US Approves Nvidia H200 Chip Exports to China With Cap

• The Trump administration green-lights Nvidia H200 AI chip sales to China with conditions.
• Shipments cannot exceed half of the US domestic sales’ volume.
• White House’s balancing act between protecting US chipmaker revenues and limiting Beijing’s abilities.

The Trump administration has approved Nvidia H200 chip exports to China, but the green light comes with significant strings attached – a move that underscores Washington’s struggle to balance national security concerns against protecting the American semiconductor industry market share.

The US Department of Commerce’s Bureau of Industry and Security (BIS) confirmed Tuesday that Nvidia’s H200 artificial intelligence processor can now be shipped to China, but only if China-bound volumes remain at or below 50% of the company’s domestic US sales.

The restriction, taking effect Thursday (January 16), marks the administration’s attempt to thread the needle between curtailing China’s AI development and maintaining Nvidia’s access to the world’s second-largest economy.

The policy represents a notable change from the Biden administration’s approach. In January 2025, the H200 was placed under blanket export restrictions to China as part of the AI Diffusion Rule aimed at limiting foreign access to advanced American technology.

Under the new framework, the licensing policy moves from presumption of denial to case-by-case review – but with the important volume limitation acting as a built-in brake on China market expansion.

The compromise: Limited access, strict oversight

Each China-bound shipment of the Nvidia H200 chip must undergo verification by a US-headquartered third-party laboratory to confirm the processors’ technical capabilities and functions, according to BIS requirements.

The approval comes months after Nvidia CEO Jensen Huang met with President Donald Trump in Washington to discuss export controls – a lobbying effort that appears to have yielded partial success.

Trump indicated in December he would permit H200 sales to approved Chinese customers, though his suggestion that 25% of chip sales would be “paid” to the US government did not materialise in this week’s official BIS policy announcement.

Market implications for Asia-Pacific AI infrastructure

The Nvidia H200 chip export approval carries significant implications for enterprise AI adoption in the Asia-Pacific region, where Chinese technology companies, data centre operators, and cloud providers have faced constraints under US semiconductor restrictions.

Huang acknowledged strong Chinese market demand for H200 chips during last week’s CES technology conference in Las Vegas, stating that Nvidia was “firing up” its supply chain to meet what he described as “very high” customer interest from China.

However, the 50% volume cap limits how aggressively Nvidia can pursue Chinese market opportunities compared to unfettered domestic expansion.

The approval faces complications on both sides of the Pacific. Chinese state media outlets previously labelled Nvidia’s H20 chip – a less powerful processor specifically designed for the Chinese market – as “unsafe” after the company announced resumed sales in July 2025.

China’s Cyberspace Administration also summoned Nvidia that month to discuss alleged tracking and remote-control risks associated with the H20 processors.

According to a Tuesday report by The Informationthe Chinese government has indicated it will approve H200 chip orders only under exceptional circumstances, like for university research and development laboratories, potentially limiting commercial uptake despite US export approval.

Walking the tightrope on technology leadership

The Trump administration’s conditional approach to Nvidia H200 chip exports reflects the competing pressures facing US policymakers. On one hand, American AI firms like Nvidia argue that overly restrictive export controls simply cede market share to competitors without meaningfully slowing China’s AI development, given Beijing’s push for semiconductor self-sufficiency. On the other, national security advocates warn that providing China access to cutting-edge AI chips risks accelerating military and surveillance applications.

The 50% volume limitation appears designed as a middle path – allowing Nvidia to maintain some presence in the Chinese market while ensuring the company’s primary focus remains domestic sales and allied markets.

For Asia-Pacific technology companies dependent on cutting-edge AI infrastructure, the policy change suggests continued uncertainty in semiconductor supply chains and regulatory frameworks governing advanced chip access.

Regional enterprises will need to monitor both evolving US export policies and Beijing’s domestic chip development initiatives as they plan AI infrastructure investments.

Want to experience the full spectrum of enterprise technology innovation? Join TechEx in Amsterdam, California, and London. Covering AI, Big Data, Cyber Security, IoT, Digital Transformation, Intelligent Automation, Edge Computing, and Data Centres, TechEx brings together global leaders to share real-world use cases and in-depth insights. Click here for more information.