
Nvidia has stopped shipping its most powerful United States-approved artificial intelligence chips to China, interim chief executive Colette Kress told investors on Tuesday, a sudden halt that throws the future of a carefully negotiated Washington-Beijing licensing deal into doubt days before President Donald Trump meets Chinese President Xi Jinping in Shenzhen. Kress said Nvidia sold $2.4 billion worth of licensed Hopper-generation data centre products to China in the quarter ended October 25, but those shipments have now ended. The news came on the same earnings call where Nvidia reported quarterly revenue of $115.8 billion, more than double a year earlier.
Kress, who is leading Nvidia as interim chief executive after Jensen Huang resigned on November 12 while also serving as chief financial officer, told analysts that Nvidia's guidance for the current quarter contains no licensed Hopper data centre products for China. "Our guidance does not have any licensed Hopper data centre products included, and we are no longer shipping licensed Hopper data centre products to China," Kress said. She offered no explanation and did not say whether the halt is temporary or permanent, or whether Washington or Beijing ordered it. The company has issued no separate statement.
The timing is striking. The $2.4 billion of licensed Hopper sales shows the China business was real and growing, even though it was a small slice of Nvidia's record data centre revenue of $107.6 billion. Now that revenue line has been switched off entirely, at least for now.
The United States first imposed export controls on advanced AI chips to China in 2022, aiming to slow Beijing's development of the most powerful artificial intelligence systems. That changed after a summit between Trump and Xi last year, when Trump said he would allow sales of the H200, the most powerful Hopper chip, to approved Chinese customers. The United States eased its H200 licensing policy in January 2026 after months of lobbying from Huang, and Washington approved H200 sales to 10 Chinese companies, including Alibaba, Tencent and TikTok owner ByteDance. The deal also imposed a 25 percent tariff on eligible chips imported into the United States before shipment to China.
Beijing, however, has moved the other way, restricting H200 imports to reduce dependence on American technology. Reuters reported earlier this month that around 600,000 H200 chips were stored in Hong Kong awaiting shipment to mainland China after Beijing withheld approval for large-scale imports, while several major Chinese internet companies had each obtained about 10,000 H200 chips in recent months. The halt disclosed on Tuesday closes the American end of that pipeline as well.
The Hopper generation sits at the heart of the global AI boom. The H200 is used to train and run advanced AI models, and demand for it has been intense from cloud providers, AI labs and internet giants around the world. Nvidia's data centre business produced a record $107.6 billion in the quarter just ended, up 110 percent from a year earlier, as part of total revenue of $115.8 billion. China was only ever a small part of that total, but it was strategically important, because every Hopper chip sold there carried an American licence and American conditions.
That is what makes the halt so significant. It is not a collapse in demand or a production problem: Nvidia's newest Vera Rubin systems are selling as fast as they can be built, and the company guided to $130 billion in revenue for the current quarter. The China halt is a policy and politics story, not a sales story.
Kress's words were the only official explanation Nvidia has given, and their brevity left analysts guessing. Treasury Secretary Scott Bessent had already accused Nvidia of knowingly selling powerful chips to China to "goose their revenues and their earnings" and demanded answers within 10 days. Beijing has denied the espionage claims surrounding Nvidia as "purely a smear", and Nvidia has said it has no reason to believe sensitive information was exposed, though its investigation is in its early stages. None of those statements explains why the shipments stopped.
The halt matters because the H200 deal was the clearest test of whether the United States and China could manage their technology rivalry through licences and conditions rather than outright bans. Washington approved the sales, Beijing restricted the imports, and now Nvidia has stopped shipping altogether. If the arrangement collapses, it would mark a return to a harder line on AI chips, with Chinese buyers pushed faster towards domestic alternatives and American chipmakers losing a licensed, legal channel into the world's second-largest economy.
All eyes now turn to Shenzhen on Thursday, November 19, where Trump and Xi are due to hold bilateral talks and a working lunch, followed by a first-ever three-way meeting with Putin. Whether the Hopper halt is raised there, and whether either leader offers a way back, will determine whether this is a pause or a permanent end to licensed H200 sales. Meanwhile, the roughly 600,000 H200 chips reported to be sitting in Hong Kong remain in limbo. Related coverage: Nvidia Q3 earnings, SEC probes Nvidia, H200 chips and China-linked firms.
Has Nvidia stopped selling AI chips to China?
Nvidia has stopped shipping its licensed Hopper data centre products to China, interim CEO Colette Kress said on November 17. Its guidance assumes no further licensed Hopper sales.
How much did Nvidia sell to China last quarter?
Nvidia sold $2.4 billion worth of licensed Hopper products to China in the quarter ended October 25, 2026, Kress said on the earnings call.
What is the Hopper H200 chip?
The H200 is a powerful processor in Nvidia's Hopper family, used to train and run advanced AI models. It is made for Nvidia in Taiwan by TSMC and was approved for sale to 10 Chinese companies, including Alibaba, Tencent and ByteDance.
Why did Nvidia stop the shipments?
Kress did not explain why Nvidia stopped shipping the chips or whether the halt is permanent. Beijing had already restricted H200 imports, and around 600,000 H200 chips were reported to be stored in Hong Kong.
What happens next?
President Donald Trump meets President Xi Jinping in Shenzhen on Thursday, November 19, followed by a three-way meeting with Russian President Vladimir Putin. The chip question is expected to feature in those talks.