NVIDIA To Claw Back Lost China Market With H200 Approval After US Curbs Crushed Its 95% Share Down To Nothing

NVIDIA Eyes China Comeback as H200 Approval Opens Door After US Curbs Wiped Out Its Dominance

NVIDIA H200 GPU sales to China get U.S. approval as 10 major firms receive clearance

The U.S. government has reportedly approved NVIDIA’s sale of Hopper H200 AI GPUs to China, marking a potentially important shift in the ongoing technology restrictions between Washington and Beijing. The decision allows around 10 Chinese companies to purchase NVIDIA’s H200 chips, although shipments have not yet begun.

The approval comes as high-level discussions between U.S. President Donald J. Trump and Chinese President Xi Jinping are taking place, with several major technology executives also involved in the broader diplomatic visit. NVIDIA CEO Jensen Huang reportedly adjusted his schedule at the last minute to join the trip, as the company continues working to regain access to one of the world’s biggest artificial intelligence markets.

According to people familiar with the matter, the U.S. has cleared several leading Chinese technology companies to buy NVIDIA’s Hopper H200 GPUs. The approved list is said to include Alibaba, Tencent, ByteDance, JD.com, and other major firms. Local distributors in China, including Lenovo and Foxconn, have also reportedly received approval to help meet market demand.

Each of the 10 approved Chinese firms may be allowed to purchase up to 75,000 NVIDIA H200 GPUs. If every company buys the maximum permitted amount, total sales could reach 750,000 units. However, despite the approval, no H200 chips have reportedly been delivered to China so far.

The NVIDIA H200 is one of the company’s most powerful AI accelerators and remains highly sought after by companies building large-scale artificial intelligence systems. While it is not NVIDIA’s newest architecture, the H200 still offers significant performance for AI training, inference, and data center workloads.

For NVIDIA, the approval could represent a small but meaningful opening in China after years of export restrictions. The company once held a dominant position in the Chinese AI chip market, with an estimated share of around 95% before U.S. trade curbs disrupted sales. Since then, NVIDIA’s presence in China has fallen sharply, while domestic chipmakers such as Huawei have gained momentum as Chinese firms were encouraged to adopt locally developed AI hardware.

Jensen Huang has previously stated that NVIDIA’s official market share in China has effectively dropped to zero. That makes the H200 approval especially important, even if the total number of GPUs allowed under the reported agreement remains limited compared with demand from large U.S. AI companies.

By comparison, some American AI firms are already operating massive GPU clusters. Elon Musk’s xAI, for example, is said to have more than half a million GPUs deployed across its Colossus and Memphis facilities. Other major U.S. AI players are also investing heavily in NVIDIA Hopper and Blackwell systems as the race for artificial intelligence infrastructure accelerates.

Even if all 750,000 H200 GPUs are eventually shipped to China, the deal may not be enough to restore NVIDIA’s former position in the country. Still, it could serve as a starting point if U.S.-China technology relations continue to improve.

There has also been speculation that NVIDIA may eventually offer China-specific Blackwell-based GPUs, such as a rumored B30 model, once newer architectures become available elsewhere. However, NVIDIA has made clear that its most advanced AI chips, including Blackwell and Rubin, are not intended for sale to China under current restrictions.

The situation remains uncertain, as government approval does not automatically guarantee immediate deliveries. Export compliance, supply availability, distributor coordination, and political developments could all affect when or whether the H200 chips reach Chinese customers.

For now, the reported approval signals a cautious reopening of NVIDIA AI GPU sales to China. It also highlights the delicate balance between national security concerns, global AI competition, and the enormous commercial demand for high-performance computing hardware.