In a significant policy shift that could reshape global AI hardware flows, the United States has granted NVIDIA approval to export its H200 AI chips to China—a move announced under a new export arrangement approved by President Donald Trump. The decision reopens a lucrative yet politically sensitive market for the world’s most valuable chipmaker, albeit with strict financial and national-security conditions.
Under the deal, NVIDIA will pay a 25% fee on every sale of H200 GPUs, imposed as an import tax when the chips are shipped from Taiwan to the U.S. for mandatory inspection prior to re-export. The fee applies not only to China but also to other eligible countries receiving the chip through this framework.
A Controlled Reopening of the AI Chip Market
The H200—NVIDIA’s second-most advanced AI processor, trailing only its next-gen Blackwell series—will now be available to vetted Chinese customers, subject to Commerce Department approval and security checks. President Trump said he informed Chinese President Xi Jinping of the decision, noting that Xi “responded positively.”
The move follows weeks of internal deliberations in Washington and direct engagement between U.S. officials and NVIDIA’s CEO Jensen Huang. It also signals a potential pathway for AMD and Intel, which may soon receive similar export permissions for their AI processors.
Earlier this year, NVIDIA regained limited access to China after the U.S. allowed the shipment of its less powerful H20 chips. AMD also resumed exports of its MI308 AI chips following regulatory clearance.
Balancing Security With Market Reality
The U.S. decision reflects a strategic middle ground—permitting the export of capable AI chips like the H200 while maintaining restrictions on top-tier processors such as NVIDIA’s Blackwell or upcoming Rubin-generation GPUs.
For China, the earlier restrictions had prompted swift policy responses. Beijing offered subsidized power and incentives to encourage domestic AI firms to shift away from NVIDIA in favor of regional alternatives like Huawei. Tech giants including Alibaba, Tencent, and ByteDance had complained of increased operational costs due to the limited access to U.S. chips.
The H200 gives China an option more powerful than the previously authorized H20, which was ultimately rejected by Chinese regulators on security grounds.
Industry Impact and Market Reaction
For NVIDIA, the approval could revive access to a massive revenue source. Analysts estimate that reopening the Chinese market— even partially—could drive billions of dollars in annual sales. Investors reacted quickly, pushing NVIDIA’s stock up about 2% in after-hours trading following the announcement.
Still, uncertainty looms. Despite U.S. leniency, it remains unclear whether China will accept the H200 under the new fee and inspection requirements. Speaking shortly before the policy change, NVIDIA CEO Jensen Huang acknowledged the ambiguity:
“We don’t know. We have no clue,” he admitted when asked whether Chinese firms would buy the chip.
A New Chapter in U.S.–China Tech Relations
The approval marks a cautious recalibration in the ongoing geopolitical tug-of-war over advanced semiconductor technology. While it signals Washington’s willingness to allow controlled exports, it also highlights the broader tension governing AI hardware supply chains.
Whether the H200 becomes a bridge or a battleground in U.S.–China tech relations now depends on how both nations—and their corporate champions—respond in the months ahead.
Read more similar post: Bill Gates’ Daughter Builds Phia, an AI Tool Revolutionising Product Discovery
