METHOD & DATA

方法与数据说明 / Method & data

先说明证据怎么分级、各板块怎么做、哪里还缺,再读内容。How evidence is graded, how each section is built and what is still missing — read this before the content.

这是什么What this site is

这是一个研究整理站点:把公司产品报告、财报问答、分析师与架构师的公开材料、Hot Chips 提问模式、政策与标准放在同一张地图里,并明确每条信息的来源与证据强度。它不是投资建议,也不是对人物私下想法的复刻。A research compilation: company product reports, earnings Q&A, public material by analysts and architects, Hot Chips question patterns, policy and standards sit on one map, with the source and strength of evidence made explicit. It is not investment advice and does not reproduce anyone’s private thinking.

证据级别Evidence levels

级别Level含义Meaning
primary-read已打开并阅读官方或公司原始文件。Official or company primary document opened and read.
primary-located已定位官方文件,但细节取自摘要。Official document located; details taken from summaries.
secondary律所、智库或媒体摘要,未核对原文。Law-firm, think-tank or press summaries; primary text not checked.
editorial / 编辑归纳本站的归纳、映射与教学模型;用于提出可检验的问题,不是已核实结论。This site’s own synthesis, mappings and teaching models; they pose testable questions and are not verified conclusions.
synthetic / 合成教学用的合成数字与情景,明确标注,不代表任何公司的真实数据。Synthetic numbers and scenarios for teaching, labelled as such; not any company’s real data.

各板块如何构建How each section is built

技能编号Skill IDs

架构师技能用 A1–A10,分析师技能用 F1–F25,两套编号相互独立;技术×投资页给出两者的编辑对应。Architect skills use A1–A10 and analyst skills F1–F25; the two numbering schemes are independent and the Technology × Investment page gives an editorial mapping between them.

语言Languages

公司产品报告有中英文两版;分析师训练、架构师思想与技术×投资目前为中文;财报分析、Hot Chips 与 Architecture Notes 为英文。导航中的标记(EN / 中文)提示每个板块的语言。政策、标准、方法与术语页为双语,可在页面顶部切换。Company reports exist in Chinese and English; analyst training, architect thought and Technology × Investment are currently Chinese; earnings analysis, Hot Chips and Architecture Notes are English. The (EN / 中文) marks in the navigation show each section’s language. The policy, standards, method and glossary pages are bilingual with a toggle at the top.

数据与更新日期Data and as-of dates

已知缺口Known gaps

如何引用与更正Citing and corrections

建议引用格式:“AI 基础设施研究(wilwan01.github.io/ai),页面名称,访问日期”。发现错误或有更好的来源,请在 GitHub Issues 提出。Suggested citation: “AI Infrastructure Research (wilwan01.github.io/ai), page title, access date”. To report an error or suggest a better source, open an issue on GitHub.

术语表 →Glossary → · 全站搜索 →Site search →