US-China Deal Lets Nvidia Ship High-End AI Chips to China

US President Donald Trump announced that he reached an agreement with Chinese President Xi Jinping allowing Nvidia to export certain advanced artificial intelligence semiconductors to China, signaling a shift in US export policy that had previously restricted high-performance chip sales.

Trump said he informed Xi that Washington would permit Nvidia to ship its H200 graphics processing units to “approved customers in China and other countries, under conditions that ensure strong national security.” He added that Xi “responded positively” and stated that “25% will be paid to the United States,” without clarifying the mechanism.

The announcement partially reverses export controls enacted under President Joe Biden, which had blocked the sale of H200 units and similar high-performance GPUs. Those rules required chipmakers to produce weaker or modified versions specifically for China to comply with security restrictions. Trump criticized that framework, saying it forced companies “to spend billions of dollars building ‘degraded’ products that nobody wanted,” which he argued slowed innovation and “harmed the American worker.”

Trump described the new decision as a measure to “support US employment, strengthen American manufacturing, and benefit US taxpayers.” However, he emphasized that Nvidia’s most advanced technologies are excluded. The Blackwell line and the upcoming Rubin processors will remain restricted to US customers. Launched 2Q24, the H200 trails Nvidia’s newest platforms by roughly 18 months and had been fully restricted under prior rules.

The Commerce Department is finalizing implementation procedures, and Trump said the same approach would apply to AMD, Intel, and other major chipmakers. The policy shift comes as the United States and China continue competing for AI leadership amid ongoing trade tensions. Nvidia CEO Jensen Huang has lobbied the White House to relax previous restrictions, despite opposition in Washington to expanding China’s access to US semiconductors. GPU technology remains central to training generative AI systems, a strategic focus for both countries since 2022.