U.S. Confirms Limited Nvidia H200 Chip Exports to China

U.S. Approves Minimal H200 Shipments Under Strict Licensing

The United States has permitted limited exports of Nvidia's H200 artificial intelligence chips to China and Hong Kong, according to testimony from a senior Commerce Department official. Under Secretary Jeffery Kessler told a congressional hearing Tuesday that "very few" H200 chips have been shipped under the licensing program established earlier this year.

"The bottom line is very few shipments against licenses for H200s and equivalents have taken place. It's a very small quantity of chips," Kessler stated, according to CNBC, which first reported the remarks.

The confirmation marks a cautious reopening of AI chip sales to China after years of escalating export restrictions. Nvidia has been caught in the crossfire of U.S.-China technology competition, with Washington blocking most of the company's advanced products from reaching Chinese buyers while the chipmaker seeks access to one of the world's largest AI development markets.

Why It Matters

This development signals that Washington is willing to permit some AI chip exports to China under controlled conditions, potentially creating a template for managed technology transfer. For Nvidia, even limited sales could provide incremental revenue in a market the company has publicly written off. More broadly, the licensing approach reflects an attempt to balance commercial interests against national security concerns—a tension that will define technology policy for years as AI capabilities advance.

Licensing Process Includes Security Reviews and Inspections

The Commerce Department evaluates each H200 license application individually, Kessler explained. Applicants must satisfy national security requirements and agree to inspections verifying compliance with export rules. The government has denied some applications, though Kessler did not specify how many.

In December, President Donald Trump announced that the U.S. would approve H200 sales to China in exchange for a 25% cut—an arrangement that drew attention for its transactional framing. Licenses began flowing earlier this year, though the quantities remain small.

The H200 belongs to Nvidia's Hopper generation of chips, which are now considered older technology. American companies have moved on to the more powerful Blackwell architecture, giving U.S. firms a performance advantage over Chinese competitors limited to less advanced hardware.

Uncertainty Remains Over Chinese Import Approvals

Whether China will ultimately permit large-scale H200 imports remains unclear. Beijing must also approve the chips' entry, adding another layer of complexity to the export process.

Without access to Nvidia's products, Chinese AI developers have turned to domestic alternatives, which industry observers consider inferior for training large models. This technology gap could widen as American firms deploy next-generation chips while Chinese companies work with older architectures.

Nvidia has publicly downplayed expectations for Chinese revenue. CEO Jensen Huang told CNBC in May that he advised investors to "expect nothing" from sales in the country. The company has excluded potential Chinese AI chip revenue from its forecasts since last year. An Nvidia representative declined to comment on the latest developments.

Details of the licensing program and shipment volumes were first reported by CNBC.