Four years ago the United States tightened export controls on advanced semiconductors used in AI, data centers and defense, aiming to curb Beijing’s ability to develop technologies that could boost its military and economic power. The restrictions accelerated China’s drive for chip self-reliance, a goal rooted in the earlier Made in China 2025 plan, and Beijing has since poured hundreds of billions of dollars into domestic semiconductor production.
Beijing offered large subsidies, tax breaks and other incentives to build local equivalents of firms like NVIDIA — maker of the Blackwell AI chip — and to develop capabilities comparable to Taiwan’s TSMC, the world’s dominant contract chipmaker and developer of N2 manufacturing technology. SMIC, central to China’s self-reliance effort, reported record revenue of $9.3 billion last year, and HuaHong, the mainland’s second-largest foundry, said it was operating at about 106% capacity due to strong demand.
Despite these gains, some analysts caution that progress is often overstated. Ryu Yongwook, an assistant professor at the National University of Singapore’s Lee Kuan Yew School of Public Policy, says China remains far from true chip self-sufficiency and lags the US in research, design and innovation, and Taiwan and South Korea in advanced production.
Still, China has made meaningful advances. The Rhodium Group estimates China now holds roughly 30% of the global market for legacy chips — widely used, less advanced semiconductors found in vehicles, industrial equipment and consumer electronics. Mass production of such chips can drive global price declines and pressure non-Chinese vendors; John Lee of the research consultancy East-West Futures points to areas like silicon carbide wafers as examples.
On cutting-edge chips, Chinese firms have achieved 7-nanometer-class processors that power Huawei’s latest phones. These wafers resemble technology TSMC delivered to Western customers in 2018, but they trail 3-nanometer and 5-nanometer chips on speed, energy efficiency and cost. Tim Rühlig of the European Union Institute for Security Studies says China faces a “brick wall” because sanctions and limited access to the most advanced US chipmaking tools constrain how quickly it can close the gap — catching up could take a decade or more.
Reflecting a shift in emphasis, China’s new Five-Year Plan downplays prior ambitions for outright chip dominance and instead highlights AI repeatedly, outlining a “model-chip-cloud-application” framework that treats advanced semiconductors as one component of a broader computing ecosystem. In practice, China is prioritizing task-oriented, industry-focused AI systems that require less compute — systems domestic chips can often support — and delivering strong performance at much lower cost.
That cost advantage is driving rapid uptake of Chinese chips and AI across the Global South, where governments and firms increasingly prefer Chinese solutions. Market intelligence firm Trendforce reported that Chinese AI platforms, including DeepSeek and Alibaba’s Qwen, captured about 15% of the global AI model market by late 2025. This growth threatens the long-term dominance of US cloud and AI giants such as Microsoft and Google, which are forecast to spend heavily on AI infrastructure — Goldman Sachs projects around $700 billion this year across the industry.
The US lead in AI and chips faces other challenges beyond manufacturing. ICIS, a global market intelligence provider, warned that US data centers reliant on high-end chips could soon be constrained by the nation’s strained power grid. By contrast, China’s rapidly expanding power sector may offer it an advantage: ICIS projects roughly 400 gigawatts of spare capacity by 2030, enabling large-scale data center rollouts even if Chinese chips are less efficient. Ryu Yongwook notes that cheaper energy helps offset relative chip inefficiencies, particularly for power-hungry AI workloads.
ICIS outlines three broad possible outcomes of the chip race: the US fixes its grid and maintains an edge; the US retains lead in advanced chips while Chinese AI systems proliferate in the Global South; or geopolitical fractures produce two separate AI ecosystems. Regardless of the endpoint, many in the industry expect Chinese competitors to keep undercutting prices while narrowing gaps in sophistication and reliability — pressuring non-Chinese vendors and reshaping global technology competition. Edited by: Tim Rooks