Recently, the tensions in China–U.S. trade had just eased when the United States rolled out two new policies on artificial intelligence (AI), throwing the situation back into uncertainty. First, it announced the removal of previous semiconductor export restrictions targeting AI development, converting them into advisory guidelines. Second, the Bureau of Industry and Security (BIS) under the U.S. Department of Commerce issued a set of detailed rules, warning that enterprises worldwide using chips such as Huawei’s Ascend series may violate U.S. export controls and face penalties from the Commerce Department.
This dual approach of loosening and tightening marks a notable shift in the framework and logic of U.S. policy toward China on AI chips.
First, the relaxation of export restrictions stems from both an adjustment in policy philosophy and practical pressures. The chip bans have drawn strong dissatisfaction from U.S. tech companies. Former President Trump has repeatedly stated the need to reduce regulation and foster a more permissive environment for innovation. As China’s domestic chips continue to achieve breakthroughs, maintaining sweeping export controls would accelerate the loss of overseas market share for high-end U.S. chips. In this sense, easing restrictions represents both a compromise by the U.S. government to reality and a reluctant move to secure market access for American tech firms.
At the same time, the tightening of targeted bans—adding Chinese AI chips to the control list and extending restrictions to third countries—makes clear that Washington’s intent to slow China’s AI progress remains unchanged. The strategy has simply shifted from blocking technology acquisition to restricting technology application. Previously, the United States primarily curbed China’s access to advanced computing power by limiting high-end chip exports, aiming to contain AI research at the source. Today, the focus has moved to the application layer: delaying the market rollout of China’s homegrown chips and hindering the creation of an independent, controllable technical ecosystem built on domestic AI chips.
Overall, U.S. policy toward China on AI chips is evolving from simple technology blockade to a more complex, multi-dimensional competitive strategy. The Biden administration’s earlier Export Control Framework for Artificial Intelligence Diffusion exemplified this approach, focusing on comprehensive restrictions on AI chips and related technologies. It sought to erect global technical barriers and exclude China and other countries from advanced technology ecosystems. While this policy partly limited access to high-end chips, it also harmed U.S. companies’ commercial interests and drew widespread criticism.
The latest adjustments reflect an attempt to strike a balance between preserving U.S. technological hegemony and meeting corporate commercial needs—while maintaining a hardline stance in core competitive areas. Washington understands that, given the pace of China’s progress in domestic AI chips, failure to act would pose a severe challenge to U.S. dominance in the field. Hence, the dual track: expanding access to mainstream markets while bringing Chinese AI chips under export controls to stifle the global expansion of China’s AI technology, thereby sustaining overall U.S. leadership in both AI technology and markets.
Nevertheless, these policies may not deliver the intended results. Industrially and technologically, while the United States leads in cutting-edge AI models, its domestic industrial base is insufficient, limiting commercialization capabilities. By contrast, China has consistently prioritized application-oriented AI development. With a complete industrial foundation, a vast consumer market, and abundant use cases, China has built a virtuous ecosystem where AI technology and industry reinforce each other. Steady progress in China’s AI sector will continue to boost competitiveness and narrow the gap with the United States.
From a market perspective, as more countries integrate AI into national strategies, a multi-polar competitive landscape is inevitable. The dynamics of innovation ensure diverse technological paths. Low-cost open-source large models like DeepSeek have galvanized self-reliant nations, making breakthroughs in alternative AI model approaches only a matter of time.
Ultimately, AI competition translates into market-driven commercialization and iterative innovation. Today, the tide has turned in China–U.S. AI rivalry. U.S. unilateralism and isolationism are increasingly unpopular. In a globalized world, China should steadily advance technological self-reliance, unlock domestic consumption potential, and leverage its complete industrial chain and massive market scale to weather external frictions.
Equally, through multilateral frameworks such as the Belt and Road Initiative, China should pursue mutually beneficial incremental cooperation to share technological progress and achieve win-win development with all nations.
(The author is an Associate Research Fellow at the Institute of World Economy, Chinese Academy of International Trade and Economic Cooperation.)