The U.S. has cleared Nvidia H200 sales to roughly 10 Chinese companies, but the politically sensitive chip deal remains stuck because no deliveries have been made, Reuters reported on May 14, 2026, citing three people familiar with the matter.
The development gives Nvidia a potential opening in a Chinese artificial intelligence market that Washington has tried to control through export restrictions. It also puts the Trump administration in the middle of a familiar fight: how to protect America’s technological edge without handing business to Chinese competitors by forcing U.S. companies out of the market.
The approved buyers include major Chinese technology firms such as Alibaba, Tencent, ByteDance and JD.com, Reuters reported, citing sources familiar with the licenses. Lenovo confirmed to Reuters that it is “one of several companies approved to sell H200 in China as part of Nvidia’s export license.”
The U.S. Commerce Department, which oversees export controls on advanced semiconductors, declined to comment to Reuters.
Nvidia H200 Sales Remain Caught Between Washington and Beijing
The approvals do not mean the chips are already moving into China. Reuters reported that not a single H200 delivery has been made so far, leaving Nvidia and its Chinese customers in limbo.
That distinction matters. In policy terms, the United States has opened a controlled door. In commercial terms, Nvidia still does not appear to have completed the sale.
The H200 is a powerful data-center chip used for artificial intelligence and high-performance computing. Nvidia says the H200 is based on its Hopper architecture and offers 141 gigabytes of HBM3e memory with 4.8 terabytes per second of memory bandwidth.
Those capabilities make the chip valuable for companies building or running large AI models. They also make it a flashpoint in the U.S.-China technology rivalry, where advanced computing power is increasingly treated as both a commercial asset and a national-security concern.
Washington’s export-control policy has tried to thread a narrow needle. A total cutoff could slow China’s AI development in the short run, but it also risks pushing Chinese firms toward domestic alternatives. A looser approach could preserve U.S. commercial influence, but critics argue it could strengthen a strategic competitor.
The Deal Reflects a Free-Market Opening With Security Strings
The Trump administration’s approach appears to be a controlled, licensed export system rather than an unrestricted sale. Reuters reported that U.S. rules issued in January require Chinese buyers to show they have “sufficient security procedures” and will not use the chips for military purposes.
That condition gives the administration a way to argue that it is not simply opening the floodgates. Still, the policy will face scrutiny from China hawks who view advanced AI chips as too important to sell into China under almost any condition.
The Wall Street Journal reported that Trump’s decision to allow Nvidia exports has drawn backlash from some Republican allies and national-security analysts who warned that the move could risk America’s lead in the technology race.
A more market-oriented view cuts the other way. Keeping U.S. firms present in China, under strict safeguards, may be preferable to forcing Chinese customers to standardize around domestic chip suppliers. That argument has become more urgent as Beijing pushes self-reliance in semiconductors.
Commerce Secretary Howard Lutnick told a Senate hearing last month that China’s central government had not yet allowed the purchases because it was trying to keep investment focused on its domestic industry, Reuters reported.
That is the central tension. Washington can authorize exports, but Beijing can still discourage or delay purchases if it wants to protect homegrown chipmakers.
Jensen Huang’s China Trip Raises the Stakes
Nvidia CEO Jensen Huang is in China as the issue moves back into the spotlight. Reuters reported on May 14 that Huang joined President Donald Trump’s state visit to China at the last minute while Nvidia seeks to preserve its presence in the world’s second-largest economy.
The Financial Times reported that Chinese President Xi Jinping told a group of American executives in Beijing that China’s door to business “will only open wider and wider,” with Huang among the U.S. business leaders present.
The optics are hard to miss. Nvidia is the world’s most important AI chip company. China is one of the most important AI markets. The U.S. government is trying to limit the strategic downside of selling advanced chips while still allowing American companies to compete.
That balancing act is not new, but the H200 approvals sharpen it. For Nvidia, each month of delay could mean lost revenue and weaker customer relationships. For Washington, each approved shipment carries political risk if critics later argue the chips strengthened Chinese military or surveillance capabilities.
Why the H200 Matters for AI Competition
Nvidia describes the H200 as a chip designed to accelerate generative AI, large language models and high-performance computing workloads. The company says the H200’s larger and faster memory improves performance for inference and scientific workloads compared with prior-generation systems.
For Chinese companies, that makes the H200 attractive even though Nvidia has already moved ahead with newer architectures. Access to H200 chips could help major Chinese internet and cloud companies support AI services, train or run models more efficiently, and remain competitive in a costly infrastructure race.
For U.S. policymakers, that same usefulness is the problem. Advanced AI chips are not ordinary consumer electronics. They can support commercial innovation, but they can also strengthen strategic capabilities when placed in the wrong hands.
A balanced policy should recognize both facts. It is reasonable for the U.S. government to restrict the most advanced chips from reaching a geopolitical rival. It is also reasonable to question whether Washington should use blunt restrictions that damage American companies while accelerating China’s push to build a rival domestic ecosystem.
The better test is whether licensing rules are enforceable, transparent and tied to real national-security risks rather than political theater. If the government allows sales, it should explain the guardrails clearly and enforce them consistently.
A Limited Win, Not a Completed Sale
For now, the H200 decision is best understood as a limited win for Nvidia, not a completed breakthrough. The U.S. approvals are important, but Reuters’ reporting that no deliveries have occurred shows the deal is still vulnerable to regulatory, diplomatic and commercial obstacles.
The market will likely treat the approvals as a sign that Washington is willing to give Nvidia some room in China. But the real test will be whether Chinese firms actually receive the chips, whether Beijing allows the imports, and whether the licensing system satisfies U.S. security concerns.
Until then, the story is not simply about Nvidia selling more hardware. It is about whether America can defend its technological lead without smothering the companies that built it.
