The United States maintains an edge on China in the artificial intelligence race. At a CEPA private roundtable, industry leaders and policy experts from the US and Europe discussed maintaining this advantage. The event covered the effectiveness of targeted export controls on high-end semiconductors and semiconductor manufacturing equipment to prevent China from catching up to the US, and why cooperation between the US and its allies is essential in this competition.

The Cold War analogy was invoked, and then debated. The first quarter of the Cold War was about running faster, while today’s competition is about running faster and slowing down the second-place competitor. This slowing comes by leveraging AI supply chain chokepoints. Many participants agreed that these tools should be a clear priority, but their value depends on what comes next. Washington must prioritize slowing China and pressing its own advantage forward as the capability gap narrows. Several participants estimated the gap between the US and China on semiconductor equipment technology closer to two years than ten.

The US has the AI Advantage, For Now

Despite China’s extensive state investment in AI, the US maintains roughly a six-to-eight-month advantage on frontier model capabilities. This edge stems from US companies’ access to the most advanced, energy-efficient AI chips. These chips rely on extreme ultraviolet (EUV) lithography equipment, manufactured exclusively by the Dutch company ASML. Western export controls on the sale of EUV machines to China have imposed a major ceiling on Chinese production capacity of advanced semiconductors, with some attendees suggesting Beijing is now a decade behind on developing similar or alternative EUV technology.

Yet, the US advantage is eroding at the margins. Chinese firms have smuggled advanced AI chips through third countries and used model distillation attacks to approximate frontier Western capabilities at far lower computational cost. To combat this, an attendee proposed country-level chip location verification using Know Your Customer (KYC) protocols with existing industry APIs. Such measures would impose meaningful roadblocks to Chinese progress. However, there are industry objections to this view: the goal is to create friction and to prevent China from reaching the scale necessary to catch up. A 50% reduction in Chinese compute aggregation would be a major win.

While some roundtable participants framed China’s catching up on designing and manufacturing AI chips as inevitable, others argued that President Xi Jinping’s belief in indigenous chip production is misguided. The miscalculation is a serious US advantage because it reduces Beijing’s pressure for access to US chips and buys Washington room to tighten controls without immediate retaliation. ‘Do not disturb this’, the argument went.

Export Controls and Constraints

Going “all in” on export controls, however, carries real retaliatory exposure. The US is deeply intertwined with China in pharmaceuticals, chemicals, machinery tools, and other vital industries. Additionally, some participants suggested that ending servicing of ASML DUV machines in China could backfire by granting the Chinese unconstrainted ability to reverse engineer the machines.

Europe is between a rock and a hard place. Without the same economic leverage as the US, Europe has felt Chinese retaliation more deeply. Beijing has weaponized its dominance in critical minerals to undermine derisking efforts.

It was suggested that stunting China’s AI development could deter Chinese aggression toward Taiwan. Buying time allows Taiwan, Japan, and Australia to bolster their defense industrial bases. It also gives Washington room to convince President Xi that an invasion would be catastrophic. Volt Typhoon demonstrates the urgency of deterrence. Beijing was willing to risk the bilateral relationship in order to be better positioned for a Taiwan conflict, even after President Biden raised the campaign directly with Xi and received no response from the Chinese leader.

Takeaways

The US and its allies still hold decisive advantages in the AI supply chain, but those advantages are neither permanent nor self-reinforcing. Export controls on semiconductor chips and manufacturing equipment will continue to feature in any response, and it is an area policymakers must monitor closely.

Jack Galloway is a Program Assistant with the Tech Policy Program at the Center for European Policy Analysis (CEPA).

Alina Kreynovich is a Program Assistant for External Affairs at the Center for European Policy Analysis (CEPA).

This event was held at the Center for European Policy Analysis under the Chatham House Rule.

CEPA is a nonpartisan, nonprofit, public policy institution. All opinions expressed are those of the author(s) alone and may not represent those of the institutions they represent or the Center for European Policy Analysis. CEPA maintains a strict intellectual independence policy across all its projects and publications.