Signaling Confidence in Its Domestic Industry, China Bans Foreign AI Chips in State-Funded Data Centers

China is pulling away from the United States to complete its own artificial intelligence (AI) ecosystem. Following a ceasefire with the United States over export controls, China is signaling its intention to remove any foreign components from its technology ecosystem. On November 5, only days after Chinese paramount leader Xi Jinping met President Donald Trump in South Korea, Beijing ordered all state-funded data centers to stop using or purchasing foreign AI chips, effectively severing one of its last remaining dependencies on U.S. semiconductors for its AI sector.

The decision underscores a shift from dependence to insulation — having long sought to develop its own domestic tech ecosystem, Beijing now increasingly believes its domestic sector can backfill without American components.

Beijing Pushes for ‘Algorithmic Control’ in Sweeping Regulations

The new legislation bans state-owned or state-paid-for computing facilities that are less than 30 percent complete from acquiring any foreign chips, while allowing more advanced projects to maintain their holdings on a case-by-case basis. While the rule does not specify guidelines for the country’s current data centers, it will likely have a significant impact on China’s AI sector, which is in the midst of a substantial construction boom following last year’s announcement of a $47 billion state-led semiconductor fund.

The new law follows previous announcements restricting the presence of foreign components within China’s AI supply chain as part of Beijing’s push to achieve “algorithmic sovereignty” and cement complete control over the country’s computing infrastructure by 2027. In September, the Chinese Cyberspace Administration (CAC) ordered large internet companies to cease buying Nvidia processors due to concerns over their vulnerability, a move which corresponded with Chinese Premier Li Qiang touting the performance of China’s domestic chips on state television during a visit to an AI lab. 

China Hopes Self-Reliance Overcomes Structural Weaknesses

China’s efforts to build its domestic semiconductor industry predate U.S. export controls and are a keystone in Beijing’s push to develop indigenous, controllable technology. China has funneled billions of dollars since 2014 into developing a fully domestic computing industry, combining its capacity to produce both legacy chips used in common electronics and high-bandwidth memory and logic chips used to run AI systems and advanced military equipment. This effort, which has been carried across multiple Five-Year plans, remains at the center of China’s efforts to both build a modern manufacturing economy and modernize the People’s Liberation Army as a near-peer adversary to the United States.

Despite Beijing’s massive investment, China’s hardware still trails American chips in both performance and scalability, forcing Chinese firms to expend greater effort to achieve similar results to their American competitors. While a study conducted in October by the National Institute of Standards and Technology found that Chinese models have rapidly improved over the past year, leading Chinese AI firms still reportedly struggle to improve model performance or expand access in the absence of higher-quality Western hardware. These shortages have forced Chinese data centers to invest in parallel computing — splitting computational tasks across a greater number of chips — and increased their energy consumption, adding new barriers to firms’ cost structures during a period of overall economic decline.

Washington Should Tighten Export Controls on Chipmaking Technology

China’s new ban shows both confidence and constraint, with new regulatory measures catching up to Beijing’s current political signaling on the quality of its domestic AI ecosystem without undermining previous investments.

In response, the United States should tighten current chokepoints while also expanding export restrictions to include services and components that enhance the viability of China’s alternative AI chips. This effort should encompass electronic design automation (EDA) software, sophisticated chip packaging apparatuses, and high-bandwidth memory — all essential facilitators of AI accelerator efficacy. Washington should also close enforcement gaps by working with key allies — Japan, South Korea, and the Netherlands — to standardize export rules, prevent third-party rerouting, and ensure consistent enforcement across allied markets.

 Jack Burnham is a research analyst in the China Program at the Foundation for Defense of Democracies (FDD), where Duncan Lazarow is an intern. For more analysis from Jack and FDD, please subscribe HERE. Follow Jack on X @JackBurnham802. Follow FDD on X @FDD. FDD is a Washington, DC-based, nonpartisan research institute focusing on national security and foreign policy.