· via TechCrunch
Nvidia stops selling H200 chips in China, closing US export ban loophole
Nvidia says it will no longer sell H200 chips in China, ending a tariff-backed workaround to US chip export controls just as new rules block foreign data centers from serving Chinese customers remotely.

Nvidia will no longer sell its H200 chips in China, acting CEO Colette Kress told investors on Tuesday, according to TechCrunch. The move ends the company's role in a contested workaround to Washington's restrictions on advanced chip exports and signals a decisive shift in how Nvidia approaches the Chinese market.
The disclosure came on Nvidia's first earnings call since co-founder Jensen Huang stepped down as chief executive the week before. The same day, the company said it had formed a board-level committee to oversee national security policy, export control compliance and global trade risks, TechCrunch reports — an acknowledgment of how heavily geopolitics now weighs on the business.
How the loophole worked
The H200 arrangement was personally negotiated by Huang and President Donald Trump, according to TechCrunch. Under the deal, chips had to be shipped through the United States, which allowed the government to collect a 25% tariff on the sales.
That pathway drew fire from China hawks in Washington who wanted to deny Chinese labs computing power and preserve the American lead in AI. Huang consistently argued the opposite, telling podcaster Dwarkesh Patel in April that conceding the Chinese market "is a disservice to our country" for the US technology industry.
Huang's exit was itself entangled with the export story: TechCrunch reports his ouster came in part because of an affair with the CEO of a company under investigation for smuggling chips into China.
New rules close the remote-compute route
The H200 withdrawal lands alongside regulations issued Monday that bar advanced US chips from foreign data centers serving Chinese customers remotely. Investment bank Jefferies estimates those rules could shave between $44 billion and $62 billion off Nvidia's data center revenue next year. Kress, however, said the backlog of demand for chips such as Vera Rubin means the rules will have no effect on the company's top-line revenue.
The longer-term exposure is harder to dismiss. TechCrunch notes Nvidia may lose potential customers in Singapore, Malaysia, Thailand and Indonesia, and critics of its China dependence claim as much as 20% of its revenue is tied to Chinese demand — with the Singapore subsidiary alone reporting $104 billion in annual revenue in 2025 before Nvidia stopped disclosing the figure.
China was already pulling away
China, for its part, had mostly refused to accept H200 deliveries even before Nvidia's decision. Jacob Feldgoise, a researcher at Georgetown's Center for Security and Emerging Technology, told TechCrunch that Beijing would rather build demand for its domestic chip industry, even if China's biggest AI compute users still prefer Nvidia's superior hardware.
Chinese frontier labs have nonetheless kept finding Nvidia silicon. TechCrunch reports that every leading Chinese AI model shows evidence of having been trained on Grace Blackwell chips, which cannot legally be exported to China. That training may have occurred in data centers outside the country — exactly the kind of facility now covered by Monday's rules. Export control attorney Dan Fisher-Owens told TechCrunch that large AI training data centers wanting Nvidia chips will have to certify they serve no Chinese customers; those that cannot will likely turn to alternative suppliers, further cutting into Nvidia's market share.
Why it matters
With the H200 program dead and remote access to US chips closing, the era of calibrated, tariff-mediated chip sales to China is effectively over. For US policymakers who saw the arrangement as a leak in the export regime, this is the outcome they pushed for. For Nvidia, a deep backlog for its next-generation chips cushions the immediate hit, but the company is quietly writing off demand that critics estimate at up to a fifth of its revenue. And for China, the shutdown removes ambiguity: the country's AI buildout will now run increasingly on domestic accelerators, while its top labs lose the offshore compute channels that, on the technical evidence, they have been relying on all along.
- #nvidia
- #export-controls
- #ai-chips
- #china
- #semiconductors