Nvidia’s Form 10-K for fiscal 2026, filed on 25 February 2026, puts its fiscal 2025 revenue from customers headquartered in China (including Hong Kong) at $25,048 million. Its fiscal 2025 Form 10-K, filed a year earlier, put China revenue for the same year at $17,108 million, counted by where customers were billed. By our arithmetic, that is 19.2% of the $130,497 million total under the new count, against 13.1% under the old one.
Counted by headquarters, China fell from 23.2% of Nvidia’s quarterly revenue in the quarter to 27 October 2024 to 4.5% in the quarter to 25 January 2026, then rose to 8.2%, or $7,880 million, in the quarter to 26 July 2026, by our arithmetic from its filings. The New York Times reported on 29 October on Megaspeed and Nvidia; our report on the Times story covers that. This piece sets out what Nvidia’s own filings say about China.
A vertical news Short narrated from the published headlines and opening lines of this piece and a related FourWeekMBA piece on the New York Times report.
Business Pill · Concentration
An evergreen Business Engineer explainer on concentration. It teaches the general idea only and does not discuss Nvidia, any company or any person in this story.
The key insight: By our arithmetic from Nvidia’s filings, China was 19.2% of its fiscal 2025 revenue counted by customer headquarters, against 13.1% counted by billing location, and 8.2% in the quarter to July 2026.
Nvidia’s fiscal 2025 10-K says geographic revenue “is based upon the billing location of the customer” and that “The end customer and shipping location may be different from our customer’s billing location”. Under that method Singapore was 18% of fiscal 2025 revenue, and the filing says “Customers use Singapore to centralize invoicing while our products are almost always shipped elsewhere”.
In the third quarter of fiscal 2026, the 10-Q for the quarter to 26 October 2025 reported the switch. Its footnote says billing location “often reflected a customer’s centralized invoicing location, even though our products were almost always shipped elsewhere”, and that prior periods had been recast.
The recast changed the China line for both earlier years. Fiscal 2024 China revenue went from $10,306 million to $12,330 million, and fiscal 2025 from $17,108 million to $25,048 million. By our arithmetic, the change added $7,940 million, or 46%, to fiscal 2025, and lifted fiscal 2024 from 16.9% to 20.2% of revenue.
Revenue from customers headquartered in China (including Hong Kong) as a share of Nvidia’s quarterly revenue, quarter to October 2024 through quarter to July 2026. Shares and the January 2026 quarter are our arithmetic. Sources: NVIDIA Forms 10-K and 10-Q.China (including Hong Kong) revenue counted by customer billing location (original fiscal 2024 and 2025 10-Ks) and by customer headquarters (fiscal 2026 10-K, prior years recast). Shares of revenue are our arithmetic. Source: NVIDIA Forms 10-K.
China, Quarter by Quarter
On the headquarters basis, the 10-Q filings give China revenue of $8,142 million in the quarter to October 2024, $9,659 million in the quarter to April 2025, $3,985 million in the quarter to July 2025 and $2,973 million in the quarter to October 2025.
For the quarter to January 2026, our arithmetic from the full-year figure of $19,677 million less nine months of $16,617 million gives $3,060 million, or 4.5% of that quarter’s revenue. Fiscal 2026 as a whole was 9.1%.
The two quarters since then show $4,550 million to April 2026 and $7,880 million to July 2026. By our arithmetic, the July quarter is 8.2% of $96,221 million of revenue, and 98% more than the $3,985 million a year earlier.
What the Filings Say About Export Licences
The fiscal 2026 10-K says that in April 2025 the US government required a licence to export Nvidia’s H20 chips to China, and that Nvidia “incurred a $4.5 billion charge in the first quarter of fiscal year 2026 associated with H20”. After licences were granted in August 2025, the filing says “We generated approximately $60 million in H20 revenue under those licenses”.
The 10-Q for the quarter to July 2026 says US licences allow shipments of “small amounts of H200 products to specific China-based customers”, but “such sales were restricted by the PRC government”. It reports a $0.4 billion H200 charge in the first half of fiscal 2027 and says the H200 shipments “account for less than 1% of Data Center revenue in our most recent quarter”.
Where a Customer Is Counted
Nvidia’s 10-Q for the quarter to July 2026 says geographic revenue “is designated based on the location of the headquarters of direct customers” and that “The end customer and shipping location may be different from our direct customers’ headquarters location”.
The filings do not name customers by country, do not split the China line by product and do not say how any individual customer is classified. Our earlier report noted Nvidia’s statement, reported by Benzinga, that Megaspeed is headquartered outside China; Nvidia’s filings do not mention Megaspeed.
The Structural Read
As we read the filings, the switch moved revenue out of billing hubs such as Singapore, which was 18% of fiscal 2025 revenue under the old count, and into the countries where customers are headquartered.
By our arithmetic, China’s share on the new count fell from 21.9% in the quarter to April 2025, when the H20 licence requirement and the $4.5 billion charge came, to 8.5% in the quarter to July 2025.
As we read the July 2026 10-Q, the rise to $7,880 million came while H200 shipments to China were less than 1% of Data Center revenue, and the filing does not say which products made up the China line.
NVIDIA Form 10-Q for the quarter to 26 October 2025, geographic revenue footnote
“often reflected a customer’s centralized invoicing location, even though our products were almost always shipped elsewhere”
Three Implications
FOR INVESTORS By our arithmetic, fiscal 2026 China revenue of $19,677 million was 9.1% of Nvidia’s $215,938 million total, against 19.2% a year earlier on the same headquarters count.
FOR EXPORT POLICY The fiscal 2026 10-K says Nvidia generated approximately $60 million of H20 revenue under licences granted in August 2025.
FOR READERS COMPARING REPORTS Figures for fiscal 2025 China revenue differ by $7,940 million between Nvidia’s fiscal 2025 and fiscal 2026 10-Ks because of the change in method, by our arithmetic.
The framework lists export controls between nation-states in its governance layer and says of that layer: “No longer a passive layer; an active priced variable.”
As we read it, Nvidia’s China line is one place where that layer shows up in reported revenue. The framework does not discuss Nvidia’s geographic reporting.
What Is Not Established
The China line covers all of Nvidia’s products sold to customers headquartered in China, not data-centre chips alone, and it does not show where the chips end up. The quarter to January 2026 is our arithmetic, not a reported quarterly figure, and Nvidia has not yet filed results for the quarter to October 2026.
The shares of revenue are our arithmetic from the reported dollar figures. Nothing in the filings refers to the Times report or to any individual named in it.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
By our arithmetic from Nvidia’s filings, counting customers by headquarters instead of billing address raised China’s share of fiscal 2025 revenue from 13.1% to 19.2%. On that count China fell to 4.5% of revenue in the quarter to January 2026 and rose to 8.2%, or $7,880 million, in the quarter to July 2026, while the filings put H200 shipments to China at less than 1% of Data Center revenue.
What happened next (29 Oct):The same recount removed Nvidia’s Singapore billing line, which held $23,684 million of fiscal 2025 revenue. By our arithmetic, the United States line rose by $16,225 million and China by $7,940 million on the headquarters count; the filings do not say which customers moved.
A note on sourcing. Revenue figures come from NVIDIA’s Forms 10-K for fiscal 2024, 2025 and 2026 and its Forms 10-Q for the quarters to October 2025, April 2026 and July 2026, read on SEC EDGAR on 29 October 2026. Shares of revenue, the January 2026 quarter and the size of the recast are our arithmetic. The New York Times report is covered in our linked earlier piece; these filings do not refer to it.
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.
