Trump has approved the export of Nvidia’s H200 chips to China, on the condition that 25% of the related revenue must be handed over to the U.S. government.
However, the latest Blackwell series and the upcoming Rubin chips are still banned from export, meaning China can only purchase second-tier products.
Given Nvidia’s extremely high gross margin of 75%, even if the company absorbs that 25% levy itself, it would still be profitable—and in most cases, the cost will likely be passed on to Chinese customers.
Jensen Huang has long argued that the more China relies on Nvidia and the CUDA ecosystem, the more the U.S. can control the standards and direction of AI technology development.
Instead of letting that money flow to Huawei’s chip R&D—strengthening a potential competitor—it’s “better” for the U.S. government to take a cut.
Comments