Sojun Park, a postdoctoral associate at CIS, argues that the US must acknowledge and offset the uneven economic costs its AI export controls impose on allied economies to sustain an effective allied export control regime.
In December 2024, the US Department of Commerce announced sweeping export controls designed to restrict China’s ability to produce advanced semiconductors used in artificial intelligence (AI). Additional guidance was issued in May 2026 clarifying the applicability of licensing requirements to advanced computing exports destined for Chinese-headquartered entities outside China. These measures aim to deny China access to the computing power needed to develop next-generation AI technologies and advanced military capabilities.
Yet the effectiveness of these controls depends on the structure of global supply. Advanced AI systems depend on specialised and geographically fragmented supply chains spanning chip design, fabrication and manufacturing equipment. Technologies needed for AI development are not produced by a single country but emerge from a complex global supply chain that runs through US allies and partners, including Japan, South Korea and the Netherlands. Washington cannot substantially slow China’s technological development without cooperation from US allies that produce memory chips and semiconductor manufacturing equipment essential for advanced AI computing.
The United States has already secured some allied cooperation on semiconductor export controls, notably from Japan. In 2023, following a surge in exports of semiconductor manufacturing equipment to China, the United States persuaded Japan to restrict exports of key chipmaking equipment. The decision generated controversy in Tokyo — some policymakers and industry leaders worried that following Washington’s lead could provoke Beijing and harm Japanese firms that rely heavily on Chinese customers.
Even when governments share strategic goals, coordinating export controls among allies has proven difficult. Recognising the limits of unilateral restrictions, former US president Joe Biden’s administration pursued a ‘small yard, high fence’ strategy, narrowly targeting the most advanced AI-related technologies while coordinating closely with allied governments. The AI Diffusion Rule differentiated countries according to their access to US technology, granting trusted allies broader access while tightening restrictions on potential channels of technology diversion.
The challenge of maintaining the prolonged diplomacy required to sustain AI export control cooperation among US allies whose firms remain deeply integrated with Chinese markets becomes even more pronounced when US policy signals appear uncertain. Reports of internal disagreements within US President Donald Trump’s administration over whether Nvidia should continue exporting AI chips designed for the Chinese market, including H20 chips, have created uncertainty about the credibility of US efforts to restrict technology flows to China.
