The H200 is an older generation of Nvidia's AI chips, built on its Hopper architecture launched in 2022.
The H200 is an older generation of Nvidia's AI chips, built on its Hopper architecture launched in 2022.US President Donald Trump on Monday said his administration will allow Nvidia to ship its H200 artificial intelligence chips to “approved customers” in China and select other countries. The clearance will come with strict conditions, including a requirement that 25% of the proceeds be paid to the United States, Trump said on his Truth Social platform.
The US Department of Commerce is now finalising the operational details, news agency Reuters reported. Trump added that the same framework would apply to other American chipmakers, including AMD and Intel, signalling a broader shift in the US tech-export stance.
However, it remains uncertain whether Beijing will allow Chinese companies to buy them.
The announcement sent Nvidia shares jumping as much as 3% on Monday, as markets briefly celebrated what looked like a reopening of a crucial revenue lane. But the optimism may be premature.
According to the Financial Times, Beijing is preparing to restrict access to Nvidia’s H200 chips despite Washington’s approval. China is reportedly seeking tighter control over foreign AI hardware to reduce dependency and accelerate its push for domestic alternatives.
The H200, based on Nvidia’s 2022 Hopper architecture, is no longer the company’s cutting-edge product. It predates the far more powerful Blackwell-generation chips, which now dominate US AI development. A recent analysis by the Institute for Progress (IFP), a non-partisan Washington think tank, shows just how wide the gap has become:
Blackwell chips are 1.5× faster than the H200 for training AI systems
They are 5× faster for inferencing – the real-world deployment of AI models
The export clearance therefore gives China access to older chips, not Nvidia’s current AI backbone.
Still, Nvidia remains caught in an increasingly intense geopolitical struggle between Washington and Beijing, with AI emerging as the new battleground for technological dominance. President Trump and Chinese President Xi Jinping have both framed AI leadership as a national priority, leaving companies like Nvidia to navigate rapidly shifting export rules.
Even before Trump’s latest announcement, China’s importance to Nvidia had already declined sharply. CFO Colette Kress said on the company’s third-quarter earnings call that the China-specific H20 chips generated only about $50 million, a tiny sliver of Nvidia’s $57 billion in revenue during the first three quarters of the fiscal year.
While Monday’s decision could reopen some supply channels, analysts point out that the limited scope of approved shipments—and China’s own tightening—means the financial upside could be modest. What’s clearer is that the US is seeking a new middle path: allowing controlled exports while extracting economic value and maintaining strategic advantage in next-generation AI hardware.