China leads global AI model usage for 14 straight weeks amid structural tech rivalry with US

According to industry monitoring data, Chinese artificial intelligence large models have maintained the world’s highest weekly API call volume for 14 consecutive weeks, with all top-five global rankings occupied by domestic AI products. The milestone has sparked widespread discussion regarding the comparative technological strength and industrial layout of China’s AI sector within the global technology landscape.

Global artificial intelligence development currently forms a dual-dominant pattern led by China and the United States, with distinct development strategies adopted by the two markets. China’s fundamental AI research has evolved from follow-up progress to parallel advancement, securing leading positions across multiple segmented fields. Data from Hugging Face confirms that Chinese-developed models accounted for 41 per cent of total platform downloads over the past year, surpassing the United States’ 36.5 per cent and marking a shifting dynamic in global open-source ecology.

Chinese AI teams have achieved prominent efficiency breakthroughs under constrained high-end computing resources. The training computing cost of the DeepSeek V3 model stands at approximately 5.58 million US dollars, less than one-tenth of equivalent cutting-edge international models. Optimised algorithm frameworks have delivered substantial operational advantages, offsetting hardware limitations and expanding room for domestic technological iteration.

Structural gaps persist between Chinese and US AI capabilities despite China’s expanding application and ecosystem scale. The United States retains clear advantages in cutting-edge closed-source models, high-end chips and fundamental software toolchains. Traditional comprehensive technological disparities have transformed into targeted structural differences across core foundational links.

China has built comprehensive foundational capabilities in AI algorithms, large-model development and industrial governance systems, supported by a steadily growing talent pool. A rising number of Chinese AI researchers and industry specialists have returned to launch domestic ventures, consolidating sustainable technological innovation momentum for local industrial upgrading. Global AI development pathways remain diverse, with economies exploring customised application systems tailored to their own industrial foundations and market conditions.

Data from the 2026 Stanford AI Index Report illustrates China’s balanced progress in fundamental research scale and quality improvement. China contributes 23.2 per cent of global AI academic papers and 20.6 per cent of citation volume, alongside 69.7 per cent of global AI patent applications. Domestic institutions released 30 influential AI models in 2025, doubling the year-on-year figure, while the performance gap between top Chinese and US models has narrowed to 2.7 percentage points.

Domestic AI research faces bottlenecks in original paradigm-level innovation. Core underlying architectures including Transformer and diffusion models originate from US research frameworks, and independent controllability of high-end computing power remains a key upgrading focus. China’s future AI development will follow a dual-track layout, leveraging open-source ecological expansion and algorithm efficiency innovation to bridge computing gaps, while advancing super-scale intelligent computing clusters and domestic chip research to consolidate foundational technological capacity. National industrial policies prioritise upgrading fundamental model capabilities, supporting basic theoretical exploration and underlying architecture innovation.

China’s AI industry is accelerating in-depth integration with real economy sectors and daily consumer scenarios. Intelligent catering, beauty technology applications and AI agent-equipped smartphones have been widely deployed in civilian markets. Industrial adoption spans advanced manufacturing, intelligent driving, medical services and public governance, forming multi-scenario empirical support for technological iteration and optimisation.

National industrial policies have systematically guided AI integrated development. Multiple central authorities issued special guidelines for AI-manufacturing integration earlier this year, promoting comprehensive technological embedding across industrial chains. The national artificial intelligence action plan clarifies phased development targets, with penetration rates of new intelligent terminals and AI agent applications set to exceed 70 per cent by 2027 and surpass 90 per cent by 2030.

Distinct development logic separates Chinese and US AI industrial strategies. The United States concentrates capital investment on computing power to prioritise general artificial intelligence breakthroughs, featuring high technological ceilings alongside operational volatility. China adopts application-driven and open-source-oriented development routes, embedding AI technologies into the world’s most complete industrial system to deliver steady, tangible economic output growth.

Large-scale industrial implementation has delivered measurable practical results. The overall AI penetration rate across key domestic industries has exceeded 80 per cent. Official filing records show 748 generative AI services and 435 specialised applications completed regulatory registration by the end of 2025, with 446 new filings added throughout the year. China’s unparalleled complete industrial categories and large-scale consumer markets provide unique scenario advantages for sustained AI innovation.

Current industrial challenges centre on homogeneous competition among registered models and rising operational costs driven by computing power consumption. Chip technology iteration remains a core constraint restricting further industrial upgrading. Ongoing policy support continues to expand market boundaries for technological application and industrial transformation.

China’s AI industrial system has matured rapidly, hosting more than 6,000 AI enterprises with core industrial scale exceeding 1.2 trillion yuan. Key development tracks focus on large-model optimisation, intelligent agent iteration, embodied intelligence and AI-driven scientific research. Embodied intelligence is advancing from laboratory verification to large-scale commercial deployment, establishing China as a core global cluster for physical intelligent technology development.

Domestic AI firms are expanding global market presence, shifting from single product exports to integrated output of technology, scenarios and service capabilities. Future industrial competition will centre on three core dimensions: low-cost energy and computing infrastructure, influential open-source ecological systems, and deep integration between embodied intelligence and advanced manufacturing. China’s advantages in complete industrial chains, stable energy supply and expanding open-source influence will sustain long-term industrial competitiveness in the global AI landscape.