U.S. allows Nvidia H200 exports to China
Washington permits controlled sales with a 25?e
The U.S. administration will permit Nvidia to export its H200 artificial intelligence chips to China under a controlled framework, while collecting a 25% fee per chip, the U.S. president said. The fee, confirmed by a White House official as 25%, is higher than a previously discussed 15% rate. The Commerce Department is finalizing details and the same approach is expected to apply to other chipmakers such as AMD and Intel. The president said he informed Chinese leader Xi Jinping, who “responded positively.”
Officials described the decision as a compromise: it stops short of allowing the most advanced Blackwell chips to go to China while avoiding a total export ban that could push Chinese firms toward domestically developed alternatives or boost competitors like Huawei. The H200 is reported to be substantially more powerful than the H20—the most advanced chip currently permitted for export—but still below Blackwell-class processors used for large-scale AI training.
Under the anticipated licensing regime, shipments would be limited to civilian buyers with requirements for end-user reporting, system integration disclosures, and measures to prevent diversion to military-linked entities; licenses may include real-time monitoring and rapid suspension mechanisms. A White House official said the fee will be collected as an import tax in Taiwan, where the chips are manufactured, before U.S. security checks and onward shipment to China.
The announcement drew mixed reactions in Washington. Industry groups welcomed a path that stabilizes global sales and helps U.S. chipmakers fund research, while national security critics and some Democratic lawmakers warned that exporting high-performance AI chips risks accelerating China’s military and surveillance capabilities and called the move a major strategic error. Analysts say Chinese tech firms are likely to seek limited H200 purchases once licensing terms are clear, but Beijing has previously instructed domestic companies not to use certain U.S. technologies, leaving near-term uptake uncertain.
Nvidia shares rose modestly after the announcement. Administration officials framed the policy as a “practical recalibration” intended to preserve U.S. technological leadership while maintaining national security controls, acknowledging the delicate balance between strategic competition and economic interdependence.




