The United States has granted approximately 10 Chinese companies authorization to purchase Nvidia’s H200 artificial intelligence chips, yet not a single delivery has reached any of them. The approved buyers include Alibaba, Tencent, ByteDance, and JD.com. Authorization has also been extended to distributors Lenovo and Foxconn, through whom Chinese customers may route their purchases rather than buying from Nvidia directly. US licensing terms cap individual purchases at 75,000 chips per approved company, according to Reuters.
Lenovo confirmed that it “is one of several companies approved to sell H200 in China as part of Nvidia’s export license.” Nvidia, Alibaba, Tencent, ByteDance, JD.com, and Foxconn did not respond to requests for comment, according to Reuters.
The approvals exist on paper. The chips are not moving. Transactions froze after Chinese companies backed away in response to direction they received from Beijing. Commerce Secretary Howard Lutnick told a Senate hearing last month that “the Chinese central government has not let them, as of yet, buy the chips, because they’re trying to keep their investment focused on their own domestic industry.”
Why Beijing Is Blocking the Purchases
A desire to protect China’s domestic chip industry from foreign competition is at the heart of Beijing’s resistance. The deal structure itself has also created friction. Under a framework assembled by President Donald Trump, Washington would take a 25% cut of chip sale revenues. Because American law prohibits levying export fees directly, the arrangement routes the chips through US territory on their way to China — an arrangement that has reportedly created additional complications on the Chinese side.
US Trade Representative Jamieson Greer, speaking on Friday after the conclusion of the Trump-Xi summit in Beijing, framed Beijing’s reluctance as a sovereign choice. “Allowing the H200 imports would be a ‘sovereign decision’ for China,” he said. He also acknowledged that Chinese policymakers view US technological leadership in AI chips as a potential threat to their own growth trajectory. “They often see US high tech sometimes as a threat to them because if we’re ahead of the game like we are on AI chips, sometimes they feel that can stop their own growth,” Greer said.
The calculus on the Chinese side is not purely political. Chinese AI firms have been increasingly vocal about their reliance on domestic chips. Alibaba, Tencent, and ByteDance are snapping up Huawei’s Ascend 950PR, which entered mass production in April. Huawei’s Ascend 950 production scaled to 750,000 units in 2026. As domestic alternatives improve, the commercial urgency of importing H200s diminishes from Beijing’s perspective, even as computing power shortages continue to force some Chinese AI models to ration user access.
Jensen Huang’s Last-Minute Trip to Beijing
Nvidia CEO Jensen Huang was added to the White House delegation at the last minute after Trump personally invited him. Huang boarded the presidential party in Alaska as Trump was en route to meet Xi Jinping. Speaking to CCTV, China’s state broadcaster, Huang expressed hope that the Beijing talks would strengthen the personal bond between Trump and Xi and lead to warmer relations between the two countries.
The value of China’s AI market this year could reach $50 billion, by Huang’s own reckoning. When US export controls were less stringent, Nvidia’s share of China’s advanced chip market stood at roughly 95%, with Chinese sales representing around 13% of the company’s overall revenue. That position has been dramatically eroded since the Biden administration’s restrictions took effect.
The Trump administration’s H200 approval was announced in December 2025 with additional conditions added in January 2026, raising hopes that the summit might produce a framework for completing actual sales. Those hopes were tempered on Friday when Greer confirmed that chip export controls were not a major topic of discussion at the bilateral meeting. “We did not talk about chip export controls at the meeting,” he said.
The Domestic Chip Strategy Complicates Everything
The H200 deadlock reflects a deeper structural tension. China’s tech giants are under pressure from Beijing to prioritize domestic chip suppliers, both to enhance supply chain security and to support the development of China’s semiconductor industry. China has set a target of 80% chip self-sufficiency by 2030, and allowing large-scale purchases of Nvidia H200s would work against that goal by reducing the commercial incentive for Chinese firms to invest in domestic alternatives.
Anthropic’s policy paper, published on the same day as the Trump-Xi summit, cited a Financial Times report that Alibaba and ByteDance now train flagship models on export-controlled US chips in data centers in Southeast Asia, suggesting that Chinese firms are finding ways to access advanced computing even without domestic deliveries. That workaround does not require Beijing’s approval and does not benefit China’s domestic chip industry, but it also does not require the kind of visible, large-scale H200 imports that would be politically sensitive at home.
For Nvidia, the outcome is a continuation of the status quo: licenses exist on paper, but no chips are moving. The company’s path back into China’s advanced AI market depends on a policy shift, on both sides, that, after this week’s summit, appears no closer than before. The Trump administration had issued Nvidia export licenses to China in April in anticipation of the summit producing a breakthrough. It did not.