Nvidia CEO Jensen Huang has firmly stated that China should not have access to its "latest and greatest" AI GPU technology, notably its Blackwell and Rubin lines, arguing that the United States should have a lead in AI technology. As per Nikkei Asia, Huang says Nvidia are "huge supporters of the United States having the first, the most, and the best."

The comments by the head of Nvidia came during the Milken Institute Global Conference in Los Angeles, an event attended by bankers, businessmen, and investors. He stated that the global reach of American AI accelerators helps increase tax income, which strengthens the economy and in turn supports national security. At the same time, Huang emphasized that China should not receive the 'latest and greatest' AI accelerators, such as the current-generation Blackwell or next-generation Rubin. This is a way for the U.S. to maintain its lead in the frontier AI models.

Back in November, the U.S. government approved sales of Nvidia's H200 processors to approved customers in China, provided that they would pay 25% cut to the U.S. government. The company confirmed in February that it had received orders from numerous customers in the People's Republic. But in its most recent conference call, Nvidia said it had not shipped any H200 processors to Chinese customers recently. By contrast, AMD, which also got permission to sell its Instinct MI308X, sold $390 million worth of AI GPUs to its clients in China.

More recently, Huang admitted that Nvidia's AI accelerator market share in China had dropped to zero percent, suggesting that the company did not ship any H200 GPUs to the PRC in the first calendar quarter of 2026.

There might be a reason for Nvidia to avoid shipping Hopper GPUs to China: Both Hopper and Blackwell processors are made at the same fabs and N4/N5-capable production lines by TSMC, and since capacity is limited, Nvidia aims to sell more advanced and expensive Blackwell processors to its customers in the U.S. This in terms means avoiding the U.S. government's 25% cut rather than selling Hopper GPUs to clients in China.

Once Nvidia ramps up production of its next-generation Rubin GPUs using TSMC's N3 process technology, the Taiwan-based foundry will be able to allocate more N4/N5 capacity to Hopper GPUs. It's possible to speculate that Nvidia will likely resume shipments of Hopper accelerators to China.

In any case, earlier this year, Jensen Huang said that Nvidia's clients in China would get Blackwell GPUs in time, but did not elaborate, which probably means that when Rubin GPUs are available widely, some cut-down versions of Blackwell may be supplied to customers in the PRC.

Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.