Commerce Official: Nvidia H200 AI Chips Shipped to China

A senior U.S. trade official indicated on Tuesday that the volume of Nvidia’s H200 artificial intelligence chips shipped to China and Hong Kong has been “very few.” This statement signals a potential resumption of these high-demand components flowing into the Chinese market, which could provide a notable boost to Nvidia’s already surging revenues.

Under Secretary of Commerce for Industry and Security, Jeffery Kessler, stated at a congressional hearing, “The bottom line is very few shipments against licenses for H200s and equivalents have taken place. It’s a very small quantity of chips.” This is a significant development, as Nvidia has previously excluded any anticipated revenue from China in its financial forecasts. In May, CEO Jensen Huang explicitly advised investors to “expect nothing” from Chinese sales, reflecting the stringent export controls in place.

An Nvidia representative declined to comment on the matter. The company has long aimed to serve the substantial Chinese market for AI development but has been navigating complex geopolitical currents and the ongoing trade and technology disputes between Washington and Beijing. Most of Nvidia’s advanced AI products have been subject to export restrictions.

The path for H200 sales to China has been a complex one. In December, former President Donald Trump reportedly proposed that the U.S. government would approve China’s purchase of H200 AI chips in exchange for a 25% revenue share. Licenses for these chips, which some within the administration have flagged for potential military applications, began to be issued earlier this year to a limited number of Chinese firms.

It’s important to note that the H200, while powerful, belongs to Nvidia’s previous Hopper generation of chips. American companies are currently transitioning to the even more advanced and potent Blackwell architecture. The U.S. government’s vetting process for H200 chip applications involves a rigorous case-by-case assessment, requiring applicants to meet stringent national security requirements and undergo inspections to ensure compliance with export regulations. Kessler confirmed that license applications are indeed denied if they do not meet these criteria.

The ultimate impact on China’s AI development capabilities remains to be seen. Without access to cutting-edge Nvidia hardware, Chinese enterprises are largely reliant on domestic alternatives. However, these locally produced chips are widely considered to be less capable for advanced AI training and inference tasks compared to their U.S. counterparts, potentially creating a performance gap.