Is China about to win the AI race?

Talent is shifting, too: in 2019, nearly 60% of the world's top AI researchers were working in the US and only 11% in China. By 2022, that had narrowed to 42% and 28%. Tightened US visa restrictions under the Trump administration have pushed more Chinese researchers home — accelerating Beijing's domestic expertise.

The power of efficiency and open models

The US still produces more frontier models — 40 of the world’s most notable AI systems in 2024 compared to China’s 15. But Chinese labs have learned to “do more with less.” Their most advanced models, like DeepSeek-V3 and Alibaba’s Qwen 2.5-Max, rival American systems in algorithmic efficiency, using far fewer computational resources.

Open-source innovation has become China's great equaliser. Domestic developers now lead global downloads of open-weight AI models, according to Air Street Capital. Chinese companies already deploy AI tools more widely than their American counterparts in sectors such as fintech, e-commerce, and logistics.

This combination of efficiency and broad application may ultimately matter more than raw computing power. As the research field evolves toward embodied AI — systems integrated with hardware such as drones and industrial robots — China’s strength in advanced manufacturing could become decisive.

A state-driven ecosystem with unique advantages

China's industrial model provides the country with a scaling advantage in innovation. Innovation models can rapidly move from research to application by both local governments and large enterprises. Already, AI reasoning tools are being used in pilot programmes within administrative systems, financial institutions, and logistics networks.

Education forms part of the strategy, too. Universities are embedding AI training across disciplines, and the Ministry of Education is set to introduce AI literacy in schools from an early age. The approach ensures a workforce prepared for the integration of AI both from engineers down to civil servants long before market demand peaks.

This top-down coordination allows for the implementation of such systems nationwide with usually less civic pushback than is common in Western democracies. The feedback loop between state planning, industry rollout, and user data then fuels further improvement-a self-reinforcing cycle of progress.

Chips, constraints, and creative responses

Hardware remains the biggest obstacle. Export controls limit China's access to the most advanced Nvidia GPUs, forcing research labs to recycle components or look to grey markets. Yet those same constraints have spurred innovation in efficiency and the domestic production of chips.
Chinese researchers are optimizing models to run on limited resources, sharing open weights to extend participation. The pressure on working around shortages has created, in effect, a much leaner and more collaborative AI environment.

Cultural optimism and global ambition

Public sentiment is also what makes China's trajectory in AI different. According to the surveys by Stanford's Human-Centered AI Institute, Chinese citizens are the most optimistic about the impact of AI on the world, even as economic growth in the country has slowed down. To many, artificial intelligence represents the country's next economic spark and also a path back to global leadership in technology.

A new generation of Chinese founders is now coming of age: globally fluent, venture-savvy, and unencumbered by older ideological divides. Many shuttle between Silicon Valley, Dubai, and Shenzhen, building transnational AI companies that blend Chinese scale with Western innovation methods.

Competing visions for the future of AI

This is no longer a US-China competition about hardware or funding. It reflects two philosophies of innovation. The US relies on private enterprise and open experimentation, favouring proprietary systems such as OpenAI's GPT models. China, by contrast, backs open-weight models, central coordination, and fast public integration.

This is not a mere race among nations, but between two technological philosophies-one decentralized, one directed. Even OpenAI's Sam Altman has admitted that his company may have underestimated open-source collaboration's power, calling it a "different side of history."

The bottom line

Though far from inevitable, China's rise in AI has a trajectory that is already fairly clear. US advantages in frontier research and semiconductor design notwithstanding, Beijing's combination of industrial policy, efficiency, and public adoption has created a formidable ecosystem.

But to the extent that the next phase in the development of AI depends not so much on breakthroughs in theory but rather on widespread application, China's system — built on state coordination, national scale, and rapid deployment — may prove to be its greatest strength. The US might continue to lead in speed and experimentation, but China increasingly shapes how artificial intelligence becomes part of everyday life.