Trump’s AI chip deal with Nvidia and AMD sets a dangerous precedent

It may also undermine Washington’s ability to win the AI race against China.

Expert comment

Published 11 August 2025

Updated 12 August 2025 — 4 minute READ

Image — US President Donald Trump listens as Nvidia CEO Jensen Huang speaks at the White House during an event on 'Investing in America' on 30 April 2025. Photo by Andrew Harnik/Getty Images

Katja Bego

Senior Research Fellow, Europe Programme

The news that the Trump administration has struck a deal with US chipmakers Nvidia and AMD, allowing exports of some of the companies’ chips to China in return for a 15 per cent cut of revenue (to be paid to the US federal government), has been met with disbelief from national security, economics and legal experts alike. 

US President Donald Trump said he brokered what he called a ‘little deal’ last week during a meeting with Nvidia CEO Jensen Huang in the White House. The significance of this decision is not primarily in China’s ability to access the chips themselves. The chips in question (Nvidia’s H20 and AMD’s MI308) are not cutting-edge and are unlikely to be gamechangers for China’s AI fortunes. 

More significant is the precedent this sets for the Trump administration to potentially weaponize national security arguments and export control licenses as leverage in deals with private companies.

This is in many ways an unprecedented move. It not only risks significantly undermining Washington’s global credibility and leverage when it comes to its economic statecraft agenda, but also reflects a more structural weakness in the US approach to its technological competition with China. 

National security risk or not?

The logic behind the scheme is highly questionable. While the deal will raise some revenue for the Trump administration, US chip policy has been driven primarily not by economic considerations, but instead by national security concerns. 

Successive US administrations – in a rare case of bipartisan consensus – have placed progressively more restrictive limitations on China’s ability to access the highest-end US-designed chips and other key components and equipment necessary for producing them. The primary justification here is that these are dual use technologies that could significantly aid China’s military build-up.

While most restrictions on the highest-end chips and equipment remain in place, slightly less advanced chips have been subject to more debate. It was in fact Trump back in April who first brought limiting the export of the H20 to the table – a chip Nvidia specifically designed for the Chinese market that adheres to the limits imposed by the US government. 

This is a clear case of either all-in or all-out. 

If the Trump administration is serious in its beliefs that the security risks that come with exporting specific chips and components, such as Nvidia’s N20, are unacceptably high, it needs to continue to restrict all such exports. This is a clear case of either all-in or all-out. 

A rather arbitrary 15 per cent levy to be made to the federal government will not help alleviate Washington’s security concerns. If anything, it may do the opposite: it could leave US companies less profitable (and thus less able to compete), while undermining the credibility of the necessity of the export regime in the process. 

Divided on China

The decision also reflects wider disarray over China within the Trump team. This deal is almost certain to leave the China hawks in the administration deeply displeased. 

Only last week, several influential Washington voices, including Matt Pottinger, one of the chief architects of the first Trump administration’s approach to China, penned an open letter calling the lifting of controls on H20 chips ‘a strategic misstep that endangers the United States’ economic and military edge in AI.’  

A rather arbitrary 15 per cent levy to be made to the federal government will not help alleviate Washington’s security concerns.

But instead of the continued restrictions they asked for, this new scheme appears to muddle national security, geoeconomic and trade objectives, fused with Trump’s own idiosyncratic approach to striking bespoke deals with individual companies and countries. This arguably also mirrors the apparent confusion underpinning the administration’s wider approach to China, which alternates between tough and conciliatory approaches. 

A concerning precedent

The deal sets a concerning precedent with long-term ramifications. It suggests that other companies active in strategic industries could potentially in future pay their way out of burdensome and complex export control regimes, even if they involve key US national security concerns.  

This kind of arrangement could conversely also set the scene for the Trump administration to more broadly exercise its ability to control export licensing to influence companies whose supply chains involve the US (which in high-end tech, is no small number). 

A move in this direction also risks undermining the credibility of Washington’s national security arguments, which could erode the effectiveness of the US’s wider approach to export controls.

Washington’s export control regime in part relies on other countries playing ball and similarly restricting Chinese companies’ ability to access the highest-end chips and equipment. These include allies like the Netherlands, Japan and South Korea, which each control their own critical rungs of chip and semiconductor supply chains. 

Successive US administrations have expended significant political capital to convince these countries to incur significant economic costs and invite the wrath of Beijing. If the Trump administration is now apparently willing to undermine the national security arguments that underpinned the core case presented to allies, it may find it difficult to credibly make similar demands in future. 

In the long run, the US may end up isolating itself from global supply chains and see its own significant advantages in the technology sector diminish. 

US allies will likely also fear that the Trump administration could soon amp up similar pressure tactics against them, perhaps dangling the threat of higher tariffs if they don’t play ball. Policymakers I have spoken to about this already quietly worry about forced tech transfers that might lie ahead – for example if companies such as the Dutch chip toolmaking giant ASML are pressured into setting up shop in the US. These kinds of coercive practices may in the short-run inflict significant pain on allies; in the long-run, they could undermine the US’s global leverage and influence. 

Always doomed to fail?

The US has staked a significant part of its future global leadership in winning the AI race with China. Export control regimes aimed at limiting Beijing’s ability to catch up have been an important pillar of this approach – but one with significant intellectual and practical pitfalls. 

This strategy has always lacked a clear endgame. Neither the Biden or the Trump administrations have ever convincingly articulated what it tangibly means for the US to win the AI race against China, nor exactly what it would take for Washington to believe it has established a sufficiently secure and durable lead in order to relax restrictions. 

In the meantime, the regime of export controls itself has proven leaky and difficult to enforce. Several voices – including within the Trump administration itself – have said that overly burdensome restrictions could be an unhelpful distraction that harms US companies’ ability to innovate in an industry that is evolving at breakneck pace. 

Most importantly, export restrictions have also created significant incentives for not just China, but also others, to develop their own sovereign chip and semiconductor supply chains.

In the long run, the US may end up isolating itself from global supply chains and see its own significant advantages in the technology sector diminish. The current administration’s transactional approach will only accelerate this trend.