、地区、分发、付费转化与单位服务经济性。", "boundary": "跨地域比较先统一用户、时间起点、可获得性与付费口径。", "counterexample": "早期地区收入高、受补贴或基础设施更好。", "test": "匹配用户群、持续留存、付费成功任务和渠道成本。", "question": "一个地区agent月活翻倍，怎样检验全球会复制其采用曲线？", "constraint": "新增用户主要免费；设定两个季度的推翻条件。", "transfer": "从mobile到agentic，再到企业工作流，说明不能迁移的变量。", "tags": [ "F1", "F6", "F10", "T3" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "MM02", "profile": "meeker-bond", "title": "技术汇合不等于赢家已经确定", "source_ids": [ "D03", "D04" ], "locator": "2009 technology convergence; 2025 slides213–214", "source_paraphrase": "比较技术组合及平台/专业化的潜在商业路径。", "attribution": "团队/材料层面的转述；模型、边界、挑战均为我们的编辑归纳，不冒充本人。", "model": "比较平台广度、专业深度、分发、数据和工作流控制。", "boundary": "主题成立仍需市场份额与付费转化；不把报告写成单一推荐。", "counterexample": "专业厂商保有合规工作流，平台未能捕获利润。", "test": "部署留存、采购主体、切换成本、合同净价与实施成本。", "question": "通用平台用户是专业工具的10倍，能否断言平台利润池更大？", "constraint": "专业工具每用户毛利20倍；重新比较可服务客户。", "transfer": "迁移通用GPU与定制ASIC，加入软件迁移/资格成本。", "tags": [ "F2", "F6", "F12", "T4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "MM03", "profile": "meeker-bond", "title": "使用、成本与利润可以反向变化", "source_ids": [ "D04" ], "locator": "slides116–119 and130", "source_paraphrase": "把AI使用扩张与商业化、成本问题并列。", "attribution": "团队/材料层面的转述；模型、边界、挑战均为我们的编辑归纳，不冒充本人。", "model": "成功任务×收入/任务−变动成本−固定成本−资本负担。", "boundary": "成本降低不必转化为价格权；需要付费与现金回收。", "counterexample": "供应商被竞争压价，或债务扩容闲置。", "test": "付费贡献、可用容量、净采购、融资期限与合同返还义务。", "question": "AI使用翻倍，成本/次下降40%，为什么公司利润可能下降？", "constraint": "价格降60%，新容量借款融资；重算贡献与现金缺口。", "transfer": "迁移云模型服务到内存供应，考虑成本库存确认与售价合同。", "tags": [ "F9", "F10", "F11", "T2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "MM04", "profile": "meeker-bond", "title": "图表密集：如何避免数据制造确定性？", "source_ids": [ "D03", "D04" ], "locator": "2009 data-rich framework; 2025 chart/source notes", "source_paraphrase": "以多指标呈现技术趋势。", "attribution": "团队/材料层面的转述；模型、边界、挑战均为我们的编辑归纳，不冒充本人。", "model": "先统一口径与来源，再构造竞争解释；不按图表数投票。", "boundary": "来源复用不是独立信号；峰值与典型、估计与实测分开。", "counterexample": "多个图都来自同一个估计，或算法/精度/任务变化造成效率跳跃。", "test": "每个决定性指标追溯原始单位、任务、时间与误差。", "question": "五张图都显示AI效率跃升，怎样判断它们是否构成独立证据？", "constraint": "都引用同一厂商演示且质量条件不同；设计可比测试。", "transfer": "迁移财报增长图到CPU benchmark与机架节能，追踪被转移成本。", "tags": [ "F1", "F10", "F13", "T6" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "HM01", "profile": "marks-oaktree", "title": "好技术不等于任何价格都值得", "source_ids": [ "D05", "D06" ], "locator": "D05 guiding principles; D06 Value", "source_paraphrase": "将经营品质与购买价格分开。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "合成：股价回报=(EPS1/EPS0)×(PE1/PE0)−1；基准口径相同，不含分红。", "boundary": "EPS与倍数时点和稀释口径匹配；不拿周期峰值充当正常盈利。", "counterexample": "业务增长但利润、股数或支付倍数恶化。", "test": "可持续盈利区间、融资成本、稀释、进入价格与反向估值。", "question": "合成：EPS增40%，P/E从50降至30，股价回报怎样？", "constraint": "若EPS增40%仅来自回购，现金和债务代价怎样进入模型？", "transfer": "迁移到高端CPU：性能更好但采购单价上升，如何判断值得？", "tags": [ "F12", "F14", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "HM02", "profile": "marks-oaktree", "title": "泡沫是心理诊断，不是顶部时钟", "source_ids": [ "D05" ], "locator": "What Is a Bubble; three stages of bull market", "source_paraphrase": "关注无视价格的乐观心理。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：价格内含假设→可证伪边界；概率情景而非精确崩盘日。", "boundary": "心理信号需与价格、经营及融资证据组合。", "counterexample": "高倍数被超预期现金流验证，热情不等于立即下跌。", "test": "反向估值要求、融资质量、失败容忍度与反例材料。", "question": "合成：公司订单增长而市场宣称没有价格太高；你能推出什么，不能推出什么？", "constraint": "空头需支付借券成本且公司一年内继续增长，改变研究表达。", "transfer": "迁移到HBM采购：紧缺和囤货叙事怎样与真实订单分开？", "tags": [ "F13", "F14", "T6" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "HM03", "profile": "marks-oaktree", "title": "价值范围怎样变成安全边际", "source_ids": [ "D06" ], "locator": "Value; fundamentals and earning power", "source_paraphrase": "价值依赖盈利能力，估值带有不确定性。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：估值区间V=[Vbear,Vbase,Vbull]；价格与下行损失、流动性联合比较。", "boundary": "情景必须包括永久损失；折价只相对于可靠且可实现的价值。", "counterexample": "资产账面价值高但过时、不能出售或债权人优先。", "test": "可实现资产价值、债务层级、现金消耗与情景假设。", "question": "合成：价值估计80/120/180，股价100；为什么不能只称便宜20？", "constraint": "下行情景还有40债务到期且不能再融资，修订权益桥。", "transfer": "迁移到旧GPU租赁：低购价能否抵消残值和利用率风险？", "tags": [ "F11", "F12", "F14", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "SD01", "profile": "druckenmiller-duquesne", "title": "先小规模验证，再随证据修正", "source_ids": [ "D07" ], "locator": "E4 NVIDIA recollection", "source_paraphrase": "回顾先建较小头寸，再因新证据增加。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：假设→有限风险试验→独立证据更新→复核规模；并非无研究下注。", "boundary": "独立信息和风险承受是必要条件；事后叙述不证明当时概率。", "counterexample": "多个专家重复同一上游故事，错误一起放大。", "test": "证据独立性、客户付费、供给兑现、事前损失预算。", "question": "合成：两位专家和一个模型演示都支持agent需求；这是三份独立证据吗？", "constraint": "发现两位专家的数据都来自同一供应商，如何调整验证计划？", "transfer": "迁移到CXL采购：从实验机到部署规模需要哪些独立证据？", "tags": [ "F1", "F6", "F14", "T6" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "SD02", "profile": "druckenmiller-duquesne", "title": "预测正确与风险规模是两件事", "source_ids": [ "D07" ], "locator": "E4 Soros sizing discussion; portfolio matrix", "source_paraphrase": "强调判断之外的盈亏规模。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "合成：期望盈亏=w[p×G−(1−p)×L]；另检验尾部、相关性与流动性。", "boundary": "p必须标为假设；均值不能代替最坏损失。", "counterexample": "共同融资冲击同时伤害股票和看似对冲的资产。", "test": "情景联合损失、变现时间、保证金和风险预算。", "question": "合成：假定p=60%，上涨50%、下跌40%；期望回报与20%权重的组合影响？", "constraint": "另一个80%组合资产与该主题共享融资风险，原结论哪里不足？", "transfer": "迁移到CPU供应商选择：预测胜率与集中采购损失怎样分开？", "tags": [ "F10", "F14", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "SD03", "profile": "druckenmiller-duquesne", "title": "旧信号被普及后，还剩多少信息", "source_ids": [ "D07" ], "locator": "E4 unlearning; technical analysis and price/news", "source_paraphrase": "说明过去有效的价格/新闻信号不再同样可靠。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：检验信号条件分布和样本外增量，而非只看历史命中率。", "boundary": "事件定义、时期、成本与选择偏差必须固定。", "counterexample": "算法提前交易或研究拥挤使公开信号迅速消失。", "test": "冻结规则，跨时期留出样本，比较基准与成本后效果。", "question": "合成：十年前坏财报后反涨的公司常继续涨；今天能直接套用吗？", "constraint": "最近20例只有公布当天上涨，无后续超额收益，如何修改？", "transfer": "迁移到存储周期：过去库存拐点信号在长期合同下是否仍有效？", "tags": [ "F8", "F13", "F14", "T6" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "DL01", "profile": "loeb-thirdpoint", "title": "HBM差异化能否变成穿越周期的利润", "source_ids": [ "D09" ], "locator": "pp3–4 SK Hynix", "source_paraphrase": "将HBM技术差异化与利润稳定性联系。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：利润=HBM/DDR数量×各自价格−晶圆、封装与研发成本；再接周期情景。", "boundary": "必须有良率、客户认证与价格持续性；溢价不等于超额利润。", "counterexample": "竞争认证追上、单位成本上升或客户转ASIC内存路径。", "test": "客户续代认证、毛利/投入资本、供给分配和替代路线。", "question": "合成：HBM占bits10%、价格为DDR4倍；收入占比是多少，能推出利润占比吗？", "constraint": "HBM每bit晶圆消耗也4倍且封装成本高，重新评价溢价。", "transfer": "迁移到高端CPU：ASP溢价与大die/封装成本怎样桥接？", "tags": [ "F2", "F7", "F10", "T4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "DL02", "profile": "loeb-thirdpoint", "title": "控股公司折价不是自动兑现的收益", "source_ids": [ "D09" ], "locator": "p4 SK Square valuation", "source_paraphrase": "讨论控股公司折价与回购催化剂。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "合成：价格=NAV×(1−d)；回报=(NAV1/NAV0)×[(1−d1)/(1−d0)]−1。", "boundary": "NAV需扣债务税费；少数股东价值取决于治理与可实现催化剂。", "counterexample": "折价长期扩大；低流动性、关联交易或底层资产下跌。", "test": "回购执行、持股比例、债务、治理及资产处置进度。", "question": "合成：折价60%，NAV增20%，折价不变，回报多少？", "constraint": "若折价扩大到70%，同样NAV增20%时回报怎样？", "transfer": "迁移到芯片集团分部估值：分拆价值为什么不等于现股价值？", "tags": [ "F11", "F12", "F14", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "DL03", "profile": "loeb-thirdpoint", "title": "行业机会与融资风险可以同时成立", "source_ids": [ "D08", "D09" ], "locator": "D08 Review and Outlook; D09 credit pp7–8", "source_paraphrase": "调整敞口，同时警惕AI资本投入的不平滑。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：客户运营现金+已承诺融资−CapEx−利息/到期债务；检验每期资金缺口。", "boundary": "回报与融资时点必须匹配；未承诺融资不是现金。", "counterexample": "最终需求增长但资金先断，供应商订单延迟或取消。", "test": "融资条款、最低流动性、项目通电日期、合同可取消性。", "question": "合成：年现金来源8，CapEx10，利息1；资金缺口多少，技术前景能填补吗？", "constraint": "融资只在通电验收后发放，建设期怎样桥接？", "transfer": "迁移到DRAM扩产：订单保底能否替代建设期融资？", "tags": [ "F9", "F11", "F14", "T1", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "JM01", "profile": "moore-ms", "title": "全球市场与可合法销售市场分开", "source_ids": [ "D10" ], "locator": "p11 Joe Moore Q&A", "source_paraphrase": "追问中国业务与新平台许可。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：可达收入=需求×合规可售比例×竞争份额×交付比例。", "boundary": "法规按地区/型号/日期核验；不能将许可等同于客户订单。", "counterexample": "市场很大但许可、产品竞争力或采购能力不足。", "test": "官方许可条款、可售型号、客户订单和实际交付。", "question": "合成：地区需求100，可合规满足40%，公司份额50%；能确认收入20吗？", "constraint": "许可获批但客户转换平台需一年，收入时间如何修订？", "transfer": "迁移到服务器CPU出口：同ISA不等于同一合规路径，如何建模？", "tags": [ "F3", "F6", "F14", "T1" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "JM02", "profile": "moore-ms", "title": "累计机会怎样桥到季度收入", "source_ids": [ "D11" ], "locator": "pp8–9 Joseph Moore Q&A", "source_paraphrase": "追问累计平台机会与已出货额。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：剩余机会=同口径累计机会−已实现额；按季度交付且区分预测/订单。", "boundary": "范围、日历/财年与重复订单必须一致。", "counterexample": "累计预测被当成不可取消订单，或交付推迟超出窗口。", "test": "已确认收入、绑定订单、增量订单、交付与取消率。", "question": "合成：两年机会500、已交付150，剩余350能平均写成每季87.5吗？", "constraint": "只有200具绑定合同，剩余150是预测，修改季度模型。", "transfer": "迁移到HBM backlog：累计合同与年收入怎样分期？", "tags": [ "F3", "F4", "F9", "T6" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "JM03", "profile": "moore-ms", "title": "押金、订单与收入不是一个账户", "source_ids": [ "D12" ], "locator": "Moore Q&A lines447–491", "source_paraphrase": "追问押金的合同作用及是否预付款。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：押金现金/负债、RPO和收入履约分别滚动；遵循合同和现金流分类。", "boundary": "根据真实条款确认可返还性和分类，不能推广所有押金。", "counterexample": "现金充裕但有返还义务；订单增长不等于履约收入。", "test": "合同、负债附注、现金流分类、履约与返还表。", "question": "合成：客户付10可返还押金；尚未交付产品。收入和自由现金流能增加10吗？", "constraint": "押金允许用于CapEx但五年后须返还，是否可全部分红？", "transfer": "迁移到云GPU预付款：履约义务与融资押金怎样区别？", "tags": [ "F11", "F14", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "TA01", "profile": "arcuri-ubs", "title": "行业CAGR不能直接复制到公司模型", "source_ids": [ "D10" ], "locator": "pp14–15 Timothy Arcuri Q&A", "source_paraphrase": "追问行业增长能否映射公司增长。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "合成：公司收入=行业可服务支出×份额×兑现比例。", "boundary": "市场定义需可比，不把总设施CapEx全当芯片市场。", "counterexample": "行业增长但客户自研、价格下降或份额受损。", "test": "同口径TAM、竞争份额、内容与当期交付。", "question": "合成：市场增50%，份额80%降60%，兑现比例不变；公司收入增长多少？", "constraint": "行业总CapEx的40%是建筑电力且比设备增得快，如何修订？", "transfer": "迁移到CPU：服务器支出增长是否等于CPU供应商增长？", "tags": [ "F4", "F6", "F9", "T4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "TA02", "profile": "arcuri-ubs", "title": "找最短板，而不是相加所有扩产", "source_ids": [ "D11" ], "locator": "pp13–14 Timothy Arcuri Q&A", "source_paraphrase": "并列追问电力、资金和芯片供应约束。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "合成：部署上限=min(算力组件、内存、供电、集成、融资所支持台数)，加入各自时滞。", "boundary": "只有归一化成同单位且同期间才能取min；是上限不是必然销量。", "counterexample": "内存供给翻倍但通电能力不变，部署没有增加。", "test": "各环节交付量、通电和验收节点、客户融资。", "question": "合成：GPU支持100架、HBM80、电力60、融资90；部署上限多少？", "constraint": "电力扩到100、HBM仍80但集成只能70；上限怎么变？", "transfer": "迁移到CPU集群：CPU、DIMM、NIC与机房资源怎样归一化？", "tags": [ "F7", "F9", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "TA03", "profile": "arcuri-ubs", "title": "合同最低额与预计收入分期", "source_ids": [ "D12" ], "locator": "Arcuri Q&A lines405–442", "source_paraphrase": "追问合同地板与逐季进入模型。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：RPO期末=期初+新约束金额−已履约±合同修订；各合同按交付时点分期。", "boundary": "RPO按可执行条款与会计定义；地板收入不是完整销售预测。", "counterexample": "同五年总额不意味着每年均匀；合同份额和收入份额分母不同。", "test": "RPO短长期间披露、量价条款、合同批次与交付。", "question": "合成：五年最低100，未来12个月披露12；为何不能把当年保底写20？", "constraint": "新合同在第四季度签且未来12个月18；会计年度收入能直接增加18吗？", "transfer": "迁移到ASIC长单：设计赢单到量产收入怎样分期？", "tags": [ "F3", "F4", "F11", "T6" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "CM01", "profile": "muse-cantor", "title": "晶圆入厂到盈利运行有几道门", "source_ids": [ "D10" ], "locator": "p8 CJ Muse Q&A", "source_paraphrase": "连接晶圆到机架与电力、冷却条件。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：晶圆→良品→封装→集成→通电→验收；按各门容量与时间核算。", "boundary": "供应商收入确认规则与客户可用时间可能不同。", "counterexample": "芯片发出但机房未通电，利用率和客户ROI为零。", "test": "各阶段在制品、验收、冷却、电力与收入政策。", "question": "合成：本季度出货100架，60通电、50验收；供应商收入和客户可用容量能都写100吗？", "constraint": "设备交付即转移控制权，但客户半年后通电；重建两方现金模型。", "transfer": "迁移到CPU服务器升级：到货和业务迁移完成为何不等价？", "tags": [ "F9", "F11", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "CM02", "profile": "muse-cantor", "title": "满负载不等于扩产永远有回报", "source_ids": [ "D11" ], "locator": "pp9–10 CJ Muse Q&A", "source_paraphrase": "追问供需追平窗口与融资/ROI背景。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：付费需求、单位资源、可用容量与资本回收联动；供需按季度而非永久叙事。", "boundary": "利用率需区分付费/内部/免费任务；融资成本与价格重要。", "counterexample": "高利用率但实现价格低、重复任务多或资本成本高。", "test": "付费成功任务、实现价格、净贡献、有效容量与资本回收期。", "question": "合成：利用率90%、单价降50%，调用增80%；收入和ROI必然提高吗？", "constraint": "客户用免费调用填满剩余容量，如何改利用率指标？", "transfer": "迁移到SSD缓存：命中率高是否必然降低端到端成本？", "tags": [ "F7", "F9", "F10", "T2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "CM03", "profile": "muse-cantor", "title": "不受限制的现金仍可能有未来返还义务", "source_ids": [ "D12" ], "locator": "Muse Q&A lines499–517", "source_paraphrase": "追问押金能否用于投资及资本回报。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：可分配现金≠银行存款；减运营缓冲、合同返还、债务和已承诺投资。", "boundary": "用途限制与负债义务分开；不是所有押金同类。", "counterexample": "把可用押金全部回购，之后返还时需高息借款。", "test": "返还时间表、最坏现金消耗、融资窗口及回购承诺。", "question": "合成：现金30含返还押金10，运营缓冲8、已承诺CapEx9；可分配上限如何估？", "constraint": "押金五年后返还但业务衰退使经营现金为负，为什么不能忽略？", "transfer": "迁移到托管采购：供应商客户资金可用性与偿付能力怎样区别？", "tags": [ "F11", "F12", "F14", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "SR01", "profile": "rasgon-bernstein", "title": "增长来自平台、网络还是重新分类", "source_ids": [ "D10" ], "locator": "p13 Stacy Rasgon Q&A", "source_paraphrase": "拆问平台与网络的季度增量。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：总收入=互斥产品桶之和；量价混合交互一次计入。", "boundary": "定义产品桶；系统总价里的网络不能再单独加一次。", "counterexample": "网络attach提高只是系统收入重分桶，不是同额新增支出。", "test": "同口径分部、系统BOM、独立与内含销售、毛利。", "question": "合成：总增量7，其中系统增6且内含网络2，独立网络增1；能说总增9吗？", "constraint": "公司把内含网络从compute改列network，怎样保持同比可比？", "transfer": "迁移到CPU+DIMM整机：芯片与服务器收入为什么不能简单相加？", "tags": [ "F2", "F4", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "SR02", "profile": "rasgon-bernstein", "title": "毛利维持靠成本、预购还是定价", "source_ids": [ "D11" ], "locator": "pp14–15 Stacy Rasgon Q&A", "source_paraphrase": "追问输入成本与毛利维护措施。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "合成：GM=1−单位成本/ASP；量、配置、成本改善与定价拆桥。", "boundary": "固定产品与会计口径；组合改善不等于同产品成本下降。", "counterexample": "预购暂时压低成本，但下一代输入涨价且无法传导。", "test": "单位BOM、预购合同、库存成本、售价与产品组合。", "question": "合成：ASP100、成本25，成本增20%而价格不变，毛利率多少？", "constraint": "ASP也增10%，GM怎样；这证明客户ROI更好吗？", "transfer": "迁移到HBM：售价涨幅与良率/封装成本怎样桥接？", "tags": [ "F2", "F4", "F10" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "SR03", "profile": "rasgon-bernstein", "title": "收入高增不能替代费用模型", "source_ids": [ "D11" ], "locator": "p15 Stacy Rasgon follow-through / OpEx", "source_paraphrase": "追问增长下的运营费用。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "合成：营业利润=收入×GM−OpEx；经营杠杆需期间匹配。", "boundary": "GAAP/非GAAP一致；SBC、资本化和一次性不可藏起。", "counterexample": "毛利率降或研发阶跃使收入增长不转为利润。", "test": "费用计划、SBC、资本化、折旧与平台研发节奏。", "question": "合成：收入100、GM75%、OpEx10；次期收入150、GM70%、OpEx20，利润增长多少？", "constraint": "新增SBC8被排除在非GAAP费用中，怎样并列呈现？", "transfer": "迁移到CPU项目：研发人力增长与量产规模如何算盈亏平衡？", "tags": [ "F2", "F4", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "MN01", "profile": "newman-bernstein", "title": "传统服务器中的CPU需求是谁买的", "source_ids": [ "D14" ], "locator": "Newman Q&A lines242–259", "source_paraphrase": "追问企业、云与AI服务器客户构成。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "编辑：按客户×工作负载×平台计数；CPU服务器标签不等于agent工作负载。", "boundary": "分类需互斥；客户类别与负载不同维度。", "counterexample": "企业更新旧服务器导致传统业务增长，被误归为agent新需求。", "test": "客户类别、应用部署、核心/台、CPU时间与旧机退役。", "question": "合成：传统服务器收入增60%，能推出agent CPU需求增60%吗？", "constraint": "新机核心/台翻倍且退役旧机更多，怎样估净CPU能力？", "transfer": "迁移到云vCPU：售出实例与底层物理核心采购怎样区分？", "tags": [ "F6", "F10", "F14", "T3" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "MN02", "profile": "newman-bernstein", "title": "台数、核心、配置与ASP分别增长多少", "source_ids": [ "D13" ], "locator": "pp15–16 Newman question", "source_paraphrase": "追问量、CPU核心和配置价格。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "合成：收入=台数×ASP；核心容量=台数×核心/台；同配置价格另列。", "boundary": "同质量性能与台数/核心/金额单位要匹配。", "counterexample": "收入高增但台数不变，主要是内存涨价。", "test": "台数、核心、DRAM/台、BOM传导、同配置价格。", "question": "合成：收入增35%、ASP增25%，台数增幅多少？", "constraint": "每台核心增50%，总核心增长多少；是否等于有效性能？", "transfer": "迁移到DRAM：金额、bits和带宽为什么是三条不同曲线？", "tags": [ "F2", "F4", "F7" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "MN03", "profile": "newman-bernstein", "title": "订单跑得更快，收入何时才能追上", "source_ids": [ "D13" ], "locator": "pp15–16 Newman orders vs revenue", "source_paraphrase": "追问订单增速与供应约束持续时间。", "attribution": "具名材料的简要转述；模型、反例和挑战是我们的编辑归纳，不是本人原话或投资建议。", "model": "合成：backlog期末=期初+新订单−交付−取消；交付=min(可执行需求,可用供应)。", "boundary": "订单金额/台数同口径，取消和涨价重估分开。", "counterexample": "客户重复下单或等待价格变化，backlog未必全部转换。", "test": "订单批次、取消、交付时滞、供给认证和库存。", "question": "合成：期初backlog100，新单80、交付50、取消10；期末多少，能预测下一季收入吗？", "constraint": "下一季供应最多40，订单涨价重估又增加20，模型如何分开？", "transfer": "迁移到HBM：配额预订和最终消耗怎样避免重复需求？", "tags": [ "F3", "F7", "F9", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "evidence_status": "指定段落核查2026-10-05；方法样本，未认证永久个人模式" }, { "id": "B1C01", "profile": "smith-fundsmith", "title": "高ROCE不保证新增AI投资能赚钱", "source_ids": [ "B1D01" ], "locator": "pp5–6 lines216–249", "source_paraphrase": "高资本回报要结合再投资机会及现金转换。", "attribution": "源材料简要转述；以下数值、模型、挑战与答案均为编辑合成，不是作者原话、持仓或投资建议。", "model": "合成，单位£m：FCF=CFO−CapEx；项目增量ROIC=稳定NOPAT/投入资本。", "boundary": "CFO已含营运资本，CapEx不含并购；稳定税后利润和投入资本匹配，不以公司平均ROCE替代项目回报。", "counterexample": "旧业务高回报掩盖新集群低利用率；项目现金回报滞后且需更新设备。", "test": "下一年度项目收入、利用率、现金成本、折旧、维护CapEx及税；若预定达产后ROIC仍低于10%则修订扩张论点。", "question": "合成，金额均为£m：CFO300、CapEx100；次年CFO330、CapEx180，利润增长能否推出FCF增长？", "constraint": "新增投入80，达产增量NOPAT6，资本成本10%；若要过门槛需多少NOPAT？", "transfer": "迁移到制药产能：已验证需求与未验证需求如何改变投产和回报窗口？", "tags": [ "F10", "F11", "F20", "F21" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "合成案例；未进行真实预测结果或投资业绩审计", "evidence_status": "指定原稿段落核查2026-10-06；跨材料编辑归纳，未认证永久个人模式" }, { "id": "B1C02", "profile": "smith-fundsmith", "title": "终端小幅下滑如何放大成设备订单骤停", "source_ids": [ "B1D01" ], "locator": "pp10–11 lines433–449", "source_paraphrase": "比较消费者业务与资本设备供应商的需求结构。", "attribution": "源材料简要转述；以下数值、模型、挑战与答案均为编辑合成，不是作者原话、持仓或投资建议。", "model": "合成：需求容量=任务量/有效产能；扩张订单=max(0,需求容量−现有容量)，更新订单另计。", "boundary": "单位一致，新增需求与替换互斥；有效容量已含利用率，不再重复折减。", "counterexample": "终端仅降5%，但上一轮超前扩产使新增订单归零；反之老机退役可维持订单。", "test": "客户既有有效容量、任务量、退役时间及已承诺订单；下一采购周期检验，不用供应商backlog替代客户需求。", "question": "合成：有效容量100，原需求110，新增需求订单10；终端任务降5%，下一期扩张订单多少？", "constraint": "若额外退役8容量且没有替代，扩张/更新合计新增需求容量多少？", "transfer": "迁移到工程机械：竣工面积小降如何穿过在用设备、库存和更新率？", "tags": [ "F5", "F6", "F8", "F9", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "合成案例；未进行真实预测结果或投资业绩审计", "evidence_status": "指定原稿段落核查2026-10-06；跨材料编辑归纳，未认证永久个人模式" }, { "id": "B1C03", "profile": "smith-fundsmith", "title": "长期正确还能否扛住短期赎回", "source_ids": [ "B1D02" ], "locator": "pp5–7 lines163–239", "source_paraphrase": "资金流约束促使调整持有过程。", "attribution": "源材料简要转述；以下数值、模型、挑战与答案均为编辑合成，不是作者原话、持仓或投资建议。", "model": "合成：现金缺口=max(0,赎回−现款)；额外价格冲击随卖出规模和深度而变，事实/论点/流动性处置分开记录。", "boundary": "只作资金结构教学，未给实际交易或仓位建议；赎回时点固定为期初，NAV未先受损。", "counterexample": "估值判断最终正确但被迫卖出；也可能基本面已经永久恶化，不能把所有卖出怪于资金流。", "test": "匹配赎回期限、可售深度、冲击成本、基本面证伪指标；未来一月压力情景与下一季经营证据分别核查。", "question": "合成，金额采用相同货币百万单位：基金NAV100、现金5，立即赎回15，无其他现金流；需要卖出多少证券？", "constraint": "卖出所获净现金只为卖出账面价值的98%；所需账面出售额和冲击损失多少？", "transfer": "迁移到无到期负债的封闭式资本：哪些操作约束消失，哪些估值与风险约束仍在？", "tags": [ "F13", "F22", "F23", "F24" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "合成案例；未进行真实预测结果或投资业绩审计", "evidence_status": "指定原稿段落核查2026-10-06；跨材料编辑归纳，未认证永久个人模式" }, { "id": "B1C04", "profile": "sleep-nomad", "title": "让利能把规模转成利润吗", "source_ids": [ "B1D03", "B1D04" ], "locator": "2008 letter p151 lines5583–5597; 2009 letter p160 lines5935–5967", "source_paraphrase": "规模节约回馈客户是两信的经营机制样本。", "attribution": "源材料简要转述；以下数值、模型、挑战与答案均为编辑合成，不是作者原话、持仓或投资建议。", "model": "合成：贡献利润=(单价−单位变动成本)×数量−固定成本；复购和新增需求须分别验证。", "boundary": "品质、服务水平和资本占用可比；成本下降不是把维护支出移出口径。", "counterexample": "成本降低全部让利但需求不回应，或流量增长伴随配送/融资成本上升。", "test": "按客户批次追踪留存、量、净价、单位成本和新增资本；下一完整年度检验量超过盈亏门槛。", "question": "合成：单价100、单位成本90、量100；成本降至80、售价85、量130，固定成本不变，贡献利润升还是降？", "constraint": "要恢复原贡献利润所需数量多少；若变动成本实际为83，门槛多少？", "transfer": "迁移到GPU云：tokens/$下降怎样穿过客户付费需求、利用率和折旧周期？", "tags": [ "F10", "F16", "F18", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "合成案例；未进行真实预测结果或投资业绩审计", "evidence_status": "指定原稿段落核查2026-10-06；跨材料编辑归纳，未认证永久个人模式" }, { "id": "B1C05", "profile": "sleep-nomad", "title": "现金余量的价值不能凭空入账", "source_ids": [ "B1D03" ], "locator": "pp149–150 lines5480–5553", "source_paraphrase": "讨论现金及思考余量的作用。", "attribution": "源材料简要转述；以下数值、模型、挑战与答案均为编辑合成，不是作者原话、持仓或投资建议。", "model": "合成：期末财富=现金利息+可兑现机会收益−机会获取成本；为分支概率模型，不把可选机会记作已赚利润。", "boundary": "一年度、无税费基础；只有确定可用现金可投入机会，机会回报是情景假设。", "counterexample": "长期等不到机会、识别错误或市场危机时现金受赎回挤占，现金拖累可能超过期权价值。", "test": "预先记录机会概率、触发条件、可投入金额和不行动基线；年度末核查机会是否出现和可执行。", "question": "合成：保留现金20利息2%；50%概率年中有机会，将20一次投入，至年末总回报20%；无机会全年2%。暂忽略年中利息，其期末期望财富多少？", "constraint": "若机会识别失败，机会分支回报改成−30%，怎样变化？", "transfer": "迁移到半导体产能：保留流动性与马上扩产的机会选择，如何同时建融资和设备交付约束？", "tags": [ "F20", "F21", "F22", "F23" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "合成案例；未进行真实预测结果或投资业绩审计", "evidence_status": "指定原稿段落核查2026-10-06；跨材料编辑归纳，未认证永久个人模式" }, { "id": "B1C06", "profile": "sleep-nomad", "title": "延长复利期限会不会遮住技术替代", "source_ids": [ "B1D04" ], "locator": "pp160–163 lines5957–6054", "source_paraphrase": "检验估值期限与均值回归假设。", "attribution": "源材料简要转述；以下数值、模型、挑战与答案均为编辑合成，不是作者原话、持仓或投资建议。", "model": "合成：明确三年现金流PV加第三年末持续价值；r=10%，永续g=0%，无债、无额外融资。", "boundary": "FCF为对股东可分配、已扣维持/成长投入的现金；不能用收入增长率代替FCF增长率。", "counterexample": "技术换代迫使追加资本且收费下降，增长反馈环在终值前失效。", "test": "续费/量价、更新资本、技术替代时间；每年复检永续假设和第三年终值，不能以企业文化代替证据。", "question": "合成，无债务、无额外融资，金额采用相同货币单位：明年末FCF10、第二年末12、第三年末14.4，此后每年末固定14.4；折现率10%、永续增长0%。当前价值多少？", "constraint": "若第三年开始技术替代，FCF只能8并永续固定8，价值多少；差额能否归为折现率变化？", "transfer": "迁移到医保专利药：专利到期、仿制药和生产资本如何决定持续价值，避免机械年份五均值回归。", "tags": [ "F14", "F15", "F16", "F21" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "合成案例；未进行真实预测结果或投资业绩审计", "evidence_status": "指定原稿段落核查2026-10-06；跨材料编辑归纳，未认证永久个人模式" }, { "id": "B2C01", "profile": "ackman-pershing", "title": "增长足够高，能否消化估值收缩？", "source_ids": [ "B2D02" ], "locator": "物理PDF第8–10页（零基P7–P9）：Investment Manager’s Report / LETTER TO SHAREHOLDERS；The Business Prospects of Our Current Portfolio Holdings；联署William A. Ackman / Ryan Israel", "source_paraphrase": "联合信把多年盈利增长与倍数波动分开。", "attribution": "仅此方法转述属于联署信；以下数字、概率、边界与证伪都是编辑合成，不是Ackman预测。", "model": "三年、无分红无摊薄合成：P3/P0=(1+g)^3×M3/M0。EPS0=$10/股，M0=24x，g=20%/年，M3=18x；股价单位$/股。", "boundary": "盈利口径匹配，增长可持续，终值倍数明确；高增长不保证指定期限回报。目标10%年回报所需g≥1.10×(24/18)^(1/3)−1。", "counterexample": "盈利增长来自一次性税益或透支回购，退出倍数随资本成本下降到12x。", "test": "合成观察期限2027-10-06：以未来四季经常性EPS/稀释股份核验增长，另测20%增长假说；2029-10-06核验终值假设，价格下跌不能单独证伪业务增长。", "question": "合成、三年持有且无分红无摊薄：初始EPS=$10/股，买入P/E=24x，EPS每年增长20%，三年退出P/E=18x。算总回报及CAGR；10%年回报目标满足了吗？不准把盈利增速直接写成投资回报。", "constraint": "新增：EPS增长实际只有10%/年且退出18x。区分业务假说失效和倍数风险；给出10%年回报所需最低盈利增速。", "transfer": "迁移到成熟数据中心REIT：用AFFO/股与P/AFFO，加入年度分配和新增发行后重新建总回报模型，不用会计EPS替代现金分配能力。", "tags": [ "F12", "F14", "F21", "F23" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计原文长期增长预测或组合回报" }, { "id": "B2C02", "profile": "ackman-pershing", "title": "折价回购增加NAV，不等于股价必涨", "source_ids": [ "B2D01", "B2D02" ], "locator": "B2D01 slides8–11；B2D02物理PDF第8–10页（P7–P9）Letter to Shareholders的NAV discount discussion；另董事长资本结构章节", "source_paraphrase": "材料把基金资产回报和基金份额折价分别处理，并列回购措施。", "attribution": "团队与联署材料层面的方法；合成基金并非PSH实际预测。", "model": "合成基金无债务、无税费、无分配：NAV总额$100m，股份1m，折价d=30%，P=$70；花$7m回购0.1m股。NAV/股=(NAV−现金)/(原股数−回购股数)，P=(NAV/股)×(1−d)。", "boundary": "NAV可实现性、可用现金、费用和流动性均需确认；NAV/股增厚与折价缩窄是两项独立假说。", "counterexample": "折价长期不变或变宽；NAV估计中非流动资产被高估，回购减少流动性缓冲。", "test": "合成期限2026-12-31核对回购注销、现金、股份及可实现NAV；若回购未注销须修订模型；2027-10-06检验折价，折价没有缩窄不能抹掉NAV增厚。", "question": "合成无债务基金，无税费无分配：总NAV=$100m，股份1m，市价较NAV/股折价30%；以当前市价花$7m回购并注销股票。算回购后NAV/股增厚及折价仍30%时股价；能否推出未来回报无风险？", "constraint": "回购后资产NAV整体再下降20%，市场折价扩大至40%，不发生新分配。算每股NAV、股价和从$70起的回报。", "transfer": "迁移到控股型电信公司SOTP折价：先剔除重复持股、少数股权、控股债务与税，再分开资产改善和控股折价变动。", "tags": [ "F11", "F12", "F20", "F21", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计PSH折价预测及回购结果" }, { "id": "B2C03", "profile": "ackman-pershing", "title": "优化价值，是否掩盖收购后的资金约束？", "source_ids": [ "B2D01" ], "locator": "slides 14–16资本配置；slide23 Core Principles的as-is与optimized value边界", "source_paraphrase": "团队原则将现状价值与优化价值分开，基础估值不依赖未来并购。", "attribution": "公开机构原则启发；收购数字和契约是独立合成练习，不套到真实保险或HHH交易。", "model": "合成工业收购，所有额$m/年除EV/债務存量：EV600=股权400+净债200，EBITDA80；维持CapEx25、现金税10、利息12、增长CapEx20、ΔWC0。FCFE=EBITDA−CapEx−现金税−利息−ΔWC。契约净债/EBITDA≤3x。", "boundary": " EBITDA不能当可分配现金；增长CapEx若支撑预测增长不可随意取消；此FCFE简化假定税及现金项目已经一致，不用于保险浮存金。", "counterexample": "增长项目先支出后收款，现金税与维持CapEx上升；所谓协同难兑现而债务利息确定。", "test": "合成期限2027-04-06用两季财报核验现金转换与项目收款；若年化FCFE≤0或契约余量耗尽，停止使用无融资约束的优化估值，重估现状资产。", "question": "合成工业收购：EV=$600m、股权成本$400m、净债$200m；全年EBITDA$80m、维持CapEx$25m、现金税$10m、利息$12m、增长CapEx$20m、营运资本变动0。算FCFE及现金股权收益率；EV/EBITDA便宜足以证明可行吗？", "constraint": "EBITDA降至60，其他现金项目不变；需要多少现金偿债才能满足3x契约？若增长CapEx能取消，须同时删除哪部分增长预测？", "transfer": "迁移到GPU云收购：将GPU更新支出、最低租赁承诺、客户回款延迟及再融资期限引入现状FCFE；不能把全部设备支出都叫可自由取消增长投资。", "tags": [ "F10", "F11", "F16", "F20", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未验证真实收购完成后的投资回报" }, { "id": "B2C04", "profile": "einhorn-greenlight", "title": "股东收益率高，现金真能持续返回吗？", "source_ids": [ "B2D03", "B2D04", "B2D07" ], "locator": "B2D03 slides 3–6、41–44；B2D04 pp.1–2机构落款p.7；B2D07 slides16、20–22现金利润/SBC边界", "source_paraphrase": "早期材料检验现金返还；后续个人演示补充SBC与可兑现盈利的边界。", "attribution": "个人演示与团队信分别归因；‘市场结构已坏’为作者假说，此卡不采用其为已证事实。", "model": "合成成熟软件公司（$m，全年）：股权市值500，净利润80，经营现金流65（已含现金税利息），维持CapEx20，计划分红30、回购30，无增长CapEx。可分配FCF=CFO−维持CapEx；分配覆盖率=FCF/(分红+回购)。", "boundary": "回购收益率非投资者收到的现金分红，也不能与EPS增长重复计回报；回购需扣SBC/增发稀释。持续可分配现金须先保证业务维护及偿债。", "counterexample": "净利润强但应收账款吞噬现金；债务融资回购暂时抬高EPS，软件研发资本化令维护成本被遗漏。", "test": "合成期限2027-01-31核验四季CFO、维护投资、净股份变化、净债；两期计划分配超过可分配现金且无可信回款改善，证伪‘靠业务现金持续分配’假说。", "question": "合成成熟软件公司，全年$m：股权市值500、净利润80、经营现金流65（已含现金税利息）、维持CapEx20、无增长CapEx、计划分红30和回购30。算headline股东收益率、现金覆盖率和不举债的可持续分配收益率；能否说不等重估也必赚12%？", "constraint": "其中10m回购用于抵消SBC摊薄；净债同时增加15m。区分净回购与债务支持的资本返还，并说明股价下跌时现金分配不保证总回报。", "transfer": "迁移到煤矿：加入修复保证金、矿山维持资本支出和价格下降；分开高煤价现金与跨周期可分配现金。", "tags": [ "F2", "F11", "F16", "F20", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计Solvay或其他真实持仓股东回报" }, { "id": "B2C05", "profile": "einhorn-greenlight", "title": "低谷估值：正常化盈利与可分配现金能否同时成立？", "source_ids": [ "B2D03" ], "locator": "slides 2 underlying measures及40–44 valuation/cash-flow/2028 targets", "source_paraphrase": "演示用当期低谷和长期目标两种盈利口径解释价值，且披露调整项。", "attribution": "来自个人演示的结构启发；合成公司数字不同于Solvay，正常化不是本次实证预测。", "model": "合成化工股：EV1800=股权900+净债700+环境/退休义务200（$m），股数100m，P=$9。EBITDA当期200、正常化300；D&A100、利息35、税率25%。EPS=(EBITDA−D&A−利息)×0.75/股数。", "boundary": "期末正常化的量价/成本证据与股数需验证；EV债务和义务应匹配，不把管理层目标当实际；环境义务存量进入EV不代表当年无需支付。", "counterexample": "新增低成本竞争产能导致周期高点永久下移；调整盈利排除的修复现金反复发生。", "test": "合成期限2027-04-06先以销量/实现价/成本独立核验正常化300路径；2027-10-06检验现金兑现。若可分配现金不足覆盖分红或市场份额结构下移，撤回‘低谷会回到旧正常值’假说。", "question": "合成化工股，$m：EV1800=股权900+净债700+环境/退休义务200；股数100m、市价$9。EBITDA当期200、假设正常化300；两种口径D&A均100、利息35、税率25%。算两种EPS、P/E和EV/EBITDA；哪个便宜倍数依赖尚未实现的盈利？", "constraint": "即便正常化300已兑现，该年ΔWC20、CapEx90、环境现金支付40，要求分红100（均$m）。CFO=净利润+D&A−ΔWC。检验分红覆盖及是否需新增债务。", "transfer": "迁移到HBM供应商：将良率改善、份额/ASP、节点投资和预付款纳入周期正常化；把强GM与可分配现金分开。", "tags": [ "F7", "F8", "F11", "F14", "F21" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未检验演示2028目标兑现" }, { "id": "B2C06", "profile": "einhorn-greenlight", "title": "AI产业收入增长，是外部付费还是融资驱动采购？", "source_ids": [ "B2D05" ], "locator": "信pp.1–4融资总量与明确标注illustrative的产业链收入例子；p.7团队署名", "source_paraphrase": "机构信用产业链示意强调外部需求、融资与各层收入不能混同。", "attribution": "历史机构批判方法，不采用其关于LLM能力或具体公司损失的表述为当前事实。下列数值为编辑合成，避免把其示意当实测统计。", "model": "合成同一年度、美元m：应用外部收入100，向云支付150；云收入150、现金运营成本30、年初设备CapEx200、寿命4年直线折旧、残值0；设备供应商当年收入200。总各层收入=100+150+200，云EBIT=150−30−50，云投资后现金=150−30−200。不含税/利息/ΔWC。", "boundary": "各实体合法确认的收入≠可相加的最终外部需求；CapEx是存量新增、服务收入为年度流量，必须同周期分析折旧和更新。融资依赖不能单独证明泡沫或技术无价值。", "counterexample": "应用收入迅速超过基础设施支出，云设备能跨年复用且利用率/更新回报高；效率下降价可扩大可付费任务需求。", "test": "合成期限2027-04-06跟踪真实外部客户收款、服务毛利和无新增融资现金；2027-10-06检查已签融资可用性。若应用收入充分覆盖服务成本且云跨周期FCF稳定，撤回‘仅靠融资循环’假说。", "question": "合成同一年度，单位$m，忽略税利息与营运资本：应用从外部客户收100、付云150；云收150、现金运营成本30、年初设备CapEx200、四年直线折旧且残值0；设备供应商当年收入200。算各层毛收入合计与外部需求、云会计EBIT和投资后现金、应用现金缺口。能否把毛收入合计称最终AI支出？", "constraint": "云下一年不新增CapEx且营运不变，应用外部收入增至160、服务费仍150。重新算两层现金；加入第5年设备须更新200的条件，别把不投资一年永久化。", "transfer": "迁移到电池产业：终端电动车付费、整车采购电池、电池采购材料、设备CapEx分开；用产能利用率和更新周期判别供给链盈利能否长期兑现。", "tags": [ "F6", "F9", "F10", "F11", "F14", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计Greenlight2025对AI投资周期或真实公司的预测" }, { "id": "B3C1", "profile": "mauboussin-counterpoint", "title": "高ROIC的公司，买入后一定高回报吗？", "source_ids": [ "B3D1", "B3D2" ], "locator": "B3D1 pp1–2公司/投资者回报与PVGO；B3D2 p8 ROICs and Total Shareholder Returns", "source_paraphrase": "共同作者区分经营资本回报与买入证券后的回报，要求识别价格已包含的预期。", "attribution": "共同署名材料的简述；全部练习数值、边界和解题模型由我们合成，不冒充作者预测。", "model": "合成零增长、期末现金流模型：EV=NOPAT/WACC；股权价值=EV−净债务；每股价值=股权价值/稀释股数。", "units": "NOPAT及净债务为百万美元，WACC为每年比例，股数为百万股，价格为美元/股；EV是企业价值。", "boundary": "仅适用经济利润和维护投资已正确处理、净债务与股数固定的无增长简化。真实MEROI需先构造价格匹配DCF，不能把公司ROIC直接当股票预期收益率。", "counterexample": "经营ROIC维持高位，价格却包含更高利润或更长护城河寿命，实际表现优秀仍可低于市场门槛。", "test": "同一时点股价、净债务、稀释股数与独立NOPAT/维护资本桥，检查估值是否依赖不现实利润。", "deadline": "合成案例：下一次季度披露后五个工作日复核；先冻结假设，禁止用未来价格反填。", "question": "合成：NOPAT=100、WACC=10%、净债务=200、股数=10；市价100。无增长价值是多少，市价内含NOPAT是多少？", "constraint": "若WACC升到12%，其余不变；维持市价100需要多少NOPAT？", "transfer": "迁移到成熟机场特许经营：期限有限时为何不能沿用永续NOPAT/WACC？", "tags": [ "F1", "F12", "F16", "F21", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行实际预测结果或投资业绩审计", "evidence_status": "指定原文段落核查2026-10-06" }, { "id": "B3C2", "profile": "mauboussin-counterpoint", "title": "更快增长，却更少自由现金流？", "source_ids": [ "B3D2" ], "locator": "pp1–2机会成本及NOPAT−增长投资的现金关系", "source_paraphrase": "共同作者把资本回报和增长投资连接到现金流，要求考虑资本机会成本。", "attribution": "材料层面简述；合成模型采用ROIIC驱动增长，不声称作者给出该案例。", "model": "稳定简化：再投资=NOPAT×g/ROIIC；FCF=NOPAT−再投资。零增长现有业务稳态价值与新增项目NPV分别计算，平均ROIC不替代新增ROIIC。", "units": "NOPAT、再投资、FCF为百万美元/年；g、ROIIC、WACC为每年比例；项目NPV为百万美元。", "boundary": "利润增长仅由新增资本驱动且时滞固定，无并购、额外杠杆和效率跳跃；投资/利润必须同税后基础。ROIIC不能用当年短窗口机械估计。", "counterexample": "存量资产ROIC很高，但新增机房利用率低，新增资本回报低于WACC；扩大规模使现金消耗更快。", "test": "分批投产资本、投产后的增量税后现金/利润、维护支出和资产寿命；辨别存量效率与新项目回报。", "deadline": "合成：新项目投产满12个月时核验ROIIC；每季追踪投产延迟，不将延迟自动当失效。", "question": "合成：NOPAT100、利润增长10%、ROIIC20%；需要再投资多少、FCF多少？改为增长20%会怎样？", "constraint": "新增项目投入50，在下一年开始每年永续产生税后5，WACC10%；若下一代项目只产生4，继续投入同样规模能否创造价值？", "transfer": "迁移到零售开店：存量同店销售增长与新店资本回报怎样拆开？", "tags": [ "F4", "F9", "F10", "F20", "F21" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行实际预测结果或投资业绩审计", "evidence_status": "指定原文段落核查2026-10-06" }, { "id": "B3C3", "profile": "mauboussin-counterpoint", "title": "把研发资本化，现金真的增加了吗？", "source_ids": [ "B3D1", "B3D2" ], "locator": "B3D1 p8无形投资调整；B3D2 p3调整ROIC并保持FCF不变", "source_paraphrase": "材料以一致的利润和投资调整解释无形资本，调整财务展示不改变现金流。", "attribution": "共同署名方法的简述；合成数字和检验为我们的教学模型，不能用来更改法定报表。", "model": "合成税后口径：NOPAT'=NOPAT+当期无形投入−无形摊销；净投资'=净投资+当期无形投入−无形摊销；FCF'=NOPAT'−净投资'=FCF。", "units": "所有金额为百万美元/年、同税后基础；投资为经济净投资而非毛CapEx；资本存量为百万美元。", "boundary": "必须一致处理税、摊销寿命和资本存量；不能随意把维持经营费用全变投资，更不能调利润却漏掉投资现金。", "counterexample": "研发失败、产品短寿命或客户留存下降导致经济资产寿命缩短，调整后高利润并不证明高价值。", "test": "研发批次对应产品收入/留存、失败率、经济寿命敏感性，以及报表FCF到调整FCF的逐项勾稽。", "deadline": "合成：下一年年度报告前重估研发寿命；每次调整同时提交FCF一致性核验。", "question": "合成税后：原NOPAT80、净投资30；当期可资本化研发40、摊销15。调整后NOPAT、净投资与FCF是多少？", "constraint": "若摊销加速至30，现金结果变化吗？需要怎样更新资本存量和ROIIC？", "transfer": "迁移到消费品牌广告：哪些支出维护本期销量，哪些可能创造跨期资产，需要什么证据？", "tags": [ "F2", "F11", "F16", "F21", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行实际预测结果或投资业绩审计", "evidence_status": "指定原文段落核查2026-10-06" }, { "id": "B3C4", "profile": "mauboussin-counterpoint", "title": "历史上没发生过，概率就是零吗？", "source_ids": [ "B3D3" ], "locator": "pp1–2先验/动态分布；pp4–6参照类、无形强度、基础设施与有机/并购增长", "source_paraphrase": "作者要求从合适参照类开始并更新，承认分布变化；历史罕见不证明不可能，业务增长也要通过价值检验。", "attribution": "共同署名材料的简述；先验、似然与场景均为教学假设，不是作者对任何公司给出的概率。", "model": "Bayes：P(H|E)=P(E|H)P(H)/[P(E|H)P(H)+P(E|非H)(1−P(H))]；收入增长、价格、付费留存与资本需求另建经营模型。", "units": "先验、似然、后验均为0–1概率，H定义为五年有机收入达标；收入为同货币、名义年度金额，run-rate与已确认收入不混用。", "boundary": "参照类须匹配起始规模、增长期限、有机/并购与名义/实际；似然必须可检验。重叠年度样本和相关证据不能假设独立；尾部分布不能机械当正态。", "counterexample": "免费用户爆发但付费留存弱，或收入靠并购跃升；前者未必支持有机达标，后者改变事件定义。", "test": "同客群付费转化、留存与净收入，容量/融资兑现；将证据对成功/失败两种假设的出现频率分开。", "deadline": "合成：90天后取得可比付费留存样本更新；五年终点2031年，仅为教学情景，不用于真实公司预测。", "question": "合成：H先验10%，证据E在H下概率60%、非H下20%；出现E后达标概率是多少？", "constraint": "再见到一项独立证据E2：P(E2|H)=50%、P(E2|非H)=25%。若E2其实是同一客户群的重复新闻，如何处理？", "transfer": "迁移到创新药：临床前成功、II期结果和同一试验多篇报道怎样防止重复更新？", "tags": [ "F14", "F17", "F19", "F23", "F15" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行实际预测结果或投资业绩审计", "evidence_status": "指定原文段落核查2026-10-06" }, { "id": "S1C01", "profile": "arya-bofa", "title": "每GW内容价值不能直接乘公告电力", "source_ids": [ "D11" ], "locator": "NVIDIA 2025-11-19 pp10–11 Arya Q&A；web lines390–449", "source_paraphrase": "询问每GW的公司内容价值及资本来源。", "attribution": "具名公开提问的简短转述；全部模型、数值、反例及训练题均由我们合成，不是分析师估计或管理层预测；回答归属管理层。", "model": "合成：当期收入=实际投产IT功率GW×公司捕获$/GW×当期验收率；公告、站点输入及IT负载功率分开。", "boundary": "仅在每GW价值与功率的IT/站点口径、机型和时间一致时成立；验收率用于已交付待确认，避免重扣产能。", "counterexample": "建设许可、电网容量和全年营收不在同一时间，乘积可虚增。", "test": "客户IT负载口径、合同BOM、投产时间、验收条款；下一季更新。", "question": "合成：公告4GW中只有2GW是已投产IT负载，公司内容$20bn/GW，当期验收75%；可确认多少？", "constraint": "原4GW是站点输入，PUE1.25；若其中50%按相同PUE投产，前述2GW数要怎样修订？", "transfer": "迁移到药品：获批人群不等于治疗人群，怎样连接渗透、发货和确认？", "tags": [ "F9", "F10", "F18", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行真实预测结果或投资业绩审计", "evidence_status": "指定Q&A和角色名录核查2026-10-06；跨材料方法样本，未认证永久个人模式" }, { "id": "S1C02", "profile": "arya-bofa", "title": "客户CapEx不增，公司收入仍可增吗", "source_ids": [ "S1D01" ], "locator": "NVIDIA 2026-02-25 p8 Arya Q&A；web lines274–298", "source_paraphrase": "将客户预算不增长作为公司增长的条件测试。", "attribution": "具名公开提问的简短转述；全部模型、数值、反例及训练题均由我们合成，不是分析师估计或管理层预测；回答归属管理层。", "model": "合成：供应商收入=C×a×s+N；C客户总预算，a相关设备占比，s本公司份额，N不重叠新客收入；利润另扣增量费用。", "boundary": "C/a/s期间币种统一；新客不是总预算内客户；总CapEx含建筑/能源，不能全算GPU。", "counterexample": "份额提升来自折价或低毛利整机，收入改善但利润下降。", "test": "客户预算分类及现金融资、实际采购份额、ASP和利润；未来两季复核。", "question": "合成：客户总预算C=$100bn且不增长；相关设备占比a由40%升50%，本公司在该设备池份额s由60%降55%。供应商该池收入变化？", "constraint": "新增不重叠客户3bn、原相关池毛利75%降65%，新增客毛利40%，OpEx增2bn；增量营业利润怎样？", "transfer": "迁移到医疗器械：医院预算固定，耗材份额增加是否必然增利？", "tags": [ "F4", "F9", "F10", "F16", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行真实预测结果或投资业绩审计", "evidence_status": "指定Q&A和角色名录核查2026-10-06；跨材料方法样本，未认证永久个人模式" }, { "id": "S1C03", "profile": "arya-bofa", "title": "同比加速：周期复苏还是涨价和新内容", "source_ids": [ "S1D03" ], "locator": "ADI 2026-08-19 p7–8 Arya Q&A；web lines210–245", "source_paraphrase": "将增长拆分为长期、周期和价格贡献。", "attribution": "具名公开提问的简短转述；全部模型、数值、反例及训练题均由我们合成，不是分析师估计或管理层预测；回答归属管理层。", "model": "合成：R1/R0=(1+数量增长)×(1+净价格增长)×(1+单件内容增长)；固定篮子口径，不将周期与数量重复相加。", "boundary": "产品和客户篮子可比；长期/周期是数量来源解释，不是独立收入加数；未剔收购时不能声称有机增长。", "counterexample": "低基数/补库存造成同比加速，终端消耗不加速。", "test": "sell-through、净价、内容、收购、渠道库存、季节性；两个季度检验。", "question": "合成：收入同比增32%，数量增10%、净价增10%；隐含内容增长多少？", "constraint": "发现同比基期库存出清压低收入5%，将基期收入恢复为原来的1.05倍；规范化增长和隐含内容怎样变？", "transfer": "迁移到付费软件：席位、ARPU、使用量和续约率怎样避免增长重复解释？", "tags": [ "F2", "F5", "F8", "F18", "F23" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行真实预测结果或投资业绩审计", "evidence_status": "指定Q&A和角色名录核查2026-10-06；跨材料方法样本，未认证永久个人模式" }, { "id": "S1C04", "profile": "sur-jpm", "title": "关税避开了主机，是否绕不过零件", "source_ids": [ "S1D02" ], "locator": "Lam 2025-04-23 pp9–10 Sur两段Q&A；web lines287–318", "source_paraphrase": "将制造调配与后续材料成本风险分开追问。", "attribution": "具名公开提问的简短转述；全部模型、数值、反例及训练题均由我们合成，不是分析师估计或管理层预测；回答归属管理层。", "model": "合成：毛利损失=应税成本×税率×企业承担比例；库存覆盖让冲击延迟，不是永久免除；转产成本单列。", "boundary": "税率为教学假设，不代表现行关税；法定纳税人与经济承担分开；库存计价/原产地规则需具体核验。", "counterexample": "美国整机装配不能自动免除进口子系统税费；供应商承担后可能调高报价。", "test": "BOM原产地与授权规则、库存批次、供应商合同、合格转产时点；按月验证。", "question": "合成：季度收入1bn，原COGS600m，其中应税进口成本200m，税率20%，企业承担75%；耗完旧库存后的毛利率？", "constraint": "Q1旧库存覆盖50%应税消耗；Q2转产减少应税成本40%，却增加固定运营费用10m；分别算Q1毛利和Q2营业利润变化。", "transfer": "迁移到电池：本地组装后的材料来源、认证与贸易成本怎样闭合？", "tags": [ "F3", "F7", "F19", "F20", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行真实预测结果或投资业绩审计", "evidence_status": "指定Q&A和角色名录核查2026-10-06；跨材料方法样本，未认证永久个人模式" }, { "id": "S1C05", "profile": "sur-jpm", "title": "网络收入占比上升：附着增长还是分母下降", "source_ids": [ "S1D01" ], "locator": "NVIDIA 2026-02-25 pp9–10 Sur Q&A；web lines338–370", "source_paraphrase": "从网络占比追问产品路线及收入运行率。", "attribution": "具名公开提问的简短转述；全部模型、数值、反例及训练题均由我们合成，不是分析师估计或管理层预测；回答归属管理层。", "model": "合成：网络R=机架数×每架网络内容×附着率；占比=网络R/(网络R+计算R)；计算端确认时滞单列。", "boundary": "scale-up、scale-out、跨站点和NIC口径分开；内容$/架已经含芯片数则不可再次乘数量。", "counterexample": "GPU延迟确认也会提高网络占比；附着100%不是对开放生态的必然结论。", "test": "每架BOM、外部网络供应商份额、验收和季末出货；下一季检验。", "question": "合成：网络收入10bn、计算40bn；下一季网络仍10bn但计算30bn，占比变化能证明网络增长吗？", "constraint": "另一个独立方案：机架数+20%、网络内容+25%、附着率由80%降64%；原网络10bn，新网络是多少？", "transfer": "迁移到支付：支付费收入占平台收入升高，怎样拆交易量、take-rate和广告分母？", "tags": [ "F2", "F6", "F15", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行真实预测结果或投资业绩审计", "evidence_status": "指定Q&A和角色名录核查2026-10-06；跨材料方法样本，未认证永久个人模式" }, { "id": "S1C06", "profile": "sur-jpm", "title": "从营业利润指引反推毛利，需要哪些未知数", "source_ids": [ "S1D03" ], "locator": "ADI 2026-08-19 p6 Sur/CFO Q&A；web lines185–200", "source_paraphrase": "用营业利润指引追问毛利改善的量、组合和价格来源；管理层校正其估算。", "attribution": "具名公开提问的简短转述；全部模型、数值、反例及训练题均由我们合成，不是分析师估计或管理层预测；回答归属管理层。", "model": "合成：GM=OM+OpEx/R；同比变化ΔGM不能仅用ΔOM替代；需区分固定成本吸收、净价与产品组合。", "boundary": "同一GAAP/non-GAAP和费用分类；研发资本化/并购摊销调整不可跨口径；来源只是研究问法。", "counterexample": "OpEx占收入下降提升OM，即使GM不动；价格上调被材料涨价抵消。", "test": "同口径费用、售价执行、成本、库存吸收与组合桥；本季度披露后校验。", "question": "合成：收入4bn，营业利润率50%，OpEx0.9bn；隐含GM？收入下季4.3bn，OM52%，OpEx仍0.9bn时GM是多少？", "constraint": "若不能取得下季OpEx，只知道0.8–1.0bn区间，GM范围？可以断言有150bp毛利改善吗？", "transfer": "迁移到物流：网络规模提高时，营业利润改善怎样分为运输毛利和管理费用率？", "tags": [ "F3", "F4", "F11", "F18", "F24" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行真实预测结果或投资业绩审计", "evidence_status": "指定Q&A和角色名录核查2026-10-06；跨材料方法样本，未认证永久个人模式" }, { "id": "S2C-B01", "profile": "rolland-susquehanna", "title": "季度强势是结构增长还是时点搬移？", "source_ids": [ "S2D-B01", "S2D-B02" ], "locator": "CRUS 2024-11-04 pp6–7；2025-08-05 p6本人首问", "source_paraphrase": "Rolland追问内容、销量和提前出货；管理层分别解释周数差异及难以量化提前订单。", "attribution": "具名问题转述；管理层回答单列。全部模型、数字、判断边界及训练题为我们编辑的合成案例，不是分析师预测。", "model": "合成：R=Q×C×P；Q为百万台终端，C为每台芯片数，P为$/芯片；收入单位$百万。预算Q=100,C=2,P=2→400；实际Q=110,C=2.1,P=2→462。顺序桥：量40，内容22。", "boundary": "同口径季度与产品篮子；供货量不等于终端卖出，量/内容交互只计一次。", "counterexample": "额外10百万台是客户库存，不是消费者需求；内容增加也可能换来降价。", "test": "未来两季库存周转、终端销售和内容拆解；若回补消失，回撤需求上修。", "question": "合成：预算季度终端100百万台、每台2颗芯片、单价$2/颗；实际110百万台、每台2.1颗、单价不变。收入单位$百万。给出量、内容不重复计交互的收入增量桥，判断能否延续至下季。", "constraint": "额外销量中8百万台是从下季提前；在同ASP/内容下修正两季净增量。", "transfer": "迁移光模块：400G升级800G造成内容/ASP变化，如何区分端口数与建设时点？", "tags": [ "F1", "F2", "F4", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计分析师预测结果；历史材料本周补录，不是本周新预测", "deadline": "合成练习：未来两个季度；真实研究以下一次公开业绩/项目公告日期为准，不预设未宣布日期", "fact_testimony_inference": "来源事实：具名有日期提问；答复：管理层证词，非独立核实；模型/反例：编辑推断与合成练习。", "inherited_assumption": "旧模型假设变量/时间口径可比，增量可延续；先冻结基准。", "changed_constraint": "额外销量中8百万台是从下季提前；在同ASP/内容下修正两季净增量。", "competing_hypotheses": [ "经营结构变化", "订单/交付/会计时点变化", "需求没有变化，口径或组合变化" ], "reversal_criterion": "若指定窗口的可比数据不能满足本卡模型边界，撤回相应增量假设；区别延期与需求消失。" }, { "id": "S2C-B02", "profile": "rolland-susquehanna", "title": "财年周数会制造怎样的假拐点？", "source_ids": [ "S2D-B01" ], "locator": "CRUS 2024-11-04 pp6–7", "source_paraphrase": "Rolland追问异常季节性；管理层答复涉及53周财年、14周基期与生产峰值周。", "attribution": "具名问题转述；管理层回答单列。全部模型、数字、判断边界及训练题为我们编辑的合成案例，不是分析师预测。", "model": "合成：旧季R=560($百万)、14周，新季R=546、13周；每周代理R/W。旧40，新42；总收入−2.5%，平均每周+5%。", "boundary": "均匀收入只为代理；高峰周的产品组合不能用平均周精确还原。", "counterexample": "多一周恰在新品高峰，平均归一化仍会低估真实差异。", "test": "未来两季按周与产品归一化；检查同长度滚动窗口，若单位销售持续降则推翻恢复。", "question": "合成：旧季度收入$560百万、14周；新季度$546百万、13周。是否足以判定周期转弱？计算总收入与平均每周收入增长，列出需要继续核实的口径。", "constraint": "被删去的额外一周收入为70，改算可比13周旧收入及增长。", "transfer": "迁移零售53周财年，区分促销峰值、门店数和同店需求。", "tags": [ "F5", "F8", "F17", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计分析师预测结果；历史材料本周补录，不是本周新预测", "deadline": "合成练习：未来两个季度；真实研究以下一次公开业绩/项目公告日期为准，不预设未宣布日期", "fact_testimony_inference": "来源事实：具名有日期提问；答复：管理层证词，非独立核实；模型/反例：编辑推断与合成练习。", "inherited_assumption": "旧模型假设变量/时间口径可比，增量可延续；先冻结基准。", "changed_constraint": "被删去的额外一周收入为70，改算可比13周旧收入及增长。", "competing_hypotheses": [ "经营结构变化", "订单/交付/会计时点变化", "需求没有变化，口径或组合变化" ], "reversal_criterion": "若指定窗口的可比数据不能满足本卡模型边界，撤回相应增量假设；区别延期与需求消失。" }, { "id": "S2C-B03", "profile": "rolland-susquehanna", "title": "多元化设计赢单怎样变成利润？", "source_ids": [ "S2D-B02" ], "locator": "CRUS 2025-08-05 pp6–7本人follow-up", "source_paraphrase": "Rolland问PC/汽车多元化；管理层强调设计周期与早期产品不能立即贡献大额收入。", "attribution": "具名问题转述；管理层回答单列。全部模型、数字、判断边界及训练题为我们编辑的合成案例，不是分析师预测。", "model": "合成：R=可服务台数×设计份额×量产转化×每台内容。PC40百万台×20%×50%×$3=$12百万/年；毛利55%，增量研发$8百万，增量经营利润−$1.4百万。", "boundary": "年化产量且所有项目能在周期内出货；转化与份额不得混成概率再重复折扣。", "counterexample": "研发可跨产品复用，单年亏损不等于项目NPV负；但获设计不等于终端量产。", "test": "未来两季样品→设计→量产→现金台账；转化低于约60.61%需重评单年盈亏假设。", "question": "合成PC项目：年可服务40百万台，设计赢单份额20%，本年量产转化50%，每台供货内容$3，毛利率55%，增量研发$8百万/年。计算本年收入与增量经营利润；设计赢单增加是否已证明客户集中风险下降？", "constraint": "认证延迟半年，仅本年一半收入而研发不变，重算。", "transfer": "迁移Arm IP到整芯片：客户认证、单位毛利和增量研发/资本分开。", "tags": [ "F10", "F15", "F16", "F18", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计分析师预测结果；历史材料本周补录，不是本周新预测", "deadline": "合成练习：未来两个季度；真实研究以下一次公开业绩/项目公告日期为准，不预设未宣布日期", "fact_testimony_inference": "来源事实：具名有日期提问；答复：管理层证词，非独立核实；模型/反例：编辑推断与合成练习。", "inherited_assumption": "旧模型假设变量/时间口径可比，增量可延续；先冻结基准。", "changed_constraint": "认证延迟半年，仅本年一半收入而研发不变，重算。", "competing_hypotheses": [ "经营结构变化", "订单/交付/会计时点变化", "需求没有变化，口径或组合变化" ], "reversal_criterion": "若指定窗口的可比数据不能满足本卡模型边界，撤回相应增量假设；区别延期与需求消失。" }, { "id": "S2C-B04", "profile": "curtis-jefferies", "title": "后半年权重怎样进入季度模型？", "source_ids": [ "S2D-B03", "S2D-B04" ], "locator": "Lam 2025-10-22 p16 Curtis follow-up；2026-01-28 p17 NAND follow-up", "source_paraphrase": "Curtis问总体投资后半年权重与NAND升级时点；管理层并未提供完整分部季度数。", "attribution": "具名问题转述；管理层回答单列。全部模型、数字、判断边界及训练题为我们编辑的合成案例，不是分析师预测。", "model": "合成年度设备收入$1000百万，H2权重60%；H1 400、H2 600。由年度总额只能得到半年合计，不能唯一分成四季。", "boundary": "WFE支出与厂商确认收入存在份额、交付和验收时滞；先固定收入模型边界。", "counterexample": "订单前置但验收推迟，支出/订单年增长仍可能伴随季度收入下滑。", "test": "未来两个季度按订单→出货→验收对账；验收率低于使全年兑现的条件则下修。", "question": "合成年度设备收入计划$1000百万，后半年占60%，前后半年合计对应同一年度确认收入。可否每季按$250百万建模？算半年合计并列出季度分配无法识别的变量。", "constraint": "H2计划600其中20%推迟至下一年，H1不变；现金提前收50，重算本年收入与现金关系。", "transfer": "迁移AI服务器积压订单：交付、通电、验收与收入年度归属不同。", "tags": [ "F3", "F4", "F9", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计分析师预测结果；历史材料本周补录，不是本周新预测", "deadline": "合成练习：未来两个季度；真实研究以下一次公开业绩/项目公告日期为准，不预设未宣布日期", "fact_testimony_inference": "来源事实：具名有日期提问；答复：管理层证词，非独立核实；模型/反例：编辑推断与合成练习。", "inherited_assumption": "旧模型假设变量/时间口径可比，增量可延续；先冻结基准。", "changed_constraint": "H2计划600其中20%推迟至下一年，H1不变；现金提前收50，重算本年收入与现金关系。", "competing_hypotheses": [ "经营结构变化", "订单/交付/会计时点变化", "需求没有变化，口径或组合变化" ], "reversal_criterion": "若指定窗口的可比数据不能满足本卡模型边界，撤回相应增量假设；区别延期与需求消失。" }, { "id": "S2C-B05", "profile": "curtis-jefferies", "title": "区域和终端增长怎样去重？", "source_ids": [ "S2D-B04" ], "locator": "Lam 2026-01-28 p17 Curtis首问Reliant", "source_paraphrase": "Curtis问中国下降时Reliant增长来自哪里；管理层答复包含跨国客户和中国。", "attribution": "具名问题转述；管理层回答单列。全部模型、数字、判断边界及训练题为我们编辑的合成案例，不是分析师预测。", "model": "合成交叉表($百万)：中国成熟80、中国先进20、其他成熟40、其他先进60，总200。新期中国成熟60/先进10，其他成熟80/先进70，总220。区域变化−30/+50；产品变化+20/0。", "boundary": "区域按安装地、产品按工艺/设备，两种维度不能相加；跨国客户国籍不等于安装地。", "counterexample": "客户总部在海外但设备安装中国，按客户国籍归类会高估非中国分散度。", "test": "未来两季核对安装地×设备类别交叉表与许可，区域不明先保留unknown，不作许可结论。", "question": "合成交叉表，单位$百万：旧期中国成熟80、中国先进20、其他地区成熟40、其他地区先进60；新期中国成熟60、中国先进10、其他地区成熟80、其他地区先进70。分解区域与产品变化；能否把其他地区增量与成熟产品增量相加？", "constraint": "其他成熟80中15安装在中国，只按总部错分类；改分类并重算区域变化。", "transfer": "迁移云算力：云厂商总部、数据中心国家、最终工作负载客户是不同分母。", "tags": [ "F6", "F7", "F19", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计分析师预测结果；历史材料本周补录，不是本周新预测", "deadline": "合成练习：未来两个季度；真实研究以下一次公开业绩/项目公告日期为准，不预设未宣布日期", "fact_testimony_inference": "来源事实：具名有日期提问；答复：管理层证词，非独立核实；模型/反例：编辑推断与合成练习。", "inherited_assumption": "旧模型假设变量/时间口径可比，增量可延续；先冻结基准。", "changed_constraint": "其他成熟80中15安装在中国，只按总部错分类；改分类并重算区域变化。", "competing_hypotheses": [ "经营结构变化", "订单/交付/会计时点变化", "需求没有变化，口径或组合变化" ], "reversal_criterion": "若指定窗口的可比数据不能满足本卡模型边界，撤回相应增量假设；区别延期与需求消失。" }, { "id": "S2C-B06", "profile": "curtis-jefferies", "title": "旧厂升级何时比新建厂更快？", "source_ids": [ "S2D-B04" ], "locator": "Lam 2026-01-28 p17 Curtis NAND升级/洁净室follow-up", "source_paraphrase": "Curtis追问NAND升级是否需等待洁净室空间；管理层对升级节奏的回答不等于所有扩产无需空间。", "attribution": "具名问题转述；管理层回答单列。全部模型、数字、判断边界及训练题为我们编辑的合成案例，不是分析师预测。", "model": "合成：旧厂可升级100线，交付80线/年，停机窗口60线/年，需求90线/年；兑现min=60。每线设备$2百万→120收入；新厂40线因洁净室晚半年今年只20线→40。总160。", "boundary": "同一年度、设备验收对应整线；升级导致原产出停机，不可只加新增产能不扣损失。", "counterexample": "升级越复杂停机越久，设备收入增加而客户比特产出短期下降。", "test": "未来两季度检查停机窗口/客户验收与层数升级；升级兑现<60线则重算，不用TAM替代订单。", "question": "合成年度：旧厂有100条可升级线，设备交付上限80条，停机窗口60条，客户需求90条，每条设备收入$2百万；新厂规划40条，但洁净室晚半年导致本年只验收20条，每条$2百万。计算本年可确认收入和约束；旧厂需求翻倍是否收入翻倍？", "constraint": "升级停机窗口降至40线，单线设备内容涨20%；新厂收入不变。", "transfer": "迁移GPU服务器升级：机架电力/维护窗口成为瓶颈，交付更多卡也不能即时上线。", "tags": [ "F7", "F9", "F10", "F15", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计分析师预测结果；历史材料本周补录，不是本周新预测", "deadline": "合成练习：未来两个季度；真实研究以下一次公开业绩/项目公告日期为准，不预设未宣布日期", "fact_testimony_inference": "来源事实：具名有日期提问；答复：管理层证词，非独立核实；模型/反例：编辑推断与合成练习。", "inherited_assumption": "旧模型假设变量/时间口径可比，增量可延续；先冻结基准。", "changed_constraint": "升级停机窗口降至40线，单线设备内容涨20%；新厂收入不变。", "competing_hypotheses": [ "经营结构变化", "订单/交付/会计时点变化", "需求没有变化，口径或组合变化" ], "reversal_criterion": "若指定窗口的可比数据不能满足本卡模型边界，撤回相应增量假设；区别延期与需求消失。" }, { "id": "S3C01", "profile": "rakers-wf", "title": "网络改善利用率，谁捕获经济价值？", "source_ids": [ "D10" ], "locator": "2025-08-27 NVIDIA FY26 Q2 PDF pp12–13; Aaron Rakers Q at lines469–474, management reply476–508", "source_paraphrase": "Rakers问Spectrum-XGS定位及增量机会；未提供个人ROI模型。", "attribution": "具名问题转述；下述模型、数值、边界和挑战均为编辑合成，管理层回答不归为Rakers观点。", "model": "合成一年模型：等效有效GPU小时=装机GPU数×8760小时/年×有效利用率；新增服务贡献=新增有效小时×每有效小时贡献美元；网络收益需减网络CapEx、年度运营成本、部署损失。供应商收入=客户新购网络美元×可实现份额×当期验收比例，与客户收益分开。", "boundary": "同模型、同精度、延迟SLA且新增有效小时可销售；网络改善不能直接按GPU利用率与收入同比放大。CapEx不是利润费用。", "counterexample": "瓶颈是付费需求或HBM，网络增加可用吞吐但没有增加付费任务；旧网络替换也可能只抵消原有销售。", "test": "未来90天合成验证窗：A/B网络故障率、有效GPU小时、SLA合格付费任务、租赁实现价格及客户验收；若无付费贡献增量，不认定客户现金回收。现实核查点2026-11-17，未披露须标未知。", "question": "合成：1万GPU，有效利用率65%升80%，每新增有效GPU小时贡献$0.20；新网络投入$1000万、每年额外Opex$100万。第一年增量贡献和简单现金回收怎样计算？", "constraint": "只有一半新增有效小时能售出，部署首年损失原有贡献$50万；修正净现金收益，勿把CapEx当年度利润费用。", "transfer": "迁移到液冷：散热升级带来的吞吐收益，如何分离电价、利用率、客户服务收入和设备厂商订单？", "tags": [ "F10", "F18", "F6", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "S3C02", "profile": "rakers-wf", "title": "独立CPU销量是否创造净利润？", "source_ids": [ "S1D01" ], "locator": "2026-02-25 NVIDIA FY26 Q4 PDF pp15–16; Rakers Q lines623–630; management reply632–653", "source_paraphrase": "Rakers问Vera独立CPU的架构角色，及异构推理是否驱动需求。", "attribution": "具名问题转述；独立CPU财务算式和合成数字为编辑构造，不是Rakers模型或管理层指引。", "model": "合成一年净经营利润增量=独立CPU销量×单位毛利−被替代原有CPU销量×原单位毛利−新增现金R&D和服务Opex−新资产折旧；收入与利润分别核算，GPU协同需单列避免重复。", "boundary": "需区分GPU主机attach、独立CPU工作负载、第三方CPU替换和原产品蚕食；CPU能运行工具负载不证明客户一定采购。", "counterexample": "客户使用已有x86池执行工具，独立CPU只能折价销售；迁移成本超过节能，或新增收入全部替代原平台收入。", "test": "未来两季度合成窗核查独立客户与GPU搭售、验收量、实现ASP、软件/移植Opex；若80%以上销量是旧平台替换，应否定全额增量收入桥。现实下一公开检查点2026-11-17，未披露不编造销量。", "question": "合成一年：独立CPU售10万颗，ASP$2000、GM50%；其中3万颗替代原CPU，原ASP$1800、GM45%；新增现金研发与支持$6000万、折旧$500万。增量收入与经营利润各是多少？", "constraint": "销量只达7万颗，替代旧CPU仍3万颗，固定新增成本不变；再算一次并解释产品逻辑为何可能成立而财务结果不佳。", "transfer": "迁移到AMD自建Arm主机：怎样比较自研、第三方兼容与EPYC优化三条路线的净GPU收益和CPU机会成本？", "tags": [ "F15", "F16", "F18", "F20", "F4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "S3C03", "profile": "rakers-wf", "title": "每GW美元增加，不等于服务产出同比增加", "source_ids": [ "S3D02" ], "locator": "2026-08-26 NVIDIA FY27 Q2 PDF pp14–15; Rakers Q lines608–616 plus fiscal correction628–630; management reply632–648", "source_paraphrase": "Rakers追问每GW美元密度后续扩张及供给部署是否线性。", "attribution": "具名提问转述；模型和$30/$40bn是编辑合成，不使用管理层回报主张作为实证。", "model": "合成一年：硬件收入=当年验收且通电GW×硬件美元/GW；GPU等效吞吐=通电GW×同SLA吞吐/GW；客户服务收入=吞吐×付费利用率×实现服务价格。订单GW、验收GW和投运GW各自单列，按收入会计政策而非通电自动确认。", "boundary": "比较一致硬件范围和同质量工作负载；美元/GW增长可以来自价格或更高算力密度，两者不可混同。合成收入验收且通电只是练习假设。", "counterexample": "单GW更贵但每任务资源成本上升；或缺电导致已订设备不能投运，供应商与客户收入时点不同。", "test": "下一两季度合成窗：验收GW、通电GW、同SLA吞吐、美元/GW、实现计费及生产良率；若验收推迟超过一季，修正当期硬件收入而非只改TAM。现实检查点2026-11-17。", "question": "合成A平台当年投运1GW、硬件$30bn/GW、吞吐指数100/GW；B平台投运0.8GW、$40bn/GW、吞吐130/GW。已验收全部同年计入收入：硬件收入、吞吐增长及单位吞吐硬件成本如何变化？", "constraint": "B的0.2GW中硬件仍未验收且未通电，当年只有0.6GW可计收入；区分未来供给能力、当年现金与收入。", "transfer": "迁移到HBM：每GPU内存美元增长时，容量增长、ASP上升与可交付GPU数量如何影响总收入及客户ROI？", "tags": [ "F9", "F10", "F15", "F18", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "S3C04", "profile": "rakers-wf", "title": "ASIC替代需要跨过迁移和利用率阈值", "source_ids": [ "D11" ], "locator": "2025-11-19 NVIDIA FY26 Q3 PDF pp15–16; Rakers Q lines646–652, management reply654–702", "source_paraphrase": "Rakers问与Anthropic合作后，ASIC和GPU在架构部署中的竞争判断是否发生变化。", "attribution": "具名问题转述；编辑合成盈亏边界，不把管理层生态优势回答写成分析师结论。", "model": "合成一年同质量任务成本：GPU美元/任务=cG；ASIC美元/任务=cA；ASIC年净节省=可迁移付费任务数×(cG−cA)−年化迁移和运维固定成本；需用同延迟、准确率、可靠性与全成本口径。", "boundary": "单负载成本优势不代表所有模型或整体份额；固定成本年化期间是合成假设，实际先现金流后折现。", "counterexample": "模型快速变化导致ASIC兼容范围缩小，低利用率与软件重写吞掉芯片便宜优势；反之稳定大负载可让专用芯片创造优势。", "test": "合成未来12个月：同SLA成功任务成本、迁移投入、失败回退、可迁移任务量与最低利用率；若可迁移任务低于成本盈亏阈值，证伪净节省。现实检查点2026-11-17并继续跟踪独立客户证据。", "question": "合成：GPU全成本$0.020/合格任务，ASIC$0.012/任务；可迁移任务100亿/年，新增固定成本$1亿/年。ASIC能否省钱？计算任务阈值。", "constraint": "agentic工具负载只能迁移总任务的60%，其余仍跑GPU；重新核算，但不能把整家云GPU需求机械下调40%。", "transfer": "迁移到CUDA转ROCm：代码生成节省编写时间后，验证、通信、上线稳定性和客户流量怎样改变迁移门槛？", "tags": [ "F10", "F16", "F15", "F18", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "B100BR01", "profile": "rogoff-polar", "title": "收入增长能抵消架构迁移引起的终值下降吗？", "source_ids": [ "B100-BR1", "B100-BR2" ], "locator": "2024 valuation Q&A; FY2026 Mag7 challenged / PC cycle lessons", "source_paraphrase": "Rogoff在2024将成长前景与相对估值结合；2026共同报告质疑部分旧龙头在AI架构中的价值捕获。", "attribution": "前句为来源转述；以下计算及反证是编辑合成，不是本人定价模型。", "model": "合成：股价=EPS×PE；分开盈利路径与终值倍数。", "boundary": "不把技术变化机械映射为固定估值倍数。", "counterexample": "拥有渠道或专有数据的应用可继续捕获价值。", "test": "客户留存、净实现价格、获客成本和替代率。", "question": "合成：EPS由10增至13，PE由30降至20；无股息，总回报多少？", "constraint": "若稀释使EPS只到12，且终值PE18，再计算。", "transfer": "CPU平台营收上升30%，毛利率从50%到35%，收入基数100；毛利变化多少？能推出股价吗？", "tags": [ "F12", "F14", "F16" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "B100TK01", "profile": "kim-blackrock", "title": "把CapEx换算成GW之后，还缺什么？", "source_ids": [ "B100-TK1", "B100-TK2" ], "locator": "AI stack; 2026 supply constraints", "source_paraphrase": "Kim将AI拆为三层，并尝试将数据中心资本开支转换为功率需求来定位供给约束。", "attribution": "来源方法转述；下列数字为编辑教学场景，不复用其预测。", "model": "合成：GW=同口径建设预算/每GW造价；可用容量再受通电进度约束。", "boundary": "建设、IT容量、设施容量和计算能力必须各自定义。", "counterexample": "预算包括设备更新而非新GW，或造价下降。", "test": "资本支出构成、PUE、通电验收、部署与利用率。", "question": "合成：新增设施预算800亿美元，每GW设施总造价400亿美元；理论新增GW多少？", "constraint": "其中200亿美元是存量设备更新，其余建设只有75%按期通电；年末新增通电GW多少？", "transfer": "新CPU集群100MW设施功率，PUE1.25；每台IT功率2kW，可容纳多少台？这能否直接推出核心利用率？", "tags": [ "F18", "F7", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "B100RC01", "profile": "clode-janus", "title": "盈利增长不等于估值回报", "source_ids": [ "B100-RC1", "B100-RC2" ], "locator": "Valuation process; skepticism about AI valuations", "source_paraphrase": "Clode保留PE与FCF收益率估值纪律；团队提醒高增长中仍可能出现循环投资和局部炒作。", "attribution": "来源观点转述；回报模型和参数是编辑合成。", "model": "合成：总回报=(终值EPS×终值PE+股息)/买入价−1。", "boundary": "盈利质量、股本变动与维持性资本支出仍须核验。", "counterexample": "增长以追加资本和稀释融资换取，现金回报较弱。", "test": "应收、SBC、现金转换和客户融资依赖。", "question": "合成：初始EPS5、PE40；一年后EPS6.5、PE30、无股息；回报多少？", "constraint": "报告FCF40，但其中10为一次性客户预收，市值800；报告及调整后FCF收益率是多少？", "transfer": "芯片供应商利润增长30%，客户CapEx却由供应商贷款支持；需要哪些额外证据？", "tags": [ "F12", "F11", "F21" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "B100PW01", "profile": "wick-seligman", "title": "高利用率下的小幅需求增长为何放大缺口？", "source_ids": [ "B100-PW1", "B100-PW2" ], "locator": "2021 first and final questions; 2023 internet versus AI distinction", "source_paraphrase": "Wick将代工集中、高利用率与供给余量减少相连；其后讨论AI对知识工作的用途。", "attribution": "来源转述；产能数字及采购练习为编辑合成。", "model": "合成：新需求=原产能×原利用率×(1+需求增幅)；缺口=max(新需求−有效产能,0)。", "boundary": "不同节点、产品及封装不能无条件互换。", "counterexample": "客户重复下单，名义短缺不代表最终消费。", "test": "终端消耗、库存、可取消订单、合格产能和交货周期。", "question": "合成：月产能100万片，利用率95%，需求增长10%；新增需求与超产能缺口各多少？", "constraint": "扩产10万片但只完成60%良率等效产能；所有产能及需求使用相同良品等效口径，需求不变，还缺多少？", "transfer": "CPU封装月产能100万颗、晶圆可供120万颗；增加20%晶圆能否提升出货？", "tags": [ "F7", "F18", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "B100AB01", "profile": "benjamin-fidelity", "title": "库存下降何时才能叫复苏？", "source_ids": [ "B100-AB1", "B100-AB2" ], "locator": "2026 sector outlook non-AI paragraph; Q&A non-AI recovery sidebar", "source_paraphrase": "Benjamin先判断非AI去库存接近结束，随后承认前两年复苏预期过早，并以库存、需求和供给重新论证。", "attribution": "来源判断按2026年1—2月时点转述；下面为编辑合成库存模型。", "model": "合成：期末库存=期初库存+进货−消耗；库存消化和终端增长分开。", "boundary": "供应商收入、渠道采购和终端消耗不是同一变量。", "counterexample": "库存天数因价格上涨或减产下降，终端需求未改善。", "test": "单位库存、sell-through、渠道补库、价格及同口径季度序列。", "question": "合成：渠道期初库存30万颗，季度进货80万，终端消耗100万；期末库存多少？能证明需求增长吗？", "constraint": "下季需求仍100万，目标期末库存20万；采购需多少，较本季增长多少？", "transfer": "AI预算从30增至50、总IT预算由100到110；非AI预算变化多少？可以直接用AI增幅预测CPU采购吗？", "tags": [ "F6", "F14", "F23" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "B100GB01", "title": "毛利率下降是否必然意味着价值缩水？", "locator": "2020 Intel假设遗漏复盘；2025 business model / AI economics", "source_paraphrase": "Baker复盘未充分检验工艺路线图假设；2025讨论AI增加计算成本、改变传统软件经济。", "model": "编辑合成：毛利=收入×毛利率；营业利润=毛利−营业费用；资本回报需单独计算。", "boundary": "毛利额上升不能证明自由现金流或资本回报上升。", "counterexample": "AI增加收入却需要更多硬件资本，可能稀释资本回报。", "test": "核验推理成本、客户付费、实际利用率及投资资本增量。", "question": "合成软件企业：原收入100、毛利率80%、营业费用50；AI后收入160、毛利率60%、费用60。两期营业利润及变化？", "constraint": "假设税率均20%，投资资本从100升至180，以期末资本算简化税后营业回报，分别多少？", "transfer": "CPU路线图成功概率70%时利润120，失败时利润40；原估计成功90%。两个期望利润及下调幅度？", "tags": [ "F14", "F18", "F23" ], "profile": "baker-atreides", "source_ids": [ "B100-GB1", "B100-GB2" ], "attribution": "来源转述与后续编辑合成计算明确区分；题设数字非本人预测。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "B100BG01", "title": "削支承诺能变成多少可分配现金？", "locator": "2022 open letter三项建议；2023 uncertainty与cash-flow valuation问答", "source_paraphrase": "Gerstner公开信把支出纪律连接现金流；次年访谈强调技术潜力并不能替代合理价格和证券条款分析。", "model": "编辑合成：增量FCF=税后现金费用节省+资本开支减少−重组现金−其他增量现金需求。", "boundary": "减少资本开支可能牺牲未来收入，不能把节省永久化。", "counterexample": "砍掉高回报产能扩张会提高当期FCF却降低未来价值。", "test": "区分一次性与经常性节省、维持性与成长性投资，核验税盾及客户流失。", "question": "合成企业拟减少年度现金费用10、资本开支5；税率20%，无其他变化。稳定年度FCF改善多少？", "constraint": "首年另付重组现金3，且维持业务需追加资本开支2；题设两项均无额外税效应，首年改善多少？", "transfer": "一项AI证券投入100，60%概率回收180，40%回收40；若合同把高情景回收限制为130，期望回收与未折现回报各多少？", "tags": [ "F11", "F12", "F21" ], "profile": "gerstner-altimeter", "source_ids": [ "B100-BG1", "B100-BG2" ], "attribution": "来源转述与后续编辑合成计算明确区分；题设数字非本人预测。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "BHS01", "profile": "hendi-susanto", "title": "AI设备投入增长，供应商就有定价权吗？", "source_ids": [ "BX-HS-20250422" ], "locator": "Semiconductor Themes；Kokusai Electric", "source_paraphrase": "共同研究指出，制造技术升级提供设备机会，客户集中也可能削弱供应商议价。", "attribution": "共同材料转述；以下数字、模型和挑战由编辑合成。", "model": "合成：收入=采购总额×可服务份额×供应商份额；利润另计价格和成本。", "boundary": "设备支出增长不能替代供应商份额和利润率证据。", "counterexample": "最大客户集中采购并压价，技术需求增加而供应商利润下降。", "test": "工艺步骤、设计导入、客户集中、实现价格和分部利润。", "question": "合成情景：客户设备预算100增至130；某供应商可服务比例40%降至30%，份额固定50%。其收入如何变化？", "constraint": "再加入平均降价10%、单位成本不变；缺少数量与成本基数时，能否给出利润点估计？", "transfer": "迁移到HBM封装设备：如何区别封装总投资与某一步骤可服务投资？", "tags": [ "F9", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "BHS02", "profile": "hendi-susanto", "title": "长期需求正确，买入估值仍会错吗？", "source_ids": [ "BX-HS-20240812" ], "locator": "AI Hype vs. Reality完整署名小节", "source_paraphrase": "AI长期资本开支机会与估值风险、传统终端库存调整可以并存。", "attribution": "署名观点转述；估值数字与案例为编辑合成。", "model": "合成：固定股数和同口径EPS下，股价比=EPS比×PE比。", "boundary": "EPS增长不等于自由现金流增长，也不保证倍数不变。", "counterexample": "盈利增长兑现，但起点估值已反映更强增长。", "test": "冻结买入时EPS预期、倍数及风险预算，再对照兑现情况。", "question": "合成：EPS从4升至5，PE从40降至28；无分红、股数不变，股价回报是多少？", "constraint": "现在EPS一半增量来自回购；你还需要哪些现金流和融资证据？", "transfer": "迁移到云服务：如何区分业务增长与每股指标改善？", "tags": [ "F12", "F14", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "BZWS01", "profile": "zheng-weishan", "title": "代工涨价，是设计公司的利好还是挤压？", "source_ids": [ "BX-ZWS-20210910", "BX-ZWS-20220609" ], "locator": "2021正文首段；2022问答最后一题", "source_paraphrase": "产品竞争力和产品梯队影响设计公司能否转嫁代工涨价，不能用一家代工厂代表全行业。", "attribution": "历史材料观点；以下财务模型为编辑合成。", "model": "合成：毛利=收入−直接成本；转嫁率以增量价格/增量成本定义。", "boundary": "维持毛利额与维持毛利率所需涨价不同。", "counterexample": "产品可替代性强，竞争者库存充足，设计商无法同步提价。", "test": "单位售价、代工成本、产品组合、销量和客户合同。", "question": "合成：每颗芯片售价100，直接成本60，其中代工30；代工涨价20%，数量固定。售价应提高多少才保持原毛利率？", "constraint": "客户只接受涨价3%，重新计算毛利额和毛利率；不能把收入增长当利润增长。", "transfer": "迁移到服务器：HBM或DDR涨价后，ODM成本加成合同能否保护百分比利润率？", "tags": [ "F7", "F10", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "BZWS02", "profile": "zheng-weishan", "title": "一个子行业下行能代表半导体整体吗？", "source_ids": [ "BX-ZWS-20210910", "BX-ZWS-20220609" ], "locator": "2021局部价格与全行业区分；2022分领域财报问答", "source_paraphrase": "把局部价格变化和全行业分开，利用细分财务数据验证产业逻辑。", "attribution": "历史材料归纳；下列权重和变化为编辑假设。", "model": "合成：固定基期权重计算可比收入变化；财务汇总与中位数分开。", "boundary": "终端、产品和统计口径不同，不可直接跨公司外推。", "counterexample": "高增速板块权重小，整体仍下滑；结构变化可能遮蔽普遍疲弱。", "test": "匹配产品、区域、基期和权重，检查收入、利润及库存。", "question": "合成行业两部分：存储基期收入60下降20%，逻辑基期收入40增长25%。总收入变化是多少？", "constraint": "逻辑的增长全来自一家龙头；还能说典型公司景气改善吗？", "transfer": "迁移到AI服务器与通用服务器，怎样避免把高ASP增长当出货量复苏？", "tags": [ "F6", "F8", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "attribution": "公开材料的编辑转述；数值、模型、约束及迁移练习均为编辑合成，不是本人原始案例或投资建议。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计", "id": "B71C01", "profile": "buffett-berkshire", "title": "利润增长来自经营，还是资产价格？", "source_ids": [ "B71D01", "B71D02" ], "locator": "2023 letter P2–3 Operating Results; 2024 letter P3–4 Last Year's Performance", "source_paraphrase": "把经营盈利与投资资产市值波动区分，并以长期经济价值判断持股。", "model": "编辑合成：报告利润=税后经营利润＋税后未实现投资损益；经营自由现金流=税后经营利润＋折旧−现金CapEx−营运资本增加。", "boundary": "此处所有项目均为同币种百万，税后数值已给；实际报表需要独立核对投资收益、税及现金分类。", "counterexample": "账面盈利上升但经营利润和现金流恶化，不能直接叫经营改善。", "test": "同口径经营利润、实际CapEx、营运资本和投资损益桥。", "question": "合成案例：去年税后经营利润100、未实现投资收益0；今年分别80和70。今年折旧20、现金CapEx60、营运资本增加10。计算报告利润增长、经营利润增长和今年经营自由现金流。", "constraint": "把70未实现投资收益改为−30，其他全部不变。重算报告利润与经营自由现金流，解释为何两者变化不同。", "transfer": "分析云厂商持股估值波动与AI机房现金支出，怎样避免把会计利润当成可自由分配现金？", "tags": [ "F3", "F4", "F8", "T2" ] }, { "attribution": "公开材料的编辑转述；数值、模型、约束及迁移练习均为编辑合成，不是本人原始案例或投资建议。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计", "id": "B71C02", "profile": "buffett-berkshire", "title": "盈利奖金是否奖励了额外投入的资本？", "source_ids": [ "B71D01", "B71D02" ], "locator": "2023 letter P3 high-return incremental capital; 2024 letter P1 Forest River incentive adjustment", "source_paraphrase": "重视增量资本的回报；Forest River案例中，额外收购投入应调整奖金基准。", "model": "编辑合成：增量NOPAT/增量投入资本；简化经济利润=增量NOPAT−资本门槛×增量资本。", "boundary": "假设进入稳定期、资本不折旧、增量NOPAT永久持续、无后续净投资、现金流年底发生。不是原信里的真实回报计算。", "counterexample": "总利润提高，新增资本回报却低于门槛；按利润增量发奖仍可能奖励毁值。", "test": "项目全生命周期现金流、资本时点、收购与有机增长拆分。", "question": "合成：假设新增投资在期初付出，此后稳定增量税后利润等于每年末自由现金流且永续，无追加净投资，折现率8%。业务税后经营利润从100升至112，期间新增资本200；资本门槛8%。利润增长率、增量资本回报、增量经济利润是多少？按新增利润10%发放的奖金是多少？按这一永续简化假设，新投资NPV是多少？", "constraint": "新增资本仍为200，稳定增量利润改为20。重算回报、经济利润、奖金及NPV；奖金只作对比，不从给定NOPAT再扣。", "transfer": "将同一检验用于GPU集群扩容：吞吐增长能否覆盖新增资本的机会成本？", "tags": [ "F8", "F12", "F14", "T2" ] }, { "attribution": "公开材料的编辑转述；数值、模型、约束及迁移练习均为编辑合成，不是本人原始案例或投资建议。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计", "id": "B71C03", "profile": "nygren-oakmark", "title": "预测正确的增长，是否已经在股价里？", "source_ids": [ "B71D03" ], "locator": "TEAM / What is the ultimate role of an analyst?", "source_paraphrase": "非共识观点必须明确哪里不同，并量化其估值影响；知道很多并不等于有投资优势。", "model": "编辑合成：一年末价格=一年末EPS×届时P/E；期间无股息，当前股数不变。", "boundary": "终值倍数是独立假设，不能用技术增长自动证明。", "counterexample": "EPS超预期但估值倍数下降，价格仍可能下跌。", "test": "共识EPS、自己EPS、倍数依据及季度证伪指标。", "question": "合成：假设没有股息且股数不变。当前EPS4、股价100。共识一年后EPS4.4、P/E20；你的研究估计EPS5、P/E20。分别算一年后价格及相对当前总回报；你的EPS判断比共识高多少？", "constraint": "你EPS5的预测兑现，但一年后P/E降到16。计算股价及回报。若目标是一年10%回报且EPS5，最低期末P/E是多少？", "transfer": "一家公司AI服务器收入超预期，怎样区分预期差、EPS转换与已计入股价的估值？", "tags": [ "F1", "F5", "F14", "T2" ] }, { "attribution": "公开材料的编辑转述；数值、模型、约束及迁移练习均为编辑合成，不是本人原始案例或投资建议。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计", "id": "B71C04", "profile": "nygren-oakmark", "title": "AI硬件高利润持续多久，才有安全边际？", "source_ids": [ "B71D03", "B71D04" ], "locator": "2026 commentary P2–3, AI hardware profitability and margin of safety", "source_paraphrase": "AI领导企业目前盈利强，但盈利能否长期持续仍需判断；不确定不等于确定看空。", "model": "编辑合成：三年FCFE逐年年底折现；第三年另收取清算剩余价值。无债务、不追加融资，无其他现金流。", "boundary": "本练习没有永续终值；第三年剩余价值是清算所得，不应再叠加终值倍数。概率未给，不计算期望价值。", "counterexample": "技术采用继续增长，竞争和供给扩张却使利润正常化。", "test": "供给与需求、价格、正常化利润率、资本支出及退出价值。", "question": "合成：买价100。高利润持续情景每年末FCFE30，第三年末另有清算剩余价值50（不含当年FCFE）；折现率10%。算股权现值、相对现值的买价折扣，以及要求20%折扣时的最高买价。", "constraint": "正常化情景改为第一年FCFE30、第二三年各15，清算剩余价值仍50。重算现值与20%折扣最高买价；未给情景概率时能否断言100有安全边际？", "transfer": "区分GPU、存储和网络的不同竞争及供给周期，不用同一个利润持续期外推整个AI供应链。", "tags": [ "F1", "F5", "F8", "F14" ] }, { "id": "EXP200-PL01", "profile": "philippe-laffont", "title": "提高现金能降低多少风险，又放弃多少上行？", "source_ids": [ "EXP200-B28-A", "EXP200-B28-B" ], "locator": "ECNY PDF printed pp.11–12 提高现金段落；CTEK PDF p.28 Risk Management / Mosaic、p.59 Primary Portfolio Manager", "source_paraphrase": "Laffont把现金视为保留后续行动能力；机构流程披露头寸、现金和模型监测的结合。", "attribution": "个人经验与机构流程分开；以下组合、收益率和估值调整全部为编辑合成，不代表实际持仓或交易建议。", "model": "组合损益=各资产初始权重×情景收益；账面不更新不等于经济价值不变。", "boundary": "不含费税、杠杆、对冲或现金利息；私人资产跌幅仅为合成风险情景。", "counterexample": "现金降低市场暴露，也可能在上涨时造成机会成本；私人估值滞后会掩盖真实共同风险。", "test": "检查标价日期、共同客户/融资暴露、现金用途和有日期的增减仓条件。", "question": "合成：净资产100，其中上市科技40、私人科技20、现金40。上市资产跌30%，私人账面暂不更新，但经济价值情景跌20%。账面净值与情景经济净值分别是多少？", "constraint": "在跌市前将上市资产20按账面卖成现金，其余不变。重算两种净值；不能卖出的私人资产如何影响风险预算？", "transfer": "另设独立上涨情景：以上两种初始配置的上市资产均涨20%，私人和现金不变。增加现金放弃多少收益？迁移到GPU、能源与云服务篮子时为何还需检查共同客户风险？", "tags": [ "F11", "F14", "F22", "F23" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "E7S01", "profile": "rasgon-bernstein", "title": "分部与价格的增长桥", "source_ids": [ "E7S01", "S3D02" ], "locator": "Q1 p.13; Q2 p.10", "source_paraphrase": "Q1澄清新云归类；Q2追问产品与价格贡献。", "attribution": "具名问题归分析师，回答归管理层；公式、假设、反例、挑战均为编辑合成，不代表本人投资立场。", "model": "用统一口径构造可加总增长桥；不是本人目标价或评级。", "boundary": "价格、销量及分部边界必须可比。", "counterexample": "披露变更或产品绑定会造成重复计算。", "test": "获取可比期间原始披露及合同数据，冻结口径后检查模型预测和实际兑现。", "question": "合成：上年收入100，本年170。按统一旧年分类，提价贡献10、新产品贡献20，其余为原产品销量及未解释项。计算残差；能否直接证明原产品销量增长？", "constraint": "复核发现新产品20中有8只是分类变化，不是新增收入；收入170不变。把新产品真正增量改为12，计算剩余未解释项，并列出验证要求。", "transfer": "迁移到CPU＋GPU整机：怎样避免把CPU从整机拆出后同时计入两项增长？", "tags": [ "F2", "F4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行投资收益或预测命中率审计" }, { "id": "E7S02", "profile": "moore-ms", "title": "供需差与可交付收入", "source_ids": [ "E7S01", "S3D02" ], "locator": "Q1 p.7; Q2 p.8", "source_paraphrase": "Q1追问分部竞争差异；Q2追问需求与供给增长差额。", "attribution": "具名问题归分析师，回答归管理层；公式、假设、反例、挑战均为编辑合成，不代表本人投资立场。", "model": "将意向需求与可交付上限分开，再检验按客户分组的竞争条件。", "boundary": "需求没有重复下单，交付口径一致。", "counterexample": "未融资或未通电的订单并非可兑现收入。", "test": "获取可比期间原始披露及合同数据，冻结口径后检查模型预测和实际兑现。", "question": "合成：上年交付100台；次年有效需求200台；供应上限170台；单价固定1。收入上限及未满足需求是多少？", "constraint": "供应提高至190台，但完成融资及上电的客户只接收160台；价格不变。重算上限。", "transfer": "迁移到HBM采购：晶圆产量、封装良率和服务器验收哪个约束当季收入？", "tags": [ "F9", "F10" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行投资收益或预测命中率审计" }, { "id": "E7S03", "profile": "arya-bofa", "title": "增量市场与生态融资分开", "source_ids": [ "E7S01", "S3D02" ], "locator": "Q1 p.11; Q2 p.11", "source_paraphrase": "Q1追问独立CPU与替代；Q2追问生态支持及竞争性自研。", "attribution": "具名问题归分析师，回答归管理层；公式、假设、反例、挑战均为编辑合成，不代表本人投资立场。", "model": "先消除产品重叠，再将销售机会与融资风险分别入模。", "boundary": "产品边界可核验；保证并不等同立即现金支出。", "counterexample": "新增CPU可能替换旧服务器；被支持客户也可能发展替代芯片。", "test": "获取可比期间原始披露及合同数据，冻结口径后检查模型预测和实际兑现。", "question": "合成：系统业务100，新增独立CPU20；其中5替代原系统销售。总收入增量和新总额是多少？", "constraint": "另有信用保证30，假设20%触发、触发损失率50%；这不是收入抵减。计算预期损失并说明现金风险。", "transfer": "迁移到AI云：设备销售、参股和保底采购怎样分别列示而不相互抵销？", "tags": [ "F9", "F10", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行投资收益或预测命中率审计" }, { "id": "E7S06", "profile": "arcuri-ubs", "title": "专用产品与开放模型的价值归属", "source_ids": [ "E7S01", "S3D02" ], "locator": "Q1 p.10–11; Q2 p.12", "source_paraphrase": "Q1问专用推理产品；Q2问开放模型与前沿公司的关系。", "attribution": "具名问题归分析师，回答归管理层；公式、假设、反例、挑战均为编辑合成，不代表本人投资立场。", "model": "把终端模型收入、计算用量及芯片份额分开，避免把模型降价直接等同硬件需求下降。", "boundary": "任务质量和成本口径一致，模型与芯片份额不能混用。", "counterexample": "开放模型也可能迁移至竞争硬件；专用芯片不覆盖所有上下文和负载。", "test": "获取可比期间原始披露及合同数据，冻结口径后检查模型预测和实际兑现。", "question": "合成：原来100个同质量付费任务，每任务2单位算力，供应商份额80%。开放模型后任务150，每任务1单位，份额仍80%。该供应商算力需求怎样变？", "constraint": "若份额提高到90%，重算；价格每算力单位不变。能否仅凭任务增长推断硬件收入增长？", "transfer": "迁移到CPU：向量加速降低单任务CPU时间，但调用增加，怎样检验净核时和收入？", "tags": [ "F6", "F9", "F10" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行投资收益或预测命中率审计" }, { "id": "E7S04", "profile": "lipacis-evercore", "title": "渠道库存与真实消耗对账", "source_ids": [ "E7S03", "E7S04" ], "locator": "ADI p.4–5; TI Mark Lipacis Q&A", "source_paraphrase": "2011追问库存周数；2023对照分销报表与渠道叙事。", "attribution": "具名问题归分析师，回答归管理层；公式、假设、反例、挑战均为编辑合成，不代表本人投资立场。", "model": "用同产品范围的库存恒等式核对出货与消耗；仅为历史方法样本。", "boundary": "期初期末、币种及产品范围一致。", "counterexample": "分销商总库存混合多个供应商，不能代表单一厂商。", "test": "获取可比期间原始披露及合同数据，冻结口径后检查模型预测和实际兑现。", "question": "合成：期初渠道库存80，本期厂商出货120，下游消耗100，无退货。期末库存和库存增量是多少？", "constraint": "加入退回厂商10，价格不变；另知分销商其他厂商库存增60。能否以分销商总库存增长证明本厂商库存恶化同样严重？", "transfer": "迁移到GPU渠道：出货、云租用率与终端付费任务怎样形成三层对账？", "tags": [ "F7", "F8", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行投资收益或预测命中率审计" }, { "attribution": "问题关注点来自具名一手材料；模型、数字、约束、反例与迁移均为编辑合成，不是本人预测。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "id": "S36C01", "profile": "charlie-chan-ms", "title": "汇率压力能否被涨价抵消？", "source_ids": [ "S36SRC1" ], "locator": "Charlie Chan Q&A, printed PDF pp.8–9 (zero-based pages 7–8)", "source_paraphrase": "Chan询问汇率冲击能否进入后续晶圆定价；未承诺涨价幅度。", "model": "合成：毛利率=(美元收入−美元成本)/美元收入；价格、销量、固定与变动成本分别变动。", "boundary": "假定全部成本为本币、无套保，美元换算成本上升10%；不代表TSMC真实成本结构。", "counterexample": "提价减少销量，固定成本无法同步下降。", "test": "同产品价格、销量、汇率、成本币种与套保分别核对。", "question": "合成工厂原收入100、固定成本20、变动成本25。销量和成本用量不变，美元成本因汇率上升10%。新毛利率是多少？若仅涨价，恢复原毛利率需涨价多少？", "constraint": "改为涨价10%后销量下降10%；固定成本不变、变动成本按销量缩放，再统一乘1.10汇率系数。重新算收入、成本和毛利率。", "transfer": "转到欧洲半导体设备：收入币种、零部件币种与金融套保不同，哪些原假设必须改？", "tags": [ "F2", "F4", "F18", "F19" ] }, { "attribution": "问题关注点来自具名一手材料；模型、数字、约束、反例与迁移均为编辑合成，不是本人预测。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "id": "S36C02", "profile": "charlie-chan-ms", "title": "移除一个市场，原增长目标还需要什么？", "source_ids": [ "S36SRC1", "S36SRC2" ], "locator": "TSMC Q2 printed pp.8–9; Q3 printed pp.8–9, Charlie Chan China-market follow-up", "source_paraphrase": "Chan跨季度回访中国限制对长期AI增长的影响；增长信心来自管理层回答。", "model": "合成：分地区收入加总，再反推剩余市场需要的增速。", "boundary": "区域互斥且会计口径不变，不把潜在市场直接当已实现收入。", "counterexample": "原基准排除了中国，则删除中国潜在机会不自动下修该基准。", "test": "冻结预测原始地域构成、交付期和许可假设；分别更新而非总额套百分比。", "question": "合成公司当前收入100，中国20、其他80。两年后总收入目标按40% CAGR设定。若两年后中国收入为0，其他地区需达到多少收入及两年CAGR才能维持总目标？", "constraint": "其他地区两年实际只能按50% CAGR增长，中国仍为0。重算总收入与整体两年CAGR；指出增长率下降是否等于收入下降。", "transfer": "把地理市场改为GPU与ASIC客户：如何避免同一笔下游CapEx在两类上游收入中重复计算？", "tags": [ "F3", "F4", "F6", "F14", "F19" ] }, { "attribution": "问题关注点来自具名一手材料；模型、数字、约束、反例与迁移均为编辑合成，不是本人预测。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "id": "S51C01", "profile": "keith-weiss-ms", "title": "最大的AI客户会不会成为竞争者？", "source_ids": [ "S51SRC2" ], "locator": "Microsoft FY25 Q4, Keith Weiss question and Nadella response", "source_paraphrase": "Weiss询问AI实验室可能自建并与云厂商竞争时的风险回报管理。", "model": "合成：收入按客户拆分；新增客户收入不自动等额替代高毛利流失收入。", "boundary": "以下客户毛利已按相同成本口径计量；未给估值和资本成本，不能推股票价值。", "counterexample": "客户流失恰好释放稀缺容量，被新客户以更高贡献填补。", "test": "客户集中、合同约束、可迁移工作负载、空置时间和替补收入质量。", "question": "合成云收入100：最大AI客户40、毛利率60%；其他客户60、毛利率40%。下一年AI客户自建使收入减半，其他客户收入增20%，毛利率不变。总收入与总毛利分别变化多少？", "constraint": "流失后额外闲置费用8，此费用尚未计入上述毛利且全部计当期成本。重算毛利及其变化；收入变化能否单独证明平台风险降低？", "transfer": "转到CPU定制采购：客户推出自研CPU后，外购核心数减少与供应商毛利减少为什么不能直接同比例映射？", "tags": [ "F4", "F10", "F14", "F16", "F20" ] }, { "attribution": "问题关注点来自具名一手材料；模型、数字、约束、反例与迁移均为编辑合成，不是本人预测。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "id": "S51C02", "profile": "keith-weiss-ms", "title": "研发能力、可售产品与付费采用不是同一阶段", "source_ids": [ "S51SRC1", "S51SRC3" ], "locator": "Morgan Stanley 2025-03-03 transcript, Weiss innovation stages; Microsoft Q2 execution question", "source_paraphrase": "Weiss区分技术研究、产品开发/建设和商业采用，并追问云收入执行偏差。", "model": "合成：收入=获准试用客户×付费转化率×付费客户用量×实现单价。", "boundary": "客户/用量期间一致，试用量未计收入；技术能力不保证采用或盈利。", "counterexample": "免费试用迅速增加但转化、付费使用率或实现价格下降。", "test": "跟踪同批次客户转化、价格、可售容量与销售渠道，拆需求与交付限制。", "question": "合成产品有1000家试用客户，30%付费，每家付费客户每月1000次任务，实现单价1元。计算月收入。次年试用客户翻倍、付费率20%、每家任务增50%、单价降20%，月收入增长多少？", "constraint": "可交付容量仅40万次付费任务/月；按同一0.8元单价全部交付且无其他收费。重算收入，并区分需求、可服务收入与已收现金。", "transfer": "转到服务器采购：把PoC、正式部署与生产利用率映射到各阶段，如何避免用试点数量代替采购订单？", "tags": [ "F3", "F5", "F7", "F10", "F14" ] }, { "attribution": "问题关注点来自具名一手材料；模型、数字、约束、反例与迁移均为编辑合成，不是本人预测。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "id": "S52C01", "profile": "mark-moerdler-bernstein", "title": "用量收入能补上席位减少吗？", "source_ids": [ "S51SRC2", "S52SRC2" ], "locator": "Microsoft FY25 Q4 Moerdler monetization; Adobe Q3 FY25 p.17", "source_paraphrase": "Moerdler问AI用量收入能否弥补潜在席位压力，并问不同SaaS计费和利润率。", "model": "合成：总收入=席位收入+用量收入；增量毛利需扣两类直接成本。", "boundary": "两类收入互斥、未重复包含AI套餐；不把合同额、调用量和收入混用。", "counterexample": "用量收入增长但推理成本增长更快，或套餐内用量并非额外收费。", "test": "同客户队列的席位数量、净价、免费额度、超额付费、推理及服务成本。", "question": "合成SaaS有10000席位，每月100元、毛利率80%。AI后席位减20%，每月增加30万次付费任务，每次1元、直接成本0.4元。新旧月收入及毛利各是多少？", "constraint": "付费任务不变，但每任务成本升至0.6元。重算毛利，并求抵消席位毛利损失至少需要多少月付费任务；其他假设不变。", "transfer": "转到推理API：若席位只是包装而账单按token收费，如何构建互斥收入桥？", "tags": [ "F2", "F4", "F10", "F11", "F16" ] }, { "attribution": "问题关注点来自具名一手材料；模型、数字、约束、反例与迁移均为编辑合成，不是本人预测。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计", "id": "S52C02", "profile": "mark-moerdler-bernstein", "title": "AI收入变重要，利润也一定变重要吗？", "source_ids": [ "S52SRC2" ], "locator": "Adobe Q3 FY25, Moerdler AI materiality question p.17", "source_paraphrase": "Moerdler追问AI何时足以影响Creative业务；未给出本人收入或利润预测。", "model": "合成：增量经营利润=AI收入×增量毛利率−新增运营费用−被替代旧业务毛利。", "boundary": "费用和替代收入均属相同期间，未给税、资本开支或营运资金，不能推出FCF。", "counterexample": "AI高速增长但替代原高毛利收入，或成本/销售费用超过增量毛利。", "test": "新增与替代分离；按用户队列量化旧收入流失、AI直接成本及新增费用。", "question": "合成公司旧业务收入1000、毛利率80%。新AI收入100、毛利率40%，新增运营费用30；AI同时替代旧业务收入20，旧业务毛利率仍80%。相对无AI基准，总收入与经营利润分别增加多少？", "constraint": "AI收入变为200，直接成本率仍60%；新增运营费用变为50，替代旧收入变为50。重算增量经营利润，是否已转正？", "transfer": "转到GPU租赁：把软件席位替代改为旧GPU租金蚕食，加入折旧与设备现金支出时怎样避免双重扣除？", "tags": [ "F2", "F4", "F10", "F11", "F18" ] }, { "id": "EXP200-ME01", "profile": "michael-elias", "title": "涨租究竟增加收益，还是只抵消更高成本？", "source_ids": [ "EXP200-S81-A" ], "locator": "Full transcription：Elias关于2021–2025成本、价格与收益率差的追问；网页lines199–217", "source_paraphrase": "Elias区分建造成本、融资成本及供需带来的回报差。", "attribution": "分析师提出拆解；收益率趋稳判断来自Martynek。下列全部数字与模型为编辑合成。", "model": "稳定NOI收益率=年NOI/建造资本；超额差=NOI收益率−匹配资本门槛。仅作筛查，不等同项目IRR。", "boundary": "相同稳定出租率、费用口径与资本化边界；融资门槛不是无条件固定。", "counterexample": "租金涨幅可能只覆盖建造成本和融资成本；爬坡与追加资本会进一步削弱回报。", "test": "逐项目核对全建造成本、稳定NOI、租约、电力交付日期及融资期限。", "question": "合成：项目建造成本100，稳定年NOI为10，资本门槛7%。新项目成本升至120，NOI升至12，资本门槛升至8%。收益率及超额差如何变化？", "constraint": "新项目还须追加资本10，稳定NOI仍12。重算收益率差；这是否足以算IRR？", "transfer": "迁移到液冷改造：原项目投资100、年净节省10，追加20但净节省仍10；门槛8%，计算筛查收益率差并列出IRR还缺什么。", "tags": [ "F9", "F10", "F14", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "EXP200-ME02", "profile": "michael-elias", "title": "四份电力申请究竟代表多少需求？", "source_ids": [ "EXP200-S81-B" ], "locator": "Full transcript：电力申请、重复需求和amplifier effect段落；网页lines250–287", "source_paraphrase": "Elias要求解释电力宣称，并识别同一最终需求通过多方申请重复出现。", "attribution": "重复申请例子与承诺标准由Greason解释；Elias指出放大效应。练习数值为编辑合成。", "model": "先按最终客户与用途去重，再按时间窗口求可交付量；需求≠申请≠已通电量。", "boundary": "互为替代的申请才去重；真实分期增量必须单列。", "counterexample": "申请可能没有接入协议、设备或劳动力，不能乘建设单价直接算可确认收入。", "test": "唯一项目键、最终客户、接入承诺、设备验收、投运时间及客户付款。", "question": "合成：同一客户需要300MW，委托4家开发商各申请300MW，互为替代。另有独立客户确定需求80MW。原始申请总数和去重需求各多少？", "constraint": "所研究地区当年可通电只有200MW。去重需求不变，年度实际交付上限多少？能否由此直接确定收入？", "transfer": "迁移到HBM：同一芯片商需求100万颗，向3家发出互为替代的100万颗询价；另一客户需求20万颗，可供货80万颗。写出去重需求和供应上限。", "tags": [ "F7", "F9", "F17", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "EXP200-JM01", "profile": "julian-mitchell", "title": "订单增长何时成为收入增长？", "source_ids": [ "EXP200-S92-A", "EXP200-S92-B" ], "locator": "nVent PDF pp.10–11 Julian Mitchell Q&A；Trane PDF pp.9–10 opening Julian Mitchell Q&A", "source_paraphrase": "Mitchell把订单、交期和收入指引放在同一时间框架检验。", "attribution": "分析师提问与管理层回答分开；以下发货比例完全是编辑假设，不是公司指引。", "model": "期内收入=期初积压×期内转化率+新订单×期内转化率；只在同一价格及确认口径下成立。", "boundary": "明确订单价值、期间、取消率、验收条款；积压不自动等于下一期收入。", "counterexample": "交期延长会提高积压但压低近期收入；客户延期或取消打断转换。", "test": "订单批次、客户请求交期、实际发货/验收及后续收入兑现。", "question": "合成：年初积压200，80%将在今年交付；今年新增订单100，30%今年交付，价格稳定且交付即确认。计算今年收入及年末积压。", "constraint": "期初积压有25%取消；余下仍80%今年交付。新增订单不变。重算收入、取消额和年末积压。", "transfer": "迁移至AI服务器：订单交付后还需客户验收。如果已发货190中40未验收且控制权未转移，应确认多少收入？", "tags": [ "F3", "F4", "F5", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "EXP200-JM02", "profile": "julian-mitchell", "title": "收入加速后，涨价是否真正保护利润率？", "source_ids": [ "EXP200-S92-A" ], "locator": "nVent PDF p.10 Julian Mitchell follow-up关于全年利润率和通胀/价格", "source_paraphrase": "Mitchell在收入增长之后另查利润率和通胀对冲。", "attribution": "Mitchell询问桥接；具体涨价、通胀与指引属于管理层回答。练习为编辑合成。", "model": "利润=数量×(售价−单位成本)−固定费用；利润率=利润/收入。", "boundary": "销量、汇率、收购与会计口径保持一致；收入涨价不等于毛利率不变。", "counterexample": "按成本金额完全转嫁仍可能稀释百分比利润率；产能投入还增加费用。", "test": "同口径价格、销量、原材料/工资、生产率及产能爬坡费用桥。", "question": "合成：销量100、售价10、单位变动成本6、固定费用200。成本涨至6.6、售价涨至10.6，销量及固定费用不变。比较利润和利润率。", "constraint": "新增产能使固定费用升至230，其余不变。利润和利润率多少？", "transfer": "迁移到GPU服务器：原售价100、购料80、其他单位成本10；购料升至100且售价升至120。比较单位毛利和毛利率，忽略其他费用。", "tags": [ "F2", "F3", "F4", "F10" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA01", "profile": "unwin-polar", "title": "收入增长能否抵消资本密集度上升？", "source_ids": [ "O8BAS05", "O8BAS06" ], "locator": "FY2024 Magnificent 7；FY2025 Mag-7 update", "source_paraphrase": "联合材料把盈利增长与自由现金流压力分别观察。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：简化FCF=经营现金流−现金资本开支。", "boundary": "同一财年、现金口径；未把折旧再次扣除。", "counterexample": "增长来自需持续重投的设备，收入高增但可分配现金下降。", "test": "未来两季查看经营现金流、CapEx、租赁与预收；现金改善则修正。", "question": "合成，亿美元：经营现金流100、CapEx40；下一年分别增长20%和75%。两年FCF及变化？", "constraint": "下一年经营现金流包含一次性预收10；剔除后FCF是多少？", "transfer": "CPU平台销售增长但客户账期从30天延至90天，怎样修正现金桥？", "tags": [ "F11", "F20", "F21" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA02", "profile": "unwin-polar", "title": "组合低于指数权重是否一定低风险？", "source_ids": [ "O8BAS06" ], "locator": "Concentration risk", "source_paraphrase": "团队关注大股权重集中，区分相对配置与绝对暴露。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：单一冲击损失贡献=权重×该资产回报。", "boundary": "只计算局部冲击，不代表组合总风险。", "counterexample": "名义持仓不同却共享云客户或融资链。", "test": "下一财报季核对客户与资本来源，做相关冲击压力测试。", "question": "合成：组合某股权重12%，指数权重20%；股价跌40%，其余资产不变。各自损失多少？", "constraint": "其余组合88%同步跌10%，总组合损失多少？", "transfer": "三家CPU供应链公司都依赖同一云客户，怎样识别看似分散的风险？", "tags": [ "F22", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA03", "profile": "porter-janus", "title": "直接税负为零为何利润仍下滑？", "source_ids": [ "O8BAS02" ], "locator": "Considerations 1–3：direct impact、inflation、event vs economic risk", "source_paraphrase": "团队把政策直接税负与终端需求冲击分开。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：营业利润=销量×(售价−变动成本)−固定费用。", "boundary": "政策情景是假设；不描述现行关税。", "counterexample": "客户推迟预算，即使供应商免税也受影响。", "test": "下一季度观察取消订单、终端消耗、价格折让；恢复则撤销需求下调。", "question": "合成：卖100万件，每件售价100美元、变动成本60美元，固定费用2000万美元；销量降15%，价格成本不变。营业利润怎样变化？", "constraint": "在新销量上再降价5%，成本不变，利润是多少？", "transfer": "软件本身无实物进口税，但客户广告预算下降，如何建立间接暴露模型？", "tags": [ "F19", "F4", "F10" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA04", "profile": "porter-janus", "title": "需求很强，收入却被容量和融资限制", "source_ids": [ "B100-RC2" ], "locator": "AI is transformative；skepticism about valuations", "source_paraphrase": "共同材料同时关注供给限制与循环投资。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：可确认收入≤min(已交付数量,客户可验收数量)×单价。", "boundary": "假设控制权在验收转移，真实合同需重核。", "counterexample": "客户的采购资金完全来自供应商，订单不等于独立需求。", "test": "当季验收、现金收款和融资来源；下一季复核逾期情况。", "question": "合成：订单100套，供货80套，客户仅60套可通电验收；每套100万美元，收入确认上限多少？", "constraint": "其中20套由供应商提供全额贷款，验收和收入条件不变，其余40套当期全额付款；当期收到现金多少？", "transfer": "迁移到HBM：产能公告、良率、客户认证、出货如何逐层限流？", "tags": [ "F9", "F11", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA05", "profile": "clark-janus", "title": "AI代理新增席位能否抵消员工席位减少？", "source_ids": [ "O8BAS01" ], "locator": "What are our views：seat-based models", "source_paraphrase": "Clark认为代理可能成为可收费用户；这是可检验的商业假设。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：收入=人类席位×人类单价+付费代理数×代理单价。", "boundary": "代理必须确实收费，且与原席位不重复计费。", "counterexample": "代理通过共享API调用，无新增付费席位。", "test": "未来两个季度看合同、付费代理数、净留存与实际折扣。", "question": "合成：原有1万人类席位，每月100美元；减少20%，新增5000代理，每月20美元。月收入变化？", "constraint": "仅40%代理可独立计费，其余包含在旧合同，月收入变多少？", "transfer": "推理API token翻倍而每token价格降60%，收入能否上升？", "tags": [ "F16", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA06", "profile": "clark-janus", "title": "核心系统留存是否等于收入安全？", "source_ids": [ "O8BAS01" ], "locator": "Vibe-coding；add-on modules；Cowork data/context", "source_paraphrase": "Clark把难以迁移的核心系统与可替代附加功能分开。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：总收入=核心收入+模块收入；系统留存不约束模块价格。", "boundary": "同一客户群与同一年度，不引入新增客户。", "counterexample": "客户保留ERP记录系统，却把价值较高的工作界面交给外部代理。", "test": "下次续约核对核心留存、模块附着率与净扩张；模块收入恢复则修正。", "question": "合成：核心收入8000万美元、附加模块2000万；核心不变，模块流失一半，总收入变化？", "constraint": "核心再提价5%，模块仍为1000万，总收入和增长是多少？", "transfer": "CPU ISA兼容性不变，但云厂商转向自研CPU；哪层价值仍被替代？", "tags": [ "F16", "F15", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA07", "profile": "narang-columbia", "title": "订单增加来自终端还是补库存？", "source_ids": [ "B100-PW1" ], "locator": "PDF第2页Narang疫情回答；第3页长期主题回答", "source_paraphrase": "Narang指出原有需求预期与家庭科技消费发生偏差。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：渠道进货=终端消耗+库存变化。", "boundary": "同质产品、相同期间；不忽略退货。", "counterexample": "短缺预期诱发重复下单，渠道进货不代表持续需求。", "test": "下一季终端销售、库存周数与取消率；库存回落而订单下降应修正。", "question": "合成：渠道原库存20万件，进货120万、终端售出100万，期末库存多少？若上期终端也是100万，能否说需求增长20%？", "constraint": "下一季终端仍100万且渠道恢复库存20万，需要进货多少？", "transfer": "服务器CPU订单激增同时客户库存积压，应先查什么？", "tags": [ "F7", "F8", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA08", "profile": "narang-columbia", "title": "平台的社会价值如何变成供应商收入？", "source_ids": [ "B100-PW2" ], "locator": "2023 Q&A：platform transition；areas of opportunity", "source_paraphrase": "Narang讨论平台转换与企业用途；宏观效益需另接公司收入桥。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：可收费节约=可实现节约×付费采纳比例；供应商收入再乘价值捕获率。", "boundary": "企业节约不等于现金释放，采用率与捕获率必须分开。", "counterexample": "工具节省工时却没有减少加班、人员或提高付费产出。", "test": "试点结束90天内看可实现现金节约、付费续约与总推理成本。", "question": "合成：客户潜在年度节约10亿美元，只有30%兑现现金，其中20%愿付给软件商。收入机会是多少？", "constraint": "只有一半收入机会最终转为实际收入；每年新增推理和服务成本2000万美元，贡献额多少？", "transfer": "芯片性能提高30%，怎样从工作负载效率推到客户愿付溢价？", "tags": [ "F18", "F10", "F16" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA09", "profile": "fish-janus", "title": "供给受限能否保证高资本回报？", "source_ids": [ "O8BAS04" ], "locator": "Transcript：natural governor；data-centre shells and utilization", "source_paraphrase": "Fish把施工和电力时滞视作供给约束；本题检验这种约束是否足够保护回报。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：经营贡献=可售GPU小时×付费利用率×(单价−变动成本)−固定费用。", "boundary": "装机满载与商业利用率不同；暂不含折旧税费。", "counterexample": "GPU被部署但用于免费试验，或降价使单位收益不足。", "test": "未来两季核对收费利用率、实现价格和现金回款；不以排队数量替代。", "question": "合成：年可售1000万GPU小时，付费利用率80%，单价4美元、变动成本1美元，固定费用1000万美元；经营贡献多少？", "constraint": "单价降至2美元，其余不变；盈亏平衡需要多少付费利用率？", "transfer": "CPU云实例核数售罄但大量折扣赠额，如何修正收入判断？", "tags": [ "F10", "F7", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA10", "profile": "fish-janus", "title": "AI节省工时怎样进入利润表？", "source_ids": [ "O8BAS03" ], "locator": "Enablers and early adopters", "source_paraphrase": "Fish讨论AI采用者利润率提升；本题区分技术效率与兑现后的利润。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：净利润增量代理=人工支出×可自动化比例×可兑现比例−AI运行费用。", "boundary": "这是税前经营利润近似，不含收入提升、重组费及税。", "counterexample": "节省时间被用于改善质量，却没有费用下降或收入增长。", "test": "部署后两个财季核对人员费用、外包成本、质量与增收；未兑现则降预期。", "question": "合成：年人工费用1亿美元，AI可节省20%工时，只有25%能转为现金费用节约，AI年费用800万美元，净影响多少？", "constraint": "要达到收支平衡，可兑现比例至少多少？", "transfer": "芯片设计验证加速，如何区分工具节约、流片风险下降与提前上市收入？", "tags": [ "F10", "F18", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA11", "profile": "rizzo-trowe", "title": "关键供应商的份额与周期要分开算", "source_ids": [ "O8BAS07", "O8BAS08" ], "locator": "Official Q&A stock selection；host summary Framework and Semiconductors", "source_paraphrase": "框架把关键技术、份额与基本面改善分开。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：收入=行业支出×可服务比例×份额。", "boundary": "收入按当期交付，不将订单与销售混同。", "counterexample": "全行业资本开支下降，份额提升不能完全抵消。", "test": "下一季可服务支出、交付额与份额；不依靠主题标签。", "question": "合成：行业支出100亿美元，公司可服务比例50%、份额20%；下一年总支出降20%，份额由20%升至25%，可服务比例不变。收入增长多少？", "constraint": "可服务比例因产品错配降至40%，收入又是多少？", "transfer": "EDA工具不可替代，但客户流片数下降；哪个变量能否抵消？", "tags": [ "F7", "F6", "F16" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA12", "profile": "rizzo-trowe", "title": "高质量增长也可能给出负回报", "source_ids": [ "O8BAS07" ], "locator": "Official Q&A fourth criterion：reasonable valuation", "source_paraphrase": "Rizzo把合理估值纳入技术选股框架。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：一年回报=(EPS1×PE1)/买入价−1。", "boundary": "无分红、股本与会计口径不变。", "counterexample": "公司成长兑现但市场预期原已过高。", "test": "下一财年检验EPS、现金转换与终值假设；涨跌本身不证明方法。", "question": "合成：买入EPS4、PE50；一年后EPS增长25%，PE降至35倍。回报多少？", "constraint": "若要保本，在终值PE35下EPS至少多少、增长率多少？", "transfer": "自研CPU大幅降成本，但供应商客户浓度上升，如何修正终值倍数？", "tags": [ "F12", "F21", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA13", "profile": "curtis-franklin", "title": "更高生产率不是等额收入", "source_ids": [ "O8BAS09", "O8BAS10" ], "locator": "Podcast：quantify knowledge-worker impact；article Why opportunity is not in tech", "source_paraphrase": "Curtis将知识工作增效与价值创造相连；本题明确TAM并非收入。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：可收费收入=劳动成本池×可兑现增益×付费比例×供给商捕获率。", "boundary": "时间节约需能转为产出或费用减少；各比例边界不重复。", "counterexample": "价值被用户保留或在竞争中让渡给客户。", "test": "试点后两个季度看现金效益、付费续约和供应商实际收费。", "question": "合成：100亿美元劳动成本，同等产出下劳动成本可减少10%，全部兑现，50%采用付费工具，供应商捕获增益20%；收入机会多少？", "constraint": "增益仅40%兑现，其他比例不变，收入机会多少？", "transfer": "AI优化数据中心能耗，节电价值怎样在云厂商、软件和客户间分配？", "tags": [ "F18", "F10", "F16" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BA14", "profile": "curtis-franklin", "title": "研究自动化节省的是时间还是判断？", "source_ids": [ "O8BAS09" ], "locator": "Podcast：financial services；tools versus alpha", "source_paraphrase": "Curtis把研究工具的效率与投资判断区分。", "attribution": "来源观点的编辑转述；以下模型、参数和挑战是合成教学案例，不是本人预测、私有模型或真实持仓。", "model": "合成：总人时=自动化后初读+复核返工；质量与产出另测。", "boundary": "同样文档复杂度与错误发现标准。", "counterexample": "处理数增加但错误污染模型，复核工作反而上升。", "test": "下一批100份材料盲测漏错率、总人时与有证据结论。", "question": "合成：每份报告原耗2小时；AI初读降至0.5小时，但平均复核1小时。100份节约多少小时？", "constraint": "20份另需每份3小时纠错，其余条件不变；净节约还剩多少？", "transfer": "架构基准自动摘要同样变快，如何确保配置口径和瓶颈没有误读？", "tags": [ "F25", "F17", "F23" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-TS1", "profile": "O8BB-tom-slater", "title": "AI用量翻倍，供应商利润一定增加吗？", "source_ids": [ "O8BB-TS24", "O8BB-TS25" ], "locator": "2024 Artificial Intelligence；2025 The Rise of AI and the Shifting Shape of Value", "source_paraphrase": "Slater把AI普及与供应商成本、利润空间和投资风险回报分开。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "收入=合格单位量×实现售价；毛利=量×(售价−单位成本)。", "boundary": "相同质量、产品组合和计量单位；此例成本是单位变动成本，非整机TCO。", "counterexample": "软件和网络壁垒可支持售价，不能预设商品化必然发生。", "test": "未来两季跟踪实际ASP、单位成本、付费采用及替代率；若售价保持且单位成本下降，修正利润压缩假设。", "question": "合成：量100→200，售价100→60，单位成本40→30。计算收入、毛利及毛利率变化。", "constraint": "新产品使单位成本只能降到35，且新增研发费用1500；相对原始无额外费用的毛利，新增后的营业贡献如何？", "transfer": "迁移到HBM：比特出货增长和每GB价格变化如何连接利润，而非只看市场容量？", "tags": [ "F10", "F18", "T4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-TS2", "profile": "O8BB-tom-slater", "title": "业务更强，股票仍可能缺少不对称回报", "source_ids": [ "O8BB-TS25" ], "locator": "Resilience as a Strategic Advantage；AI risk/reward paragraph", "source_paraphrase": "Slater强调适应力及估值下的风险回报，而非仅以公司优秀决定持仓。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "P=EPS×PE；无分红、无增发的期限末情景。", "boundary": "终值倍数与增长不能同时无依据乐观设定。", "counterexample": "盈利持续性改善可支撑较高倍数，应与竞争情景对照。", "test": "未来两个年报检查现金转化、每股盈利与市场隐含增长；若收入增长依赖外部融资，降低持续性假设。", "question": "合成：现价300、EPS10，三年EPS年增20%，期末PE20倍；算累计及年化价格回报。", "constraint": "竞争使EPS年增仅10%，期末PE仍20；若要求10%年回报，最高买入价是多少？", "transfer": "迁移到数据中心REIT：把分配现金流、增长CapEx与终值分别建模。", "tags": [ "F12", "F21", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-LB1", "profile": "O8BB-lawrence-burns", "title": "价值链收入不能相加为终端AI需求", "source_ids": [ "O8BB-LB24" ], "locator": "Hardware / Infrastructure / Applications；Conclusion", "source_paraphrase": "Burns分三层识别AI机会，并提醒应用形成较晚及硬件消化期。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "最终外部支出与链内采购分别记账；各层收入不是新增总需求。", "boundary": "假设明确同一期间且没有期初期末存货或预付变化。", "counterexample": "云服务还可能服务非AI用户，不能把所有云收入都归为AI。", "test": "未来两季对照应用外部付费、云利用率和硬件投产；若应用支出不升而闲置容量升，撤销同步扩张假设。", "question": "合成：终端客户向AI应用支付100；应用向云支付35；云向芯片公司支付20。可否称生态需求155？怎样报告三层收入？", "constraint": "云支付的20中12由新增股权融资支持；终端支付仍100。能否把12算新增外部客户需求？", "transfer": "迁移到手机供应链：手机零售额、代工和零件收入如何防止重复计数？", "tags": [ "F6", "F9", "F11", "T4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-LB2", "profile": "O8BB-lawrence-burns", "title": "第二增长业务应按成功概率，还是按故事估值？", "source_ids": [ "O8BB-LB25" ], "locator": "Outliers Address Large Opportunities；Outliers are Adaptable", "source_paraphrase": "Burns认为企业可借既有能力拓展新市场，但把新增业务价值明确视为可能而非确定。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "项目NPV=−初始投入+概率加权期末现金流现值。", "boundary": "概率是合成假设，不以成功公司样本推定真实基率。", "counterexample": "第二业务可能分散管理层注意并损害主业。", "test": "未来12个月检查付费转化、留存及主业损失；未达预设里程碑即下调成功概率。", "question": "合成：现付30，三年后成功收到120、失败收到0；成功率30%，折现率10%。计算NPV和盈亏平衡成功率。", "constraint": "失败时除30初始投入，还会在第三年损害主业现金10。重新计算NPV和成功概率门槛。", "transfer": "迁移到CPU厂商推出AI加速器：渠道复用与主业机会成本怎样共同进入模型？", "tags": [ "F10", "F14", "F20", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-DB1", "profile": "O8BB-douglas-brodie", "title": "增长波折还是商业逻辑失效？", "source_ids": [ "O8BB-DB24", "O8BB-DB25" ], "locator": "2024 pp2–3 Sprout Social / new holdings；2025 p2 Sweetgreen", "source_paraphrase": "Brodie区分执行波折与增长前景变弱，并在后续保留需求和竞争的不利证据。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "收入=客户数×每客户收入；贡献=收入×贡献率−固定费用。", "boundary": "此例假设客户同质；真实分析需按留存队列和客户规模拆开。", "counterexample": "提价期间小客户流失可能合理，但高价值客户流失会改变结论。", "test": "未来四季度观察同队列留存、折扣和现金贡献；若改善仅来自新客补贴，不能称原问题解决。", "question": "合成：原客户1000、年收入/客100、贡献率60%、固定费用50000；调整后客户800、收入/客140，其他不变。收入和营业贡献如何变化？", "constraint": "收入/客140含一次性实施费20，次年客户仍800、无新实施费。重新计算可持续贡献。", "transfer": "迁移到云实例迁移：提价后收入增长能否掩盖高质量负载流失？", "tags": [ "F2", "F14", "F23", "T6" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-DB2", "profile": "O8BB-douglas-brodie", "title": "开发者熟悉度如何变成工业收入？", "source_ids": [ "O8BB-DB24" ], "locator": "2024 p3 Raspberry Pi", "source_paraphrase": "Brodie把教育/爱好者使用与进入工业设计联系，提出边缘AI机会。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "工业收入=开发者数×进入工业设计比例×成功率×每设计采购量×ASP。", "boundary": "每一环均需分母和时间窗口；兴趣、设计导入与量产互不等同。", "counterexample": "工业认证、生命周期和供应支持可能阻碍熟悉产品进入量产。", "test": "未来18个月核对设计赢单、认证和量产采购；若设计量增而生产转换下降，撤销线性用户乘数。", "question": "合成：100000名开发者中2%各提出一个工业设计，其中10%的设计进入量产，每个量产设计全年采购1000块、ASP50。潜在全年销售额是多少（假定全部交付并满足收入确认条件）？", "constraint": "认证延迟使本年仅25%的量产设计开始采购；每个开始采购的设计本年只完成一半全年用量。按同一交付和收入确认假设，修改本年销售额。", "transfer": "迁移到RISC-V：开源下载、工程设计采用和商业芯片收入如何区分？", "tags": [ "F3", "F15", "F18", "T3" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-BW1", "profile": "O8BB-brett-winton", "title": "AI生成热度能覆盖推理成本吗？", "source_ids": [ "O8BB-BW22" ], "locator": "Generative AI Tops the App Store", "source_paraphrase": "Winton的ARK评论区分流量、持久优势与广告/付费模式的单位经济。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "贡献=付费收入+广告收入−推理成本；免费使用也消耗资源。", "boundary": "不含研发、获客、支付和审核费用，贡献不是利润。", "counterexample": "多模态小模型或缓存能降成本；同时用户可能只为一次性新奇付费。", "test": "未来90天按用户队列测复购和每成功任务成本；若免费占比上升，重新检验盈亏边界。", "question": "合成：月生成1000万次，5%付费，每次收入0.20；其余广告每次0.002；所有推理成本每次0.01。计算月贡献。", "constraint": "付费比例不变、总量翻倍，推理成本降至0.008，但广告单价减半。贡献如何变化？", "transfer": "迁移到游戏AI NPC：互动次数增长、留存提升和单位算力成本怎样共同决定投入价值？", "tags": [ "F10", "F16", "F18", "T2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-BW2", "profile": "O8BB-brett-winton", "title": "供应商投资回流，怎样检验收入质量？", "source_ids": [ "O8BB-BW23" ], "locator": "When Investment Dollars Become Revenue Dollars", "source_paraphrase": "Winton指出科技供应商投资客户后形成采购回流，并把持续性与竞争限制列为待观察问题。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "供应商销售现金与股权投资现金分开；客户外部收入与融资分开。", "boundary": "投资关系本身不否定有效销售；需合同、控制权和关联方披露。", "counterexample": "有独立付费需求的早期公司也可能合理使用供应商融资。", "test": "未来12个月检查客户非融资现金收入、续约及供应商限制；若复购只能靠同一供应商继续注资，降低收入持续性假设。", "question": "合成：供应商向初创客户投资100，客户本期用80向其买设备；供应商生产成本50，全部现金结算。供应商毛利与两项交易合计现金变化分别多少？", "constraint": "客户本次投资交易余下20现金已在当年用于其他运营。下一年期初现金为0，客户只有外部经营可用现金20，却需另购设备60，无其他资金。收入增长预测还缺什么？", "transfer": "迁移到电信设备商融资建网：装机增长和运营商自我造血能力如何分开？", "tags": [ "F2", "F9", "F11", "T5" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-TK1", "profile": "O8BB-tasha-keeney", "title": "模拟次数越多，推出时间就越可信？", "source_ids": [ "O8BB-TK23" ], "locator": "Key Model Inputs / Share Price Forecast / Risks and Limits", "source_paraphrase": "联合模型对robotaxi推出日期高度敏感，并公开均值和分位数的定义。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "场景价值=成功概率×成功价值+(1−概率)×未成功价值；模拟精度与输入真实性分开。", "boundary": "未来价值与今天价值必须用同一时点；题设均为已折现今天的价值。", "counterexample": "相关的监管、融资和技术失败可同时压低成功率与成功后利润。", "test": "每半年对照原始推出定义和商业化里程碑；延迟时保留原预测并重估，不只滚动终点。", "question": "合成：按时商业化概率70%，成功与未成功情景在今天的现值分别为500和100；计算期望现值。100万次模拟可否证明70%准确？", "constraint": "只有成功情景的现金流再延期两年，故其原现值500需再按10%折现两年；成功概率降至40%，失败情景今天的现值仍为100。重算期望现值。", "transfer": "迁移到先进封装量产：良率、客户认证和融资同受延期影响时如何避免独立输入假设？", "tags": [ "F14", "F21", "F25", "T6" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-TK2", "profile": "O8BB-tasha-keeney", "title": "兼容车队不等于可收费运力", "source_ids": [ "O8BB-TK24" ], "locator": "How would Tesla’s robotaxi service work?；utilization paragraph", "source_paraphrase": "Keeney把潜在兼容车辆、扩张与利用率连接，但兼容数量仍需跨过运营与商业化条件。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "平台净收入=车辆×资格比例×参与比例×付费里程/车×每里程价格×抽成。", "boundary": "题设所有资格比例互相条件化；抽成收入不等于平台利润。", "counterexample": "车主可能不愿加入，清洁、远程协助、空驶和保险可能抵消成本优势。", "test": "未来四季核验可无监督收费车辆、付费里程、空驶和净抽成；不能以监督驾驶里程替代商业运力。", "question": "合成：100000辆潜在车，60%可合规服务、其中20%参与；每车年付费30000英里、单价1美元、平台抽成30%。净收入多少？", "constraint": "年付费里程降至20000；平台每辆参与车承担年度运营支持费2000美元。收入和扣该支持费后的贡献多少？", "transfer": "迁移到GPU云：已购买、已通电、已分配、付费利用率与有效任务产出如何逐层衔接？", "tags": [ "F10", "F18", "T3" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BB-TK3", "profile": "O8BB-tasha-keeney", "title": "监督驾驶里程能与无人驾驶里程直接比较吗？", "source_ids": [ "O8BB-TK23", "O8BB-TK24" ], "locator": "2023 five safety caveats；2024 note below real-world miles chart", "source_paraphrase": "来源比较行驶数据，也明确FSD需要驾驶注意而Waymo样本无驾驶员；2023列出人工接管等局限。", "attribution": "来源观点仅作具名历史转述；以下数值、模型、问题和反证均为编辑合成，不是作者原模型或真实预测。", "model": "事故率=事故数/暴露里程；先匹配道路、天气、监督、报告门槛和接管规则。", "boundary": "统计上可比不等于随机对照或因果识别；不得用教学例证明任何真实系统安全。", "counterexample": "人工在困难场景接管会选择性移除高风险暴露，低事故率未必来自系统能力。", "test": "下一次安全披露必须同时报告接管、运行域和事故定义；缺任何关键分母则只作描述。", "question": "合成：A系统有驾驶监督，1亿英里10起事故；B系统无人监督，1000万英里2起事故。计算表面事故率，可否说A安全一倍？", "constraint": "A还记录10000次接管，但不知道不接管会发生什么。能否把所有接管算事故，或一律忽略？", "transfer": "迁移到AI代理：有人兜底的任务成功率能否和完全自主完成率直接比较？", "tags": [ "F17", "F25", "T6" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BCC01", "profile": "qiu-gaoyi", "title": "集中度提高，定价权是否随之提高？", "source_ids": [ "O8BCS02" ], "locator": "行业格局及“定义天空”段", "source_paraphrase": "集中度须对应真实竞争边界与议价能力。", "attribution": "公开材料转述；以下模型、数字与挑战均为编辑合成，不是本人预测或真实交易。", "model": "贡献=销量×(售价−单位变动成本)", "boundary": "需纳入替代技术和买方议价。", "counterexample": "双寡头也可能遭大客户压价。", "test": "合同续签价、份额与替代产品采用。", "question": "合成：销量100，售价10，单位成本6；下一期销量增20%，售价降15%，成本降10%。贡献变化多少？", "constraint": "新增竞争使销量只增5%，其他不变，重算。", "transfer": "迁移至HBM与普通DRAM的竞争边界。", "tags": [ "F16", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BCC02", "profile": "qiu-gaoyi", "title": "利润激励与回款激励怎样区分？", "source_ids": [ "O8BCS01" ], "locator": "管理层维度：中层KPI与回款", "source_paraphrase": "管理考核可影响现金质量。", "attribution": "公开材料转述；以下模型、数字与挑战均为编辑合成，不是本人预测或真实交易。", "model": "简化经营现金=利润−应收增加", "boundary": "假设其他营运资本、非现金项、税费调整均为零。", "counterexample": "回款暂好可能来自保理融资。", "test": "账龄、追索权与回款后退款。", "question": "合成：A、B利润均10，应收增加分别8和2；简化现金各多少？", "constraint": "B现金中有无追索权应收出售3；剔除此一次性释放后重算持续现金。", "transfer": "迁移至GPU服务器验收后的应收回收。", "tags": [ "F11", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BCC03", "profile": "deng-gaoyi", "title": "产业快速扩容为何未必利好龙头？", "source_ids": [ "O8BCS03" ], "locator": "行业演进与迭代；成熟行业回报", "source_paraphrase": "技术进步可能淘汰旧资产和既有企业。", "attribution": "公开材料转述；以下模型、数字与挑战均为编辑合成，不是本人预测或真实交易。", "model": "收入=行业量×份额×价格；毛利=收入×毛利率", "boundary": "口径保持同质；不代表股权回报。", "counterexample": "旧龙头错过技术路线转换。", "test": "良率、成本、技术份额与减值。", "question": "合成：行业量从100增至200，公司份额30%降至15%，售价从10降至6，毛利率从30%降至20%；收入和毛利？", "constraint": "公司另需一次性资产减值20；给出毛利减该减值的简化比较，不称净利润。", "transfer": "迁移至GPU向专用加速器的替代。", "tags": [ "F15", "F18", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BCC04", "profile": "deng-gaoyi", "title": "正确方向如何承受等待时间？", "source_ids": [ "O8BCS04" ], "locator": "回顾投资策略：长期与时间风险", "source_paraphrase": "兑现时间不确定需要组合安排。", "attribution": "公开材料转述；以下模型、数字与挑战均为编辑合成，不是本人预测或真实交易。", "model": "年化=(终值/初值)^(1/t)−1", "boundary": "零分红、无税费；终值只是情景。", "counterexample": "价值正确但融资/赎回迫使提前出售。", "test": "现金期限与催化剂时间。", "question": "合成：支付100，情景终值150；两年或五年兑现的年化收益各多少？", "constraint": "资金成本每年10%，五年后的最低保本终值是多少？", "transfer": "迁移至晶圆厂投产延期。", "tags": [ "F21", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BCC05", "profile": "feng-gaoyi", "title": "公开减持能证明转空吗？", "source_ids": [ "O8BCS05", "O8BCS06" ], "locator": "2019采访：高仓位资金调动、长期客户资金段；2020采访：公开持仓不完整及“抄作业”问答", "source_paraphrase": "组合流动性需要可导致看好标的被减持。", "attribution": "公开材料转述；以下模型、数字与挑战均为编辑合成，不是本人预测或真实交易。", "model": "现金缺口=赎回−现金；减持行为与观点分开", "boundary": "不知道全组合时不能反推信念。", "counterexample": "减仓可能支付赎回或买入更优标的。", "test": "完整资金流、时点、交易理由。", "question": "合成：组合100，现金2，赎回10；若其余资产不变，至少需卖出多少？", "constraint": "所售资产必须以账面价值95%成交，需卖出多少账面额？", "transfer": "迁移至机构出售债券应付保证金。", "tags": [ "F22", "F23" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BCC06", "profile": "feng-gaoyi", "title": "估值保护是否保证重估？", "source_ids": [ "O8BCS06" ], "locator": "成长/价值股与估值保护问答", "source_paraphrase": "便宜不保证市场在可承受期限内重估。", "attribution": "公开材料转述；以下模型、数字与挑战均为编辑合成，不是本人预测或真实交易。", "model": "总回报=(期末价+分红)/买价−1", "boundary": "终值和分红需证据。", "counterexample": "盈利恶化使低倍数合理化。", "test": "分红覆盖、盈利稳定与资本要求。", "question": "合成：买价80，年度分红4，三年后价格仍80；不再投资，累计回报多少？", "constraint": "第三年股价60且三年总分红仍12，累计回报？", "transfer": "迁移至成熟网络设备商。", "tags": [ "F21", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BCC07", "profile": "liu-igw", "title": "实际增长怎样被价格下跌抵消？", "source_ids": [ "O8BCS07", "O8BCS08" ], "locator": "季报§4.4；年报§4.4–4.5", "source_paraphrase": "量价分离是判断名义需求与盈利的起点。", "attribution": "公开材料转述；以下模型、数字与挑战均为编辑合成，不是本人预测或真实交易。", "model": "名义增长=(1+实际增长)×(1+价格变化)−1", "boundary": "宏观GDP价格不等于公司售价。", "counterexample": "份额下滑或成本上涨抵消宏观恢复。", "test": "公司实现售价、订单量、成本。", "question": "合成：实际量增5%，价格降2%，名义收入增长多少？", "constraint": "单位成本不变，初始售价100、成本80，销量100→105；毛利变化？", "transfer": "迁移至存储bit增长和ASP下降。", "tags": [ "F18", "F19", "F10" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BCC08", "profile": "liu-igw", "title": "库存减少是需求回升还是供给收缩？", "source_ids": [ "O8BCS08" ], "locator": "年报§4.5：库存和产能利用率", "source_paraphrase": "库存出清不单独证明终端需求复苏。", "attribution": "公开材料转述；以下模型、数字与挑战均为编辑合成，不是本人预测或真实交易。", "model": "期末库存=期初+生产−销售", "boundary": "单位与时间窗口一致。", "counterexample": "减产去库存而销售继续恶化。", "test": "销量、生产、价格和未交订单联看。", "question": "合成：期初库存30，生产80，销售90，期末多少？能否断言需求增长？", "constraint": "上期销售100，当前销售90，怎样修订判断？", "transfer": "迁移至CPU渠道库存清理。", "tags": [ "F7", "F8", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BCC09", "profile": "fu-foresight", "title": "量产、良率与ASP如何共同影响收入？", "source_ids": [ "O8BCS09" ], "locator": "中期报告§4.4.1：PCB与光模块", "source_paraphrase": "量产进度与制造改善是产业链增长的条件。", "attribution": "公开材料转述；以下模型、数字与挑战均为编辑合成，不是本人预测或真实交易。", "model": "可售量=投入量×良率；收入=可售量×售价", "boundary": "同产品质量且良品全部售出；收入确认另核。", "counterexample": "客户认证延迟导致良品库存积压。", "test": "认证、良率、出货与回款。", "question": "合成：投入100、良率80%、单价10；新投入120、良率90%、单价降10%，收入变化？", "constraint": "只有90%的新良品被验收并确认收入，重算。", "transfer": "迁移至先进封装交付与验收。", "tags": [ "F18", "F2", "F15" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8BCC10", "profile": "fu-foresight", "title": "持仓占比增加是否等于主动加仓？", "source_ids": [ "O8BCS10" ], "locator": "年度报告§4.4.1：集中度与资本回报疑问", "source_paraphrase": "集中度及AI暴露增加需要分清价格与交易贡献。", "attribution": "公开材料转述；以下模型、数字与挑战均为编辑合成，不是本人预测或真实交易。", "model": "期末权重=初始权重×(1+标的回报)/(1+组合回报)", "boundary": "无现金流和交易的反事实。", "counterexample": "股价上涨本身造成权重上升。", "test": "持股数量、申赎、交易及行业共同风险。", "question": "合成：股票权重10%，上涨50%，其他资产价格不变且无交易；新权重？", "constraint": "股票随后从新价下跌40%，其他不变，相对最初组合损失多少？", "transfer": "迁移至云厂商资本开支集中于同一供应商。", "tags": [ "F22", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SAC01", "profile": "omalley-barclays", "title": "增长最快的业务是否提升整体毛利？", "source_ids": [ "O8SAD1", "O8SAD2" ], "locator": "AMD Q1 2026 Earnings Call pp8–9，Thomas O’Malley 问答；AMD Q2 2026 Earnings Call pp7–8，Tom O’Malley 问答", "source_paraphrase": "问CPU/GPU季度节奏与供给、电力限制；管理层澄清AI收入因中国业务环比下降，给出Helios爬坡节奏。；问CPU/GPU收入交叉时点与利润组合；管理层强调两者增长及不同毛利贡献，没有给确定交叉日期。", "attribution": "具名问题归分析师，管理层回答单列于证据；公式、数值、反例和练习均为编辑合成，不是本人估计或建议。", "model": "分业务收入×各自毛利率，再扣增量费用。", "boundary": "只在题目假设及单位、时点和范围一致时成立；不能代替真实公司估计。", "counterexample": "高增长业务稀释毛利率但仍增加毛利额。", "test": "按产品收入/毛利与研发费用桥。", "question": "合成：CPU收入60、毛利率65%；GPU收入40、毛利率45%。下一年CPU增长50%、GPU增长100%，毛利率各自不变。求总收入、毛利和综合毛利率。", "constraint": "新增约束：GPU爬坡令其毛利率降至35%，增量研发费用8。计算营业利润相对原基期的变化（其他费用不变）。", "transfer": "迁移至高毛利交换芯片与低毛利整机：为何销售额增长不能代替利润桥？", "tags": [ "F2", "F4", "F18", "T2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SAC02", "profile": "buchalter-td-cowen", "title": "核心数增长能否解释全部收入增长？", "source_ids": [ "O8SAD1", "O8SAD2" ], "locator": "AMD Q1 2026 Earnings Call pp6–8，Joshua Buchalter 问答；AMD Q2 2026 Earnings Call pp10–11，Joshua Buchalter 问答；主持人口述转录为Buckwalter", "source_paraphrase": "追问CPU市场预测变化、竞争份额和MI450客户范围；管理层给出工作负载、客户预测及推理部署解释。；问2027 CPU增长的量价及核心数代理，并区分Helios超计划是需求还是良率；管理层澄清指需求量，良率改善为后续预期。", "attribution": "具名问题归分析师，管理层回答单列于证据；公式、数值、反例和练习均为编辑合成，不是本人估计或建议。", "model": "收入=出货量×核心数/颗×每核心价格，代理指标不等于可比性能。", "boundary": "只在题目假设及单位、时点和范围一致时成立；不能代替真实公司估计。", "counterexample": "核心数增加伴随频率与内存带宽瓶颈。", "test": "相同配置ASP、代际份额和工作负载实测。", "question": "合成：CPU收入增长60%，出货量增长25%，每颗核心数增长20%。假设价格可拆为核心数×每核心价格，推算ASP和每核心价格变化。", "constraint": "新增约束：低端与高端核心性能差异显著。能否将每核心价格变化解释成相同性能提价？给出需要新增的证据。", "transfer": "迁移至HBM容量：每颗容量增长与每GB价格、良率和收入怎样分开？", "tags": [ "F2", "F18", "F25", "T2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SAC03", "profile": "malik-citi", "title": "吞吐提升何时变成利润改善？", "source_ids": [ "S1D01", "O8SAD2" ], "locator": "NVIDIA FY2026 Q4 Earnings Call — issuer-hosted corrected transcript pp11–12，Atif Malik 问答；AMD Q2 2026 Earnings Call p16，Atif Malik 问答", "source_paraphrase": "问推理投资增加后CUDA的重要性；管理层给出软件栈及系统性能论证，非分析师背书。；问Cerebras合作的快速推理收入前景与费用增速；管理层给出产品可用时间和费用慢于收入增长的方向，未量化收入。", "attribution": "具名问题归分析师，管理层回答单列于证据；公式、数值、反例和练习均为编辑合成，不是本人估计或建议。", "model": "利润=付费任务×价格−全部任务×单位成本−固定费用。", "boundary": "只在题目假设及单位、时点和范围一致时成立；不能代替真实公司估计。", "counterexample": "免费用量上升掩盖商业转化恶化。", "test": "付费率、任务成功率及全成本。", "question": "合成：推理服务付费任务量100、每任务价格1、每任务变动成本0.4、固定研发费用30。架构升级后付费任务增50%、售价降20%、单位成本降25%、研发增20%。计算两期营业利润。", "constraint": "新增约束：总任务仍为150，其中新增50任务的三分之一改为免费试用，但全部任务仍产生成本。重新计算利润，并指出算力利用率为何不能替代付费率。", "transfer": "迁移到存储分层：命中率与SSD节约怎样经由付费业务和迁移成本影响现金？", "tags": [ "F10", "F16", "F18", "T2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SAC04", "profile": "reitzes-melius", "title": "批准的市场机会可以直接加到下一季收入吗？", "source_ids": [ "O8SAD3", "S1D01" ], "locator": "AMD Q2 2025 Earnings Call p16，Ben Reitzes 问答；NVIDIA FY2026 Q4 Earnings Call — issuer-hosted corrected transcript pp12–13，Ben Reitzes 问答", "source_paraphrase": "追问季度增量是否主要GPU及中国许可恢复后的爬坡；管理层确认MI355为主要驱动，但提示许可后仍需时间。；追问高毛利可持续性与内存创新；管理层将答案连接代际客户价值，没有提供保证毛利的成本桥。", "attribution": "具名问题归分析师，管理层回答单列于证据；公式、数值、反例和练习均为编辑合成，不是本人估计或建议。", "model": "当季确认收入=已签订单×当季交付比例×确认比例；机会不等于订单。", "boundary": "只在题目假设及单位、时点和范围一致时成立；不能代替真实公司估计。", "counterexample": "许可获批后物流、需求或验收仍延迟。", "test": "批准、订单、交付、验收与确认逐阶段台账。", "question": "合成：原预计下一季收入10，其中GPU为4。新增许可涉及已签GPU订单2，但下一季仅60%可交付且收入确认比例为已交付的75%。其余业务不变，算当季收入与新增未确认金额。", "constraint": "新增约束：新增许可产品毛利率30%，原业务毛利率60%，额外合规费用0.1。计算总毛利率及增量营业利润。", "transfer": "迁移到电网并网许可：获批MW、通电MW、可计费容量为何不能相等？", "tags": [ "F3", "F4", "F11", "T2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SAC05", "profile": "kumar-piper", "title": "更快芯片为什么可能仍带来更少可用算力？", "source_ids": [ "O8SAD4", "O8SAD5" ], "locator": "O8SAD4 p12，Harsh Kumar 问答；O8SAD5 pp15–16，Harsh Kumar 问答", "source_paraphrase": "问Instinct机架级差距和UALink的补齐时点。 管理层谈互连、网络和系统投入，没有给差距关闭的确定日期。；问客户采用障碍及Instinct毛利达到公司水平的条件。 Lisa Su强调软件熟悉度和多代客户信任，Jean Hu给出规模改善方向，未给到达日期。", "attribution": "分析师提问与管理层回答分开；本卡公式、数值与情景均为编辑合成。", "model": "有效算力=芯片数量×单芯片性能×集群效率×按期可用比例。", "boundary": "只在题目给定期间、良率和成本范围内成立，未给参数不能推定。", "counterexample": "系统和供应链的约束抵消单一技术指标改善。", "test": "按同一口径核查交付、性能、良率、成本和实际部署，建立可复算桥。", "question": "合成：原方案100颗芯片，每颗性能1，集群效率80%，全部按期可用。新方案单颗性能提升50%，集群效率70%，只有70%按期可用；数量仍100。比较当期有效算力。", "constraint": "新增约束：新方案效率仍70%，需要达到原方案有效算力，最少多少比例必须按期可用？另有一次软件迁移费用，是否能仅凭算力宣布TCO改善？", "transfer": "迁移至液冷机架：芯片峰值、机架供电与通电率怎样共同限制有效产出？", "tags": [ "F10", "F15", "F18", "T2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SAC06", "profile": "schafer-oppenheimer", "title": "制程缩小一定降低每颗合格芯片成本吗？", "source_ids": [ "O8SAD7", "O8SAD6" ], "locator": "O8SAD7 SEC正文开头：backlog and supplies问答，HTML lines28–58；O8SAD6 pp7–9 PC市场与架构；pp12–13 工艺/ASP问答", "source_paraphrase": "问消费市场放缓释放的晶圆与基板供应能多快帮助MaxLinear，及订单可见度。 管理层称供应改善仍取决于节点与厂商，较长交期要求相应backlog；这不是已兑现收入。；问PC可服务机会及Arm/x86影响，再问22nm转移对毛利和ASP的作用。 管理层区分可服务高端市场与总量，指出新节点成本、良率和模拟电路缩放限制，不提供通用提价规则。", "attribution": "分析师提问与管理层回答分开；本卡公式、数值与情景均为编辑合成。", "model": "裸片成本=晶圆成本/(每片毛裸片数×良率)，再加封装测试。", "boundary": "只在题目给定期间、良率和成本范围内成立，未给参数不能推定。", "counterexample": "系统和供应链的约束抵消单一技术指标改善。", "test": "按同一口径核查交付、性能、良率、成本和实际部署，建立可复算桥。", "question": "合成：旧节点晶圆成本4000，每片1000颗毛裸片，良率90%；新节点晶圆成本6000，每片1500颗，良率75%。封测每颗均1，售价均10。计算两种单位总成本与毛利率。", "constraint": "新增约束：新节点良率提升到90%，但每颗封测升至1.4。新节点毛利率能否达到旧节点？若要保持旧毛利率，最低售价是多少？", "transfer": "迁移至HBM堆叠：制程密度、堆叠良率、封装成本为何不能各自孤立比较？", "tags": [ "F10", "F15", "F18", "T2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SB-BL01", "profile": "bruce-lu-gs", "title": "token翻倍，晶圆收入一定翻倍吗？", "source_ids": [ "S36SRC2" ], "locator": "印刷页11–13，Bruce Lu问答", "source_paraphrase": "Lu追问token与收入增速的差距。", "attribution": "原始提问归于具名分析师，回答归于管理层。下列模型、数字、阈值与练习均为编辑合成，不是分析师预测。", "model": "所需设备量比=token量比/每设备有效吞吐比；采购收入还须调整闲置容量、库存、价格与供应份额。", "boundary": "任务质量、时间窗口及有效吞吐口径一致；设备需求不是当期订单。", "counterexample": "软件改善、旧设备利用率提升可吸收需求；低质token不等于有价值工作量。", "test": "下次财报核对吞吐、利用率、新增安装与采购确认时点；若旧容量足够则撤销新增采购推断。", "question": "合成：token需求翻倍，每设备有效吞吐提高60%，原设备量100且无闲置。需要多少设备、增幅多少？", "constraint": "原100台只能利用80%，通过调度升到100%，同时吞吐提高60%；需求翻倍还缺多少台？以最终每台满载吞吐计。", "transfer": "迁移至物流：包裹增50%，每车吞吐增25%，原100辆满载；所需车辆及增幅？", "tags": [ "F6", "F7", "F18", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计预测结果或投资业绩" }, { "id": "O8SB-GH01", "profile": "gokul-hariharan-jpm", "title": "收入跑赢CapEx说明什么？", "source_ids": [ "S36SRC2" ], "locator": "印刷页6–7，Gokul Hariharan", "source_paraphrase": "Hariharan比较收入与CapEx增长。", "attribution": "原始提问归于具名分析师，回答归于管理层。下列模型、数字、阈值与练习均为编辑合成，不是分析师预测。", "model": "资本强度=当期CapEx/收入；不能代替跨期ROIC或可交付产能。", "boundary": "可比币种、期间与资本化边界；利用存量设备可使当期比值下降。", "counterexample": "收入增长由涨价而非晶圆量推动，随后新厂大额投入才发生。", "test": "下一年度核对现金CapEx、在建工程转固、利用率与收入量价；若只是支出后移，撤销效率改善结论。", "question": "合成：收入100增至150，CapEx 30增至33，资本强度如何变化？", "constraint": "其中10的计划支出从本期推迟下期。若按原计划本期支出43，资本强度是多少？能确认永久效率提升吗？", "transfer": "迁移至云平台：销售额200升到240、服务器采购40升到60；资本强度变化并列出还需哪些容量数据。", "tags": [ "F9", "F11", "F14", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计预测结果或投资业绩" }, { "id": "O8SB-KK01", "profile": "karl-keirstead-ubs", "title": "AI贡献增长不等于AI贡献超预期", "source_ids": [ "O8SB-MS3" ], "locator": "HTML：KARL KEIRSTEAD, UBS 问答", "source_paraphrase": "Keirstead追问AI增量，Hood区分非AI超预期与供给提前。", "attribution": "原始提问归于具名分析师，回答归于管理层。下列模型、数字、阈值与练习均为编辑合成，不是分析师预测。", "model": "总超预期=实际AI−预期AI+实际非AI−预期非AI。", "boundary": "基期、币种与分类一致；增长贡献和超预期贡献不同。", "counterexample": "AI增长很快但符合预期，超预期来自非AI；提前交付又可能挪用下季收入。", "test": "下次财报核对分部收入与交付时间；若提前交付抵消后季收入，不外推新增需求。", "question": "合成：预期AI收入20、非AI80；实际AI22、非AI88。总超预期多少？AI占多少？", "constraint": "AI超出的2全部是从下季提前确认；本季非AI超出的8不会在下季被冲销，下季除AI提前确认造成少2外，其余均符合原预测。两个季度合计相对原预测新增收入多少？", "transfer": "迁移至设备商：预期新品30、旧品70；实际新品30、旧品80。新品高速增长能解释超预期吗？", "tags": [ "F1", "F2", "F3", "F4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计预测结果或投资业绩" }, { "id": "O8SB-BT01", "profile": "brent-thill-jefferies", "title": "成本下降能保证供应商收入增长吗？", "source_ids": [ "S51SRC1" ], "locator": "HTML：BRENT THILL, Jefferies 问答", "source_paraphrase": "Thill询问低成本AI规模化。", "attribution": "原始提问归于具名分析师，回答归于管理层。下列模型、数字、阈值与练习均为编辑合成，不是分析师预测。", "model": "收入=付费用量×实现价格；总算力=任务量×每任务计算量。", "boundary": "质量、期限、折扣与免费量相同；价格下降不必然创造足够弹性。", "counterexample": "使用量增加不足以抵消降价；单位成本下降只扩大消费者剩余。", "test": "下一期核对付费用量、每任务算力、实现价格与毛利；若降价未被付费量抵消，不提高收入估计。", "question": "合成：token价格降50%，付费量增80%；收入指数由100变多少？", "constraint": "每token算力消耗也降40%。总算力指数原100，新指数多少？能从收入减少推出算力减少吗？", "transfer": "迁移至数据压缩：每GB收费降20%、付费GB增30%；收入指数及打平所需用量增长？", "tags": [ "F10", "F18", "F7", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计预测结果或投资业绩" }, { "id": "O8SB-KR01", "profile": "kash-rangan-historical", "title": "CapEx增速下降，现金压力会下降吗？", "source_ids": [ "S51SRC2" ], "locator": "HTML：KASH RANGAN, Goldman Sachs 问答", "source_paraphrase": "Rangan追问CapEx与Azure增长路径。", "attribution": "原始提问归于具名分析师，回答归于管理层。下列模型、数字、阈值与练习均为编辑合成，不是分析师预测。", "model": "自由现金流=经营现金流−现金资本支出；增长率放缓不等于绝对额减少。", "boundary": "现金支出与包含融资租赁的资本新增不可混用；此处全部现金口径。", "counterexample": "新增资本仍高于现金增长，即使支出增速放缓，FCF仍下降。", "test": "下次财报核对现金PP&E、租赁及经营现金流，若合同不转现金就不能以积压替代融资。", "question": "合成：去年现金CapEx由40升到60，今年升到72；经营现金流去年90、今年100。CapEx增速及FCF如何变？", "constraint": "今年另需偿还租赁本金5；用扣除此本金后的可支配现金衡量，结果是多少？", "transfer": "迁移至晶圆厂：现金流80升至90，现金CapEx50升至65。FCF变化，能否因收入增长而断言资金更宽裕？", "tags": [ "F9", "F11", "F20", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计预测结果或投资业绩" }, { "id": "O8SB-BN01", "profile": "brian-nowak-ms", "title": "注册规模能代替重复使用吗？", "source_ids": [ "O8SB-META1" ], "locator": "印刷页9–11，Brian Nowak", "source_paraphrase": "Nowak追问重复使用行为。", "attribution": "原始提问归于具名分析师，回答归于管理层。下列模型、数字、阈值与练习均为编辑合成，不是分析师预测。", "model": "留存活跃量=初始试用队列×同口径留存率；用户规模和有价值任务完成率分别衡量。", "boundary": "同一观察窗口与用户去重，不把跨应用曝光当主动使用。", "counterexample": "入口扩张带来大量试用但留存下降；总体MAU上升掩盖同队列退化。", "test": "下次产品披露按入口核对队列留存与完成率；若只有曝光增加，不升级产品市场匹配判断。", "question": "合成：旧队列100万，四周留存40%；新队列200万，四周留存25%。留存人数与留存率各怎样变化？", "constraint": "新队列留存者中只有50%完成目标任务，旧队列为80%。成功用户人数怎样变化？", "transfer": "迁移到开发工具：试用者10万增到15万，付费转化20%降到10%；付费人数变化？", "tags": [ "F5", "F10", "F14", "F17" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计预测结果或投资业绩" }, { "id": "O8SB-ES01", "profile": "eric-sheridan-gs", "title": "同样CapEx为何带来不同利润压力？", "source_ids": [ "O8SB-META2" ], "locator": "印刷页9–12，Eric Sheridan；Susan Li回应", "source_paraphrase": "Sheridan追问人才和算力的费用传导。", "attribution": "原始提问归于具名分析师，回答归于管理层。下列模型、数字、阈值与练习均为编辑合成，不是分析师预测。", "model": "折旧=投用资产成本/寿命×当期投用时间；员工费用按实际任职时间确认。", "boundary": "简化直线折旧，无残值；资本新增不等于当期折旧，现金与利润分开。", "counterexample": "年末投用延后本年费用，但次年全年费用上升；混合更短寿命资产会加速压力。", "test": "下一期核对投用日期、使用寿命、资产类型及人员全年成本；若投用延后不误判单位效率提升。", "question": "合成：设备120，寿命4年，7月1日投用，无残值；全年人员成本12亦从7月1日开始。首年折旧加人员费用多少？", "constraint": "次年全年运营、人员数量不变；合计费用多少？若设备寿命改为3年，仅作前瞻敏感性，稳定全年费用多少？", "transfer": "迁移至物流：分拣线60、寿命5年、10月1日投用，首年折旧及下一完整年度折旧是多少？", "tags": [ "F9", "F11", "F18", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计预测结果或投资业绩" }, { "id": "O8SB-DA01", "profile": "doug-anmuth-jpm", "title": "合作出资减少了成本，还是改变了付款时间？", "source_ids": [ "O8SB-META2" ], "locator": "印刷页13–14，Douglas Anmuth", "source_paraphrase": "Anmuth追问是否引入融资伙伴。", "attribution": "原始提问归于具名分析师，回答归于管理层。下列模型、数字、阈值与练习均为编辑合成，不是分析师预测。", "model": "比较相同服务与风险的全生命周期现值；前期支出减少并不等于承诺义务消失。", "boundary": "合成固定支付、无税、残值和运维差异；真实合同含担保、最低采购及退出条款。", "counterexample": "伙伴承担前期建造，但用户签署高额长期保底，风险并未全部转移。", "test": "交易披露后核对最低付款、担保、终止权及容量控制；未签协议时不计确定融资收益。", "question": "合成：自建立即支付100；合作立即付40并在未来3年末各付25，折现率10%。哪个现值更低？", "constraint": "合作每年另有5的保底费用；重算现值。前期少付60还能称总成本更低吗？", "transfer": "迁移到设备租用：购买60；每年末租金24，共3年，折现10%，服务相同无残值。选择现值更低方案并说明合同风险。", "tags": [ "F9", "F11", "F20", "F21" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计预测结果或投资业绩" }, { "id": "O8SB-JP01", "profile": "justin-post-bofa", "title": "不出售云服务，算力投资就没有回报吗？", "source_ids": [ "O8SB-META2" ], "locator": "印刷页14–16，Justin Post", "source_paraphrase": "Post询问内部与外售算力的回报。", "attribution": "原始提问归于具名分析师，回答归于管理层。下列模型、数字、阈值与练习均为编辑合成，不是分析师预测。", "model": "内部增量现金贡献=可归因增量收益（未扣下述运行成本）−运行现金成本；与外售净贡献及资本机会成本比较。", "boundary": "同一容量、期间与现金边界；需因果实验而非把全部广告增长归给AI。", "counterexample": "内部广告毛利增长来自宏观复苏；外售又可能需要额外网络和服务投入。", "test": "下一轮实验核对对照组、增量收益、运行成本与外售改造成本；收益须按同一现金边界计量，若因果增益不足就下调内部回报。", "question": "合成：内部使用带来年增量收益18（尚未扣除下述运行成本），运行现金成本6；外售同批容量年收入20、全部运行现金成本11。两种年贡献多少？", "constraint": "实验发现内部年增量收益18中只有三分之二可归因于该投入，运行成本仍为6；重新比较。资本成本相同，可否直接由贡献算IRR？", "transfer": "迁移到工厂自动化：年节省人工12、运维4；对外出租年收入10、成本3。若人工节省只能实现50%，如何选择？", "tags": [ "F10", "F14", "F18", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计预测结果或投资业绩" }, { "id": "O8SC-S72-01", "profile": "meta-marshall", "title": "毛利率改善有多少可延续？", "source_ids": [ "O8SC-CIENQ125", "O8SC-CIENQ325" ], "locator": "Ciena Q1 2025 Financial Results pp.6–7；Ciena Q3 2025 Financial Results pp.11–12", "source_paraphrase": "检查云客户季节性与关税假设。 / 区分产品组合、规模降本与关税对毛利的贡献。", "attribution": "公开提问体现检验方向；以下模型、数字和迁移题为编辑合成，不是该分析师预测或原问答。", "model": "同口径毛利率变化=组合+单位成本+临时项目；分解是估计，不是天然可观测。", "boundary": "必须说明交互项分配方式。", "counterexample": "一次性退款与规模降本同时发生。", "test": "要求单位成本、组合和临时项目分桥，下一季复核。", "question": "合成，金额单位相同：收入100、毛利40；次季收入120、毛利52，其中包含供应商一次性成本返还4，已减少销售成本，未改变收入。次季报告与剔除该返还后的毛利率各多少？", "constraint": "进一步证实，按相同收入120的桥接口径，产品组合变化另贡献毛利2；再剔除后毛利率多少？可否断言剩余变化全由规模降本？", "transfer": "迁移到CPU基准：总分升12%，编译器贡献5个百分点，测试集变化贡献3个百分点。剩余4个百分点能否直接认定微架构贡献？", "tags": [ "F2", "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SC-S74-01", "profile": "simon-leopold", "title": "新增互连市场是否被重复计算？", "source_ids": [ "O8SC-CIENQ125", "O8SC-CIENQ325" ], "locator": "Ciena Q1 2025 Financial Results pp.9–10；Ciena Q3 2025 Financial Results pp.13–14", "source_paraphrase": "问制造地调整、关税敞口及供应链迁移。 / 区分新型训练互连与传统DCI，追问路由增长来源。", "attribution": "公开提问体现检验方向；以下模型、数字和迁移题为编辑合成，不是该分析师预测或原问答。", "model": "净新增收入=新用途收入−被替代旧用途收入。", "boundary": "项目、客户及时间窗去重；先确认同一供应商边界。", "counterexample": "新网络也替代旧传输预算。", "test": "逐项目设备清单、网络用途和旧预算取消额。", "question": "合成：某供应商原计划本年传统DCI收入100；新增训练互连订单60，导致原计划DCI采购取消20。假定收入在验收时确认，剩余原订单及新订单均本年验收。总收入多少？", "constraint": "若新订单仅有金额30在本年验收，但原20采购已取消，其余条件不变，本年收入多少？", "transfer": "迁移到CPU：新型号售出30万颗，其中18万替代原型号，原计划100万颗。总颗数增长多少？", "tags": [ "F6", "F9", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SC-S76-01", "profile": "amit-daryanani", "title": "一个分部加速，其他分部必须减速吗？", "source_ids": [ "O8SC-CIENQ125", "O8SC-CIENQ325" ], "locator": "Ciena Q1 2025 Financial Results pp.7–8；Ciena Q3 2025 Financial Results p.17", "source_paraphrase": "追查一次性毛利收益及订单强而短期收入温和的原因。 / 检验云加速与总体指引是否隐含其他业务减速。", "attribution": "公开提问体现检验方向；以下模型、数字和迁移题为编辑合成，不是该分析师预测或原问答。", "model": "总增长=基期权重×各分部增长的加权和。", "boundary": "权重用基期收入；币种、并购和定义一致。", "counterexample": "使用期末权重会失真。", "test": "取同口径基期收入与分部指引。", "question": "合成：云与其他收入各100，总收入目标增20%，云增35%。其他增速必须多少？", "constraint": "若云占基期40%，其他60%，目标总增长20%，云仍增35%，其他需多少？", "transfer": "迁移到内存：HBM占基期收入20%、增长100%，其余增长0。总增长多少？能推出总bit增长吗？", "tags": [ "F3", "F4", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SC-S91-01", "profile": "andy-kaplowitz", "title": "有机利润改善为何被并购稀释？", "source_ids": [ "O8SC-TTQ325", "EXP200-S92-B" ], "locator": "Trane Q3 2025 Earnings pp.10–12；Trane Q4 2025 Earnings pp.12–13", "source_paraphrase": "检查订单加速、收入节奏与渠道库存。 / 追问有机增量利润率、价格成本及未来制冷架构。", "attribution": "公开提问体现检验方向；以下模型、数字和迁移题为编辑合成，不是该分析师预测或原问答。", "model": "增量利润率=新增经营利润/新增收入；报告与有机分开。", "boundary": "固定可比范围，不混EBITDA与营业利润。", "counterexample": "低利润收购提高收入却拉低增量率。", "test": "取得收购收入/利润、有机桥及整合费用。", "question": "合成：原收入1000、利润200；有机收入增100、利润增30；收购新增收入100、利润10。报告增量利润率和有机增量利润率多少？", "constraint": "整合费8尚未计入。报告增量率变多少？", "transfer": "迁移到服务器：系统收入增加200，其中代采GPU收入180、毛利9，自有部件收入20、毛利10。新增毛利率多少？", "tags": [ "F2", "F4", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SC-S94-01", "profile": "scott-davis", "title": "行业增长面广等于收入增长分散吗？", "source_ids": [ "O8SC-TTQ325", "EXP200-S92-B" ], "locator": "Trane Q3 2025 Earnings pp.15–16；Trane Q4 2025 Earnings pp.10–11", "source_paraphrase": "问订单交期是否延至后年，并追问参考设计新增内容。 / 问订单增长是扩散到多个行业还是集中在数据中心。", "attribution": "公开提问体现检验方向；以下模型、数字和迁移题为编辑合成，不是该分析师预测或原问答。", "model": "增长贡献=基期收入×增长率；广度与权重分开。", "boundary": "行业数不是收入权重。", "counterexample": "许多小行业增长无法抵消单一大行业萎缩。", "test": "披露行业权重、订单取消与客户集中度。", "question": "合成：10个行业，数据中心收入700增40%，另9行业合计300增2%。总增长率及数据中心对新增收入贡献多少？", "constraint": "数据中心次期下降20%，其余合计增10%，从本题初始700/300基期算，总收入多少？", "transfer": "迁移到股票组合：9个仓位各1%并涨20%，一个91%仓位跌10%，组合回报多少？", "tags": [ "F6", "F14", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SC-S96-01", "profile": "jeff-sprague", "title": "成本冲击何时传到利润？", "source_ids": [ "O8SC-TTQ325", "EXP200-S92-B" ], "locator": "Trane Q3 2025 Earnings pp.19–21；Trane Q4 2025 Earnings pp.21–22", "source_paraphrase": "问延期是项目噪声还是施工系统瓶颈。 / 问关税跨年影响、金属成本和调价速度。", "attribution": "公开提问体现检验方向；以下模型、数字和迁移题为编辑合成，不是该分析师预测或原问答。", "model": "未覆盖采购额×涨幅−同期提价收入=短期利润损失。", "boundary": "固定采购量、库存计价和销售量；忽略税。", "counterexample": "套保到期让下一期成本反而更高。", "test": "采购合同到期表、涨价生效日和库存消耗。", "question": "合成：年采购100，60%全年锁价，40%现货涨25%；当年提价增加收入6。利润净影响多少？", "constraint": "次年锁价全部到期，采购量不变、全部成本相对原基期涨25%，提价收入增至15。相对原基期利润影响多少？", "transfer": "迁移到云实例：年度收入120、算力成本80，其中一半固定、一半现货涨20%，客户合同不变。利润由多少变多少？", "tags": [ "F4", "F10", "F19" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SC-S97-01", "profile": "joe-ritchie", "title": "项目数量不能代替可交付容量", "source_ids": [ "O8SC-TTQ325", "EXP200-S92-B" ], "locator": "Trane Q3 2025 Earnings pp.18–19；Trane Q4 2025 Earnings pp.20–21", "source_paraphrase": "问模块化项目变化与不同供应链交期。 / 问欧洲数据中心机会及区域配置。", "attribution": "公开提问体现检验方向；以下模型、数字和迁移题为编辑合成，不是该分析师预测或原问答。", "model": "可确认收入=项目数量×单项目容量×设备金额/MW×当期完成比例。", "boundary": "项目去重；合同及验收口径一致。", "counterexample": "多小项目不如少大项目；施工延迟改变确认。", "test": "容量、交付日、模块化成本及现场人员。", "question": "合成：A区10项目各100MW，B区40项目各10MW；设备收入每MW为2。两区总机会额各多少？", "constraint": "A当年完成20%，B完成80%，均验收；当年收入各多少？", "transfer": "迁移到CPU云：A有100客户各买1000核时，B有1000客户各买50核时，单价相同。哪组收入更高，几倍？", "tags": [ "F9", "F18", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SC-S93-01", "profile": "stephen-tusa", "title": "增长对账反推出的权重可靠吗？", "source_ids": [ "O8SC-TTQ325", "EXP200-S92-B" ], "locator": "Trane Q3 2025 Earnings pp.26–27；Trane Q4 2025 Earnings pp.24–25", "source_paraphrase": "追问积压同比基数与渠道进销差异。 / 要求子项与总体订单增长对账，并区分订单和收入。", "attribution": "公开提问体现检验方向；以下模型、数字和迁移题为编辑合成，不是该分析师预测或原问答。", "model": "G=w×gA+(1−w)×gB。", "boundary": "只有两个同口径互斥部分才能直接解权重。", "counterexample": "漏掉服务或用收入权重代替订单权重。", "test": "拿到基期订单分母及所有子项。", "question": "合成：订单总增长35%，设备A增长100%，其余增长5%；只有两个互斥部分。A的基期权重是多少？", "constraint": "现在发现还有服务占基期20%、增长10%，B增长5%，总增长仍35%。A权重是多少？", "transfer": "迁移到CPU测试：总分提高20%，整数子集提高40%，浮点0%，若总分为几何平均能否按线性加权解出整数占50%？", "tags": [ "F2", "F4", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SC-S99-01", "profile": "nicole-deblase", "title": "设备增长怎样转成服务收入？", "source_ids": [ "O8SC-NVTQ225", "O8SC-TTQ325" ], "locator": "nVent Q2 2025 Earnings Call pp.14–15；Trane Q3 2025 Earnings p.23", "source_paraphrase": "追问非数据中心订单与液冷服务能力。 / 检查欧洲订单驱动及住宅价格、组合、销量。", "attribution": "公开提问体现检验方向；以下模型、数字和迁移题为编辑合成，不是该分析师预测或原问答。", "model": "服务收入=存量设备×服务附着率×年费；新机与存量去重。", "boundary": "同一服务期，排除设备内含保修收入重复。", "counterexample": "设备增长但服务附着降低。", "test": "客户合同、附着率、续约率、服务人员容量。", "question": "合成：存量1000台，付费服务附着60%，年费2。存量增至1300台、附着降至45%、年费不变。收入怎样变化？", "constraint": "服务团队最多承接500台，需求不变。收入上限是多少？", "transfer": "迁移到软件：用户100万增至150万，付费率10%降至6%，年费不变。付费用户及收入同比怎样？", "tags": [ "F10", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O8SC-S100-01", "profile": "deane-dray", "title": "客户自研到底拿走多少供应商价值？", "source_ids": [ "O8SC-NVTQ225", "O8SC-TTQ325" ], "locator": "nVent Q2 2025 Earnings Call pp.7–8；Trane Q3 2025 Earnings pp.25–26", "source_paraphrase": "追问积压转化及客户自研液冷是否绕过供应商。 / 质疑地区项目规模概括，另问液冷合作。", "attribution": "公开提问体现检验方向；以下模型、数字和迁移题为编辑合成，不是该分析师预测或原问答。", "model": "供应商收入=项目数×每项目可售内容×份额。", "boundary": "自研设计、制造外包及服务边界分开。", "counterexample": "客户自研反而扩大外包制造量，但压缩单机内容。", "test": "逐项BOM、产权/采购合同、制造责任及毛利。", "question": "合成：100个项目，每项目供应商内容10、份额50%；客户自研后内容降至6，但份额升至70%，项目数不变。收入如何变化？", "constraint": "项目数增长30%，其余不变；收入相对原来多少？", "transfer": "迁移到ASIC：原100万颗每颗IP授权10，新量150万颗、授权降至6；授权收入怎样变化？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B1-B02-01", "profile": "nick-evans", "title": "AI加价能覆盖席位流失与推理成本吗？", "source_ids": [ "R2B1-EV24", "R2B1-EV26" ], "locator": "The Magnificent 7 and positioning for a market broadening：2026-10-08 读取访谈PDF第1页实质正文。；Global Technology video transcript：2026-10-08 读取PDF第1–2页访谈正文。", "source_paraphrase": "区分AI基础设施、受益者与采用者；看企业之间增长差异而非统一大盘标签。 / 以AI重新划分利润池，关注软件替代风险、基础设施需求与资本开支对自由现金流的影响。", "attribution": "来源为具名公开论证或显式团队共著；以下数字、模型和三阶段练习完全为编辑合成，不是本人预测或原问答。", "model": "收入=席位×价格+AI收入；现金贡献=收入−对应现金成本。", "boundary": "本题现金成本口径，不能当公司会计毛利预测。", "counterexample": "AI收入增长仍可能被推理费用和席位收缩抵消。", "test": "匹配客户队列的席位、AI付费、留存和单位推理成本。", "question": "合成：原100席位×100，成本4000。新80席位×110，另有AI收入2000；原业务成本降到3200，新增AI成本1500。收入与现金贡献（收入减列示现金成本）如何变化？", "constraint": "新席位价格降到90，其他不变；现金贡献多少？", "transfer": "存储公司销售量增50%、单位价格降30%，成本资料未知；收入与利润可作何结论？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B1-B06-01", "profile": "reid-menge", "title": "更早获得算力值多少现金？", "source_ids": [ "R2B1-ME26Q2", "R2B1-ME26SP" ], "locator": "BlackRock Technology Opportunities Fund commentary：2026-10-08 读取PDF第1–2页；采用文件内Q2 2026日期而非搜索索引日期。；The Bid: Space — The Final Investing Frontier：2026-10-08 读取第272集官方完整文字稿中Menge发言；栏目滚动页面，标题用于定位。", "source_paraphrase": "团队关注推理效率、代理工作负载和可度量资本回报，偏好已显现商业化及持久优势。 / 把成本下降、能力提升和需求结合，同时强调发射执行、太空环境和地面通电时间；太空算力判断属预测。", "attribution": "来源为具名公开论证或显式团队共著；以下数字、模型和三阶段练习完全为编辑合成，不是本人预测或原问答。", "model": "NPV=折现后净现金流−初始成本。", "boundary": "纯合成工程项目，非太空计算实际成本或可行性证明。", "counterexample": "更早投产的收入可能不足补偿额外资本投入。", "test": "核实可用时间、达标率、网络成本和可比工作负载。", "question": "地面方案今天成本100，第2、3年末各获净现金流80（已扣运维、税费及后续资本开支）；另一方案今天成本130，第1、2年末各获同口径净现金流80。折现率10%，无残值，求两方案NPV。", "constraint": "第二方案只有80%概率获得全部所述收入，失败无收入且成本不返还；期望NPV多少？", "transfer": "云迁移今天花20，未来两年末各省12，折现率10%；仅从这些现金流看是否创造价值？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B1-B17-01", "profile": "hyun-ho-sohn", "title": "工作流黏性能否转成续费收入？", "source_ids": [ "R2B1-SO26", "R2B1-SO26Q" ], "locator": "Fidelity International Outlook 2026：2026-10-08 读取印刷第13页/PDF索引12；准备年份2025，确切发布日期未核。；Global technology: Overdone risks and diverse opportunities：2026-10-08 读取全文具名问答；非RSMR独立研究，未比对原始录音。", "source_paraphrase": "担心资本密集度、外部融资与半导体估值透支周期，要求区分基础设施增长和投资回报。 / 指出企业软件工作流、数据、安全与集成形成阻力；用管理层、专家和内部IT调研检验AI替代速度。", "attribution": "来源为具名公开论证或显式团队共著；以下数字、模型和三阶段练习完全为编辑合成，不是本人预测或原问答。", "model": "续费收入=到期客户×留存率×基础价格+附加产品收入。", "boundary": "留存、附加付费及成本必须分别测量；不能用功能演示替代。", "counterexample": "AI功能很强但企业审批延长，客户也可能压缩席位。", "test": "匹配续约队列、集成工作、安全审查和实际付费附加率。", "question": "原100个合同，每合同100；次年95%续约，续约者20%另买20的AI产品。总收入多少，相对原收入如何？", "constraint": "留存变成85%，其余不变；总收入多少？", "transfer": "工业服务有200个到期客户、续约90%，其中30%买50的升级包，基础价200；总收入多少？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B1-B19-01", "profile": "ken-allen", "title": "再投资降低当期现金，是否反而创造价值？", "source_ids": [ "R2B1-KE18", "R2B1-KE17" ], "locator": "Technology’s Golden Era：2026-10-08 读取PDF第1–3页Ken Allen问答。；Artificial Intelligence: A New Era of Opportunity：2026-10-08 读取第2–3页Allen具名段落；文件名NOV但采用正文December 2017。", "source_paraphrase": "重视持久业务、合理价格和中长期现金；看云规模优势同时保留本地部署需求。 / Allen强调云分发以及数据、算力和人才组合；其他经理的判断不归给Allen。", "attribution": "来源为具名公开论证或显式团队共著；以下数字、模型和三阶段练习完全为编辑合成，不是本人预测或原问答。", "model": "新增投资价值=新增现金流现值−投入。", "boundary": "只比较明确增量现金；永续假设极强且不包含竞争变化。", "counterexample": "增长投入若回报低于资本成本会侵蚀价值。", "test": "新增资本、归属利润、营运资金及维持性资本开支。", "question": "合成公司当前经营现金20，原无投资。今天新增投资50，从下一年起永续每年增加现金5，折现率8%。新增投资NPV多少？当期净现金多少？", "constraint": "新增现金只有3且其他不变；NPV多少？", "transfer": "电力设施今天投入80，下一年起永续现金8、折现率10%，无残值之外的其他现金；是否创造超额价值？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B1-B21-01", "profile": "alan-tu", "title": "低摩擦获客是否改善客户经济性？", "source_ids": [ "R2B1-TU22", "R2B1-TU21" ], "locator": "Investing in Companies at the Start of the Runway：2026-10-08 读取PDF第1–4页。；Investing in the Infrastructure Behind the Digital Economy（日文官方译本）：2026-10-08 读取日文官方PDF第1–4页，保留共同署名；不把团队观点拆为Tu独著。", "source_paraphrase": "从产品进入市场方式、客户扩张和竞争优势寻找长增长跑道，同时要求估值纪律。 / 团队从软件、云、半导体设备与安全切入数字化基础设施，讨论竞争壁垒、增长基数及估值。", "attribution": "来源为具名公开论证或显式团队共著；以下数字、模型和三阶段练习完全为编辑合成，不是本人预测或原问答。", "model": "简化客户价值=未来客户毛贡献现值−获客成本。", "boundary": "本题只有两年且现金已扣交付成本，不可外推终身LTV。", "counterexample": "便宜获客若引入低留存客户，会低估替换成本。", "test": "同队列留存、客户扩张、支持成本与渠道费用。", "question": "产品自助获客成本40，第一年末毛现金30；80%客户第二年仍付毛现金30，其余无收入。折现率10%，两年后无价值；单客户NPV多少？", "constraint": "获客成本降到25，但第二年留存仅30%；新NPV更好吗？", "transfer": "设备商获客成本50，第一年末贡献40，第二年留存70%、贡献40，折现率10%；两年客户NPV？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B1-B22-01", "profile": "kyle-weaver", "title": "客服省下的时间何时变成现金？", "source_ids": [ "R2B1-WE26", "R2B1-WE26Q2" ], "locator": "Investing in companies harnessing AI to transform their business：2026-10-08 读取署名文章正文。；Fidelity Advisor Growth Opportunities Fund quarterly commentary：2026-10-08 读取PDF第2页投资理念、第5页展望；确认Q2 2026。", "source_paraphrase": "把AI机会扩展到采用者，用客服和数字开户案例连接运营变化；案例陈述不等于已证明因果收益。 / 团队采用3–7年视角，结合现金增长、资本配置及不同增长类型，追踪真实AI使用与供给约束。", "attribution": "来源为具名公开论证或显式团队共著；以下数字、模型和三阶段练习完全为编辑合成，不是本人预测或原问答。", "model": "可兑现节省=节省工时×工资×兑现比例−新增AI成本。", "boundary": "工时并非自动可裁减或可避免的现金支出。", "counterexample": "业务量不变且人员固定时，效率改善可能仅提高服务质量。", "test": "验收后的自动完成率、转人工比例、人员预算和客户满意度。", "question": "合成：每年50000次客服、每次10分钟、工资30/小时。AI覆盖50%请求，其中20%仍需完整人工处理；AI年费40000。全部节省工时可减少现金工资时，净省多少？", "constraint": "只有50%的节省工时能变成现金，其余重新部署；净省多少？", "transfer": "工厂每年节省2000工时、工资40/小时，人员固定且没有避免招聘，自动化运维费10000；仅按给定现金支出，现金变化多少？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B1-B23-01", "profile": "sonu-kalra", "title": "未上市资产占比为什么会被动上升？", "source_ids": [ "R2B1-KA26", "R2B1-KA24" ], "locator": "Fidelity Blue Chip Growth Fund: Portfolio Manager Q&A：2026-10-08 读取PDF第1–4页；滚动PDF实际日期July 31, 2026。；Exploring the AI investment opportunity across public and private markets：2026-10-08 读取公开PDF实质正文；不采用图表中无效4/31/2024脚注作为日期。", "source_paraphrase": "以竞争优势、定价能力和管理层判断增长持久性，关注AI半导体机会并保留宏观及估值风险。 / 团队跨公私市场观察AI价值链，区分流动性、采用速度及长期生产率兑现的不确定性。", "attribution": "来源为具名公开论证或显式团队共著；以下数字、模型和三阶段练习完全为编辑合成，不是本人预测或原问答。", "model": "流动性与账面净值分开；赎回会改变剩余资产构成。", "boundary": "账面未变不等于可按该价格及时出售。", "counterexample": "表面资产分散，在公开市场跌价和赎回时转成私募集中。", "test": "估值时点、赎回条款、现金结算和真实成交折价。", "question": "合成：基金上市资产80、非上市20，均无债。上市资产跌20%，非上市账面不变，再从上市资产支付赎回30。剩余净值与非上市占比多少？", "constraint": "随后将非上市账面10以折价20%售出，现金留在基金；净值和非上市占比多少？", "transfer": "不发生任何赎回或交易，只把原80上市和20非上市组合中的非上市重估下调50%；净值与非上市占比多少？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B1-B29-01", "profile": "thomas-laffont", "title": "公开代码活动能证明收入增长吗？", "source_ids": [ "R2B1-LA26D", "R2B1-LA26A" ], "locator": "Our Partnership with Databricks：2026-10-08 读取官网完整实质正文及利益披露。；Our Partnership with Anthropic: Enterprise Intelligence has Liftoff：2026-10-08 读取官网完整实质正文及利益披露；未独立验证营收与工程效率宣传数字。", "source_paraphrase": "团队以分析、交易和代理应用扩张解释增长，强调数据治理与研发周期；Coatue持有经济利益。 / 团队用代码活动和工作流扩展支持企业采用论点；从开发能力延伸到平台价值，存在被投企业利益冲突。", "attribution": "来源为具名公开论证或显式团队共著；以下数字、模型和三阶段练习完全为编辑合成，不是本人预测或原问答。", "model": "使用量代理信号须除以样本覆盖并校验商业转换。", "boundary": "两文是Coatue投资论证，有被投企业利益；不认证Thomas独立方法。", "counterexample": "仓库增长来自采样扩张、免费使用或自动生成，并不带来付费。", "test": "同仓库队列、真实活跃开发者、付费转换和净收入留存。", "question": "合成：平台可见AI提交从1000升到2400，但采样仓库从100升到200。每仓库提交怎样变，能否据此称收入增140%？", "constraint": "若同口径付费比例从50%降到30%，每次付费提交单价不变；用全部提交×付费比例作收入代理，变化多少？", "transfer": "数据平台客户数从100到150、每客户收入从20到16；收入怎样变化？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B1-B33-01", "profile": "lee-ainslie", "title": "渠道总量增长是否只是样本变大？", "source_ids": [ "R2B1-AI06", "R2B1-AI21" ], "locator": "Inside a hedge fund: An interview with the managing partner of Maverick Capital：2026-10-08 读取原发布者完整访谈正文。；A Conversation with Lee Ainslie, Founder and Managing Partner of Maverick Capital：2026-10-08 读取原采访全文。", "source_paraphrase": "以管理层、持续增长和资本回报筛选，以可持续自由现金流/企业价值衡量价格，持仓按未来回报重审。 / 说明基本面与因素风险互补；用统计化替代数据替代过度加权的零散观察，并保留资本保护目标。", "attribution": "来源为具名公开论证或显式团队共著；以下数字、模型和三阶段练习完全为编辑合成，不是本人预测或原问答。", "model": "总量=覆盖单位数×单位产出；投资收益另需估值与风险。", "boundary": "渠道样本不代表全部公司，收入信号不能直接当超额回报。", "counterexample": "网站订单号或新增门店样本会制造虚假加速。", "test": "稳定队列、去重、地域与店龄结构，以及价格/销量拆分。", "question": "合成：渠道样本销售100增至120，覆盖门店1000增至1500；平均店销售增速是多少？", "constraint": "第二期1500家中新增500家贡献销售40；原1000家贡献80。原店同店增速多少？", "transfer": "服务器采购监测量从200到300，但覆盖客户40到75；每客户采购量变多少，能否直接推出行业增长50%？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B1-B24-01", "profile": "will-danoff", "title": "单店扩张回报能否覆盖总部负担？", "source_ids": [ "R2B1-DA20", "R2B1-DA26I" ], "locator": "Will Danoff: Succeeding at Scale — The Long View：2026-10-08 读取Morningstar官方Simplecast托管逐字稿中管理层复盘、竞争优势、每股现金增长及持仓重估段落。；Will Danoff: Be Very Careful of Unprofitable Companies — The Long View：2026-10-08 读取Morningstar节目官方Simplecast托管完整逐字稿，尤其unit economics、管理层核验、竞争优势及亏损公司段落；Morningstar主域对应页403，合法公开托管可读。", "source_paraphrase": "强调隔半年至一年追踪管理层承诺；偏好高资本回报、现金增长和客户价值，并用新证据逐步建立信心。 / 从门店单位经济性检验扩张；将管理层许诺与后续执行对照，并要求竞争压力下的盈利记录而非短期收入火箭。", "attribution": "来源为具名公开论证或显式团队共著；以下数字、模型和三阶段练习完全为编辑合成，不是本人预测或原问答。", "model": "增量回报=增量现金收益/投入，另扣总部与维持资本。", "boundary": "单店利润不是股东自由现金流；扩张并不确保估值合理。", "counterexample": "新店增长遮蔽成熟店疲软，总部和维护成本吞噬店面回报。", "test": "核对开店成本、成熟期、同店销售、总部费用与维护资本。", "question": "合成：今天新开10店，每店投入100，从下一年起每店每年产生30店面现金。新增总部成本每年50、维护支出每年40。稳定年净现金及投入回报是多少？", "constraint": "竞争使每店现金降至20，总部和维护不变；若所需回报12%，高于或低于门槛多少个百分点？", "transfer": "数据中心投资500，稳定年客户现金贡献100，网络支持20、维护25；稳定净现金回报率多少，是否超过12%门槛？", "tags": [ "F6", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B2-B43-01", "profile": "ryan-jacob", "title": "平台黏性如何转成股东回报？", "source_ids": [ "R2B2-J1", "R2B2-J2" ], "locator": "两信Internet Fund段，2025-10第13–17段；2026-01第13–15段", "source_paraphrase": "Internet Fund部分讨论Circle支付基础设施与Bed Bath & Beyond的tZERO资产及重组价值，不能把整个合署信归于Ryan一人。；Internet Fund新增Snowflake的逻辑是跨云数据层、客户黏性与AI能力；竞争与估值仍需检验。", "attribution": "编辑据公开材料提炼；数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "收入=客户数×单客收入；毛利=收入×毛利率", "boundary": "合署信只读取Ryan所管Internet Fund段，Discovery Fund观点不单独归于Ryan。", "counterexample": "强黏性伴随降价与高推理成本，收入增而毛利降。", "test": "未来两个财季的续费、用量、价格和单位服务成本。", "question": "合成：数据平台客户1000家，每家年收入10万元、毛利率70%；客户增长20%、单客收入下降10%，毛利额多少？", "constraint": "AI服务使毛利率降至55%；毛利额相对原值如何？", "transfer": "迁移支付平台：交易额增长30%、费率降20%，收入变化？", "tags": [ "F10", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "R2B2-B44-01", "profile": "kevin-landis", "title": "IPO与账面价值怎样变成可用现金？", "source_ids": [ "R2B2-L1", "R2B2-L2" ], "locator": "2023 PDF pp.4–6；2015 Letter: Road Ahead与IPO lock-up段", "source_paraphrase": "信件把未配置大市值股的相对拖累与Chegg被生成式AI替代的新客风险分开。；区分私企经营进度、IPO锁定、退出现金与基金NAV折价；成熟需要多年，上市不等于可兑现。", "attribution": "编辑据公开材料提炼；数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "可用现金=现金+可实现出售款−到期支出", "boundary": "历史信件和不同基金产品不能拼成当前组合；本卡侧重可迁移流动性约束。", "counterexample": "IPO后锁定延长或市价下降，退出故事不能兑现。", "test": "核对锁定条款、解禁日、可卖量和未来三个月付款表。", "question": "合成：基金NAV100，其中20为已IPO但锁定资产，现金5，其余75在付款期限内均不可变现；不可借款、无其他现金流。到期需支付12，锁定期内能否偿付？", "constraint": "付款前的独立替代情景：该20资产已解锁但价格跌25%，卖出其中40%，其余条件不变。付款前现金合计多少、支付12仍缺多少？", "transfer": "迁移AI初创：已签合同30，预付款10%，未来一季支出8，无其他现金。缺口多少？", "tags": [ "F11", "F20", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "R2B2-B59-01", "profile": "peter-seilern", "title": "高ROIC与可持续再投资", "source_ids": [ "R2B2-S1", "R2B2-S2" ], "locator": "交接信Focus/Discipline；机构Ten golden rules与Research", "source_paraphrase": "退休交接信强调清晰规则、可复制流程、团队独立性与长期激励；不承诺每个周期跑赢。；十项质量标准包含低资本强度、高投入资本回报和低净负债；机构以研究与集中组合实施。", "attribution": "编辑据公开材料提炼；数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "利润增量=再投资额×增量ROIC；经济利润=NOPAT−资本×资本成本", "boundary": "一份个人交接信加无日期机构流程页；不视为两份个人独著或跨期方法认证，按退休前历史身份。", "counterexample": "存量护城河掩盖新增项目低回报。", "test": "未来年度跟踪增量投入、投产期、分部NOPAT与资本成本。", "question": "合成：企业NOPAT20，投入资本100，把利润的50%再投，增量ROIC20%。静态模型下一年利润多少？", "constraint": "增量ROIC降至8%，原有资本回报不变；新利润多少？若资本成本10%，新增资本创造还是损害价值？", "transfer": "迁移数据中心：新增投资80、稳定NOPAT6，门槛10%；需增加多少NOPAT才刚达门槛？", "tags": [ "F10", "F20", "F21" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "R2B2-B47-01", "profile": "walter-price", "title": "采用增长与估值扩张分拆", "source_ids": [ "R2B2-A16", "R2B2-A20" ], "locator": "2016 printed pp.12–14；2020 printed pp.38–44；负责人见PDF p.20", "source_paraphrase": "管理人报告将新技术高增长与成熟业务整合、回购分开，并以产品周期和子行业分散风险。；管理人报告分拆疫情下远程采用、付费转换与低利率带来的估值扩张；绝对上涨与相对收益不同。", "attribution": "编辑据公开材料提炼；数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "P=EPS×PE，总回报须另加股息", "boundary": "两份为共同负责团队的机构报告；不能分配为Walter独立观点，不验证现职。", "counterexample": "低利率重估消退时盈利增长也可能无法抵消估值压缩。", "test": "下一财季分拆盈利修订、无风险利率与前瞻估值。", "question": "合成：EPS2、PE25；次年EPS增30%、PE升20%，股价总涨幅多少？", "constraint": "从初始EPS2、PE25出发的替代情景：EPS只增长10%、PE下降25%，相对初始价回报多少？", "transfer": "迁移成熟工业：EPS10、PE15，EPS增5%、PE不变，并收到每股股息2，初价150，总回报多少？", "tags": [ "F12", "F13", "F19" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "R2B2-B48-01", "profile": "huachen-chen", "title": "科技生命周期分散是否真能降低风险？", "source_ids": [ "R2B2-A16", "R2B2-A20" ], "locator": "2016 printed pp.13–14；2020 printed pp.43–44；团队身份PDF p.20/23", "source_paraphrase": "管理人报告将新技术高增长与成熟业务整合、回购分开，并以产品周期和子行业分散风险。；管理人报告分拆疫情下远程采用、付费转换与低利率带来的估值扩张；绝对上涨与相对收益不同。", "attribution": "编辑据公开材料提炼；数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "组合情景损失=Σ权重×情景跌幅", "boundary": "机构团队资料，Huachen为共同负责人，不将共同文本包装成其个人独有方法；历史样本。", "counterexample": "不同技术子行业共享同一客户资本开支预算。", "test": "穿透客户收入与预算来源，在下一季度复核共同因子。", "question": "合成：成长技术60%、成熟技术40%；需求冲击下分别跌40%、跌10%，组合损失多少？", "constraint": "发现二者依赖同一云客户，成熟技术也跌30%；损失如何变化？", "transfer": "迁移AI供应链：GPU、HBM、网络各三分之一，同一客户缩CapEx使各跌20%，分三行业能否消除共同风险？", "tags": [ "F6", "F14", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "R2B2-B42-01", "profile": "anthony-ginsberg", "title": "等权主题背后的共同客户风险", "source_ids": [ "R2B2-G1", "R2B2-G2" ], "locator": "2025访谈equal weighting；2026 outlook“10 leading tech themes”与pre-leased段", "source_paraphrase": "访谈以八主题各15股等权回应巨头集中；同时承认技术主题交叉与轮动难预测。；机构展望以现金融资和盈利支持AI投资，并称十主题等权；与2025八主题不同，定义变更需核对。", "attribution": "编辑据公开材料提炼；数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "穿透暴露=Σ主题权重×共同客户比重", "boundary": "一份个人访谈加一份机构无署名展望；八到十主题变化未核定生效时间，不自动称当前规则。", "counterexample": "等权降低单股权重，却未去掉共享客户和估值因子。", "test": "核对指数方法变更日期、当前成份收入暴露和租约通电里程碑。", "question": "合成：四主题各25%，各有50%收入来自同一云客户；组合穿透客户暴露多少？", "constraint": "该客户预算砍20%，假设完全线性传导、其他客户不变，收入代理减少多少？", "transfer": "迁移数据中心：容量100MW、90%预租、其中70%按期通电；已通电预租容量多少？是否等于实际使用？", "tags": [ "F9", "F17", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "R2B2-B58-01", "profile": "mark-denham", "title": "防御性标签不能代替需求与利润模型", "source_ids": [ "R2B2-M1", "R2B2-M2" ], "locator": "2020访谈robust approach；2025 PDF pp.1–2", "source_paraphrase": "以资本回报、再投资和经营利润率稳定性作筛选，再做基本面与ESG研究，持有期3–5年。；将日常消费标签拆为食品与个人护理；从需求变化、有机增长与经营杠杆推导不同盈利及估值。", "attribution": "编辑据公开材料提炼；数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "利润=原利润+收入变化×增量利润率", "boundary": "跨年具名方法样本，不外推实际未来收益；个人护理不是天然无周期，原文需求预测尚待验证。", "counterexample": "技术/健康行为改变打破历史消费稳定性。", "test": "下一年对比同客群销量、单价、产品组合及固定成本吸收。", "question": "合成：消费品收入100、经营利润20；销量增5%、价格不变、固定成本不变且增量利润率40%，利润多少？", "constraint": "新疗法使销量变为下降5%，其他不变，利润多少？", "transfer": "迁移AI软件：原收入100，固定成本30、变动成本收入40%；收入降10%，经营利润变化？", "tags": [ "F5", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "R2B2-B51-01", "profile": "wood-ark", "title": "成本下降能否兑现为股东收益？", "source_ids": [ "D01", "D02" ], "locator": "D01 Sizing The Opportunity / Stock Selection；D02标题Is AI A Bubble? The Data Say No首段：2026 hyperscaler capex >$800 billion，与2021 $135 billion同句并列，却写nearly 3x。", "source_paraphrase": "机构成本曲线与需求弹性框架；五年模型纳入销量、成本、稀释和估值。；2026摘要把AI成本下降与token增长联系；800/135与“近3倍”表述不一致，应重算而非沿用。", "attribution": "编辑据公开材料提炼；数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "收入=单位价格×付费任务；EPS=收入×净利率/稀释股数", "boundary": "两份机构材料；无日期流程页不构成两次具名个人研究，保留既有wood-ark。", "counterexample": "成本降得快但付费需求与公司价值捕获未同步。", "test": "核对可比付费用量、价格、毛利与稀释；先消除摘要单位/倍数不一致。", "question": "合成：单任务价格由1降到0.2，付费任务量达到原来的3倍，收入变化多少？", "constraint": "任务量达到原来的6倍但稀释后股数增20%，净利率不变；每股盈利相对原值多少？", "transfer": "按原文同句给出的美元十亿单位，合成算术校验：从135到800是几倍、增长多少？若均为可比hyperscaler CapEx，能否写近3倍？口径不同应怎样处理？", "tags": [ "F10", "F12", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "R2B2-B41-01", "profile": "mark-hawtin", "title": "正确的AI趋势为何仍可能买贵？", "source_ids": [ "R2B2-H1", "R2B2-H2" ], "locator": "Liontrust 2025 outlook Mark Hawtin段；DeepSeek文章“What next – where to invest?”与前文CapEx段", "source_paraphrase": "Hawtin关注指数集中和AI受益向使用者扩散；其他团队发言没有归给他。；文章提出AI普及与基础设施股估值可分离，并说明此前削减CapEx相关敞口；成本与技术比较为作者当时主张，未独立核实。", "attribution": "编辑据公开材料提炼；数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "供应商毛利=收入×毛利率；用户净收益=节省−新增成本", "boundary": "两份具名样本；技术成本原文跨模型口径未独立核实，不采其点估计；源时期角色。", "counterexample": "效率普及使总需求上升，却可能削弱供应商定价与利润。", "test": "下一财季跟踪同口径价格/成本/采用量与用户付费ROI。", "question": "合成：AI基础设施供应商收入100、毛利率70%；降价后收入120、毛利率55%，毛利额变化多少？", "constraint": "用户公司因AI每年节省10，新增年化AI费用6、其他成本不变；年经营利润增多少？", "transfer": "迁移云服务：任务量达到原来4倍，单位价格变原来20%，收入变化？", "tags": [ "F9", "F10", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "R2B2-B60-01", "profile": "david-rolfe", "title": "利润增长来自核心效率还是高毛利副业？", "source_ids": [ "R2B2-W2", "R2B2-W4" ], "locator": "Q2 PDF pp.5–8、署名p.16；Q4 PDF pp.6–8、署名p.24", "source_paraphrase": "Zoetis案例从宠物医疗渗透、产品延伸研发和资本回报构造投资逻辑，同时区分疫情提升与正常化增长。；Amazon新仓论证分拆广告增长与零售生产率对利润贡献，再讨论AWS硬件、资本开支与估值；分部利润不能直接当核心零售效率。", "attribution": "编辑据公开材料提炼；数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "利润增量桥=分部总增量−可识别业务贡献；残差不等于因果解释", "boundary": "两份均为三人合署客户信，不称Rolfe独著；两份均以公开原文PDF镜像读取，未核验官方字节。", "counterexample": "高毛利广告或补贴贡献掩盖基础业务回报不足。", "test": "下一财年核对业务利润分摊、固定成本吸收与不含补贴的投入资本回报。", "question": "合成：零售分部经营利润增加30，其中广告新增收入50、假设增量利润率40%；扣除广告后的利润增量多少？", "constraint": "若广告增量利润率实际为50%，剩余利润增量多少？能否直接称其全部来自核心零售效率？", "transfer": "迁移GPU云：经营利润增加6，包含本期新增能源补贴4，其他一次性项目为零；扣除补贴的利润增量多少？", "tags": [ "F2", "F11", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "R2B3-B66-C1", "profile": "r2b3-b66", "title": "价格下跌是否扩大安全边际？", "source_ids": [ "R2B3-B66C", "R2B3-B66D" ], "locator": "§4.1经理表、§4.5投资运作分析，印刷页5–8；§4.1经理表、§4.5投资运作分析，印刷页5–8", "source_paraphrase": "管理人解释AI应用和自主科技机会，并按价格透支及经营压力调整持仓。；管理人讨论风险偏好和重资产交易，调减部分生物科技并增加AI数据中心相关资产。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "折价率=1−价格/估值；估值与价格须同时更新。", "boundary": "估值为情景估计，非已知事实。", "counterexample": "价格跌幅大而价值下降更多。", "test": "冻结估值假设，核查需求、毛利和再融资。", "question": "合成：股价120、价值估计150；随后股价90、价值估计100。比较折价率。", "constraint": "新证据把价值估计改为80，股价仍90。如何更新？", "transfer": "迁移：机房资产购价80、可回收价值60，算价格相对价值溢价。", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B3-B70-C1", "profile": "r2b3-b70", "title": "行业分散能否抵消共同现金流风险？", "source_ids": [ "R2B3-B70A", "R2B3-B70B" ], "locator": "正文问答：现金流与PEG；广发官网转载点拾投资朱昂访谈；正文：风险模型与2022年组合反思", "source_paraphrase": "问答把经营现金流、资本开支、定价权和管理层资本配置连接起来，反思机械PEG。；第一人称反思风险虽被预见却未充分进入组合；行业分散不等于风格因子分散。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "组合损益=各权重×各回报；共同因子需独立识别。", "boundary": "简化场景不包含相关性估计和动态交易。", "counterexample": "不同产业都依赖同一融资与贴现率。", "test": "核查现金回收、资金成本和因子集中度。", "question": "合成：四个行业各25%仓位，共同冲击令各跌20%。组合损失多少？", "constraint": "其中一家公司本来利润100、经营现金流110、资本开支80；冲击后经营现金流70，资本开支不变。算两期自由现金流。", "transfer": "迁移：四家云客户各占25%收入，均削减30%采购。收入总计100变多少？", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B3-B72-C1", "profile": "r2b3-b72", "title": "长期判断遇到流动性期限", "source_ids": [ "R2B3-B72A", "R2B3-B72B" ], "locator": "问答5、9、16、19、20；正文日期12月14日，URL路径12月28日；正文投资理念及择时段落", "source_paraphrase": "问答讨论自下而上、行业均衡、三年可见度和先研究后交易；宣传业绩未审计。；谈价格波动与基本面差异，通过公司质量和成长寻找机会。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "长期预期价值与短期支付义务分别约束仓位。", "boundary": "预期兑现时间并不保证中间价格。", "counterexample": "最终回报不错但被迫低价卖出。", "test": "同时审查公司逻辑、估值和投资者现金义务。", "question": "合成：100元持股预期3年后价值133.10，无分红，算年化回报。", "constraint": "一年后须支付40现金，此时股价跌30%；初始100股、初始每股1，无其他现金。需卖多少股、剩多少？", "transfer": "迁移：设备三年后残值133.10，明年却须偿债40；能否仅用三年年化回报批准项目？", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B3-B75-C1", "profile": "r2b3-b75", "title": "EPS增加能证明资本回报改善吗？", "source_ids": [ "R2B3-B75B", "R2B3-B75C" ], "locator": "§4.1经理表、§4.4–4.5，印刷页9–12；期间截止2023-12-31；§4.1经理表、§4.5，印刷页9–13；期间截止2024-12-31", "source_paraphrase": "管理人讨论资本配置、持续高增长的低基准概率，以及现金产生能力对AI投入的支撑。；管理人质疑二元市场叙事，强调效率改善、自由现金流与股东回报。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "每股收益=净利润/股数，回购融资成本与可持续FCF分开。", "boundary": "忽略税和动态利率的合成算例。", "counterexample": "借债回购提高EPS却增加负债风险。", "test": "看现金产生、利息覆盖和回购价格。", "question": "合成：净利润100、100股；借债200以每股10回购20股，先忽略利息。EPS变化？", "constraint": "借款年利息10、直接扣净利润且无税盾；付息前经营现金流110、资本开支90。算EPS及付息后简化FCF。", "transfer": "迁移：服务器业务净利100、经营现金流80、维持性资本开支60。可用于分配现金是否100？", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B3-B77-C1", "profile": "r2b3-b77", "title": "好公司是否在任何价格都创造回报？", "source_ids": [ "R2B3-B77B", "R2B3-B77A" ], "locator": "全文演讲；演讲发生2019-03-21，官网刊载2019-03-28；官方刊载的本人书序正文；不审计其中历史收益表述", "source_paraphrase": "演讲把长期回报拆为分红、业绩增长和估值，并强调公司质量、治理及长期资金制度。；署名前言讨论价格与价值、安全边际及从廉价资产走向再投资能力。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "股价=每股盈利×估值倍数，收益依赖成长和买入价格。", "boundary": "盈利增长不保证终端倍数不变。", "counterexample": "公司经营增长被估值收缩抵消。", "test": "分别验证增量资本回报与购买价格。", "question": "合成：EPS5、买入PE30；三年EPS按10%增长，终端PE仍30，无分红。算价格与回报。", "constraint": "终端PE降至20，其他不变，算三年总回报。", "transfer": "迁移：项目投入100，新增年利润12；资本成本15%。利润正数是否证明经济增值？", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B3-B82-C1", "profile": "r2b3-b82", "title": "浮存金免费吗？", "source_ids": [ "R2B3-B82B", "R2B3-B82A" ], "locator": "Charles T. Munger署名信，印刷页1–7；仅Munger署名附文印刷页39–43；2014为财年，2015年发布，不把Buffett正文归给Munger", "source_paraphrase": "讨论浮存金成本、承保不确定性及周期和结构压力的区别。；讨论分权经营、集中资本配置、留存收益检验及保守负债。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "浮存金资金成本=承保亏损/平均浮存金，投资收益与承保结果合并。", "boundary": "平均余额、索赔和损失准备金必须可信。", "counterexample": "投资收益掩盖承保亏损或准备金不足。", "test": "核查损失准备金发展与期限错配。", "question": "合成：平均浮存金1000，承保亏损30，投资收入50。算资金成本和合计收益。", "constraint": "补提准备金40，其他不变。重新算资金成本和收益。", "transfer": "迁移：客户预付款500，履约亏损25，利息收益10。净收益多少？", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B3-B87-C1", "profile": "r2b3-b87", "title": "净敞口100%是否只有100%的风险？", "source_ids": [ "R2B3-B87A", "R2B3-B87B" ], "locator": "官方托管授权转载，两页；官方托管WSJ授权转载，两页", "source_paraphrase": "访谈比较集中与分散价值策略，解释多空组合总敞口与净敞口。；访谈以正常化现金流和资本回报讨论价值，而非把低账面倍数等同低估。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "净敞口=多头−空头；总敞口=多头+空头。", "boundary": "忽略非线性、融资和逐日保证金，不能作VaR。", "counterexample": "多头跌、空头涨导致双侧亏损。", "test": "核查杠杆、借券成本及行业因子。", "question": "合成：净资产100，多头170，空头70，算净/总敞口。", "constraint": "多头跌10%，空头标的涨20%，另付空头名义3%借券费；算损益。", "transfer": "迁移：采购承诺170、固定价销售对冲70，两侧基差损失分别17与14。净采购100是否限定损失？", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B3-B89-C1", "profile": "r2b3-b89", "title": "坏消息已经计价了吗？", "source_ids": [ "R2B3-B89A", "R2B3-B89B" ], "locator": "官网全文；本人署名2021-07-05，发布2021-07-09；官网全文；本人署名2022-03-20，发布2022-03-22", "source_paraphrase": "市场信区分事实、解释与概率，讨论公开风险和已计价预期；不是风险必被充分定价的证明。；讨论战争、利率和市场预期中的不确定性，提出价值轮动判断。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "期望终值=概率加权终值；已知风险不等于合理定价。", "boundary": "概率和终值是可错假设，不是事实。", "counterexample": "公开坏消息仍可能远未充分计价。", "test": "记录概率依据、反证与风险预算。", "question": "合成：价格80，一年后好情景120概率60%，坏情景50概率40%，无分红。算期望终值和收益率。", "constraint": "证据令好情景概率降到30%，其他不变。重新算。", "transfer": "迁移：采购项目成功时回收终值120概率30%，失败回收50概率70%，投入80。期望净收益多少？", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B3-B99-C1", "profile": "r2b3-b99", "title": "名义财富增长是否意味着购买力增长？", "source_ids": [ "R2B3-B99B", "R2B3-B99A" ], "locator": "官方网站链接英文翻译PDF，印刷页1–3、13–15；日期为演讲日；官方链接PDF，导言及第四部分印刷页15–17；日期为演讲日", "source_paraphrase": "讲座讨论受托责任、激励、能力圈及对判断前提的证伪。；讲座连接微观企业行动与宏观约束，区分名义资产和购买力财富。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "真实增长=(1+名义增长)/(1+物价增长)−1。", "boundary": "消费篮子和真实可支配价值可能与官方通胀不同。", "counterexample": "名义赚钱但购买力下降。", "test": "验证能力圈假设、费用和资本保全。", "question": "合成：资产一年由100到110，物价上涨5%；算真实增长。", "constraint": "同年另发生资产期末值3的费用，算净名义和真实增长。", "transfer": "迁移：计算预算增长10%，同质量任务单价增长20%；可买任务量变化？", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B3-B100-C1", "profile": "r2b3-b100", "title": "让赢家奔跑之后如何检查集中风险？", "source_ids": [ "R2B3-B100A", "R2B3-B100B" ], "locator": "官方链接逐字稿印刷页1–3；日期为活动日，文件名20260528为发布版本；官方链接逐字稿印刷页1–5；活动日2026-05-01；两讲主题重叠不算独立验证", "source_paraphrase": "讲座使用能力圈和有限决策次数解释选股纪律；高确信度不自动证明低风险。；讲座讨论错误常见、少数赢家贡献及不机械卖出赢家；克隆作为线索有核验边界。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "权重由相对价格变化内生改变；持有纪律仍受下行预算约束。", "boundary": "涨幅本身不能证明未来价值或卖出必要性。", "counterexample": "机械持有和机械止盈都可能忽略基本面。", "test": "独立复核竞争、价格与组合损失预算。", "question": "合成：组合100，A占20其余80；A翻四倍、其余不变。新总值和A权重？", "constraint": "随后A下跌50%，其余不变；组合损失多少？", "transfer": "迁移：从别人季度披露复制20%持仓，但披露滞后45天。可否据此复制其当前确信度？列两项必要核查。", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2B3-B91-C1", "profile": "druckenmiller-duquesne", "title": "流动性可以替代损失预算吗？", "source_ids": [ "R2B3-B91A", "D07" ], "locator": "网页精选引语正文；活动日2021-01-21，发布2021-02-03；未观看完整视频；官方节目E4逐字稿；既有审阅记录", "source_paraphrase": "主办方精选本人引语讨论跨资产寻找机会、流动性带来的纠错能力及集中仓位。；回顾分阶段加仓、仓位规模及旧信号失效。", "attribution": "材料观点的编辑转述；以下数字、问题、反例及迁移场景均为编辑合成，不是本人案例或原话。", "model": "可交易性缩短退出时间，不消除退出价格下跌和缺口风险。", "boundary": "线性场景忽略市场冲击。", "counterexample": "集中仓位在流动市场仍发生大幅跳空。", "test": "验证仓位大小、止损执行、交易量与市场冲击。", "question": "合成：净资产100，对某资产类持仓70；资产跳跌15%，其他不动。组合亏损多少？", "constraint": "希望将该风险对组合的损失限制到5%，同样15%压力跌幅下最大仓位多少？", "transfer": "迁移：供应商报价可每日调整，但采购70的设备残值跌15%；每日有报价是否避免损失？", "tags": [ "F6", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S10-C1", "profile": "r2s1-vijay-rakesh", "title": "双收入引擎不能套同一个增速", "source_ids": [ "R2S1-ARM426", "R2S1-IONQ126" ], "locator": "R2S1-ARM426 pp.12–13；R2S1-IONQ126 pp.25–26", "source_paraphrase": "追问agentic CPU需求如何传导至数据中心版税、再单独询问授权收入增长。；询问硬件和平台服务收入组合及后续增速；管理层按客户起点回答而未承诺固定占比。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "总收入=版税+授权；分别增长后相加。", "boundary": "CPU需求增加不等于版税费率或授权合同同比增加。", "counterexample": "授权大单提前确认掩盖终端出货疲弱。", "test": "分开取得出货、费率、签约和收入确认表。", "question": "合成：版税80、授权20；各增长50%和10%，总增长多少？", "constraint": "授权增长改为-20%，总增长如何？", "transfer": "迁移到量子计算：硬件60、服务40，各增长0%和50%；总增多少？", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S11-C1", "profile": "r2s1-john-vinh", "title": "内容增长能抵消销量下降多少？", "source_ids": [ "R2S1-CRUS22", "R2S1-CRUS26" ], "locator": "R2S1-CRUS22 pp.3–4；R2S1-CRUS26 pp.3–7", "source_paraphrase": "以季节性差异追问供需，并拆分下一代内容增长来自装配率、单价或新增芯片。；追问手机内容增长、PC标准迁移延迟及AI PC真正形成需求的条件。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "收入=终端量×装配率×单机内容价格。", "boundary": "新平台功能可选和标准延期会阻止装配率提升。", "counterexample": "新品只是替换旧内容而未带来净增。", "test": "核对每平台BOM、实际装配率与终端出货。", "question": "合成：终端量100、装配率50%、单机内容4；新量90、装配率60%、内容5。收入变化？", "constraint": "新装配率仅40%，其余新参数不变，增长是否仍成立？", "transfer": "迁移到服务器电源：100万台×60%×20美元，改为120万台×50%×24美元，收入变化？", "tags": [ "F18", "F7", "F15" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S14-C1", "profile": "r2s1-srini-pajjuri", "title": "涨价后的BOM挤压如何穿透版税？", "source_ids": [ "R2S1-ARM425", "R2S1-ARM326" ], "locator": "R2S1-ARM425 pp.9–10；R2S1-ARM326 pp.12–13", "source_paraphrase": "追问版税增长的端市场构成及云与网络占比；管理层答复不等于他的预测。；追问内存成本是否已影响手机销量，以及BOM压力会否拖慢CSS/v9采用。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "版税收入=终端量×[CSS采用率×CSS单位费率+(1−CSS采用率)×普通单位费率]。", "boundary": "高端与低端手机不可混为同一暴露。", "counterexample": "新增高费率内容被整机减配抵消。", "test": "分别跟踪量、采用率、单位费率，比较内存涨价前后。", "question": "合成：100台、CSS采用率20%，普通版税1、CSS版税2。收入多少？", "constraint": "销量降10%，CSS采用率升至30%，费率不变；较基准增减？", "transfer": "迁移到电动车：100辆每辆芯片内容10，销量80辆、内容12；供应商收入变多少？", "tags": [ "F18", "F6", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S17-C1", "profile": "r2s1-ruben-roy", "title": "新增网络市场属于哪个产品？", "source_ids": [ "O8SC-CIENQ125", "O8SC-CIENQ325" ], "locator": "O8SC-CIENQ125 pp.11–12；O8SC-CIENQ325 pp.10–11", "source_paraphrase": "询问华为替换周期及CPO对Ciena是机会、威胁还是中性。；询问scale-across是否超出Coherent Lite范围，以及宽带业务是出售还是减少投资。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "可实现收入=项目数×适配比例×每项目可售内容。", "boundary": "距离、能耗与时延约束决定产品，市场总额不能全分配给某一方案。", "counterexample": "供电位置迫使链路更长，入门方案失去适用性。", "test": "按链路长度与功耗分类核对设计及产品资格。", "question": "合成：100项目，60%适配产品A，每项目内容5；其余产品B每项目内容8。两类收入各多少？", "constraint": "供电选址变化令A适配率降至30%，项目总数不变，组合收入如何？", "transfer": "迁移到冷却：80机柜，50%用方案X单柜10，余下用Y单柜20；X适配率降到25%，总内容如何变化？", "tags": [ "F15", "F18", "F16" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S18-C1", "profile": "r2s1-tristan-gerra", "title": "容量增长与长期合同能保证收入吗？", "source_ids": [ "R2S1-STX223", "R2S1-STX126" ], "locator": "R2S1-STX223 p.23；R2S1-STX126 pp.19–20", "source_paraphrase": "询问NAND降价会否传导HDD价格或引起份额替代。；追问HAMR长期协议期限、价格条款和收入占比，及密度升级后交期。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "收入=出货容量×每TB价格；合同覆盖率与价格锁定率分别核对。", "boundary": "长期协议可能仅约定量而不锁价。", "counterexample": "密度增加带来容量增长，但每TB价格下降更快。", "test": "读取合同量价条款及客户存储层级、迁移成本。", "question": "合成：100PB×每TB20美元；容量增20%、单价降10%，收入增长多少？", "constraint": "若单价跌20%，容量仍增20%，增长变多少？", "transfer": "迁移HBM：bit量增30%、每bit价格降15%，收入增速？", "tags": [ "F7", "F8", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S22-C1", "profile": "r2s1-chris-caso", "title": "赢家都能供货，收入就不受组合变化影响吗？", "source_ids": [ "R2S1-MPWR126", "R2S1-MPWR226" ], "locator": "R2S1-MPWR126 pp.14–15；R2S1-MPWR226 pp.12–13", "source_paraphrase": "询问商用GPU与ASIC解决方案增量，并检验汽车下半年增长可见性。；询问CPU电源份额、agentic需求传导与高速增长如何影响业务多元化。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "收入=CPU数×单CPU内容×份额；再与其他业务合并。", "boundary": "横跨x86/Arm不能证明两边内容或份额相同。", "counterexample": "客户采用更多自研芯片但供应商单机价值下降。", "test": "逐平台认证及内容、份额、终端量桥。", "question": "合成：CPU相关收入40、其他60；CPU增80%、其他增10%，总增多少，CPU占比多少？", "constraint": "CPU次年回落30%、其他持平，以上一年新总额为基准总收入跌多少？", "transfer": "迁移设备：先进封装收入30、传统70；各增100%与0%，总增长及先进封装占比？", "tags": [ "F6", "F18", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S25-C1", "profile": "r2s1-william-stein", "title": "短交期意味着需求疲软还是产能充足？", "source_ids": [ "R2S1-MPWR225", "R2S1-MPWR226" ], "locator": "R2S1-MPWR225 pp.12–13；R2S1-MPWR226 pp.8–9", "source_paraphrase": "以收入相对名义产能的差距追问短交期和订货模式。；直接追问价格对收入增长的贡献，继而检查机器人和楼宇自动化落地。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "利用率=出货量/可用产能；收入=量×价。", "boundary": "产能金额随产品组合变化，不能当作物理利用率精确值。", "counterexample": "产能扩得更快导致交期缩短，需求仍在增长。", "test": "核对同口径单位产能、出货与价格。", "question": "合成：假设同期间产量等于出货，无库存或外包变化；产能100、出货60、单价10；次年产能150、出货90、单价10。收入与利用率如何？", "constraint": "产品涨价20%，出货仍90；把收入增长全归于需求量错多少？", "transfer": "迁移晶圆厂：产能200、出货160，扩到300、出货210；量增长与利用率变化？", "tags": [ "F7", "F2", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S28-C1", "profile": "r2s1-richard-windsor", "title": "增长是否覆盖资本和稀释成本？", "source_ids": [ "R2S1-RFM1", "R2S1-RFM2" ], "locator": "R2S1-RFM1 正文收入/GW与运营成本推算段；R2S1-RFM2 正文warrants及条件条款段", "source_paraphrase": "将收入预测除以GW检验单位容量货币化，再加入研发和销售成本；作者估计仍需核验。；将销售增长与客户权证稀释放在一起讨论；原文推测的行权门槛不作为事实。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "经营盈余=GW×每GW收入×扣除变动成本后的贡献率（未扣固定成本）−固定成本；每股经营盈余=盈余/稀释后股数。", "boundary": "采购额、产能资本支出和股权稀释不能重复计算；权证公允价值不等于零行权价。", "counterexample": "收入高增长但资本更新与稀释吞掉每股回报。", "test": "已投运GW、付费利用率、token价格、运营费和真实权证条款。", "question": "合成：2GW，每GW年收入10，贡献率50%，固定成本8；经营盈余多少？", "constraint": "每GW收入升至12，贡献率不变；股数从1增至1.25，盈余每股较原来多少？", "transfer": "迁移软件：经营盈余从20增至24，股数从10增至12；每股经营盈余变化？", "tags": [ "F10", "F11", "F12" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S29-C1", "profile": "r2s1-jim-kelleher", "title": "客户更多是否真的降低集中度？", "source_ids": [ "R2S1-ARGUS25", "R2S1-ARGUS26" ], "locator": "R2S1-ARGUS25 pp.22–23（Apple/NVIDIA具名页）；R2S1-ARGUS26 pp.21–22（Apple/NVIDIA具名页）", "source_paraphrase": "以设备更新、服务和AI平台产品集合组织投资论点；不将目标价当估值模型。；将客户类型扩展到企业、主权与neocloud并考察非数据中心业务。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "最终需求暴露按资金来源归并；收入×毛利率映射利润。", "boundary": "客户名称多不等于经济风险独立；具名两页仅为公开样本。", "counterexample": "新客户主要依赖同一上游客户的融资和租赁合同。", "test": "穿透订单、资金来源和转租关系，去除重复终端需求。", "question": "合成：四客户收入各25，其中三家最终由同一买家承担需求。名义最大客户占比与穿透后占比各多少？", "constraint": "该最终买家需求下降20%，其他不变；收入变化？", "transfer": "迁移软件渠道：五经销商各20，其中四家卖给同一集团；该集团需求减25%，总收入减少多少？", "tags": [ "F16", "F6", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S30-C1", "profile": "r2s1-jay-goldberg", "title": "防守既有利润能否解释AI投资？", "source_ids": [ "R2S1-D2D1", "R2S1-D2D2" ], "locator": "R2S1-D2D1 Everything is an AI Trade Now；Semiconductor IP章节；R2S1-D2D2 Strategy or Faith?章节", "source_paraphrase": "将跨行业增长还原为少数资本开支买家，并讨论开放ISA不等于免费设计。；提出保护既有云利润的替代解释，同时质疑份额变化、融资能力和收入重复计算。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "防守价值=避免损失的现金贡献×持续年数（简化，未折现）；与额外投入比较。", "boundary": "新增AI利润与保住的既有利润不能算作同一现金流两次。", "counterexample": "对手融资受限，原本假定的份额损失并不会发生。", "test": "反事实份额、分部利润、客户迁移及互相采购净额。", "question": "合成：额外投资30可避免每年现金贡献损失8、持续5年，不折现无残值。净价值多少？", "constraint": "避免的损失仅每年4，净价值如何？", "transfer": "迁移通信：网络升级花12，避免每年流失现金贡献3，持续3年；是否覆盖投入？", "tags": [ "F9", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S32-C1", "profile": "r2s1-andrew-gardiner", "title": "收入占比增加能推算单位采用率吗？", "source_ids": [ "R2S1-ARM425", "R2S1-ARM426" ], "locator": "R2S1-ARM425 pp.6–7；R2S1-ARM426 pp.6–7", "source_paraphrase": "追问v9占比提升来自哪些端市场及CSS费率。；追问新CPU需求增加来自原有客户加量还是新增客户，并检查供应交付约束。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "某类收入占比=该类量×费率/总版税。", "boundary": "高费率产品收入占比高于数量占比，不可直接混用。", "counterexample": "价格翻倍导致收入占比增长，实际采用量不变。", "test": "获取单位出货、费率与同一期间合同结构。", "question": "合成：100芯片，20个高费率芯片每个版税2，其余1。高费率收入占比多少？", "constraint": "高费率单价变3、数量不变，收入占比多少？能否称采用率增加？", "transfer": "迁移软件：100席位，10高级席位每席5，其余每席1，高级收入占比及席位占比？", "tags": [ "F2", "F18", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S33-C1", "profile": "r2s1-didier-scemama", "title": "占比增长不是绝对收入增长", "source_ids": [ "R2S1-AIX423", "R2S1-AIX325" ], "locator": "R2S1-AIX423 PDF pp.33–35，印刷pp.37–39；R2S1-AIX325 pp.9–11", "source_paraphrase": "询问客户出售旧MicroLED设备会否形成竞争及是否能改作其他工艺，另问研发费用弹性。；将产品占比变动换算金额，追问光电子需求能否跨过单轮AI资本开支周期。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "分部收入=公司总收入×分部占比。", "boundary": "总收入预测与占比必须同期；中古设备影响订单而非必然影响当期收入。", "counterexample": "分部占比增加只是其他业务衰退。", "test": "结合订单金额、设备安装、利用率及客户二手设备处置。", "question": "合成：公司收入500，光电子占20%；次年总收入400，占30%。分部收入与增长？", "constraint": "若总收入只有300，占比仍30%，结论如何？", "transfer": "迁移GPU供应商：总收入1000、网络占10%；次年800、占15%，网络收入怎样？", "tags": [ "F4", "F8", "F15" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S37-C1", "profile": "r2s1-randy-abrams", "title": "建成产能怎样变成确认收入？", "source_ids": [ "R2S1-TSM422", "R2S1-TSM124" ], "locator": "R2S1-TSM422 pp.7–9；R2S1-TSM124 pp.11–13", "source_paraphrase": "比较美国和台湾每单位晶圆产能投资，主动询问口径差异与新节点成本加速。；追问N2量产到有意义收入的时差、tape-out及资本开支节奏。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "已装产能×利用率×良率×单位收入；量产到收入有周期延迟。", "boundary": "跨地区比较须统一工艺节点、阶段、基础设施范围与产能单位。", "counterexample": "低利用率或客户认证延期使新增产能暂不贡献利润。", "test": "同口径CapEx/千片月产能、量产日期、周期与良率。", "question": "合成：两地投资分别30和40，月产能分别100和80（同单位同范围）。单位产能投资溢价多少？", "constraint": "第二地有效利用率仅80%，第一地100%；按有效产出比较溢价？", "transfer": "迁移云：A投入20提供100有效实例，B投入24提供80；B单位有效实例成本溢价？", "tags": [ "F9", "F18", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S23-C1", "profile": "r2s1-gus-richard", "title": "调整后增长仍需要进度证据", "source_ids": [ "R2S1-POWI305", "R2S1-ALTR410" ], "locator": "R2S1-POWI305 Gus Richard具名问答，增长/准备金/毛利段；R2S1-ALTR410 Gus Richard具名问答末段", "source_paraphrase": "先剔除准备金转回与项目延迟再比较季节增长，接着区分降本留存和客户让利。；把28nm送样计划拆成仍在晶圆厂与客户已经拿到产品，核对实际进度。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "同口径增长=(本期经常收入/上期经常收入)-1；计划送样不等于完成送样。", "boundary": "旧公开样本不能证明当前AI观点；消除一次性项不能消除终端需求不确定性。", "counterexample": "收入变化来自准备金和交付错期，非持续需求。", "test": "准备金明细、延迟订单确认及客户收货记录。", "question": "合成：上季收入100含一次性转回10；本季108无一次性。报告与经常收入增速各多少？", "constraint": "本季另有18来自上季延迟交付。仅为时点敏感性，从本季剔除此项但不回补上季，经常收入增速多少？", "transfer": "迁移项目验证：20个设计胜出，只有5个已交付客户样品。若验证率定义为已交付样品/设计胜出，是多少？", "tags": [ "F3", "F7", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S26-C1", "profile": "r2s1-pierre-ferragu", "title": "费用桥不能只靠收入增速", "source_ids": [ "R2S1-MSI412", "R2S1-ARM224" ], "locator": "R2S1-MSI412 p.8 Pierre Ferragu问答；R2S1-ARM224 pp.13–14 Pierre Ferragu问答", "source_paraphrase": "由收入和营业利润率指导反推费用增长，并询问新品上市后的采用节奏。；追问费用增加对应研发、客户工程或业务拓展，并区分共识误建模与新增支出。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "营业利润=收入×毛利率−经营费用；开发支出先于收入。", "boundary": "管理层对费用原因的解释不是分析师既有预测；GAAP与non-GAAP须同口径。", "counterexample": "前瞻研发扩招造成当前费用增长，但后续合同无法变现。", "test": "工资人数、研发项目、股权费用、合同和量产兑现时间。", "question": "合成：收入100、毛利率60%、费用40。下一年收入120、毛利率60%、目标营业利润率20%，隐含费用多少及增幅？", "constraint": "若毛利率降至55%，仍要20%营业利润率，允许费用多少？", "transfer": "迁移软件：收入200、毛利率80%、费用120；收入250、毛利率80%、费用140，营业利润率如何变化？", "tags": [ "F18", "F3", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S31-C1", "profile": "r2s1-sandeep-deshpande", "title": "技术验证中的订单能全部当收入吗？", "source_ids": [ "R2S2-ASML424", "R2S2-ASML125" ], "locator": "R2S2-ASML424 Sandeep Deshpande末段Q&A；R2S2-ASML125 Sandeep Deshpande末段Q&A", "source_paraphrase": "追问已下单工具与尚未确认量产导入时点的矛盾，检验延期风险。；把下一年订单覆盖与上一年同期比较，并追问第三阶段High-NA订单时点。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "当期可实现收入=积压订单×当期可交付比例×验收比例。", "boundary": "公开镜像转录尚未逐字核对官方录音；积压订单并非不可延期的现金。", "counterexample": "客户验证不达标，已在订单簿工具延迟验收。", "test": "按研发/试产/量产阶段核对合同、客户验证与收入确认。", "question": "合成：积压订单100，其中60计划本年交付，本年交付部分80%可验收。可确认多少，未确认收入的原订单金额多少（含已交付未验收，按验收确认且订单金额与收入一比一）？", "constraint": "验收比例下降到50%，其他不变。可确认收入和未确认收入的原订单金额如何？假设无取消、新增或汇率变化。", "transfer": "迁移服务器：订单200，60%本年交付，交付中75%验收。确认额？", "tags": [ "F7", "F25", "F4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S1-S34-C1", "profile": "r2s1-sara-russo", "title": "总增长能否证明老客户加购？", "source_ids": [ "R2S1-ARM224", "R2S2-ASML424" ], "locator": "R2S1-ARM224 pp.17–18 Sara Russo问答；R2S2-ASML424 Sara Russo Q&A（首次出现）", "source_paraphrase": "问同一续约客户IP采用与花费变化，再问订阅方案是否改善版税预测可见性。；追问减少订单披露的理由，并将EUV订单与年度指导覆盖和次年交付分开。", "attribution": "具名分析师提问/署名论证样本；管理层回答只作为被挑战的答复。以下挑战及数字由研究者合成，非原作者预测。", "model": "群组净收入留存=老客户本期收入/老客户基期收入；订单须按交付期拆分。", "boundary": "ASML为公开转录镜像，未听原录音；机构简介登录墙，不能用历史call认证现职。", "counterexample": "新客户贡献掩盖老客户收缩，长交期订单掩盖当期兑现不足。", "test": "同群组续约账单、IP采用、合同期限及计划交付季度。", "question": "合成：去年老客户收入100；今年老客户90，新客户30。总增长和老客户净留存各多少？", "constraint": "新客户收入变60，老客户仍90。总增长能否证明老客户加购？", "transfer": "迁移设备：新增订单100，只有40本年交付并验收，余下明年；能作为本年订单转收入的金额多少？", "tags": [ "F4", "F6", "F7" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S2-S41-01", "profile": "mehdi-hosseini", "title": "资本开支强度下降就等于少花钱？", "source_ids": [ "R2S2-ASML424", "S36SRC1" ], "locator": "ASML Q4 2024 earnings Q&A — public transcript mirror：Hosseini具名问答；TSMC Q2 2025 earnings call — Charlie Chan Q&A：pp.17–18", "source_paraphrase": "追问设备吞吐、产能弹性与客户承诺。 / 追问N2流片增长是否需要更高资本强度。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "CapEx=收入×资本强度；吞吐改善影响所需设备数量。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：收入100、资本强度40%；下一年收入130、强度35%，CapEx变多少？", "constraint": "再假设设备有效吞吐提升20%，完成相同产出原需120台，现在需多少台？", "transfer": "迁移到云：收入增长30%，资本强度从50%到40%，CapEx同比怎样？", "tags": [ "F9", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S42-01", "profile": "krish-sankar", "title": "1GW公布容量如何变成收入？", "source_ids": [ "R2S2-ASML125", "S36SRC2" ], "locator": "ASML Q1 2025 earnings Q&A — public transcript mirror：Sankar具名问答；TSMC Q3 2025 earnings call — Charlie Chan Q&A：pp.15–16", "source_paraphrase": "追问次年增长需要的订单及区域构成。 / 询问1GW AI容量如何映射晶圆需求及GPU/ASIC组合。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "交付收入=公布GW×落地率×每GW芯片量×单芯片晶圆收入。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：1GW，落地率60%，每GW 10万颗，代工收入每颗2000美元，收入多少？", "constraint": "若落地率降40%，每GW芯片量增25%，新收入多少？", "transfer": "迁移到设备：100座厂公告，兑现50%，每厂设备内容20，供应份额30%，收入代理多少？", "tags": [ "F9", "F18", "F7" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S43-01", "profile": "toshiya-hari", "title": "GPU增长是否必然抬高公司毛利？", "source_ids": [ "O8SAD4", "O8SAD5" ], "locator": "AMD Q2 2024 Earnings Call：pp.10–11；AMD Q3 2024 Earnings Call：pp.7–8", "source_paraphrase": "检验MI300指引是需求还是供给限制，并问盈利变化。 / 追问GPU指引上调、客户扩展与毛利组合。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "综合毛利率=各业务收入权重×各自毛利率之和。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：GPU收入40毛利50%，其他60毛利60%，整体毛利多少？", "constraint": "GPU翻倍，其他不变，GPU毛利仍50%；整体毛利怎样？", "transfer": "迁移到存储：高端产品占比从20%到40%，毛利分别60%和30%，综合毛利怎样？", "tags": [ "F2", "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S44-01", "profile": "joe-quatrochi", "title": "提前交货会创造全年需求吗？", "source_ids": [ "R2S2-ASML424", "R2S2-ASML125" ], "locator": "ASML Q4 2024 earnings Q&A — public transcript mirror：Quatrochi具名问答；ASML Q1 2025 earnings Q&A — public transcript mirror：Quatrochi具名问答", "source_paraphrase": "追问收入指引上端与预建库存、交期。 / 检验关税是否促使提前交付及旧机库存可售性。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "全年收入=sum季度收入；交付上限=min备机,客户已备空间)。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：Q1/Q2各计划20台，5台由Q2前移到Q1，单价2，两个季度收入怎样？", "constraint": "客户Q1仅有22台空间，前移最多多少？", "transfer": "迁移到服务器：原H1/H2各100台，30台前移，全年台数增长多少？", "tags": [ "F2", "F3", "F7" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S45-01", "profile": "brian-chin", "title": "升级和扩产能否重复算？", "source_ids": [ "R2S2-LAM425", "S2D-B03" ], "locator": "Lam Research FY2025 Q4 earnings Q&A — public transcript mirror：Brian Edward Chin具名问答；Lam Research FY2026 Q1 Earnings Call：Brian Chin具名问答；本轮读取Roic同场镜像", "source_paraphrase": "追问刻蚀工具跨代适用及服务业务拆分。 / 要求拆分AI支出到WFE构成，并检验升级与新建是否并行。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "升级扩产后总bit供给=保留原晶圆量×升级后bit密度+新增晶圆量×新bit密度。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：100片产能，每片bit指数1，升级后1.5，另新增20片同样1.5；总bit多少？", "constraint": "升级导致原产线晶圆量损失10%，再算。", "transfer": "迁移到CPU：100台每台吞吐1，升级后1.4且新增20台1.4，总吞吐多少？", "tags": [ "F7", "F9", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S46-01", "profile": "brett-simpson", "title": "成本上涨需要涨价多少才保毛利？", "source_ids": [ "R2S2-TS419", "S36SRC1" ], "locator": "TSMC Q4 2019 earnings call：p.11；TSMC Q2 2025 earnings call — Charlie Chan Q&A：pp.11–12", "source_paraphrase": "拆分中国增长来源及5G/HPC贡献。 / 追问结构性成本、涨价与AI效率能否抵消毛利压力。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "目标售价=新单位成本/(1−目标毛利率)。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：售价100、成本40；成本升到44，维持60%毛利率需要售价多少？", "constraint": "若仅能涨价5%，毛利率多少？", "transfer": "迁移到云：每任务成本0.04→0.05，目标毛利50%，价格应从多少到多少？", "tags": [ "F18", "F4" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S47-01", "profile": "janardan-menon", "title": "同业增长能直接外推吗？", "source_ids": [ "R2S2-STM125", "R2S1-AIX325" ], "locator": "STMicroelectronics Q1 2025 earnings Q&A — public transcript mirror：Menon具名问答；Aixtron Q3 2025 earnings call：pp.6–7", "source_paraphrase": "追问与TI工业复苏差异、渠道和毛利组合。 / 拆分Opto/GaN/SiC走势，问晶圆尺寸转换。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "组合增长=Σ初始收入权重×分部增长。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：A工业占80%、汽车20%；B各50%。工业增10%、汽车降20%，两公司增速多少？", "constraint": "B工业通过渠道，终端增10%但渠道去库存使工业出货仅增2%，B总增速多少？", "transfer": "迁移到云：AI占20%增50%，非AI占80%增5%，整体增多少？", "tags": [ "F6", "F2", "F8" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S48-01", "profile": "stephane-houri", "title": "单季好成绩可以乘四吗？", "source_ids": [ "R2S2-ASML424", "R2S2-ASML125" ], "locator": "ASML Q4 2024 earnings Q&A — public transcript mirror：Houri具名问答；ASML Q1 2025 earnings Q&A — public transcript mirror：Houri具名问答", "source_paraphrase": "检验服务业务单季年化与全年指引差异。 / 追问订单波动是否来自关税，以及指引订单覆盖。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "全年=首季+剩余三季；升级收入与经常服务拆开。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：Q1收入2.1，全年目标7.5，后三季平均需要多少？", "constraint": "Q1中一次性升级0.6，其余每季稳定1.5，全年多少？", "transfer": "迁移到订阅：Q1收入30含一次性实施6，后三季订阅各24，全年多少？", "tags": [ "F3", "F2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S49-01", "profile": "francois-xavier-bouvignies", "title": "更贵设备什么时候更便宜？", "source_ids": [ "R2S2-ASML424", "R2S2-ASML125" ], "locator": "ASML Q4 2024 earnings Q&A — public transcript mirror：Bouvignies具名问答；ASML Q1 2025 earnings Q&A — public transcript mirror：Bouvignies具名问答", "source_paraphrase": "检验2nm扩产是否前置及AI能否抵消手机弱势。 / 问High-NA定价能否促进采用及单次曝光替代节奏。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "每合格曝光成本=年度成本/(名义吞吐×可用率×良率)。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：旧机年成本100、吞吐100、可用率90%、良率95%；新机成本150、吞吐160、可用率80%、良率95%，比较单位成本。", "constraint": "新机可用率仅60%，是否仍便宜？", "transfer": "迁移到GPU：价格成本2倍、吞吐3倍、有效利用率只有旧卡一半，单位成本比多少？", "tags": [ "F10", "F15", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S50-01", "profile": "alexander-duval", "title": "曝光层数增加等于设备量同幅增加？", "source_ids": [ "R2S2-ASML125", "R2S2-ASML225" ], "locator": "ASML Q1 2025 earnings Q&A — public transcript mirror：Duval具名问答；ASML Q2 2025 earnings Q&A — public transcript mirror：Duval具名问答", "source_paraphrase": "区分EUV研发消息与可扩展可靠产品，问推理需求。 / 追问DRAM光刻强度和通用平台时间线。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "设备需求∝晶圆量×曝光层数/有效吞吐。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：晶圆量增20%、曝光层数增25%、设备吞吐增50%，台数需求增长多少？", "constraint": "若设备可用率另下降10%，台数需求相对基线多少？", "transfer": "迁移到推理：请求量2倍、每请求算力1.5倍、芯片有效性能2.5倍，芯片数增长多少？", "tags": [ "F15", "F18", "F7" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S54-01", "profile": "brad-zelnick", "title": "客户省钱等于供应商利润吗？", "source_ids": [ "S51SRC1", "S52SRC2" ], "locator": "Microsoft FY2025 Q2 earnings call — Keith Weiss Q&A：Brad Zelnick具名问答；Adobe Q3 FY2025 earnings call — Mark Moerdler Q&A：pp.18–19", "source_paraphrase": "询问推理成本下降后产品包装与覆盖客户。 / 问AEP agents给客户创造的价值如何被Adobe捕获。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "增量毛利=付费客户×收费−推理成本−服务成本。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：100客户每户节省100，供应商捕获20%，总推理服务成本800，增量毛利多少？", "constraint": "只有60客户愿付费，但成本仍800，毛利多少？", "transfer": "迁移到设备：客户总节省1000，合同分享30%，实施成本200，供应商利润多少？", "tags": [ "F10", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S55-01", "profile": "kirk-materne", "title": "新品套餐的全年收入怎么桥接？", "source_ids": [ "O8SB-MS3", "R2S2-ADBE225" ], "locator": "Microsoft FY2025 Q3 earnings call — selected Q&A：Kirk Materne具名问答；Adobe FY2025 Q2 earnings call：pp.13–14", "source_paraphrase": "区分Azure非AI改善与此前优化周期。 / 询问Creative Cloud Pro改变的客户反应及下半年收入。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "收入=客户数×价格×生效月份/12。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：1000客户，年费100，7月起半数转年费120，其余不变；全年收入多少？", "constraint": "升级客户中10%因此流失，均7月生效，全年收入多少？", "transfer": "迁移到云：40收入AI增50%，60非AI增5%，合计增长多少？", "tags": [ "F3", "F4", "F2" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S58-01", "profile": "raimo-lenschow", "title": "跨售是否带来净新增？", "source_ids": [ "R2S2-UI324", "S51SRC2" ], "locator": "UiPath FY2024 Q3 earnings call：p.13；Microsoft FY2025 Q4 earnings call — Weiss and Moerdler Q&A：Raimo Lenschow具名问答", "source_paraphrase": "询问测试自动化与SAP转型的需求。 / 检验Copilot如何扩展到数据与agent工作流。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "净新增ARR=新模块ARR−被替代旧模块ARR−流失ARR。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：新模块ARR30，替代旧模块10，额外流失5，净新增多少？", "constraint": "若新模块附带算力成本12，旧模块停用节约成本3，增量贡献多少？", "transfer": "迁移到网络：新交换机销售100，其中替换原计划60，其他业务流失10，净新增多少？", "tags": [ "F16", "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S59-01", "profile": "alex-zukin", "title": "AI-first增长能与核心增长相加吗？", "source_ids": [ "R2S2-ADBE225", "S52SRC2" ], "locator": "Adobe FY2025 Q2 earnings call：pp.15–16；Adobe Q3 FY2025 earnings call — Mark Moerdler Q&A：pp.15–16", "source_paraphrase": "问Express竞争和Meta广告自动化的竞合。 / 拆分AI-first采用与传统价量对ARR改善的贡献。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "总ARR=互斥核心ARR+新增AI ARR；内含AI标签不可重复。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：总ARR100，其中AI标签20。总ARR升至115，其中AI标签35；能否称AI增15又加总增15？", "constraint": "若AI标签增加15中10只是原有产品重新分类，真实新AI ARR多少？", "transfer": "迁移到GPU：数据中心100内含AI80，AI增至100但数据中心到120，合计收入增量多少？", "tags": [ "F2", "F11", "F16" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S60-01", "profile": "derrick-wood", "title": "超预期多少来自提前续约？", "source_ids": [ "R2S2-ADBE422", "R2S2-NOW225" ], "locator": "Adobe FY2022 Q4 earnings call：p.12；ServiceNow Q2 2025 earnings Q&A — public transcript mirror：operator Derrick Wood / speaker James Wood问答", "source_paraphrase": "追问Creative/Document ARR季节性差异。 / 问收入超指引是否由自托管比例、线性及业务构成驱动。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "剔除已识别跨期因素后的超额=报表超额−已识别提前确认；未来持续性另行验证。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：收入指引100，实际102，其中1为下季续约提前确认；剔除这项跨期因素后的超额多少？", "constraint": "原下季收入110且其他不变，提前确认后下季多少？", "transfer": "迁移到服务器：交付105超计划100，其中8台从下季前移，剔除前移后较计划怎样？", "tags": [ "F2", "F11", "F3" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S61-01", "profile": "rishi-jaluria", "title": "编程提效多少进入利润？", "source_ids": [ "R2S2-COUR123", "R2S2-MS125" ], "locator": "Coursera Q1 2023 earnings call：p.10；Microsoft FY2025 Q1 earnings call：Rishi Jaluria具名问答", "source_paraphrase": "问GenAI如何改善内部工程和产品效率。 / 询问Copilot数字口径及企业/消费战略。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "现金节省=受影响成本×可自动化比例×该环节单位工时下降率×实际兑现率。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：工程成本100，可自动化40%，该环节单位工时降低25%，兑现率50%，现金节省多少？", "constraint": "新增推理成本6，净现金收益多少？", "transfer": "迁移到呼叫中心：成本80，自动化50%，可兑现70%，新算力成本12，净收益多少？", "tags": [ "F10", "F18", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S62-01", "profile": "tyler-radke", "title": "增长来自价格还是用户？", "source_ids": [ "R2S2-ADBE225", "S52SRC2" ], "locator": "Adobe FY2025 Q2 earnings call：pp.14–15；Adobe Q3 FY2025 earnings call — Mark Moerdler Q&A：p.14", "source_paraphrase": "问Firefly生成量与货币化是否同步。 / 要求拆分提价后的真实业务量。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "收入倍数=价格倍数×付费数量倍数。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：收入增长15%、价格提高10%，付费数量增长多少？", "constraint": "若活跃总量增30%，付费数量仅按前题增长，付费渗透率相对变化多少？", "transfer": "迁移到DRAM：收入增40%，ASP增25%，bit出货增多少？", "tags": [ "F2", "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S63-01", "profile": "gregg-moskowitz", "title": "销售扩张是否掩盖生产率下滑？", "source_ids": [ "R2S2-ADBE224", "R2S2-NOW225" ], "locator": "Adobe FY2024 Q2 earnings Q&A — public transcript mirror：Moskowitz具名问答；ServiceNow Q2 2025 earnings Q&A — public transcript mirror：Moskowitz具名问答", "source_paraphrase": "追问付费用户增长及数字渠道获客。 / 检验不同地区销售领导变化后的执行能力。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "新增ARR=成熟销售人数×每人成交额+新人成交额。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：100成熟销售各产1，新增20人仅达成熟水平25%，总产出及人均多少？", "constraint": "原100人成交额下降10%，总产出及同比多少？", "transfer": "迁移到工厂：100熟练工各10件，20新人各5件，总产出与全员平均多少？", "tags": [ "F4", "F20", "F17" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S2-S64-01", "profile": "michael-turits", "title": "部署提速会立即增加多少使用收入？", "source_ids": [ "R2S2-SPLK123", "R2S2-UI324" ], "locator": "Splunk FY2023 Q1 earnings Q&A — public transcript mirror：Michael Turits具名问答；UiPath FY2024 Q3 earnings call：pp.17–18", "source_paraphrase": "检验计费方式价值与宏观延迟对消费的影响。 / 追问部署提速及固定汇率NRR仍下降。", "attribution": "具名公开提问体现待检验方向；模型、数字和迁移题均为编辑合成，不是该人的预测或原问答。", "model": "年度使用收入=客户数×投产后每月消费×可收费月数。", "boundary": "仅在题设口径、时期和其他条件不变时成立；历史材料不代表现时观点。", "counterexample": "表面增长可能来自时点、组合或口径变化，需排除后再推断持续增长。", "test": "下一次披露核对分项驱动、可比口径与兑现时点；偏离题设则撤回简单外推。", "question": "合成：100客户1月签约，每月消费1，部署4个月后5月起收费，全年收入多少？", "constraint": "部署缩为2个月，3月起收费，但每月消费降10%，全年多少？", "transfer": "迁移到机房：100MW从7月投产提前至4月，月收入每MW 2，年度增量多少？", "tags": [ "F10", "F4", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "R2S3-S65-01", "profile": "r-scott-berg", "title": "新增功能能否覆盖销售成本？", "source_ids": [ "R2S3-FRSHQ325", "R2S3-NAVANQ426" ], "locator": "Freshworks Q3 2025 Earnings Call pp.10–11；Navan Q4 FY2026 Earnings Call pp.12–13", "source_paraphrase": "询问独立ESM销售是否另设团队、如何定价。 / 询问餐厅预订经济性及对指引贡献。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "增量贡献=客户数×付费率×年价×贡献率−增量销售费。", "boundary": "收入贡献需付费、部署及确认；发布不等于变现。", "counterexample": "采用上升但免费使用且销售费用增加。", "test": "付费客户cohort、销售编制、增量合同和续费。", "question": "合成：1000客户，付费率20%，年价1000元，贡献率70%，增量销售费10万元，年贡献多少？", "constraint": "若必须另设团队使增量销售费升至20万元，盈亏平衡付费率多少？", "transfer": "迁移到CPU特性：100万颗，付费启用10%，每颗授权20元，增量支持成本300万元，净贡献？", "tags": [ "F10", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S66-01", "profile": "mark-mahaney", "title": "扩产何时解除收入上限？", "source_ids": [ "R2S3-AMZNQ424", "O8SB-META1" ], "locator": "Amazon Q4 2024 Earnings Call — public transcript mirror Q&A首位Mark Mahaney；Meta Q1 2025 earnings call — selected Q&A pp.18–19", "source_paraphrase": "追问资本开支年化规模与AWS增长是否受供给限制。 / 追問Reality Labs亏损何时及如何下降。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "收入=min(需求,可用容量)×单价。", "boundary": "资本开支与当期可用容量之间存在建设时滞。", "counterexample": "资本开支增长但未通电或需求低于计划。", "test": "分期上线MW与实际使用量、利用率及折旧。", "question": "合成：需求140单位，可用100单位，单价2；投入能新增50但本年只交付20，收入上限？", "constraint": "新增容量50全部交付但需求降至110；相对需求140且150容量均可用的方案，收入少多少？", "transfer": "迁移到封装：订单120万，设备名义150万，良率70%，可交付多少？", "tags": [ "F9", "F7", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S71-01", "profile": "james-fish", "title": "集成设备的溢价还能保住吗？", "source_ids": [ "R2S3-MSIQ325", "R2S3-MSICONF26" ], "locator": "Motorola Solutions Q3 2025 Earnings Call pp.19–20；Motorola Solutions at Piper Sandler Growth Frontiers pp.1–2", "source_paraphrase": "询问SVX上市后的竞争者定价反应。 / 追问AI策略及适合救援人员工作流的原因。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "客户TCO=硬件+期限内服务费；供应商利润另算。", "boundary": "客户价值不自动等于供应商定价权。", "counterexample": "对手降价使集成便利仍不能覆盖溢价。", "test": "竞标净价、续约、客户流程实测与转换成本。", "question": "合成：旧方案两设备各600元，年服务各100元，用3年；集成设备1500元、年服务80元，节省多少？", "constraint": "对手两设备硬件总价降至800元，其他不变；集成设备可接受硬件价上限？", "transfer": "迁移到服务器：分立系统硬件100万、三年运维60万；融合系统硬件120万、运维30万，总成本差？", "tags": [ "F10", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S75-01", "profile": "samik-chatterjee", "title": "附加收入能把整机利润拉高多少？", "source_ids": [ "R2S3-DELLQ225", "R2S3-JBLQ424" ], "locator": "Dell Q2 FY2025 Earnings Call pp.11–12；Jabil Q4 FY2024 Earnings Call pp.24–25", "source_paraphrase": "询问企业AI客户能带来多少服务、存储附加及毛利差异。 / 以收入规模、产能利用与费用率检验6%利润率路径。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "毛利=硬件收入×硬件毛利率+附加收入×附加毛利率。", "boundary": "附加服务须独立定价；收入确认期限匹配。", "counterexample": "附加增长但交付成本或折扣吞噬利润。", "test": "客户层附加率、合同毛利、服务确认年限。", "question": "合成：硬件收入100、毛利率5%；服务附加收入20、毛利率40%；合并毛利率？", "constraint": "硬件收入翻倍为200，服务仍20，其余不变；毛利率？", "transfer": "迁移到云：计算收入80、毛利20%，数据库收入20、毛利60%；合并毛利率？", "tags": [ "F10", "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S77-01", "profile": "wamsi-mohan", "title": "出货放缓能否证明需求减弱？", "source_ids": [ "R2S3-DELLQ125", "R2S3-DELLQ225" ], "locator": "Dell Q1 FY2025 Earnings Call pp.9–10；Dell Q2 FY2025 Earnings Call p.14", "source_paraphrase": "区分主动降价抢份额与竞争应对，并拆解毛利压力。 / 追问AI积压不增是否反映Blackwell或液冷供给时点。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "期末积压=期初积压+订单−出货−取消。", "boundary": "管线不是已签订单；订单不是收入。", "counterexample": "供给解除而客户延期，积压仍增加。", "test": "订单取消、验收准备、部件交付和出货桥。", "question": "合成：期初积压40、订单30、出货25、取消5，期末多少？", "constraint": "供给可发40但客户仅接20，订单30取消5；期末多少？能推断需求更强吗？", "transfer": "迁移到晶圆：期初50万片订单，新增30、交付20、取消10；期末及交付瓶颈还需什么证据？", "tags": [ "F7", "F3", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S78-01", "profile": "david-vogt", "title": "EPS改善有多少只是分母缩小？", "source_ids": [ "R2S3-JBLQ424", "R2S3-JBLQ124" ], "locator": "Jabil Q4 FY2024 Earnings Call pp.20–21；Jabil Q1 FY2025 Earnings Call pp.13–14", "source_paraphrase": "分辨后半年EPS提升来自利用率、组合还是回购。 / 追问数据中心新增收入的利润率及并购取向。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "EPS=净利润/加权平均稀释股数。", "boundary": "回购时点和融资成本必须对齐。", "counterexample": "利润下降也可能因股份下降而EPS上升。", "test": "加权股数、营业利润桥、净利与净债。", "question": "合成：净利100、股数50；下期净利105、股数45，EPS增多少？", "constraint": "保持股数50时EPS是多少；实际EPS提升中有多少绝对额来自股数减少？", "transfer": "迁移到CPU：吞吐100、功耗50；新吞吐95、功耗40，性能/瓦改善可否说性能提升？", "tags": [ "F2", "F4", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S79-01", "profile": "steven-fox", "title": "现有产能能否支撑迁移订单？", "source_ids": [ "R2S3-JBLQ424", "R2S3-JBLQ124" ], "locator": "Jabil Q4 FY2024 Earnings Call pp.17–18；Jabil Q1 FY2025 Earnings Call pp.8–9", "source_paraphrase": "追问汽车需求变化的短期负杠杆及FCF改善桥。 / 区分现有美国产能承接与后续扩产能力。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "短期产出=min(订单,已认证有效产能)。", "boundary": "产能扩建与客户认证不能当即生效。", "counterexample": "全球有闲置厂房却不满足当地认证。", "test": "分厂余量、认证进度、投产日期及现金投入。", "question": "合成：美国有效产能100、已占90，新转入订单30，短期能额外承接多少？", "constraint": "扩建30但仅一半完成客户认证，总共可承接多少新增订单？", "transfer": "迁移到HBM：名义产能100，良率80%，已承诺70，新订单20，短期剩余可交付多少？", "tags": [ "F7", "F9", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S80-01", "profile": "george-wang", "title": "总增长掩盖了哪个子行业？", "source_ids": [ "R2S3-JBLQ424", "R2S3-JBLQ124" ], "locator": "Jabil Q4 FY2024 Earnings Call pp.15–16；Jabil Q1 FY2025 Earnings Call pp.10–11", "source_paraphrase": "拆解半导体设备增长并核对AI收入跨分部定义。 / 再问ATE/WFE节奏及硅光产品演进。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "总增速=上期收入权重加权的各分部增速。", "boundary": "AI标签可跨分部，不能重复相加。", "counterexample": "高增长小基数无法抵消大业务衰退。", "test": "按一致分部口径重述历史收入、剔除交叉标签。", "question": "合成：设备收入100，ATE占60并增长40%，WFE占40下降10%，总增长？", "constraint": "新增AI分类收入50其中30已含在ATE，若另加50会高估多少？", "transfer": "迁移到内存：HBM20增长100%，传统DRAM80下降20%，总体增长？", "tags": [ "F6", "F5", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S84-01", "profile": "ari-klein", "title": "预售增长是需求还是口径？", "source_ids": [ "R2S3-EQIXQ225", "R2S3-EQIXQ325" ], "locator": "Equinix Q2 2025 Earnings Call — public transcript mirror Aryeh Klein Q&A；Equinix Q3 2025 Earnings Call — public transcript mirror Ari Klein Q&A", "source_paraphrase": "问追加资本支出能否提前交付。 / 追问预售增长与销售窗口延长关系。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "可比预售需固定交付窗口。", "boundary": "长期预售不等于当期收入。", "counterexample": "扩大销售窗口带来订单增长而同窗口需求不变。", "test": "同交付窗口预售和实际上电日期。", "question": "合成：今年12个月窗口预售120，去年6个月窗口80；可否断言需求增长50%？", "constraint": "今年前6个月预售70，后6个月50；可比增速？", "transfer": "迁移到芯片：交期由3月延至6月，积压由30升至50；每月订单流量不变，能否确认需求扩张？", "tags": [ "F1", "F3", "F7" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S85-01", "profile": "eric-luebchow", "title": "更快出租是否改善投资回收？", "source_ids": [ "R2S3-EQIXQ225", "R2S3-EQIXQ325" ], "locator": "Equinix Q2 2025 Earnings Call — public transcript mirror Eric Luebchow Q&A；Equinix Q3 2025 Earnings Call — public transcript mirror Eric Luebchow Q&A", "source_paraphrase": "问稳定期能否缩短及预租降低风险。 / 问供给紧张定价与次年AFFO融资。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "稳定收益率=年稳定NOI/投入资本。", "boundary": "预租签约须转为付租；融资与项目收益分开。", "counterexample": "签约高但租户延期，现金回收仍后移。", "test": "免租期、投产时间、出租率和现金利息。", "question": "合成：投资100，满租年NOI12；假设NOI与全年平均出租率线性（忽略固定成本导致的非线性），首年平均出租50%、第二年100%，两年累计NOI？", "constraint": "首年预租80%但半数延至第二年起租，首年全年平均实际出租40%，第二年全年平均100%；沿用线性NOI假设，两年累计NOI？", "transfer": "迁移到GPU云：设备100、满负荷年贡献40；假设贡献随全年平均可计费利用率线性，首年该利用率25%，贡献多少？", "tags": [ "F9", "F14", "F11" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S88-01", "profile": "nick-del-deo", "title": "生态入口如何变成增量收益？", "source_ids": [ "R2S3-EQIXQ225", "R2S3-EQIXQ325" ], "locator": "Equinix Q2 2025 Earnings Call — public transcript mirror Nick Del Deo Q&A；Equinix Q3 2025 Earnings Call — public transcript mirror Nick Del Deo Q&A", "source_paraphrase": "询问互联新增锯齿波动及持续性。 / 问新云入口相对传统云的战略价值。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "互联收入=计费连接数×平均单价。", "boundary": "入口宣布不等于付费连接，新增不等于净增。", "counterexample": "新入口吸引用户却替代旧连接。", "test": "净连接、取消、付费率与同客户支出。", "question": "合成：期初1000连接，新建200、取消150，每条年费100；期末年化收入？", "constraint": "新建中100替代旧连接且已经包含在取消150中，是否再扣100？", "transfer": "迁移到AI API：新增调用200万，原产品少150万，同价每万10元，净年化收入增加？", "tags": [ "F3", "F10", "F25" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S90-01", "profile": "frank-louthan-iv", "title": "利息资本化改善了什么？", "source_ids": [ "R2S3-EQIXQ225", "R2S3-EQIXQ325" ], "locator": "Equinix Q2 2025 Earnings Call — public transcript mirror Frank Louthan Q&A；Equinix Q3 2025 Earnings Call — public transcript mirror Frank Louthan Q&A", "source_paraphrase": "追问资本化利息变化及历史处理。 / 问互联ARPU与电力密度、地区复制性。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "现金利息=费用化利息+资本化利息。", "boundary": "资本化改变确认时点，不免除现金支出。", "counterexample": "利润提升而当期现金流未改善。", "test": "项目支出、资本化比率、折旧年限和现金流分类。", "question": "合成：现金利息10，资本化6、费用化4；若资本化升至8但现金利息不变，当期税前利润提高多少？假定新增资本化当期尚未开始折旧。", "constraint": "新增资本化2对应资产，5年直线折旧无残值，之后每年新增折旧？", "transfer": "迁移到软件：研发现金20，资本化12而原来8；不计税与摊销，当期利润变化及现金支出变化？", "tags": [ "F11", "F2", "F1" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S95-01", "profile": "andrew-obin", "title": "高增长是否必然赢得份额？", "source_ids": [ "R2S3-ETNQ125", "R2S3-ETNQ225" ], "locator": "Eaton Q1 2025 Earnings Call — public transcript mirror Andrew Obin Q&A；Eaton Q2 2025 Earnings Call — public transcript mirror Andrew Obin本人自我确认后Q&A", "source_paraphrase": "问高基数下订单与电力需求。 / 问增长是否说明份额提升及数据中心产品布局。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "份额变动比率=(1+公司增长)/(1+市场增长)。", "boundary": "产品、地区和确认期间必须一致。", "counterexample": "公司并购或提前出货使增长不可比。", "test": "同口径有机销售与终端市场交付。", "question": "合成：公司销售100增长50%，市场1000增长30%，份额变动？", "constraint": "公司新增50中20来自并购，基准市场不含被购公司；可比公司收入130，份额怎样？", "transfer": "迁移到服务器：厂商出货增长20%，市场增长25%，相对份额变化？", "tags": [ "F5", "F6", "F1" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S98-01", "profile": "christopher-glynn", "title": "订单拐点何时进入收入？", "source_ids": [ "R2S3-AMEQ424", "R2S3-AMEQ225" ], "locator": "AMETEK Q4 2024 Earnings Call — public transcript mirror Christopher Glynn Q&A；AMETEK Q2 2025 Earnings Call — public transcript mirror Christopher Glynn Q&A", "source_paraphrase": "问Kern并购竞争定位及组合退出。 / 问航天并购优先级及自动化周期。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "收入期由订单、交付时滞和取消决定。", "boundary": "工业样本仅迁移分析方法，不证明其直接AI覆盖。", "counterexample": "订单上升但交付或认证延期。", "test": "订单cohort、确认政策、交付日期与取消。", "question": "合成：一月新订单120全部四个月后交付确认，二月新订单100同样时滞；五月、六月收入各多少？", "constraint": "一月订单25%取消且其余一半再延一月，五月与六月收入多少（其余不变）？", "transfer": "迁移到芯片：设计胜出100项目、20%流片、其中50%量产；量产项目多少？", "tags": [ "F3", "F14", "F7" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S73-01", "profile": "tal-liani", "title": "并购现金利润率是否被调整项美化？", "source_ids": [ "O8SC-CIENQ125", "R2S3-PANWQ126" ], "locator": "Ciena Q1 2025 Financial Results pp.10–12；Palo Alto Networks Q1 FY2026 — public transcript mirror Tal Liani Q&A", "source_paraphrase": "追问相干光应用区间及服务增速。 / 问并购过渡期利润稀释与协同时点。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "实际FCF=调整后FCF−被剔除现金支出。", "boundary": "调整后指标和真实现金流必须桥接。", "counterexample": "协同承诺实现而整合现金费仍高。", "test": "一次性支出、实现协同和现金利润率时间表。", "question": "合成：收入100，调整后FCF率37%，剔除整合现金费5，实际FCF及率？", "constraint": "次年收入110、调整后率40%，整合现金费8，实际FCF率相对首年改善多少个百分点？", "transfer": "迁移到光模块：收入100、毛利40，研发投入15；新增短距业务毛利10但额外研发12，经营贡献变化？", "tags": [ "F11", "F14", "F10" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S82-01", "profile": "david-guarino", "title": "电力押金改变了多少投入？", "source_ids": [ "R2S3-DLRQ225", "R2S3-EQIXQ325" ], "locator": "Digital Realty Q2 2025 Earnings Call — public transcript mirror David Guarino Q&A；Equinix Q3 2025 Earnings Call — public transcript mirror David Guarino Q&A", "source_paraphrase": "问公用事业预付款对成本和竞争影响。 / 追问预售与年化签约正常运行水平。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "经济投入包括设备资本和受限现金。", "boundary": "可退押金与费用不同但占用资金。", "counterexample": "报告建设成本不变而融资需求上升。", "test": "押金条款、退款日期、签约转起租。", "question": "合成：建设100，电力可退押金20，年NOI12；仅按建设及按全部占款收益率各多少？", "constraint": "押金三年后退回，不计利息；前3年累计NOI36加退回20，是否等于利润56？", "transfer": "迁移到GPU采购：设备100、预付款30已包含在设备价中，是否总投入130？", "tags": [ "F11", "F9", "F1" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S83-01", "profile": "jonathan-petersen", "title": "土地储备如何转成可用容量？", "source_ids": [ "R2S3-DLRQ225", "R2S3-EQIXQ325" ], "locator": "Digital Realty Q2 2025 Earnings Call — public transcript mirror 首位Jon Petersen Q&A；Equinix Q3 2025 Earnings Call — public transcript mirror Jonathan Petersen Q&A", "source_paraphrase": "问小容量增长来自市场还是份额及销售改变。 / 问新增土地分配零售与xScale。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "可建容量受土地、电力与用途许可共同约束。", "boundary": "拥有土地不等于可出租或已签约。", "counterexample": "名义MW大但电力和用途不匹配。", "test": "逐址电力承诺、许可、资本和回报。", "question": "合成：土地可建100MW，电力已落实60MW，许可只允许50MW，当前可建上限？", "constraint": "许可升至80MW，零售计划40MW加xScale计划40MW，电力仍60MW；缺口？", "transfer": "迁移到芯片市场：公司收入由10升12，市场100升120；份额提升多少？", "tags": [ "F7", "F5", "F9" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S86-01", "profile": "timothy-horan", "title": "一次性收费能否年化？", "source_ids": [ "R2S3-EQIXQ324", "R2S3-EQIXQ125" ], "locator": "Equinix Q3 2024 Earnings Call — public transcript mirror Timothy Horan Q&A；Equinix Q1 2025 Earnings Call — public transcript mirror Tim Horan Q&A", "source_paraphrase": "问xScale收益贡献何时进入运行水平。 / 问供给约束下提价和产能匹配。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "年化可持续收入仅包括可重复合同流。", "boundary": "建设服务收费和持续管理费不同。", "counterexample": "竣工费推高单季而未来回落。", "test": "合同续期、收费条款和项目交付。", "question": "合成：季度管理费10、一次性竣工费30，总40；可持续年化收入多少？", "constraint": "明年管理费每季12且无竣工费，全年收入相对今年四季管理费加一次竣工费变化？", "transfer": "迁移到云：年度订阅80、一次迁移费20；把季度末单月迁移费年化会产生什么错误？", "tags": [ "F14", "F1", "F10" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S87-01", "profile": "richard-choe", "title": "合资开发规模等于公司投入吗？", "source_ids": [ "R2S3-EQIXQ423", "R2S3-DLRQ225" ], "locator": "Equinix Q4 2023 Earnings Call — public transcript mirror Richard Choe Q&A；Digital Realty Q2 2025 Earnings Call — public transcript mirror Operator Richard Choe/转录Yong Choe，管理层称Richard", "source_paraphrase": "问竞争流失和价格变化。 / 问开发CapEx规模持续多久。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "归属股本承诺=项目所需股本总额×约定股本承担比例；债务担保另计。", "boundary": "承担担保和管理义务另计，不由少数股权自动消除。", "counterexample": "少数股权仍承担全部融资担保。", "test": "JV资本承诺、担保、资金来源及项目回报。", "question": "合成：项目需股本总投入100，公司承担其中20%，无额外担保，公司股本投入多少？", "constraint": "除上述100股本外另有60贷款，公司提供贷款50%担保；股本加担保敞口上限多少？该数等于已支出现金吗？", "transfer": "迁移到联合晶圆厂：公司承担30%股本、项目需股本总投入200，已支付40，还需支付多少股本？", "tags": [ "F9", "F11", "F7" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "R2S3-S89-01", "profile": "brendan-lynch", "title": "AI需求是否都必须放在核心城市？", "source_ids": [ "R2S3-EQIXQ122", "R2S3-EQIXBARC26" ], "locator": "Equinix Q1 2022 Earnings Call — public transcript mirror Brendan Lynch Q&A；Equinix at Barclays Financial Services Conference — public speaker transcript Speaker segments #12–15", "source_paraphrase": "问长期利润率目标与可控成本杠杆。 / 追问延迟敏感负载占需求的角色。", "attribution": "公开问题仅体现检验方向；下面模型、数字、约束及迁移题均为编辑合成，非原分析师观点或预测。", "model": "部署可行性同时受延迟预算和成本约束。", "boundary": "AI标签不能替代逐工作负载SLO。", "counterexample": "离用户更近却计算排队导致更高总延迟。", "test": "端到端P99、队列、网络和成本分解。", "question": "合成：核心城网络5ms、计算15ms；远端网络25ms、计算10ms。SLO30ms，哪处达标？", "constraint": "核心城拥塞使计算30ms，远端优化网络至15ms，计算仍10ms；哪处达标？", "transfer": "迁移到CPU：方案A内存延迟60ns、排队50ns；B延迟80ns、排队10ns；总预算100ns，选谁？", "tags": [ "F7", "F18", "F19" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G1-B04-01", "profile": "xuesong-zhao", "title": "数据独特性还能换来利润吗", "source_ids": [ "O9G1-B04-S1", "O9G1-B04-S2" ], "locator": "AI: We are only scraping the surface：2026-10-09 原发布者PDF第1、4–5页，末页正文明确本人署名与日期。；Artificial Intelligence Fund video transcript：2026-10-09 官方PDF下载转文本，实读第1–2页；日期来自文件名，正文无具体录制日。", "source_paraphrase": "把独有数据及非科技采用者纳入AI受益研究，考察技术如何改变成本和商业模式。 / 软件功能可被复制时重估原有数据优势；区分盈利修正与估值重估。", "attribution": "来源为本人具名论述；下列数字与三阶段练习为编辑合成，不是本人预测或原问答。", "model": "现金贡献=收入−现金成本；数据优势必须体现为留存、定价或成本差异。", "boundary": "仅提炼作者的可变假设；2023文中GPT-4参数等未经核验技术断言不作为事实采用。", "counterexample": "数据独有但买家可用较便宜替代品，护城河未必能兑现。", "test": "观察同队列留存、净价、实际现金节约，不能仅用AI宣传。", "question": "合成练习：数据独特性还能换来利润吗：某数据服务原收入100、现金运营成本60；AI竞品出现后客户保留80%，同价且客户均质，成本仍60。经营现金贡献变多少？", "constraint": "企业采用AI后成本减少12，但降价10%留住客户；客户仍为原来的80%，现金贡献多少？", "transfer": "另一工业企业收入1000、现金成本950；AI使现金成本降低1%，另需向软件供应商支付年费5。现金贡献净增多少？", "tags": [ "F14", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G1-B10-01", "profile": "mike-seidenberg", "title": "低配巨头的代价与保护", "source_ids": [ "O9G1-B10-S1", "O9G1-B10-S2" ], "locator": "I couldn’t sleep with full index weights to Magnificent Seven：2026-10-09 AIC公开转载Citywire原访谈摘段；已读本人完整回答，仅该问答，不冒充一小时全片。；371. Finding real value in tech investing：2026-10-09 原采访机构公开完整机器转录；实读mega-cap、AI、cyber和valuation段；2025-10-09录制，10-16发布；未核音频，转录有错误。", "source_paraphrase": "承认低配巨头可能拖累相对表现，但把股东承受的集中风险纳入组合选择。 / 在高增长、合理估值增长和低估值可选性间配置；网络安全需求持续不保证供应商不会过时。", "attribution": "来源为本人具名论述；下列数字与三阶段练习为编辑合成，不是本人预测或原问答。", "model": "组合收益=初始权重×各资产收益之和；低配产生机会成本与下行缓冲。", "boundary": "机器转录未作音频复核；组合情景仅为教学，不是经理风险预测。", "counterexample": "名称和风格不同的股票仍可能暴露于同一AI资本开支风险。", "test": "检查底层因子、单股权重及相对和绝对损失。", "question": "合成练习：低配巨头的代价与保护：基准中单股权重20%，基金权重10%，其余资产均零收益；该股上涨50%，基金和基准收益及相对差多少？", "constraint": "独立压力情景：该股跌40%、其余零收益；两个组合损失多少？能凭这一情景认定低配更优吗？", "transfer": "另一组合三类资产各占1/3，同期回报分别30%、10%、−10%，不调仓、无费用。组合收益多少？是否证明风险分散充分？", "tags": [ "F22", "F23", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G1-B18-01", "profile": "josh-spencer", "title": "云的弹性值多少钱", "source_ids": [ "O9G1-B18-S1", "O9G1-B18-S2" ], "locator": "Cloud computing: still in its early stages and spreading globally：2026-10-09 官方PDF第1–3页实质正文，本人具名；PDF封面2017-09，非搜索抓取年份。；Semiconductor stocks: buying when the chips are down：2026-10-09 官方PDF第1–3页，本人具名，2018-11；历史判断未作事后成功认证。", "source_paraphrase": "将客户成本、扩容弹性与SaaS稳态利润率结合，同时保留估值约束。 / 用供给集中度、扩产资本强度和终端内容增长检验周期底部利润，而非只外推峰值。", "attribution": "来源为本人具名论述；下列数字与三阶段练习为编辑合成，不是本人预测或原问答。", "model": "固定成本与按量收费的交叉点；营业利润=量×(价−单位变动成本)−固定成本。", "boundary": "历史技术基金方法样本，不把2017/2018职务冒充现职；原文看多不等于事后验证。", "counterexample": "单位需求增长可能被价格和成本杠杆吞没。", "test": "同服务水平比较全成本，分别审计需求、价格、供给与利用率。", "question": "合成练习：云的弹性值多少钱：某年自建容量成本固定100，最多服务100单位；云同等服务每实际单位收费1.5，无其他费用。实际需求40，选哪种更便宜，利用率交叉点多少？", "constraint": "需求变为90，且自建容量仍足够；成本如何？所以云必然更省钱吗？", "transfer": "迁移到芯片供应商：原卖100单位、售价2、单位变动成本1、固定成本50。量增20%但售价降25%，固定和单位变动成本不变，营业利润如何变化？", "tags": [ "F10", "F8", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G1-B25-01", "profile": "steve-wymer", "title": "增长持续期值多少", "source_ids": [ "O9G1-B25-S1", "O9G1-B25-S2" ], "locator": "Fidelity Growth Company Fund PM Q&A — May 2018：2026-10-09 Fidelity原问答公开第三方全文镜像，第2–3页正文与经理姓名可见；未取得2018官方存档逐字校验。日期是报告截至日。；Fidelity Growth Company Fund PM Q&A — May 2026：2026-10-09 官方滚动PDF第2–4页，实际头部为2026-05-31；截至日不等于发布日；未复制来源未独立核验的业绩或IPO数字。", "source_paraphrase": "看三至五年增长机会和商业模式延展性，同时因组合构建调整高确信持仓。 / 把持续AI采用与企业投资回报及基建约束相连；增长持仓仍需管理集中度。", "attribution": "来源为本人具名论述；下列数字与三阶段练习为编辑合成，不是本人预测或原问答。", "model": "有限期现金流现值；增长持续期必须扣除维持它的现金投入。", "boundary": "2018为未逐字核验的公开镜像，2026为滚动官方文档；日期是报告期间终点。", "counterexample": "增长年限更长但再投资需求更大，股东可得现金反而更低。", "test": "独立估计增量资本需求、现金转化及持仓集中度。", "question": "合成练习：增长持续期值多少：某企业未来第1–3年每年末自由现金流10，第4–5年为零，无残值，贴现率10%；另一情景是第1–5年每年10。两种现值与差额多少？", "constraint": "五年情景的新证据发现，原估每年自由现金流10遗漏了每年维持增长所必需的现金投入4；重新算现值，是否仍优于三年情景？", "transfer": "另一仓位初始占组合10%，股价翻倍，其余持仓不变；新权重多少？长期看好能否替代集中风险管理？", "tags": [ "F21", "F10", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G1-B30-01", "profile": "alex-sacerdote", "title": "S曲线陡峭就等于盈利增长吗", "source_ids": [ "O9G1-B30-S1", "O9G1-B30-S2" ], "locator": "Graham & Doddsville Issue XXV — Alex Sacerdote：2026-10-09 Columbia原始刊物，2015年秋刊，第6–12页访谈；重点读第7–9页S曲线和退出案例。；Invest Like the Best EP477 — Investing on the S-Curve：2026-10-09 公开第三方机器逐字稿，实读19:23–29:40；官方Colossus节目页核对身份、节目号和日期，官方完整稿需登录，未访问。未核音频；明显转录错误不作为事实。", "source_paraphrase": "把采用拐点与竞争优势、未被价格反映的盈利能力联合考察；市场增长不保证每家厂商获利。 / 区分采用速度与市场上限，并在渗透路径偏离时修正假设；企业整合难度也影响采用速度。", "attribution": "来源为本人具名论述；下列数字与三阶段练习为编辑合成，不是本人预测或原问答。", "model": "企业收入=可达市场×采用率×企业份额×单位收入；曲线高度与份额不可混同。", "boundary": "2026第三方机器稿未核音频；只取清楚的框架段，不将转录中的具体估值、公司历史或投资业绩当作已核事实。", "counterexample": "采用高速增长时，价格下跌或份额流失仍会减少公司收入。", "test": "分别跟踪可达市场、队列采用、企业份额、单位价格与现金利润。", "question": "合成练习：S曲线陡峭就等于盈利增长吗：固定潜在市场1000客户，采用率从10%升至20%，企业份额50%不变，每客户收入2。企业收入多少、增长多少？", "constraint": "下一阶段采用率升到30%，企业份额降到30%，每客户收入降为1.8；收入相比上一阶段如何？", "transfer": "另一服务潜在市场原估1000、实际客户100；新证据把可达市场降至400。渗透率如何修正？剩余未采用客户数改变多少？", "tags": [ "F5", "F16", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G1-B13-01", "profile": "shekhar-pramanick", "title": "订单强是否代表终端需求强", "source_ids": [ "O9G1-B13-S1", "O9G1-B13-S2" ], "locator": "TSMC Q2 2004 Earnings Conference Call — Pramanick Q&A：2026-10-09 TSMC IR官方承载的Thomson完整逐字稿，下载后读第7–8页本人具名提问；2004-07-29为会议日；当时Schwab Soundview。管理层回答不归给提问人。；Investing in the metaverse: 13 stocks to buy for 2022 — attributed interview portions：2026-10-09 Fortune原采访正文The infrastructure bet节，仅使用Pramanick具名段；其他受访者观点排除。本人为Columbia Seligman买方经理。", "source_paraphrase": "卖方时期追问出货指引是否已考虑库存风险，并追问收入增长能否转成毛利率改善。 / 考察芯片需求与公司盈利。", "attribution": "来源为本人具名论述；下列数字与三阶段练习为编辑合成，不是本人预测或原问答。", "model": "库存期末=期初+进货−终端销售；利润=出货×单位贡献−固定成本。", "boundary": "跨角色的有限方法样本：一份2004卖方问答、一份2022买方采访；不声称两份均为买方或论证跨时期稳定。", "counterexample": "渠道补库可使供货增长先于真实终端需求，去库存又会放大回落。", "test": "分开终端销售、渠道库存、供应商出货、价格和确认时点。", "question": "合成练习：订单强是否代表终端需求强：某渠道期初库存20、本期进货140、终端销售120；期末库存多少？若下期终端销售100、目标期末库存10，下期应向供应商进货多少？", "constraint": "供应商将进货量视为其本期出货且同季确认收入，无其他客户；上一期140单位每单位价格2、变动成本1，固定成本40。下一期按第一阶段算出的70单位、价格成本不变；收入和营业利润各变化多少？", "transfer": "设备公司新增订单80、当期交付100、期初积压订单200，无取消。期末积压多少？能否仅凭当期交付增加就确认终端扩产需求在加速？", "tags": [ "F7", "F4", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G2-B34-01", "profile": "stephen-mandel", "source_ids": [ "O9G2-M21", "O9G2-M25" ], "locator": "2026-10-09 公开完整转录已读00:09–00:22投资方法段；节目原页核日期与人物；自动转录存在姓名错字。；2026-10-09 公开完整转录已读Valuation Frameworks and Non-Linear Thinking等章节；镜像发布日期2025-10-21，节目日期由Colossus原页核为2025-09-15。", "source_paraphrase": "从可复制的单位经济与管理质量理解业务安全边际；由具体公司发现多年变迁，而非从主题标签倒推股票。；以自由现金流收益率、再投资回报与融资依赖共同看估值；高回报扩张导致的低FCF不能直接视为劣质，AI赢家也可能因买价过高给投资者低回报。", "attribution": "上述论证归于具名材料中的本人；所有模型数值与三阶段情景为编辑合成，非本人预测或原问答。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计实际预测或投资收益", "title": "自由现金流低，是增长投资还是价值消耗？", "model": "当期FCF=经营现金流−维护资本开支−扩张资本开支；扩张NPV=新增稳定年现金流/r−初始投资。", "boundary": "假设新增税后现金流从一年后永久持续、无残值或追加投资；永续简化不能替代真实门店寿命模型。", "counterexample": "高FCF可能只是停止有价值扩张；高扩张回报也可能因竞争而消失。", "test": "对新老门店分队列核现金回报、维护支出与资本成本。", "question": "合成：公司年经营现金流120，维护开支20，扩张开支80，市值1000。扩张项目从次年起每年末新增税后、扣维护后的净现金流12且永续；资本成本10%。计算当期FCF收益率和该扩张NPV。", "constraint": "竞争使新增年现金流降至6，其他不变；NPV与当期FCF收益率各如何变化？", "transfer": "数据中心初始投资200，从次年起每年末现金流30永久持续，同期每年末维护投入10。折现率12%。NPV多少？为何不能用30/200直接判断价值？", "tags": [ "F10", "F20", "F21" ] }, { "id": "O9G2-B35-01", "profile": "andreas-halvorsen", "source_ids": [ "O9G2-H09", "O9G2-H10" ], "locator": "2026-10-09 公开信件转录正文已读页1–3，抬头明确From: O. Andreas Halvorsen、January 9, 2009；附件Dave Ott邮件不是Halvorsen个人原话；未认证上传者身份或官方字节一致。；2026-10-09 公开长篇信件主文转载已读开头From: O. Andreas Halvorsen及集中度、做空激励、总敞口段；原信无可确认具体写信日，date用镜像公布日，报告期2010Q2；未取得原附件及官方原件字节核验。", "source_paraphrase": "把多空选股价差、净敞口与总敞口分开；降低总杠杆可能牺牲账面收益，却保留融资安全与危机中继续研究交易的能力。；检视高确信度持仓、分散度和多空激励；多头与空头的流动性及损失不对称，不能按同样集中规则处理。", "attribution": "上述论证归于具名材料中的本人；所有模型数值与三阶段情景为编辑合成，非本人预测或原问答。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计实际预测或投资收益", "title": "净敞口不变，总敞口翻倍的代价", "model": "组合贡献=L×rL−S×rS；gross=L+S；net=L−S；L、S为相对期初净资产的多空名义比例。", "boundary": "单期持仓不再平衡；忽略利息、借券费及交易成本；净敞口并不等于真实beta。", "counterexample": "低净敞口仍可因多空价差反向或空头挤压出现大亏损。", "test": "压力测试多空各自收益、借券可用性、追加保证金与变现速度。", "question": "合成：期初净资产100，多头80、空头60。多头涨5%，被做空股票涨2%。求总/净敞口和净资产收益。", "constraint": "独立压力情景，期初净资产仍为100：维持净敞口20%，把总敞口提高至200%。求多空名义头寸；若多头跌10%、被做空股票涨10%，损失多少？", "transfer": "另一组合多头60、空头40，仅被做空的一只股票上涨100%，该股票空头名义为净资产的5%，其他股票价格不动。不计费用的净资产损失是多少？净敞口20%为何不能阻止该损失？", "tags": [ "F22", "F14", "F23" ] }, { "id": "O9G2-B36-01", "profile": "dan-sundheim", "source_ids": [ "O9G2-D25", "O9G2-D26" ], "locator": "2026-10-09 节目官方完整逐字稿，已读00:07:02–00:11:44的IRR、退出估值及Netflix投资段。；2026-10-09 公开完整转录已读约00:12–00:20资本回报、Netflix/Spotify类比与业务聚焦段；动态广告造成时间戳与官方show notes略有差别，未逐秒核音频。", "source_paraphrase": "三年预期回报依赖退出倍数；股价提前兑现会降低未来IRR，但过早卖出长期优秀企业同样有机会成本。；把LLM训练视为前置固定投资，并检验采用速度、后续资本供给与回报；Netflix式成本摊薄与Spotify式个性化黏性是类比，模型商品化及延迟采用可能破坏回报。", "attribution": "上述论证归于具名材料中的本人；所有模型数值与三阶段情景为编辑合成，非本人预测或原问答。", "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计实际预测或投资收益", "title": "训练资本的回报经得起采用延迟吗？", "model": "NPV=−初始训练投入+Σ逐年净现金流/(1+r)^t；延迟采用减少现金回收现值。", "boundary": "现金流已扣推理、运营、税费和维护；三年后无残值，无新一轮训练投入；仅为简化项目现金流模型。", "counterexample": "用户增长未必覆盖持续模型训练和推理支出；采用放缓可能早于技术路线失败耗尽融资。", "test": "核付费采用、续费、每任务成本、下一代训练预算与现金期限。", "question": "合成：模型项目现在投入100，未来第1、2、3年末净现金流各50，折现率10%，三年后无残值。NPV是多少？", "constraint": "采用晚一年，原三笔50依次推迟到第2、3、4年末，且第1年需额外支付净维持费用10。NPV多少，结论改变吗？", "transfer": "零售公司现在股价100，三年后预期售价172.8，无分红。一年后股价先升到144，但第三年售价预期不变。起初三年年化回报与之后两年的年化预期回报各是多少？", "tags": [ "F9", "F10", "F21" ] }, { "id": "O9G2-B32-01", "profile": "scott-shleifer", "title": "渗透率翻倍之前，公司能撑多久？", "source_ids": [ "O9G2-SH22", "O9G2-SH23" ], "locator": "FII完整转录的渗透率与成本段；2023原报道本人直接引语选段（正文119–125行），非完整通话。", "source_paraphrase": "2022材料将低渗透市场、增量利润和融资依赖相连；2023选段提示增长需转为利润。", "attribution": "第一来源为本人完整访谈转录，第二仅本人被原报道摘录的窄方法样本；下述数字、模型及案例全部编辑合成。", "model": "收入=潜在客户×采用率×份额×单位收入；现金经营贡献=收入×单位贡献率−固定现金成本；续航=现金/年烧钱。", "boundary": "无税、资本开支、营运资金变化或融资；现金贡献率已扣随收入变化的成本；续航只在烧钱速率恒定下有效。", "counterexample": "市场采用上升可能被价格下降抵消；业务增长不保证在现金耗尽前达到盈亏平衡。", "test": "分别追踪渗透率、份额、净价、单位贡献及每季度现金余额。", "question": "合成：潜在1000客户，采用率10%，公司份额50%，每客户年收入2，变动成本率40%，固定现金成本80。现金余额30。求年现金经营贡献、盈亏平衡采用率及恒定烧钱下的续航年数。", "constraint": "独立情景：采用率提高到20%，份额仍50%，但每客户年收入下降到1.2，成本率和固定成本不变。年现金贡献及盈亏平衡采用率多少？", "transfer": "新领域一企业年收入300，贡献率30%，固定现金成本120，手头现金60，收入成本维持不变。它有多少年现金续航？只有市场增长预测能否证明无需再融资？", "tags": [ "F10", "F16", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未认证来源基金业绩或预测结果" }, { "id": "O9G3-B38-01", "profile": "dan-niles", "title": "盈利改善如何与估值和仓位上限联动？", "source_ids": [ "O9G3-N1", "O9G3-N2" ], "locator": "O9G3-N1：NILES具名段，约18:25前后；网页L203–246；O9G3-N2：NILES具名段L103–119，19:20前后", "source_paraphrase": "Niles从汽车毛利、交付指引过渡至估值与小仓位风险控制。；Niles讨论政策持续时间、市场准入及科技企业的跨境风险。", "attribution": "编辑根据已注明归属的公开材料编写；所有数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "价格=EPS×PE；组合损失=仓位×个股损失。", "boundary": "两次不同情景的公开发言；政策判断是当时观点，不是已证因果。", "counterexample": "利润上涨可被估值收缩抵消；监管改变也未必按市场叙事影响最终现金流。", "test": "下一财报检查可比EPS、估值基期、地域成本和价格转嫁；组合须补相关性与流动性。", "question": "合成：EPS由2增至2.4，PE由50降至35。股价从多少变为多少，回报多少？", "constraint": "该股原占组合8%，其余资产价格不变。该情景拖累组合多少？若另设该股下跌40%的压力情景，组合最大容许损失2%，仓位上限多少？", "transfer": "迁移进口商：销售100，成本70，另有固定经营费10；成本中30需进口，关税增加20%，无转嫁或销量变化。经营利润及降幅？", "tags": [ "F12", "F19", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "O9G3-B40-01", "profile": "eric-jackson", "title": "困境平台的价值能否熬过现金期限？", "source_ids": [ "O9G3-J1", "O9G3-J2" ], "locator": "O9G3-J1：首筛、不对称、Carvana及Opendoor四项问答L56–103；O9G3-J2：Full transcript内Eric Jackson video submission段L165–171", "source_paraphrase": "Jackson以生存能力、资产与执行的区别检验困境平台的上行空间。；Jackson预录问题追问人员规模及iBuying之外的附加服务收入路径。", "attribution": "编辑根据已注明归属的公开材料编写；所有数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "现金期限=可用现金/期间现金净流出；融资稀释后的每股价值=股权价值/总股数。", "boundary": "公开持仓倡议与预录问题；管理层服务收入回答不归给Jackson；没有审核其百倍收益说法。", "counterexample": "资产优秀但融资前已耗尽现金，原股东仍可能被严重稀释或损失全部本金。", "test": "未来两个季度核经营现金流、融资条款、服务付费率和固定成本，不用社交热度验证经营。", "question": "合成：平台可用现金60，每季现金净流出15，商业转型需6季，无其他现金来源。还能存续几季，资金缺口多少？", "constraint": "为跨过6季，公司以每股3增发10股筹资30；原20股。假定完成后总股权价值150，稀释后每股价值多少？不能直接使用150/20的原因？", "transfer": "迁移云软件：客户10万家、附加服务付费率20%、每年每家收入1000元、服务毛利率60%。付费率升至30%但单价降20%，其他不变，新毛利及增幅？", "tags": [ "F10", "F20", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "O9G3-B45-01", "profile": "paul-meeks", "title": "AI资本开支如何转成供应商可兑现收入？", "source_ids": [ "O9G3-M1", "O9G3-M2" ], "locator": "O9G3-M1：Paul Meeks 01:35、04:01、06:16、08:20；检索精确标题；O9G3-M2：Paul Meeks 00:36、01:50、05:25、07:01；Lisa Su 03:53插播不归Meeks", "source_paraphrase": "Meeks区分AI资本开支支持的业务判断与财报期暂停增仓的时点决策。；Meeks将消费与企业需求分开，要求随后财报确认AI支出，并区分基建投入与应用变现。", "attribution": "编辑根据已注明归属的公开材料编写；所有数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "供应商收入=同范围资本开支×设备占比×供应商份额×当期兑现比例。", "boundary": "两次采访属历史买方时期；现任卖方另记。搜索返回带时间戳正文，直连429；Lisa Su插播和主持人提问不归Meeks。", "counterexample": "宣布投入增长但转向建筑、内部芯片或晚交付，外部供应商收入并不同步。", "test": "下两个财报季核同范围CapEx、设备结构、订单及实际交付；应用收入不能由基建投入直接推出。", "question": "合成：同一客户组AI资本开支100，设备占60%，供应商份额25%，均当期兑现。供应商收入多少？次年总额120、设备占比50%、份额不变，收入增长多少？", "constraint": "次年设备部分仅80%当期交付确认，余下延至以后。次年收入多少，相对原年变化多少？", "transfer": "迁移医疗：诊疗量增20%，每次实际收费降10%，净利润率由20%降至15%，其他不变。净利润相对原年变化？", "tags": [ "F6", "F9", "F23" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "O9G3-B49-01", "profile": "alex-ely", "title": "长期替代趋势是否足够补偿估值风险？", "source_ids": [ "O9G3-E1", "O9G3-E2" ], "locator": "O9G3-E1：投资流程及数字银行段L181–202；个股列表不复述；O9G3-E2：流程及期限段L182–204", "source_paraphrase": "两份采访的简短观点摘要见来源记录；以下是编辑独立设计的估值与份额反例。", "attribution": "编辑根据已注明归属的公开材料编写；所有数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "营收=行业市场×渗透率×厂商份额；价格=EPS×PE。", "boundary": "原发布方两篇采访报道；Ely个人明确引语和记者转述分开；不复制个股清单或未核收益，非现任确认。", "counterexample": "行业替代持续，但竞争稀释份额、价格下降或估值压缩，投资仍可亏损。", "test": "未来两个报告期观察渗透率、公司份额与单位价值；检验行业增长是否被该公司现金利润捕获。", "question": "合成：行业规模1000，数字化渗透率20%，供应商占数字化市场10%。渗透率增至30%，但该商份额跌至7%，行业规模不变。该商营收增幅多少？", "constraint": "假定该商EPS与营收同比变化，原EPS2、PE40，随后PE25。每股价格及回报？", "transfer": "迁移电动车零部件：车市100万辆、电动车占比由20%增至40%，公司供货份额30%不变，但单车价值从1000降至400元。公司收入如何变化？", "tags": [ "F12", "F15", "F16" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "O9G3-B54-01", "profile": "james-anderson", "title": "少数大赢家与无法量化的不确定性如何共存？", "source_ids": [ "O9G3-A1", "O9G3-A2" ], "locator": "O9G3-A1：JA发言L104、111、114、124；O9G3-A2：Company and return concentration及结尾L119–139", "source_paraphrase": "Anderson质疑风险可被波动率完全度量，提出少数公司的上行偏斜及长期技术机会问题。；Anderson署名文章讨论少数公司主导回报、规模收益递增及低估Microsoft的反思。", "attribution": "编辑根据已注明归属的公开材料编写；所有数字与三阶段问题均为合成教学案例，不是原作者预测或真实问答。", "model": "终值=各持仓初值×终值倍数之和；路径流动性与终值回报分开。", "boundary": "2021仅JA发言归Anderson，John Kay的回答保留其本人归属；日期只知月份；2022自我反思不等于完整收益审计。", "counterexample": "极端赢家分布只是事后描述，盲目集中或现金耗尽可能错失正偏回报。", "test": "逐期追踪研发／商业里程碑及融资余量，记录失败样本与机会成本；不以股价波动单独判定永久损失。", "question": "合成：10家公司各投1，五年后其中1家变为15，另9家各剩0.5，无分红。组合终值及累计回报？", "constraint": "若事先过度集中，把5投给最终失败的某家、其余9家平分剩余5；各家终值倍数仍是前述15和0.5。组合终值及回报？这证明能提前选中赢家吗？", "transfer": "迁移研发项目：预算100，必须预留次年运营现金40；10个项目每个成本10，成功回收30、失败0，但成功概率未知。最多立项几个？能从30的回收值计算可信期望收益吗？", "tags": [ "F14", "F16", "F22" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计投资结果" }, { "id": "O9G4-B57-C1", "profile": "o9g4-b57", "title": "现金能买来多少适应时间？", "source_ids": [ "O9G4-B57-A", "O9G4-B57-B" ], "locator": "PDF第1–2页，MP回答财务韧性和适应能力；PDF第1–3页图像，MP回答持仓增长、自由现金流及规模效应", "source_paraphrase": "Pye把现金储备与逆周期研发及人才投入的能力相联系。；Pye用增量收入成本与现金生成能力解释规模化，提醒价格与经营变化可以不同步。", "attribution": "上述为公开材料转述；下列模型、数字、反例和三个阶段均为编辑合成，非本人原话、真实案例或投资建议。", "model": "跑道=可用现金/年度净现金消耗；先扣短期确定偿债。", "boundary": "假设现金消耗恒定，收入改善未计入，现金必须可动用。", "counterexample": "净现金充裕但监管冻结或研发失败仍会破坏适应能力。", "test": "未来四季跟踪现金消耗与研发里程碑，若跑道短于验证周期则修正。", "question": "合成：现金60亿元，未来一年到期债务12亿元；每年经营净现金消耗12亿元。先偿债后可支撑几年？", "constraint": "新增研发每年消耗6亿元，其他不变。研发验证需3年，现金够吗？", "transfer": "迁移到工厂：可用现金30亿元，每年经营净流出5亿元且每年维持性资本支出5亿元，回报验证需4年。资金缺口多少？", "tags": [ "F10", "F11", "F14", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计真实预测与投资回报" }, { "id": "O9G4-B61-C1", "profile": "o9g4-b61", "title": "研发亏损还是竞争力衰退？", "source_ids": [ "O9G4-B61-A", "O9G4-B61-B" ], "locator": "投资与研究两节：规模、卖出条件、亏损原因、产业演进；寻找好公司一节；五因素、产业链研究和分散", "source_paraphrase": "冯明远区分研发投入和竞争力衰退造成的亏损，并讨论规模增长对纠错和流动性的约束。；冯明远把企业、行业、财务及动态估值结合，解释技术迭代下分散持股的用途；李淑彦周期评论不归于冯。", "attribution": "上述为公开材料转述；下列模型、数字、反例和三个阶段均为编辑合成，非本人原话、真实案例或投资建议。", "model": "经营利润=收入×毛利率−现金经营费用（题设无折旧差异）；亏损拆解须保留增长投入。", "boundary": "研发回报未知，不因加回研发得到正数就视其无成本。", "counterexample": "所谓研发投入实际是无效开发，或竞争导致持续价格侵蚀。", "test": "下两季核对毛利、留存及研发里程碑；毛利继续降且无交付改进则推翻暂时亏损解释。", "question": "合成：收入100亿元、毛利率40%，其他经营费用25亿元、研发20亿元。经营利润多少？若研发中10亿元属新增项目，机械剔除后是多少，能否据此证明优质？", "constraint": "下期收入100亿元，毛利率降至25%，研发仍20、其他费用25。亏损变化来自哪里？", "transfer": "迁移到连锁店：收入80亿元，毛利率30%，正常费用18亿元，新店开办费8亿元。报表经营结果和剔除开办费结果分别是多少？", "tags": [ "F2", "F11", "F14", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计真实预测与投资回报" }, { "id": "O9G4-B62-C1", "profile": "o9g4-b62", "title": "需求增长会留给供应商多少利润？", "source_ids": [ "O9G4-B62-A", "O9G4-B62-B" ], "locator": "PDF第6页经理表；第7–8页§4.4；原报告第11–12页经理表、第14页§4.4和§4.5", "source_paraphrase": "管理人用壁垒、可落地需求和供给约束筛选，估值上升后更严格检查供需。；管理人将AI需求与受限供给、价格弹性及企业利润相接，商业闭环是当时预期而非已实现事实。", "attribution": "上述为公开材料转述；下列模型、数字、反例和三个阶段均为编辑合成，非本人原话、真实案例或投资建议。", "model": "经营利润=销量×(售价−单位变动成本)−固定成本。", "boundary": "销量必须为交付验收量；题设其他费用及税忽略，不能称净利润。", "counterexample": "新增供给让量增与降价同时发生，需求景气仍不足维持利润。", "test": "未来两季比较新增产能、合格交付、ASP与客户库存；ASP跌穿保本点需修改。", "question": "合成：供应商年售100万件、单价100元、单位变动成本60元、固定成本3000万元。经营利润多少？", "constraint": "需求增加30%，全量可售；竞争使售价85元，成本仍60元，固定成本仍3000万元。利润多少？保持原利润所需售价是多少？", "transfer": "迁移到酒店：房晚10万、房价500元、变动成本200元、固定成本2000万元；房晚增20%而房价降至400元。利润怎样变？", "tags": [ "F7", "F10", "F14", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计真实预测与投资回报" }, { "id": "O9G4-B64-C1", "profile": "o9g4-b64", "title": "终端AI渗透怎样变成应用收入？", "source_ids": [ "O9G4-B64-A", "O9G4-B64-B" ], "locator": "印刷第13页经理表、第15–16页§4.4和§4.5；印刷第13页经理表、第16–17页§4.4和§4.5", "source_paraphrase": "管理人提出终端AI渗透向企业应用传导的假设，并在成长性和估值之间筛选。；管理人把模型、产业数据与应用场景结合视为机会，同时保留成长和低估值两个筛选方向。", "attribution": "上述为公开材料转述；下列模型、数字、反例和三个阶段均为编辑合成，非本人原话、真实案例或投资建议。", "model": "收入=可触达终端数×真实活跃率×付费率×年ARPU。", "boundary": "各率分母须一致，付费率以活跃用户为分母；同一人多设备须去重。", "counterexample": "设备增长来自换机或同人多端，应用留存和支付意愿下降。", "test": "未来12个月跟踪去重活跃、付费留存及获客回收；仅设备出货增长不通过验证。", "question": "合成：1000万台可触达设备，各对应一个独立用户；活跃率40%，其中付费率10%，年ARPU 200元。收入多少？", "constraint": "设备增至2000万且仍独立用户，活跃率降至25%，付费率降至6%，ARPU不变。与原收入比怎样？", "transfer": "迁移到企业软件：1000家试用企业，60%实际部署，其中20%付费，每年收费10万元。收入及付费企业数多少？", "tags": [ "F4", "F10", "F14", "T3" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计真实预测与投资回报" }, { "id": "O9G4-B65-C1", "profile": "o9g4-b65", "title": "投入前置能否支持今天的价格？", "source_ids": [ "O9G4-B65-A", "O9G4-B65-B" ], "locator": "投资框架问答：四类资产、现金流、DCF、组合管理；周克平回答科创板结构因素、制度流动性、盈利估值三问", "source_paraphrase": "周克平按现金流和企业演化区分项目、产品、平台和生态，并强调前瞻自由现金流。；周克平区分周期扰动和结构增长，并用前置投入、后置利润解释估值判断；其他八位回答排除。", "attribution": "上述为公开材料转述；下列模型、数字、反例和三个阶段均为编辑合成，非本人原话、真实案例或投资建议。", "model": "项目NPV=未来现金流折现值−初始投资，情景概率需显式。", "boundary": "题设无终值和其他现金流；经营价值不自动等于每股价格。", "counterexample": "将所有亏损包装成未来平台化，成功概率未经验证。", "test": "两年内以付费留存、交付和再融资条件更新概率；验证失败即降值。", "question": "合成项目现在投100亿元，第1年现金流0，第2年末回收总现金144亿元，此后无价值；折现率20%。NPV多少？", "constraint": "只有60%概率回收144，40%概率回收20，两情景均在第2年末；概率加权NPV多少？", "transfer": "迁移到收费桥：现在投资60亿元，一年后确定回收69亿元后结束，折现率10%。NPV多少？为什么不能用科技项目概率直接套桥？", "tags": [ "F12", "F14", "F21", "T6" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计真实预测与投资回报" }, { "id": "O9G4-B67-C1", "profile": "o9g4-b67", "title": "价格上涨来自升级还是短缺？", "source_ids": [ "O9G4-B67-A", "O9G4-B67-B" ], "locator": "第10页经理表、第12–13页§4.4和§4.5；§4.1.2经理表，§4.4.1与§4.5（原报告第10、12–13页）", "source_paraphrase": "管理人把海外与国内算力投资节奏作比较，同时以租金优势解释线下折扣业态机会。；管理人将量价变化拆成产品材料升级与供不应求，讨论PCB产业链供给和竞争约束。", "attribution": "上述为公开材料转述；下列模型、数字、反例和三个阶段均为编辑合成，非本人原话、真实案例或投资建议。", "model": "ASP=各产品价格×销量占比之和；经营利润=收入−变动成本−固定成本。", "boundary": "不同产品单位可比较，销量与成本同期间；竞争可能先压旧品或新品。", "counterexample": "行业ASP上升完全来自产品组合，而同产品价格下跌。", "test": "下一季度核对同型号价格、升级占比、客户认证与竞争产能；拆开mix与同品涨价。", "question": "合成：旧品80万件、价100元，新品20万件、价150元；单位成本各60与90元，固定成本3000万元。算ASP及经营利润。", "constraint": "总量100万不变，新品占50%；旧品价降至90、新品价降至135，成本不变。ASP虽升，利润怎样变化？", "transfer": "迁移到餐厅：普通餐80份×50元、高端餐20份×100元；变为各50份，价格各下降10%。收入分别多少？能否称同品涨价？", "tags": [ "F2", "F4", "F7", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未审计真实预测与投资回报" }, { "id": "O9G5-B68-C1", "profile": "o9g5-b68", "title": "行业销量复苏会带来利润复苏吗？", "source_ids": [ "O9G5-B68A", "O9G5-B68B" ], "locator": "§4.1.2经理表；§4.4.1及§4.5，印刷页14–18；§4.1，页6–7；§4.4，页8", "source_paraphrase": "报告复盘新能源估值切换未兑现及错过调整窗口，区分终端增长和产能过剩导致的增量不增利。；报告区分锂电竞争格局、光伏控产价格及风电需求，使用产能利用率与盈利优势判断出清。", "attribution": "来源为材料观点的编辑转述；数字、问题、模型、反例和迁移全部为编辑合成，不是本人原话或实际投资案例。", "model": "毛利=销量×(单价−单位成本)。", "boundary": "同口径单位、期间；只算毛利，不等于净利润或现金流。", "counterexample": "销量提高但售价下跌超过降本。", "test": "下一季同时核销量、成交价、单位成本和闲置产能，不只查订单。", "question": "合成：销量100万件，售价100元，单位成本80元。下一年销量120万件，售价90元，单位成本75元。计算两年毛利。", "constraint": "若单位成本降到72元，毛利如何变化？在120万销量、90元售价下，保住原毛利要求成本不高于多少？", "transfer": "迁移：云服务调用量从100万到150万次，每次价格从1元降至0.7元，成本从0.6元降至0.5元。计算贡献额变化。", "tags": [ "F7", "F8", "F10" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "O9G5-B69-C1", "profile": "o9g5-b69", "title": "从算力主题切换到应用，哪一步产生收入？", "source_ids": [ "O9G5-B69A", "O9G5-B69B" ], "locator": "§4.1.2及§4.4.1，页11–13；§4.1.2页10；§4.4.1页11", "source_paraphrase": "报告说明从国产算力到AI终端、智能驾驶及应用布局的变化，将行业趋势与预期落地分开。；报告解释开源模型推动应用布局，并将关税敏感性、军工业绩与事件催化共同用于配置判断。", "attribution": "来源为材料观点的编辑转述；数字、问题、模型、反例和迁移全部为编辑合成，不是本人原话或实际投资案例。", "model": "付费用户=活跃用户×付费率；收入=付费用户×每用户收入。", "boundary": "比较同一期间、同一用户去重定义；活跃不等于付费。", "counterexample": "开源降低门槛，同时拉低收费率和价格。", "test": "两季内核付费留存、真实收款与获客成本，不能用事件热度替代。", "question": "合成：某应用活跃用户100万、付费率4%、年均付费500元；次年活跃用户200万、付费率3%、年均付费400元。收入增长多少？", "constraint": "新证据显示次年付费率仅2%，其余不变；收入是否仍增长？保住原收入的最低付费率？", "transfer": "迁移：工业机器人试用客户200家、转采购率20%、单客户年订单100万元；试用增至300家、转化率10%、订单仍100万元，比较订单额。", "tags": [ "F9", "F10", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "O9G5-B73-C1", "profile": "o9g5-b73", "title": "模型效率提高，硬件需求一定下降吗？", "source_ids": [ "O9G5-B73A", "O9G5-B73B" ], "locator": "§4.1页4–5；§4.4页6；§4.1页5–6；§4.4页7", "source_paraphrase": "报告将技术变革的非线性增长、细分景气反转与风险收益条件结合，讨论AI终端及资本开支传导。；报告对AI资本开支热潮保留谨慎，要求检验终端渗透、模型效率与硬件迭代对长期价值的影响。", "attribution": "来源为材料观点的编辑转述；数字、问题、模型、反例和迁移全部为编辑合成，不是本人原话或实际投资案例。", "model": "所需算力=任务量×单位任务算力；收入另需乘实现单价。", "boundary": "只讨论相同工作负载质量及利用率，不跨模型能力直接比较。", "counterexample": "任务反弹、利用率和单价压缩可能抵消技术效率收益。", "test": "下一财年核付费任务、单位成本、供给利用率及订单取消，分清假设与证据。", "question": "合成：任务量基期100、单位任务算力1；次期任务量180、单位算力0.5。算力需求变化多少？", "constraint": "若单位算力0.5不变，任务量达到250；另算供应商每算力收入下降20%时，其收入相对基期如何？", "transfer": "迁移：仓库订单数增长60%，每单机器人分钟减少25%，总机器人分钟改变多少？", "tags": [ "F9", "F18", "F14" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "O9G5-B74-C1", "profile": "o9g5-b74", "title": "好公司增长如何穿过估值压缩？", "source_ids": [ "O9G5-B74A", "O9G5-B74B" ], "locator": "三个维度精选好公司；深度研究创造持续回报；§4.1页4–5；§4.4页5–6", "source_paraphrase": "归凯将商业模式、竞争优势及产业趋势拆开评估，并提醒成长行业仍有库存周期和回报率约束。；管理人区分宏观修复与企业成本价格压力，并披露增持芯片、消费电子和面板、减持医疗器械的操作。", "attribution": "来源为材料观点的编辑转述；数字、问题、模型、反例和迁移全部为编辑合成，不是本人原话或实际投资案例。", "model": "无分红价格回报=(终期EPS×终期PE)/(初期EPS×初期PE)−1。", "boundary": "同股本及一致盈利口径；这是情景，不是目标价建议。", "counterexample": "公司盈利增长，但买入估值过高导致投资回报负值。", "test": "逐季更新盈利持续性、库存及竞争优势，重算回报而非只复述产业空间。", "question": "合成：EPS为2元，买入PE为40倍；两年后EPS为3元，PE降至25倍。无分红，累计回报多少？", "constraint": "若两年后EPS仅2.4元、PE仍25倍，回报多少？想在两年后保本，EPS至少多少？", "transfer": "迁移：医疗设备企业每股盈利5元、PE20倍；三年后EPS6元、PE18倍，无分红累计回报多少？", "tags": [ "F12", "F16", "F21" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "O9G5-B83-C1", "profile": "o9g5-b83", "title": "研发投入何时成为股东价值？", "source_ids": [ "O9G5-B83A", "O9G5-B83B" ], "locator": "Excerpt / Chapter One / Clues from the Past，研发实验室及商业化段落；Editorial Reviews / Excerpt / Chapter One，Adjusting to Key Influences of the 1960s开篇", "source_paraphrase": "Fisher强调研发需转化为商业利润，失败项目和长时间滞后令当期研发金额难以直接衡量价值。；Fisher将管理能力及商业研发进步与市场缓慢重估联系起来，要求区分价值变化和价格已计入程度。", "attribution": "来源为材料观点的编辑转述；数字、问题、模型、反例和迁移全部为编辑合成，不是本人原话或实际投资案例。", "model": "项目NPV=成功概率×成功终值/(1+r)^T−初始投入。", "boundary": "成功时终值现金回收已扣后续全部成本、未扣最初投入；失败回收为零；概率与期限均是假设。", "counterexample": "研发金额提高但商业化概率下降或延期，价值反而下降。", "test": "里程碑复核技术可行性、客户支付与部署时间；不以研发预算替代回报。", "question": "合成：现在投入100，3年后成功时终值现金回收180（已扣后续成本，未扣最初投入）、成功率80%、折现率10%，失败回收0。NPV多少？", "constraint": "保持其他条件，兑现延至第5年，NPV多少？第5年兑现时保本所需成功净现金多少？", "transfer": "迁移：药物项目现在投入50，两年后成功时终值现金回收120（已扣后续成本，未扣最初投入）、成功率50%、折现率10%，NPV多少？", "tags": [ "F10", "F15", "F20" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "O9G5-B84-C1", "profile": "o9g5-b84", "title": "熟悉产品之后，先检查能否撑过低谷", "source_ids": [ "O9G5-B84A", "O9G5-B84B" ], "locator": "Was that your secret?；Magellan早期投资与Chrysler问答；转录时间戳11:20–16:52、23:17–28:48", "source_paraphrase": "Lynch强调公司研究和保留基本面改善的赢家，容许部分判断错误；市场下跌不能替代业务分析。；演讲将熟悉业务与资产负债表研究结合，以行业库存、占用率等可观察指标替代宏观择时，反对仅因跌价而买入。", "attribution": "来源为材料观点的编辑转述；数字、问题、模型、反例和迁移全部为编辑合成，不是本人原话或实际投资案例。", "model": "现金跑道=可用现金/每月净消耗；行业量价指标不能替代偿债能力。", "boundary": "简化为消耗恒定、无新增融资；到期债务另扣除，现金流为已扣经营支出的净流量。", "counterexample": "客户喜欢产品，但现金流和债务到期使企业无法等到复苏。", "test": "每月核可用现金、净消耗、债务期限及真实订单，区分观察到的事实和预测。", "question": "合成：零售商可用现金120，每月净消耗10，预期9个月后经营恢复盈亏平衡。无债务到期，可否撑到复苏？", "constraint": "新增证据：第6个月需偿还债务50，其余不变。还能撑多久，9个月复苏前资金缺口多少？", "transfer": "迁移：AI服务商现金80，每月净消耗8，第4个月需偿债20，第8个月后才转盈亏平衡。能否撑到第8个月末？", "tags": [ "F11", "F14", "F17" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果或投资业绩审计" }, { "id": "O9G6-S35-C1", "profile": "o9g6-francois-meunier", "title": "毛利率桥接：组合与收购会计分开", "source_ids": [ "O9G6-ASML416", "O9G6-ASML117" ], "locator": "ASML Q4 2016 earnings call: Francois Meunier Q&A; lines 605–636；ASML Q1 2017 earnings call: Francois Meunier Q&A; lines 888–907", "source_paraphrase": "Meunier追问收购价格分摊对毛利率的影响；管理层明确不能把该调整直接称现金毛利率。；Meunier拆分毛利率同比提升来自服务/升级产品组合还是工具自身利润率。", "attribution": "下述方法由公开具名材料抽取，管理层回复不归为分析师主张。全部题目和数字为研究者合成，非本人预测。", "model": "毛利率=各业务收入权重×业务毛利率之和；PPA调整另列。", "boundary": "加回PPA摊销后的毛利率不是现金流指标。", "counterexample": "组合改善被误认为每条产品线提价。", "test": "取得分部收入/毛利率、PPA及现金流调节表。", "question": "合成：设备收入80、毛利率40%；服务20、毛利率70%。总毛利率多少？", "constraint": "总收入仍100，服务占比升至30%，两业务毛利率不变；新增PPA费用2计入成本。报告毛利率和剔除PPA毛利率各多少？", "transfer": "迁移：收入200，报告毛利率45%，含PPA费用4；加回后是多少？可否称现金毛利率？", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G6-S15-C1", "profile": "o9g6-matt-bryson", "title": "把去库存与交付延期从需求中拆开", "source_ids": [ "O9G6-AXT124", "O9G6-DBD226" ], "locator": "AXT Q1 2024 earnings call: Matthew Bryson Q&A; lines 652–751；Diebold Nixdorf Q2 2026 earnings call: Matthew Bryson Q&A; lines 311–320", "source_paraphrase": "Bryson纠正管理层把传统磷化铟去库存问题答成AI需求，追问库存压制收入的量和恢复时点。；Bryson追问银行设备延期具体原因，再问高毛利来自季度特殊项目还是产品组合改善。", "attribution": "下述方法由公开具名材料抽取，管理层回复不归为分析师主张。全部题目和数字为研究者合成，非本人预测。", "model": "发货=终端消耗+渠道库存变化；订单转收入另受交付时点影响。", "boundary": "去库存结束和项目延期逆转均不自动等于终端需求加速。", "counterexample": "补库存造成出货跃升，终端消耗却不变。", "test": "对照终端消耗、渠道库存、订单与交付确认。", "question": "合成：终端消耗100，渠道库存从40降到20，供应商本期发货多少？", "constraint": "下期终端消耗仍100、库存维持20，发货较本期增多少？能否说终端需求增长？", "transfer": "迁移设备项目：本期原预计收入120，其中30延期到下期；下期原基线120，延期项目全兑现，无取消。两期收入和下期环比增幅？", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G6-S24-C1", "profile": "o9g6-cody-acree", "title": "技术替代必须落到供应商份额与内容量", "source_ids": [ "O9G6-ISIL16", "O9G6-SMTC424" ], "locator": "Renesas acquisition of Intersil conference call: Cody Acree Q&A, transcript pp.4–5；Semtech Q4 FY2024 earnings call: Cody Acree Q&A, p.8", "source_paraphrase": "Acree挑战收购方此前市场份额下滑与转型能力，追问跨地域管理团队保留。；Acree追问铜与光的架构替代及Semtech是否因某一方案获益，管理层回答需看具体应用和组合。", "attribution": "下述方法由公开具名材料抽取，管理层回复不归为分析师主张。全部题目和数字为研究者合成，非本人预测。", "model": "供应商收入=各方案系统量×单系统内容量×份额；并购后还需验证组织执行。", "boundary": "技术路线获胜不保证特定供应商收入增长；并购互补也不保证执行。", "counterexample": "铜占比提高但本公司铜方案份额低，收入反降。", "test": "分别验证方案部署量、内容量、份额及团队和产品路线执行。", "question": "合成：市场100套系统，光60套/铜40套；本公司单套内容光10、铜4，份额均50%，收入多少？", "constraint": "总量不变，铜占比升至70%、光30%，份额不变，收入和变幅多少？", "transfer": "迁移并购交叉销售：新增20套，单套内容8，管理层计划份额50%；因整合只兑现计划的一半，新增收入多少？", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G6-B86-C1", "profile": "o9g6-seth-klarman", "title": "拆分价值：便宜之外还需兑现路径", "source_ids": [ "O9G6-KLGS25", "O9G6-KLCO25" ], "locator": "Seth Klarman — Goldman Sachs Talks: 02:13–04:45；Seth Klarman — Value Investing with Legends: 04:22–06:47", "source_paraphrase": "Klarman将折价与价值兑现催化剂并列，要求预先设想投资失败的原因。；Klarman回忆拆分交易：先扣清算现金，再计算所获分拆权益的隐含成本；这是本人回忆而非审计后的历史回报。", "attribution": "下述方法由公开具名材料抽取，管理层回复不归为分析师主张。全部题目和数字为研究者合成，非本人预测。", "model": "剩余权益隐含成本=(买价−清算分配)/获配权益数量；再做分配下行情景。", "boundary": "自动转录公开镜像，未复听；历史案例是访谈回忆。折价不消除清算延迟、负债与现金流缩水风险。", "counterexample": "预计现金分配减少使看似免费分拆权益变昂贵。", "test": "读取交易法律文件、债权顺序、现金分配条件与分拆现金流。", "question": "合成：每股母公司买价12，可收清算现金10并获0.5股新公司，新公司年自由现金流每股1；隐含新公司成本及现金流倍数？", "constraint": "清算现金只有8，其他不变，倍数多少？", "transfer": "迁移：买债券70，一年后有保证现金回收50及1份权益，权益年现金流2，忽略折现。权益隐含成本/倍数？如果现金回收尚未保证，应补什么证据？", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G6-B92-C1", "profile": "o9g6-david-tepper", "title": "政策情景与净敞口分别计算", "source_ids": [ "O9G6-TEP10", "O9G6-TEP18" ], "locator": "CNBC Squawk Box — David Tepper excerpts: pp.1,3；CNBC发布的unofficial transcript excerpts之公开镜像；CNBC interview — David Tepper with Scott Wapner: pp.4–6；CNBC unofficial transcript公开镜像", "source_paraphrase": "Tepper分两条宏观路径：自主复苏或经济不振引发QE；同时说明长期持续依赖QE并非好事。；Tepper把关税、汇率、美元盈利与估值联动；说明用个股多头和市场空头表达约25%的S&P等效敞口。", "attribution": "下述方法由公开具名材料抽取，管理层回复不归为分析师主张。全部题目和数字为研究者合成，非本人预测。", "model": "情景收益=概率×各情景收益之和；市场等效净敞口=多头beta敞口−空头beta敞口。", "boundary": "宏观路径不穷尽尾部风险，历史政策反应不保证重复；净敞口小不等于个股基差风险小。", "counterexample": "政策没有如期宽松，且多头持股落后指数对冲。", "test": "列出未覆盖情景，核对头寸beta、对冲与个股超额收益。", "question": "合成：复苏概率60%回报15%；不复苏但宽松概率40%回报10%，期望回报多少？", "constraint": "新增政策失灵20%概率回报−30%；原两情景概率变为48%和32%，期望回报多少？", "transfer": "迁移：资产100，个股多头80且beta1，指数空头55且beta1。净beta敞口多少？指数涨10%、个股只涨5%，组合损益多少？", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "O9G6-B96-C1", "profile": "o9g6-chris-hohn", "title": "成本削减须计入过渡成本与收入反作用", "source_ids": [ "O9G6-HOHN22", "O9G6-HOHN23" ], "locator": "TCI letter to Alphabet CEO: pp.1–4；TCI follow-up letter to Alphabet CEO: p.1", "source_paraphrase": "Hohn要求约束成本及Other Bets损失，设置利润率目标，以现金流支持回购；信中估值与预测仅为当时主张。；Hohn认可12000减员但认为不足，建议约150000目标人数并降低薪酬/SBC；不是实际已实现的削减结果。", "attribution": "下述方法由公开具名材料抽取，管理层回复不归为分析师主张。全部题目和数字为研究者合成，非本人预测。", "model": "经常性成本节约=人数减少×平均全负担成本；首年净收益再扣遣散费及业务损失。", "boundary": "信中median薪酬不能直接作全员平均全负担成本；SBC经济成本与现金支出不同。", "counterexample": "裁员减少研发和销售能力，使营收/毛利流失抵消节约。", "test": "取得岗位级平均成本、遣散费、SBC稀释和业务流失情景。", "question": "合成：人数200降至160，平均年全负担现金成本每人0.3，总经常性年节约多少？", "constraint": "每个离职员工一次性遣散成本0.15，裁员立即生效且持续一年；首年净现金节约多少？", "transfer": "迁移：裁员额外造成年度收入减少20，现金贡献率40%（已扣随收入变化的现金成本），用上述首年节约衡量，首年净收益多少？", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "FM-LA-C1", "profile": "leopold-aschenbrenner", "title": "算力情景能否通过电力交付约束", "source_ids": [ "FM-LA-ESSAY24" ], "locator": "Training compute；Power requirements；Overall compute；AI revenue", "source_paraphrase": "作者把算力外推落实为集群建设与供电问题。", "attribution": "公开原文的方法启发；下列数字、公式与反例为研究者合成，不是基金模型或预测。", "model": "先把设备负载与设施开销换成容量需求，再用已交付电力约束可部署设备数。", "boundary": "容量不是年度用电量；设备峰值、平均利用率、PUE和并网日期必须同口径。", "counterexample": "GPU已交货但电力尚未交付，采购规模无法立即转成可运行算力。", "test": "查验接入协议、分期并网时间、设施实测开销及可运行设备数。", "question": "合成：计划部署100000台加速器，每台IT侧功率1 kW，全部其他设施功率使PUE为1.25；同时满载需多少MW设施容量？", "constraint": "按期实际交付容量只有75 MW；假定PUE及单台功率不变、无其他负载，最多可同时运行多少台？相对计划缺多少台？", "transfer": "若下一代每台IT功率降到0.75 kW但单台有效吞吐相同，75 MW可支持多少台？这是否足以证明投资回报为正？", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "FM-LA-C2", "profile": "leopold-aschenbrenner", "title": "主题判断正确仍须过时序与资本约束", "source_ids": [ "FM-LA-DW24" ], "locator": "04:10:30–04:10:58（主题收入敞口论述）；04:25:58", "source_paraphrase": "受访者强调AI收入敞口与投资顺序，并承认个别判断延迟需要承受能力。", "attribution": "仅将受访者本人观点作方法起点；案例与估值、融资挑战均为研究者合成。", "model": "先衡量主题对总收入的边际贡献，再分别检查估值、资金成本和等待期限。", "boundary": "主题收入增长不是利润、现金流或股价增长；访谈不提供可复刻组合规则。", "counterexample": "产业增长兑现，但估值回落或融资提前到期使投资损失。", "test": "按季度核对分部收入、利润转化、估值及现金到期表，不用技术方向替代仓位约束。", "question": "合成：公司总收入100，其中AI业务20；AI收入翻倍，其他收入不变。新总收入及增幅是多少？", "constraint": "为单独检验估值，假设利润也增长20%、股数不变，但市盈率由30降至20。股价变化比例是多少？", "transfer": "合成组合初始权益100，另借款50，合计买入150；持有一年资产价格不变，借款年息10%，到期必须还本付息，且无新融资。需出售多少资产偿债？剩余权益是多少？为什么方向性信念不能替代现金安排？", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "FM-RJ-C1", "profile": "rob-jesudason", "title": "永久资本能解决哪一种流动性问题", "source_ids": [ "FM-RJ-EURO", "FM-RJ-POD26" ], "locator": "Euromoney永久资本段；播客Transcript永久资本定义", "source_paraphrase": "资本结构可以影响被迫出售资产的时点。", "attribution": "访谈方法启发；全部数字、融资压力与算式为研究者合成，不是Serendipity实际财务。", "model": "将投资者赎回、公司现金需求和资产变现折价分别列账。", "boundary": "永久资本可减少赎回约束，但不保证公司无需再融资或股东随时能按净值退出。", "counterexample": "没有基金赎回，仍可能因经营开支耗尽现金而折价融资。", "test": "核对现金期限、承诺出资、股东退出机制与资产可售性。", "question": "合成：可赎回载体净资产100，其中现金10、非流动资产90。须支付赎回30，非流动资产只能按账面值的80%卖出，忽略费用。需卖出多少账面资产？折价损失多少？", "constraint": "另一永久资本公司拥有同样资产，无赎回要求，但未来两年每年现金开支8、无收入。现有现金够多久？两年资金缺口多少？", "transfer": "若只能按同样80%价格出售资产填补两年缺口，需卖多少账面资产？说明永久资本解决了什么、没有解决什么。", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" }, { "id": "FM-RJ-C2", "profile": "rob-jesudason", "title": "从技术演示到可重复交付", "source_ids": [ "FM-RJ-POD26" ], "locator": "Transcript：triangulate；managerial execution and engineering scale", "source_paraphrase": "受访者要求跨路线求证，并重视工程交付。", "attribution": "方法启发来自本人发言；案例、量化规则及反证为研究者设计，不是其技术排名。", "model": "把展示成绩、有效运行条件和可重复制造拆成不同证据。", "boundary": "专家数量不等于独立性；本题数字不能证明真实量子优势。", "counterexample": "演示成功却无法稳定批量交货，或多名顾问都依赖同一未经验证的厂商数据。", "test": "索取同任务同精度测试、独立复现、良率与交付记录，并揭示顾问利益关系。", "question": "合成：方案A每小时完成120个有效任务，在线率50%；方案B每小时80个，在线率90%。任务口径相同，离线时零产出。各自每个日历小时的预期有效任务数是多少？", "constraint": "A每批装配10台、交付合格率40%；B每批6台、合格率90%。假定所有合格设备均上线，比较每批新增设备的预期有效任务总吞吐。", "transfer": "A有三位顾问背书，但都引用同一个厂商演示；B有一份独立重复测试。是否可以按顾问人数投票？还需要什么证据才能判断成本与商业回报？", "tags": [ "F4", "F18" ], "scenario_type": "synthetic teaching case", "forecast_outcome": "未进行预测结果/投资业绩审计" } ], "comparison": "买方连接研究、估值与组合表达；卖方连接问题、经营变量与预测修订。角色不同不意味着谁更正确；比较仅限所读样本。", "coverage": "200份方法档案、310条来源记录、279张合成挑战卡；另有200人固定候选池，不计入已完成方法档案。包含历史角色、团队材料和公开问答，非现任覆盖全名单；真实QA数未因合成卡增加。", "role_counts": { "buy_side": 100, "sell_side": 100 }, "research_cutoff": "2026-10-09 (Europe/London; source-specific dates retained)", "candidate_pipeline": "candidate-pipeline.json" }