AI

AI will be everywhere. The harder question is who still earns a margin when intelligence gets cheap. By Sinclair I have watched AI move from conversation and search into data processing, automation, vehicles, robots, industrial tools, and factories. That makes me sceptical of debates that begin with whether AI demand will exist. Much of the eventual demand will not look like a deliberate purchase of “AI.” Intelligence will be built into a product or workflow, and the customer will pay for the outcome. ...

本文為繁體中文版。English version: The AI Capex Money Map v0.2 — America Spends. Who Actually Keeps the Margin? 本站為文字版；含完整圖表的版本請見 Medium 原文。 大多數人現在都在問：「AI 是不是泡沫？」 這個問題重要，但不夠可操作。對投資人與供應鏈觀察者來說，更好的問題只有一句： 美國大型科技公司花出去的 AI Capex，最後流到誰的收入、誰的毛利、誰的護城河？ 這篇不是拆到每一根電纜的投行 BOM 模型，而是一張用公開資料建立的 v0.2 錢流地圖。它的目的不是假裝精準，而是把三件常被混在一起的事情分開來看：錢在哪裡變成營收、營收在哪裡變成毛利、瓶頸何時鬆解。 因為這三題的答案，往往不是同一批公司。 一、AI Capex 不是一個數字，是一組水管 Reuters 引述 Bridgewater 的估計：Alphabet、Amazon、Meta、Microsoft 四大 2026 年的 AI 基礎設施投資約 US$650B，高於 2025 年約 US$410B。先講清楚一件事——這是四大的地板數，不含 Oracle、Stargate、CoreWeave 這類 neocloud、xAI，也不含任何非美系業者。所以 $650B 是四家公司的下限，不是市場總額。 錢不會平均落下，而會進入性質不同的系統層支出。設備與材料則屬二階供應商 capex，不計入 hyperscaler 直接支出，以避免重複計算。 關鍵是——設備與材料（ASML、AMAT、TEL、Advantest…）不在這 $650B 裡。 那是台積電、SK Hynix 的資本支出，是二階供應商支出、是另一個分母。把它加進 hyperscaler 支出裡相加，就是 v0.1 犯的、也是很多市場圖表仍在犯的 double counting。 二、把箱子打開：成本重心已經從邏輯移到 HBM + 封裝 把「Compute Systems」這條水管打開，你會看到這一輪最反直覺的結構變化。 ...

繁體中文版：AI Capex 錢流地圖 v0.2｜美國出錢，誰真的收得到毛利？ The AI Capex Money Map v0.2 — America Spends. Who Actually Keeps the Margin? From $650B to HBM, CoWoS and power — mapping who gets paid, who keeps margin, and when the bottlenecks move. (The capex waterfall, cost-stack, bottleneck-clock, and regional margin-capture charts are figures in the original Medium version; this text edition keeps the surrounding reasoning — see Medium for the full visual breakdowns.) Most people are asking whether AI is a bubble. It is an important question, but not a very operational one. ...

Why the second AI revolution may be bigger — and riskier — than the firstSinclair Huang AI Infrastructure Notes｜Article 5* Field Note v4 — updated with GTC Taipei and statistical-science reference materials, using public and user-provided sources available as of June 6, 2026.* AI is acquiring three things it never had at scale: a ticker, an agent, and a body. A ticker makes AI liquid. An agent makes AI operational. A body makes AI physical. ...

** In a world of infinite content and scalable intelligence, the scarcest resource may no longer be intelligence itself — but the ability to truly see another human being, and to be seen in return.** By Sinclair Huang I walked into a bookstore — and realised something uncomfortable A few days ago, I walked into a bookstore. Surrounded by shelves of books — many of them thoughtful, deeply human, and difficult to summarise — I had a quiet realisation: ...

Work, ownership, and the new architecture of economic security By Po-Sung(Sinclair) Huang For decades, people believed that working hard was enough. In the AI era, that assumption is quietly breaking. Not because work disappears, but because ownership matters more than ever. In the AI era, relying on labour income alone is no longer a neutral choice. It is an active form of risk. That sentence may sound harsh. But it captures a structural shift that many people can already feel, even if they do not yet have the language for it. They feel it in markets. In career anxiety. In the fear of being left behind by a technology wave they did not ask for, do not fully control, and may not directly benefit from. ...

From dBase and enterprise systems to the internet revolution and generative AI, I have come to see AI not as a sudden break, but as the latest step in a long slope of automation now reaching human cognitive work itself.* By Po-Sung(Sinclair) Huang I did not decide to write this essay because AI suddenly became fashionable. It came out of two images that collided in my mind. One was a group of younger people trying to imagine new ventures built around AI. The other was the story of a father facing a rare disease so obscure it seemed to leave almost no path forward, and yet continuing, late into the night, to search for a way through with the help of computation, search, and structured reasoning. ...

From compute bottlenecks to industrial consequences — where value may actually concentrate through 2030 Series: AI Compute Supply Chain | Part 5 of 5 Author: Po-Sung (Sinclair) Huang For the past four articles in this series, I have written about CoWoS, HBM, ABF substrates, SEC filings, and the fault lines that could eventually crack today’s moats. On the surface, that may look like a semiconductor series. It is not. What these articles really reveal is something larger: AI is no longer just a software story, and no longer just a model race. It is becoming an industrial system — one that depends on power, cooling, capital expenditure, advanced packaging, memory bandwidth, substrate materials, qualification cycles, and the physical discipline of manufacturing scale. ...

A new study in Nature Machine Intelligence suggests AI may do more than summarise knowledge or accelerate discovery. It may begin to shape which scientific questions are noticed first. When AI Starts Predicting the Next Scientific QuestionWe have grown used to thinking about AI as an answer machine. It summarises papers, organises data, generates hypotheses, and accelerates analysis. In that familiar picture, AI helps scientists move faster toward results that humans still define. ...

Why is everyone suddenly talking about CoWoS, HBM, and ABF whenever AI, NVIDIA, or AI servers come up? Many people know they are important, but still get stuck the first time they run into these terms. This essay is a plain‑language walkthrough of what they actually are, why they are always mentioned together, and how they map onto Taiwan’s role in the global AI supply chain. When Google, Amazon, and Tesla are all designing their own chips, is Taiwan’s manufacturing ecosystem still structurally important — or just enjoying a temporary window? ...