Nvidia H200 chip shipments to China have resumed at a small scale

A senior U.S. trade official said Tuesday that Nvidia has shipped only a small number of H200 artificial intelligence chips to China and Hong Kong. For everyday investors, the comment matters because it suggests a restricted revenue channel for Nvidia has reopened, even if the volume is still limited.

Jeffery Kessler, the under secretary of commerce for industry and security, told lawmakers at a congressional hearing that shipments tied to licenses for H200 chips and similar products have been minimal. “It’s a very small quantity of chips,” Kessler said.

Nvidia declined to comment, according to CNBC.

Why the H200 matters for Nvidia

The H200 is part of Nvidia’s Hopper generation of AI processors. These chips are used in data centers to train and run artificial intelligence models, a market that has powered much of Nvidia’s recent growth.

The chip is no longer Nvidia’s top U.S. product line. American customers are now using Blackwell chips, which CNBC described as faster and more powerful. Still, access to the H200 matters because China is one of the world’s biggest markets for AI development.

Nvidia has been caught between that demand and U.S. export controls, which are government restrictions on selling certain technology overseas. Washington has limited many advanced chip sales to China because of national security concerns, including worries that some products could support military uses.

Since last year, Nvidia has left possible China AI chip revenue out of its forecasts. CEO Jensen Huang told CNBC in May that he had told investors to “expect nothing” from China sales.

How the licensing process works

President Donald Trump said in December that the U.S. government would allow H200 sales to China in return for a 25% cut. CNBC reported that licenses for the chips were issued earlier this year.

Kessler told lawmakers the government reviews companies seeking H200 chips individually. He said applicants must satisfy national security requirements and agree to inspections designed to confirm compliance with the rules.

“There are cases where we deny the license applications we receive,” Kessler said.

A license is government permission to ship a restricted product. In this case, it does not mean any company can freely import Nvidia chips. It means approved buyers can receive shipments only if they meet the conditions set by U.S. officials.

China approval remains an open issue

Kessler’s testimony indicates some shipments have taken place, but it does not settle how large the market could become. CNBC reported that it remains unclear whether China will allow large volumes of H200 chips to be imported.

If Chinese companies cannot obtain Nvidia chips, CNBC reported they would have to rely on domestic alternatives that are viewed as weaker for AI training. AI training is the process of feeding data into a model so it can learn patterns and produce useful outputs.

For Nvidia investors, the key takeaway is narrow but meaningful: H200 sales to China and Hong Kong appear to have restarted under tight U.S. controls, while the scale of any future shipments depends on approvals from both Washington and Beijing.

This story draws on original reporting from CNBC.