China and the United States agree on a basic premise: the country with the most powerful artificial intelligence will gain geopolitical and economic advantage. Their approaches diverge sharply. The U.S. prioritizes moving faster on frontier capabilities, whereas China emphasizes centralized control and broad, monitorable domestic adoption.
Key elements of the strategy
China’s 15th Five-Year Plan, released in March, reflects this approach: the term “AI” appears 52 times in the document — five times more than five years earlier — and computing power receives a standalone chapter. Beijing’s stated objective is to integrate AI across 90% of the economy by 2030. The United States has no comparable economy-wide numeric target.
China’s leadership is less preoccupied with white-collar job losses or apocalyptic AI scenarios than with the possibility that AI could be turned against the regime. Chen Yixin, China’s spy chief, warned publicly this month that AI can carry out public-opinion campaigns and “cognitive warfare against China,” threatening political, institutional and ideological security.
Domestic weaknesses shaping the strategy
Much of China’s AI policy responds to internal problems that the official sales pitch often downplays:
- Demographics: China’s population has declined for four consecutive years and is substantially older; the number of people over 60 approaches the total U.S. population. This helps explain the country’s heavy investment in robotics.
- High youth unemployment: Youth unemployment currently sits at about 19%, roughly double the rate in the United States. A large, disaffected, digitally equipped youth cohort poses political risk.
- Rural poverty and housing crisis: A large portion of the population outside major cities faces economic hardship, exacerbated by a five-year housing crisis that has depressed the value of many citizens’ most valuable assets.
- Inferior chip production: In advanced chips, Nvidia’s offerings are significantly ahead; China’s domestic high-end chip capabilities lag despite state support.
- Export-driven economy: China excels at exports, often flooding foreign markets with inexpensive goods, while domestic demand remains weak.
Structural advantages
Despite those weaknesses, China holds structural strengths that favor rapid, large-scale AI deployment:
- Power: Electricity is a major constraint on AI buildout in the U.S., while China is expanding generation capacity rapidly. BloombergNEF estimates China will add almost six times as much generating capacity as the U.S. over the next five years.
- Critical materials: The International Energy Agency tracks 20 minerals essential for AI and adjacent technologies; China dominates all but one and has applied export controls to about half of them to slow foreign competitors.
- Robot rollout: In 2024, more than half of global industrial robot installations went to China. Superior AI will amplify the productivity of those robots.
- Cheap, widespread adoption: Chinese open-source models are commonly used by developers building cheaply; five of the six most popular models on OpenRouter come from Chinese firms.
- Land and air autonomy: China leads in self-driving cars and self-flying drones; DJI alone holds over 70% of the global drone market.
Models and technical position
Most AI experts consulted conclude that China produces the world’s best inexpensive, open models but trails U.S. frontier systems by about half a year while spending far less. These estimates shift with each major model release and are not exact, but they capture the current assessment that Chinese models are good, improving rapidly and cheaper to run.
Notable Chinese models to know:
- Kimi K3 (Moonshot): Widely rated as the best all-around open Chinese model and strong on long multi-step agentic tasks.
- Qwen3.8-Max (Alibaba): Also highly regarded and leading in some rankings.
- GLM-5.3 (Zhipu / Z.ai): China’s coding specialist; its predecessor scored within a point of Anthropic’s Claude Opus 4.8 on a key coding test at under a fifth of the cost.
- DeepSeek V4 / V4.1-Flash: Among the cheapest to use, MIT-licensed and truly open.
U.S. officials and companies accuse Chinese labs of “distillation,” covertly querying American models at scale and training on the outputs. Washington has threatened sanctions and blacklisting but has not implemented them so far. In response, some U.S. frontier labs have shifted toward building powerful lower-cost models to prevent Chinese systems from undercutting them on price.
Use versus fear
Chinese public attitudes toward AI are far more positive than American ones, reflecting wider domestic deployment in schools, public services and commerce:
- 80% of Chinese people use AI weekly, compared with 54% of Americans.
- 84% of Chinese say AI excites them, versus 38% of Americans.
- 87% of Chinese trust AI, while only 32% of Americans do.
- 54% want more AI in daily life, compared with 17% in the United States.
These figures indicate widespread Chinese adoption and acceptance of AI, but analysts such as ChinaTalk caution that high optimism may mask growing concerns about job displacement.
Conclusion
Both China and the United States recognize that leadership in AI will shape the coming decades. China’s path emphasizes state control, rapid, economy-wide integration and leveraging advantages in power, raw materials and robotics to deploy AI at scale, even as demographic, technological and economic weaknesses influence its priorities and risks.
