Nvidia will stop selling H200 AI chips in China as U.S. tightens access
Acting CEO Colette Kress says strong chip demand will shield total revenue, while Jefferies forecasts a multibillion-dollar hit to the data-center business from new restrictions.
Nvidia acting CEO Colette Kress said on November 17 that the company will stop selling H200 AI chips in China, closing a tariff-backed route previously negotiated by Jensen Huang and President Donald Trump. New U.S. rules also bar advanced chips from overseas data centers serving Chinese customers, putting facilities’ customer relationships under scrutiny. The financial impact remains contested: Jefferies estimates a $44 billion–$62 billion data-center revenue hit in 2027, while Kress says chip demand could protect total revenue.
01
The former H200 route required shipments to pass through the United States, where the government could collect a 25% tariff.
02
China had largely declined H200 deliveries, despite demand for Nvidia equipment among some major computing users.
03
TechCrunch reported evidence that leading Chinese AI models trained on Grace Blackwell chips, but the training location is unconfirmed.
Nvidia will stop selling H200 AI chips in China, acting CEO Colette Kress said November 17, according to TechCrunch. Yet Kress maintained that demand for advanced processors would keep new U.S. export restrictions from hurting total revenue—a confident assessment alongside Jefferies’ forecast of a multibillion-dollar reduction in data-center revenue next year.
The disclosure came during Nvidia’s first earnings call since co-founder Jensen Huang stepped down as chief executive the previous week. It ends a sales arrangement Huang had brokered with President Donald Trump to allow H200 shipments into China.
A tariff-backed sales route closes
Under that arrangement, the chips had to pass through the United States, where the government could collect a 25% tariff. The pathway allowed Nvidia to sell H200s in China despite wider U.S. restrictions on the country’s access to advanced processors.
The arrangement exposed a policy disagreement. China hawks wanted to deny Chinese laboratories computing power to preserve the U.S. lead in AI. Huang argued instead that American technology companies needed to compete in China, telling podcaster Dwarkesh Patel in April that conceding the market would disserve the United States.
Permission to sell had not translated into broad acceptance. China had largely refused H200 deliveries before Nvidia’s decision, TechCrunch reported. Jacob Feldgoise, a researcher at Georgetown’s Center for Security and Emerging Technology, said Beijing preferred to support domestic chip demand, even as major computing users wanted Nvidia equipment.
Restrictions follow customers overseas
The withdrawal follows U.S. rules issued November 16 barring advanced American chips from foreign data centers that serve Chinese customers remotely. Those restrictions reach beyond hardware shipped into China: they also target facilities abroad where Chinese customers can use computing power without importing the processors themselves.
TechCrunch reported that leading Chinese AI models show evidence of training on Nvidia Grace Blackwell chips, which cannot legally be exported to China. That training may have occurred at overseas data centers; the location was not established.
Kress points to the order backlog
Kress offered a different assessment of the restrictions. She said the backlog of demand for advanced chips, including Vera Rubin, meant the rules would have no effect on Nvidia’s top-line revenue.
The two assessments use different measures. Jefferies described a possible reduction in data-center revenue; Kress addressed total company revenue. Neither is a realized result.
For large AI-training data centers seeking Nvidia chips, the practical test is their customer base. Export-control attorney Dan Fisher-Owens told TechCrunch that buyers will have to certify they are not serving Chinese customers. Facilities unable to do so may have to turn to other chips.
Sources
techcrunch.comNvidia is abandoning China after Jensen Huang's ouster | TechCrunch
Reader comments
Newest comments first. Replies stay oldest first.