Taiwan’s Commercial Times reported on 10 November that supply-chain sources say Nvidia wants to slow its chip orders. The paper tied the risk of order cuts to tighter export reviews and H200 processors piling up in Hong Kong, and said TSMC and ASE would be first in line. It gave no size or timing for any cut.
Nvidia’s own filing shows what such a slowdown would act on. Its latest 10-Q put supply and capacity commitments at $279 billion as of 26 July 2026, up from $119 billion a quarter earlier. Of that, $92 billion falls in the rest of fiscal 2027, which ends in January 2027: about $46 billion a quarter, by our arithmetic.
Business Pill · GPUs Explained
An evergreen Business Engineer explainer on GPUs and AI hardware. It teaches the general idea only and does not discuss the order-cut report or Nvidia’s commitments.
The key insight: Supply-chain sources told Taiwan’s Commercial Times that Nvidia wants to slow its chip orders, while the same paper quoted industry people saying orders and capacity are unchanged. Nvidia’s 10-Q puts its supply and capacity commitments at $279 billion, $92 billion of it in the rest of fiscal 2027: about $46 billion a quarter, by our arithmetic.
According to the Commercial Times, the foundry side has also heard talk of Nvidia order cuts. The paper said any capacity Nvidia released would help other chipmakers, including AMD and the in-house chips of large cloud companies.
ASIC makers told the paper that end customers would take all the capacity TSMC can offer. Analysts it cited named Alchip, Global Unichip and MediaTek as likely to gain capacity if Nvidia freed some up.
The paper also reported that MediaTek targets about $16 billion from its first AI accelerator in 2027 and has a second customer, and that Alchip has won an ASIC order from a second large US company. It opened the article with the pressure on chief executive Jensen Huang, whose board opened an independent investigation before Nvidia’s 17 November results.
Nvidia supply and capacity commitments by fiscal year as of 26 July 2026, $ billion, from Nvidia’s Form 10-Q (filed 26 August 2026). The 95.7% share falling in fiscal 2027 to 2029 is our arithmetic.
The Same Paper’s Second Story
A second Commercial Times article the same morning quoted industry people saying Nvidia’s existing work with Taiwanese suppliers is going ahead as normal, with orders and capacity unchanged. They named two things to watch: whether reviews delay deliveries, and Huang’s future at the company.
That article reported that about 600,000 H200 chips are held in Hong Kong because Beijing has not approved their import, and that some Asian customers’ AI server orders reportedly have to be reviewed again. Industry people told the paper the H200 is a previous-generation product and a very small share of Taiwanese assemblers’ shipments.
According to the same article, Taiwan’s server makers have shifted their focus to Nvidia’s B300, GB300 and Vera Rubin platforms, and several report results this week.
What Nvidia’s Filing Shows
The 10-Q spreads the $279 billion across fiscal years: $92 billion in the rest of fiscal 2027, $87 billion in fiscal 2028, $88 billion in fiscal 2029, then $6 billion, $5 billion and $1 billion. By our arithmetic, $267 billion, or 95.7%, falls in fiscal 2027 to 2029, and the total rose 2.34 times in one quarter.
Nvidia describes what the money buys: “These supply commitments are for our data center infrastructure systems, primarily memory and manufacturing facilities, to produce our products for long-term demand across current and future product architectures.”
The filing also says the commitments are not all fixed. In certain instances, it says, “these agreements may be cancelable, rescheduled, or adjustable for our businessneeds prior to placing firm orders. Changes to these agreements may result in additional costs.”
On the H200 itself, Nvidia took a $0.4 billion charge in the first half of fiscal 2027 for excess inventory and purchase obligations, “as the demand for H200 products diminished.” Shipments under the US licensing program, it said, “account for less than 1% of Data Center revenue in our most recent quarter.”
Why It Matters
The order-cut talk and the filing point at the same supply chain. TSMC, the first supplier the paper named, reported October revenue of NT$586.89 billion, up 59.7% from a year earlier and 14.7% from September.
As we read it, the filing sets the scale for the report: any slowdown would act on a book of commitments that rose 2.34 times in one quarter, by our arithmetic, and that Nvidia says may in certain instances be rescheduled or adjusted before firm orders are placed. Neither article says how large a cut would be.
Nvidia, Form 10-Q for the quarter ended 26 July 2026
“We have partnered with our extensive network to secure the necessary supply and critical components needed to meet demand for the next several years”
Three Implications
FOR THE SUPPLIERS The Commercial Times named TSMC and ASE as first in line if orders slowed; industry people in its second article said Taiwanese orders and capacity are unchanged.
FOR THE H200 Nvidia’s 10-Q says shipments under the US licensing program account for less than 1% of Data Center revenue, after a $0.4 billion charge for excess H200 inventory and purchase obligations.
FOR THE COMMITMENTS By our arithmetic, 95.7% of the $279 billion, or $267 billion, falls in fiscal 2027 to 2029.
The framework says: “Allocation power at the foundry is now more consequential than chip design itself.”
As we read it, a report about who gets TSMC capacity if Nvidia slows orders is a report about that allocation power. The framework predates these reports and does not discuss Nvidia’s orders.
What Is Not Established
The order-cut claim rests on unnamed supply-chain sources in one outlet, and the same paper’s second article quotes industry people saying orders and capacity are unchanged. We found relays of the report but no second outlet with its own reporting, and no comment from Nvidia, TSMC or ASE.
The 10-Q figures are as of 26 July 2026 and do not show any change since. The filing does not split the $279 billion by supplier, so it does not show how much sits with TSMC, ASE or memory makers.
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.
Supply-chain sources told the Commercial Times that Nvidia wants to slow chip orders; the same paper quoted industry people saying orders are unchanged. Nvidia’s filing puts the commitments at $279 billion, with $92 billion in the rest of fiscal 2027, or about $46 billion a quarter, by our arithmetic.
A note on sourcing. The order-cut report and the second report on orders and capacity are from two Commercial Times (工商時報) articles of 10 November 2026, read in Chinese and paraphrased in English; translations are ours. Commitment figures, quotations and the H200 figures are from Nvidia’s Form 10-Q for the quarter ended 26 July 2026. The shares, multiples and per-quarter average are our arithmetic.
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.
