TSMC is seeing “insatiable” demand for any chips it can make to power artificial intelligence, senior vice president and deputy co-chief operating officer Kevin Zhang told the Nikkei Forum Global Management Dialogue 2026 in Tokyo on 11 November, Nikkei Asia reported.
He spoke a day after TSMC reported October revenue of about NT$586.89 billion, up 59.7% from October 2025 and 14.7% from September, according to its monthly revenue report. October was 46.3% above January’s NT$401.26 billion, by our arithmetic from TSMC’s monthly table.
Business Pill · Physical AI Explained
An evergreen Business Engineer explainer on physical AI, one of the demand avenues Zhang named. It teaches the general idea only and does not discuss TSMC’s remarks or revenue.
The key insight: As we read it, the word matters because of who said it and when: a TSMC senior vice president called demand “insatiable” a day after the company reported October revenue up 59.7%, and October was 46.3% above January, by our arithmetic.
Nikkei Asia quotes Zhang: “The demand on our leading node [most advanced] silicon is insatiable”. He added that customers, and their customers, “keep telling us the biggest commodity today is [the capacity to] compute when it comes to AI”.
He went on: “They want to get [their] hands around any compute capability they can have. It’s truly an insatiable environment.” The bracketed words are Nikkei’s.
Zhang said customers are continuing to migrate to TSMC’s most advanced levels of miniaturization, from 5 nanometer to its latest 2nm chips, according to Nikkei. He suggested that advanced packaging, rather than transistor or wafer technology alone, would be key to meeting future needs to drive AI.
TSMC consolidated monthly revenue in 2026, from NT$401.26 billion in January to NT$586.89 billion in October, per TSMC’s monthly revenue table and October report (unaudited). The 46% rise from January to October is our arithmetic.
The Revenue Behind the Word
TSMC’s monthly table shows consolidated revenue of NT$401.26 billion in January, NT$317.66 billion in February and NT$514.81 billion in August, before October’s NT$586.89 billion. October was about 35.5% above the January-to-September monthly average of NT$433.19 billion, by our arithmetic.
Revenue for January through October totaled NT$4,485.62 billion, up 43.3% on the same period of 2025, TSMC said. We covered the October report when it came out on 10 November; CoreWeave set out its own spending plans earlier on 11 November.
TSMC’s year-on-year monthly revenuegrowth in 2026, from its monthly revenue table and October report. Only June’s 67.9% was higher than October’s 59.7%; the ranking is ours.
Japan, Physical AI and Microsoft’s Overhang
Japan represents around 4% of TSMC’s global revenue, Nikkei reports. Applying that share to January-to-October revenue gives roughly NT$179 billion, by our arithmetic; Nikkei does not say which period the 4% covers.
Zhang said: “although the revenue percentage is not big … in terms of diversity of the technology and the application, we think that there is a great potential”. He singled out physical AI, systems that combine AI models with robotics, as an avenue of demand growth, Nikkei reports.
At the same event, Microsoft chief technology officer Kevin Scott said AI models have become “far more capable now than what they’re actually being used for”, according to Nikkei. Scott called the challenge a “capability overhang”.
The Structural Read
As we read Nikkei’s report, Zhang placed the constraint beyond the wafer: he suggested advanced packaging, rather than transistor or wafer technology alone, would be key to meeting future AI needs.
As we read the two Tokyo sessions together, they describe two different gaps: Zhang described demand for compute running ahead of what TSMC can make, while Microsoft’s Kevin Scott described models running ahead of how they are used, which he called a “capability overhang”.
Kevin Zhang, TSMC, as reported by Nikkei Asia
“They want to get [their] hands around any compute capability they can have. It’s truly an insatiable environment.”
Three Implications
FOR THE RUN-RATE October’s NT$586.89 billion was about 35.5% above the January-to-September monthly average of NT$433.19 billion, by our arithmetic on TSMC’s table.
FOR THE NODE MIX Zhang said customers are moving from 5 nanometer to TSMC’s latest 2nm chips, and TSMC’s CFO said in October that the fourth quarter would include the steep ramp-up of its 2-nanometer technology.
FOR JAPAN Nikkei reports Japan is around 4% of TSMC’s global revenue. Zhang named physical AI as an avenue of demand growth, and Nikkei notes TSMC’s May partnership with Sony on image sensors that could be used for physical AI.
The framework argues: “In any product cycle, there are two curves to follow.” One is what the technology can do; the other is what users can actually do with it.
As we read it, the Tokyo forum put both curves on one stage: TSMC described compute demand it calls insatiable, and Microsoft’s CTO described capability running ahead of use. The framework predates the forum and does not discuss TSMC or Microsoft.
What Is Not Established
Nikkei’s article gives no figure for how far demand exceeds TSMC’s capacity, and TSMC’s monthly report does not break revenue down by node, platform or customer. Posts on X attributed further remarks to TSMC executives at the forum that we could not match to a source we read, so we have left them out.
We did not watch the session and did not contact TSMC or Microsoft. The 46.3% and 35.5% comparisons are our arithmetic on TSMC’s unaudited monthly figures.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
A TSMC executive called AI chip demand “insatiable” in Tokyo, Nikkei reports, a day after the company reported October revenue up 59.7%; by our arithmetic October was 46.3% above January.
A note on sourcing. Zhang’s and Scott’s remarks are as reported by Nikkei Asia on 11 November 2026; we read the full article text. Revenue figures are from TSMC’s October 2026 revenue report, its 2026 monthly revenue table and its third-quarter release. We did not watch the session and did not contact TSMC or Microsoft.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
