Why Trump is letting Nvidia sell advanced chips to China – and why it’s controversial

A White House move to let Nvidia sell advanced chips to China is blurring the line between national security and commercial gain.

For decades, US technology policy towards rivals like the Soviet Union and then China followed a simple idea: do not export your most advanced computing power to potential adversaries. That logic underpinned export controls on chips, supercomputers and tools used for weapons design and intelligence gathering. When Donald Trump returned to the White House, his administration initially stayed within that long tradition, the New York Times reported.

In recent months, however, that approach has shifted sharply. A group of powerful Silicon Valley figures, along with the president’s artificial intelligence adviser David Sacks, urged a different strategy. Rather than denying China access, they argued, Washington should pull it deeper into an “American tech stack” built on US hardware and software. That pitch has now produced one of the most controversial decisions of Trump’s second term.

On social media, Trump announced that Nvidia would be allowed to sell its H200 accelerator chip to “approved customers” in China and other countries, under conditions the White House has not spelled out. For Nvidia, now the world’s most valuable listed company by market value, it is a major commercial victory. For China, whose leading AI firms complain their progress is throttled by a shortage of high-end compute, it promises a powerful boost in the race with US tech groups.

Trump added a flourish that immediately set off alarm bells among policy veterans. He declared that a quarter of all revenues from these sales would flow to the US government. In effect, export control choices that were once justified purely on security grounds are being packaged as a revenue opportunity for Washington. Legal experts point out that current law does not recognise the sale of export licences in this way.

This is not the first time the administration has tried to split the difference. Earlier, the White House had approved a cut-down Nvidia product called the H20 for the Chinese market, pitched as a way to let US companies compete with Huawei without handing over their very best silicon. Sacks defended that compromise as a way to stop Huawei from capturing China’s data centre market and recycling those profits into research that would erode US advantages.

Beijing was unimpressed. Chinese buyers quietly balked at a deliberately weakened chip and treated the offer as insulting. Whether by design or not, that refusal increased pressure from Nvidia and its chief executive Jensen Huang for approval of something more powerful. The eventual green light for H200 suggests that the attempted middle path has given way to a much more permissive stance.

Critics of the move see it as a strategic mistake dressed up as a commercial win. Former Biden national security adviser Jake Sullivan, who helped design the previous generation of chip restrictions, argues that Washington is solving Beijing’s core problem in AI by handing over exactly the compute it needs. In his view, the notion that China will remain “hooked” on US chips is wishful thinking. The more plausible outcome is that advanced American hardware accelerates Chinese progress while domestic champions race to replace foreign suppliers.

There is also concern about what the decision means for allied coordination. Dutch equipment maker ASML agreed under US pressure to stop selling its most advanced chipmaking tools to China. If the United States now profits from selling high-end chips while asking partners to accept painful limits, governments in Europe and Asia may start to question why they should keep absorbing the economic cost of restraint.

Experts warn that a perceived double standard could undermine the coalition that sustains today’s export control regime. If Washington’s message is effectively that it can sell top-level chips while others must refrain from enabling Chinese fabs, resentment is likely to build. Analysts such as Rush Doshi note that allied discipline has been central to keeping the most sensitive tools out of Chinese hands. Looser US practice risks eroding that unity.

Inside the administration, officials describe the H200 decision as a compromise between security hardliners who wanted to ship nothing and corporate voices who wanted access to the whole Chinese market. For many national security traditionalists, it looks less like balance and more like a slide toward treating technology dominance as just another bargaining chip in a cash deal.

Trump has framed the move as smart deal-making that delivers money for both US companies and taxpayers. Nvidia keeps a lucrative growth engine, the Treasury gets a slice, and China’s access is supposedly managed through conditions that remain vague. Yet the long-term risks are harder to quantify and easier to ignore.

If China uses this breathing space to close the gap in advanced chips while friends and allies lose faith in US leadership on controls, the decision may come to be seen not as a clever stroke of transactional statecraft but as a turning point where Washington began to sell off one of its last clear technological advantages in return for near-term revenue.