In China, no Nvidia H200 units were delivered despite the approval, said the U.S. Department of Commerce.

A New Perspective on Nvidia H200 Accelerator Shipments to China

At the end of last year, U.S. President Donald Trump officially authorized the export of Nvidia H200 computing accelerators to China, yet no such unit has reached Chinese shores.

During a meeting of the U.S. Parliament’s International Affairs Committee, Deputy Trade Secretary for Export Control David Peters clarified: “No accelerator has been shipped to China.”

It is believed that the decision to allow shipments was made under pressure from Nvidia founder Jensen Huang. In his view, maintaining Chinese AI developers’ dependence on American component supplies would let the United States keep influence over its main geopolitical rival. A full ban on shipments, according to their calculations, would accelerate import substitution in China and push the country out of industry leadership. The H200 accelerators are built on Hopper architecture and offer moderate performance, giving the U.S. an “acceptable edge” in developing its own AI sector.

Opponents of this policy argue that American accelerators could become a key to enhancing China’s military capabilities. In fact, H200 shipments face several restrictions from both the U.S. and China. The Chinese authorities are wary of local developers’ reliance on foreign components and are ready to impose limits on subsidizing projects that depend too heavily on imports.

In response to recent reports of Nvidia accelerator smuggling into China, companies such as DeepSeek and Deputy Trade Secretary Peters confirmed: “Smuggling is indeed occurring. We are actively working on this issue and consider it one of the priorities of our restriction policy.”