Nvidia CEO Jensen Huang said a US policy mistake on selling chips to China is “hard to unwind”, in interview footage Bloomberg released on 9 November. Bloomberg anchor Emily Chang filmed the interview last month for an upcoming episode of The Circuit (Chang’s post), and Bloomberg said the CEO “defended engaging with China and criticized US chip restrictions”.
Nvidia’s filings record two China-related charges: a $4.5 billion charge tied to the H20 chip after the April 2025 US licence rule, and a $0.4 billion charge tied to the H200 in the first half of fiscal 2027, after the PRC government restricted licensed sales, the 10-Q says. By our arithmetic, that is $4.9 billion in charges tied to H20 and H200 sales to China.
Business Pill · GPUs
A one-minute explainer of what GPUs are and why AI training depends on them. It teaches the general idea only and does not discuss Nvidia’s China sales.
The key insight: Nvidia CEO Jensen Huang said a US policy mistake on China is “hard to unwind”, in footage Bloomberg released on 9 November. Nvidia’s filings record a $4.5 billion H20 charge and a $0.4 billion H200 charge, $4.9 billion by our arithmetic, and China revenue down 21% in fiscal 2026.
Asked whether the promise of the Chinese market is slipping away as companies such as DeepSeek build on Chinese hardware, Huang said: “I did my best to inform the policymakers in the United States of the importance of engaging the Chinese market and making sure that we serve that market for as long as we can”. Huang added that “this mistake is hard to unwind” and: “But I did my best. That’s all I could do.”
Huang also said “There are more Chinese AI researchers than any other nation in the world”, and that while Nvidia competes against China, “they’re not our enemies”. On AI safety, Huang said “we must open dialogues”. Asked what could change that view, Huang said: “We want America to be rich. They want China to be rich and we should respect that aspiration”.
The quotes are from our transcript of the 122-second clip Bloomberg posted; Bloomberg has not published a full transcript of the interview.
China (including Hong Kong) as a share of Nvidia revenue by customer headquarters, from Nvidia’s 10-K and 10-Q, with the two China-related charges recorded in those filings. Shares and the $4.9 billion total are FourWeekMBA’s arithmetic.
What Was Posted on X
JUST IN: ๐บ๐ธ๐จ๐ณ Nvidia $NVDA CEO Jensen Huang says China is not an enemy and US chip policy is a “mistake.”
Watcher Guru posted this at 02:58 UTC on 9 November; within minutes it had drawn tens of thousands of views. Its wording, that Huang called US chip policy a โmistakeโ and said China is not an enemy, is the accountโs own; the quotes in this piece are from our transcript of Bloombergโs clip.
What the Filings Show
Nvidia’s 10-Q for the quarter to 26 July 2026 says the US government told the company in April 2025 that H20 exports to China need a licence. The filing adds: “As a result of these requirements, we incurred a $4.5 billion charge in the first quarter of fiscal year 2026 associated with H20 for excess inventory and purchase obligations”.
The filing says US licences for some H20 sales followed from August 2025, “but such sales were restricted by the PRC government, and we were unable to sell our H20 inventory”. From February 2026, licences allowed “small amounts of H200 products” to specific China-based customers; Nvidia then booked “a $0.4 billion charge associated with H200” and says those shipments “account for less than 1% of Data Center revenue in our most recent quarter”, with a 25% tariff on licensed H200s.
The footage surfaced a day after Nvidia’s board said it is opening an independent investigation, covered in Nvidia Board Opens Independent Probe 9 Days Before Results. The clip was filmed before that announcement and does not address it.
Three House Republican committee chairs have asked Huang to brief their committees in a closed-door session on 18 November, the day after Nvidia reports third-quarter results, per House Chairs Ask Huang to Brief Day After Nvidia Earnings. Nvidia’s third-quarter outlook assumes no Data Center compute revenue from China, the company says.
Update (9 Nov): Congress. Bloomberg reported on 6 November that a new Democratic majority in Congress is set to demand answers from Nvidia’s CEO on keeping its AI chips from China and other US adversaries. The CEO declined a Senate Banking invitation in June, CNBC reported; our follow-up is Nvidia Declined a June Hearing; Democrats Win 2027 Gavels.
Update (10 Nov): House Democratic leader backs the Senate chip provisions. Hakeem Jeffries said on 9 November that he “strongly supports” restrictions on sales of US chips to China in the final version of the annual defense bill, Reuters reported. Jeffries told reporters there are provisions to restrict China’s access to chip technology in the Senate’s version of the bill, and “we’re going to work hard to make sure that those provisions make it into the final version,” according to CNN. Reuters says the Senate version contains provisions to give Congress more authority on AI chips going to China, to prod US allies to restrict sales of semiconductor manufacturing equipment to China, and to require businesses to track their AI chips abroad to combat smuggling. The Senate bill is S. 4784, the National Defense Authorization Act for Fiscal Year 2027. On 14 July, Sen. Jim Banks’s office said his AI OVERWATCH Act was included in it; the release says that act extends the prohibition on exports of the most advanced AI chips to countries of concern for 18 months. In the committee-reported text, Sec. 1248 requires an annual assessment of the contribution of certain AI chips to China’s military capabilities. The congress.gov record shows no Senate passage vote on S. 4784: cloture on the motion to proceed failed 50-46 on 14 July, and the latest action listed is a motion to proceed on 27 July.
Update (10 Nov): Commerce staff revisit remote-access rules.The Information reported on 9 November (US time) that in recent weeks staffers at the Commerce Department “have made a renewed push to create rules around two key areas of export control policy,” according to current and former Commerce officials: tooling for lithography machines, and “restrictions intended to prevent Chinese companies from remotely accessing advanced AI chips made by companies like Nvidia.” The same officials said Commerce leadership “hasn’t followed through on efforts to create new export control policies in the past and could hold off once again.” The Information also reported that in late October White House officials shared feedback with Congressional staff on export control measures under consideration in the National Defense Authorization Act, according to several people familiar with the matter.
Nvidia’s position, as relayed. In a summary of the paywalled article posted on X, Rohan Paul wrote that The Information cites two current Commerce officials as saying Nvidia “lobbied especially hard against any rules on remote access,” which the summary describes as “the route Chinese AI labs use to reach restricted chips in Singapore and Malaysia data centers.”
We could read only the opening of The Information’s article, not the passage on lobbying, and could not see whether Nvidia commented to the publication.
As we read the clip, the CEO frames the China question as a market Nvidia should keep serving “for as long as we can”, and argues the cost is hard to reverse once lost.
As we read the filings, both charges sit on chips subject to licence rules: the H20 after the April 2025 US requirement, and the H200 after the PRC government restricted licensed sales, the 10-Q says.
As we read the timing, the views were on record before the board’s investigation and ahead of the 18 November House briefing request, which puts the company’s China position in front of lawmakers.
Jensen Huang, Nvidia CEO, Bloomberg footage released 9 November 2026 (our transcript)
“this mistake is hard to unwind”
Three Implications
FOR NVIDIA’S RESULTS Nvidia says its third-quarter outlook assumes no Data Center compute revenue from China; results are due 17 November.
FOR THE LICENCE REGIME Nvidia says licensed H200 shipments account for less than 1% of Data Center revenue in its most recent quarter, with a 25% tariff on them.
FOR LAWMAKERS Three House committee chairs have asked Huang to brief them on 18 November in a closed-door session.
The framework says: “It’s a fundamental constraint on the speed of AI transformation, and it’s creating geopolitical leverage that operates independently of who has the best models.”
As we read it, licences for chips are one such lever: Nvidia’s filings tie its two China charges to licence rules in Washington and Beijing. The framework predates the interview and does not discuss Nvidia.
What Is Not Established
The quotations come from our machine transcript of Bloomberg’s clip; we restored sentence punctuation only. Bloomberg said the interview was filmed less than a month ago and has not given an exact date. Nvidia counts geographic revenue by where customers are headquartered, and says end customers and shipping locations may differ. The $4.9 billion total and the 21% fall are our arithmetic from Nvidia’s filings. We did not contact Nvidia.
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.
Huang told Bloomberg that a US policy mistake on China is “hard to unwind” and that China is not an enemy. Nvidia’s filings record $4.9 billion in charges tied to H20 and H200 sales to China, by our arithmetic, and China revenue down 21% in fiscal 2026, ahead of the 17 November results and the 18 November House briefing request.
A note on sourcing. Quotations are from our machine transcript of the 122-second clip Bloomberg posted on 9 November 2026, with sentence punctuation restored. Figures are from Nvidia’s Form 10-K for fiscal 2026 and Form 10-Q for the quarter to 26 July 2026, read via SEC EDGAR; the $4.9 billion total and the 21% fall are our arithmetic. We did not contact Nvidia.
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.
