China may be cooling on Nvidia’s H200 exports just days after the Trump administration signaled it would allow the chips to ship—an early sign that Washington’s latest attempt to undercut Chinese AI hardware rivals could be harder to pull off than expected.
White House AI and crypto czar David Sacks said on Dec. 12 that he had seen a news report suggesting China is rejecting Nvidia’s H200 in favor of domestically developed semiconductors. Crucially, Sacks cited an unspecified article, and Beijing has not publicly confirmed it is blocking imports—or formally rejecting them.
“They’re rejecting our chips,” Sacks said on Bloomberg Tech, arguing the motivation is straightforward: China wants semiconductor independence.
The strategy: sell “lagging” chips to compete inside China
The comments land just four days after President Donald Trump said on Dec. 8 that his administration would allow shipments of Nvidia’s H200 to China as part of a strategy to bring U.S. competition into China’s AI market—specifically to pressure Chinese champions like Huawei. The move was widely described as a compromise: H200 is part of Nvidia’s Hopper generation, behind Blackwell, and not considered cutting-edge relative to Nvidia’s newest roadmap.
Sacks has defended the logic as selling “lagging” technology rather than state-of-the-art hardware. If Chinese customers buy H200, the thinking goes, they’ll spend less time and money on domestic alternatives.
But Sacks’ latest remarks suggest China may be responding in the opposite direction—by doubling down on local suppliers.
Why Nvidia cares: a market it stopped counting on
Nvidia has already effectively written China out of its forward-looking narrative, removing the country’s data center market from forecasts as export rules tightened. Still, the upside remains hard to ignore.
CEO Jensen Huang has repeatedly described China as a massive opportunity (often referenced as a roughly $50 billion market in public comments), and Bloomberg Intelligence has estimated that H200 sales in China could represent a multi-billion-dollar annual opportunity—if China actually takes the chips.
In response to Sacks’ remarks, Nvidia said it is still working with the administration on H200 export licenses for vetted customers, but acknowledged there’s nothing to report yet. Nvidia also used the moment to argue that years of broad export controls have strengthened foreign competitors while costing the U.S. money.
China’s parallel push: incentives for domestic chips
Sacks’ skepticism also aligns with broader reporting that Beijing is preparing to spend heavily to reduce reliance on foreign silicon. Bloomberg reported on Dec. 12 that China is weighing incentives worth as much as $70 billion to support local chipmaking—an unmistakable signal that the country plans to keep backing its homegrown ecosystem even if Nvidia exports resume.
That context matters: even if the U.S. clears H200 shipments, China’s industrial policy may still steer customers toward domestic hardware—especially for politically sensitive or government-linked workloads.
The Huawei factor, and why it’s complicated
Sacks argued that China’s hesitation isn’t just “patriotism,” but policy—an effort to prop up Huawei. He pointed to Chinese support for local champions as the key reason H200 might struggle, even with U.S. approval.
The administration’s broader rationale reportedly included the idea that Huawei now offers AI systems with comparable “horsepower” via scale-out designs—linking large numbers of processors together to compensate for weaker performance per chip. If that’s the target, H200 exports were meant to be a wedge: not the best Nvidia can build, but good enough to compete and take share.
Sacks now suggests China has “figured that out” and may be blocking the strategy by steering demand away from Nvidia.
What’s confirmed—and what isn’t
Right now, the evidence is murky:
- Trump publicly said the U.S. would allow H200 exports (with licenses/conditions).
- Sacks says he saw reporting that China is rejecting the chips.
- China has not publicly agreed to accept H200 imports—and it also hasn’t publicly rejected them.
This isn’t the first time China has been cautious about U.S.-approved Nvidia products. Earlier in 2025, reporting suggested Chinese authorities discouraged domestic firms from buying the Nvidia H20, a less capable chip that became a focal point in the export-control tug-of-war.
For Nvidia, the near-term question is simple: can it sell meaningful volumes of H200 in China at all? For the White House, the bigger question is whether “controlled competition” works—or whether it simply accelerates the outcome Washington fears most: China building a full-stack AI supply chain that doesn’t need U.S. chips.