Nvidia Weighs H200 Production Ramp as China Orders Surge After U.S. Export Green Light

Nvidia is considering ramping up production of its H200 graphics processing units as demand from Chinese customers accelerates following fresh approval from the Trump administration to resume sales of the chips to China, according to a Reuters report citing people familiar with the matter.

The H200 is the most powerful chip from Nvidia’s previous Hopper generation and is widely used to train large language models and other advanced artificial intelligence systems. Until recently, the chip was effectively off-limits to China after the Biden administration proposed tighter export controls aimed at restricting the flow of cutting-edge AI hardware to Chinese firms.

That position shifted last week when the U.S. Department of Commerce cleared Nvidia to sell the H200 in China, under an arrangement that requires the company to remit 25% of sales from those chips. The approval reopened a critical market for Nvidia at a time when Chinese technology companies are racing to secure compute capacity to remain competitive in AI development.

According to Reuters, interest from China has been strong enough that Nvidia is now examining whether to add manufacturing capacity for the H200, which is currently produced in limited volumes. Chinese firms have been moving quickly to place orders, reflecting pent-up demand created by years of export restrictions that forced many developers to rely on less capable hardware.

Chinese regulators, however, have not yet given final clearance for the chips to be imported, and discussions are ongoing within Beijing over whether to allow large-scale purchases. If approved, the H200 would represent a significant step up from the H20 chips that Nvidia previously tailored for the Chinese market to comply with U.S. restrictions. The H20, while compliant, is widely seen as a compromised alternative with reduced performance.

Several major Chinese technology companies are already in talks with Nvidia about potential orders. Firms such as Alibaba and ByteDance, both of which are building proprietary AI models, are said to be assessing how many H200 chips they can secure if imports are approved. For these companies, access to more capable GPUs could shorten training cycles and narrow the performance gap with U.S.-based rivals that have had uninterrupted access to Nvidia’s latest hardware.

“We are managing our supply chain to ensure that licensed sales of the H200 to authorized customers in China will have no impact on our ability to supply customers in the United States,” an Nvidia spokesperson said in an emailed statement.

The possible production increase highlights Nvidia’s delicate balancing act. On one hand, China remains one of the world’s largest markets for data center and AI hardware, and renewed access offers Nvidia a chance to unlock substantial incremental revenue. On the other hand, the company must navigate U.S. national security concerns and reassure policymakers that expanded China sales will not undermine domestic supply or strategic objectives.

For China’s AI sector, the development points to how export controls have reshaped priorities. With access to top-tier hardware constrained, many Chinese firms have focused on improving model efficiency and software optimization rather than brute-force scaling. The return of a chip as capable as the H200 could shift that balance, at least temporarily, even as Beijing continues to push for homegrown alternatives to reduce long-term reliance on U.S. suppliers.

More broadly, the episode illustrates how geopolitics is now directly shaping supply chains in the AI industry. Decisions about chip production, allocation, and pricing are no longer driven solely by market demand but by negotiations between governments, regulators, and corporate giants. Nvidia’s consideration of a production ramp suggests that, for now, demand from China remains too large to ignore — even as the rules governing that demand continue to evolve.