Key Takeaways
- China is using human workers to train AI robots, generating valuable real-world data.
- US private AI investment is 23 times higher than China’s, but the model performance gap has narrowed.
- China’s industrial robot installations account for 59% of global new installations in 2025.
In a unique approach to advancing artificial intelligence (AI), China is employing human workers to train robots, generating invaluable real-world data. These workers, wearing virtual reality headsets and exoskeletons, perform mundane tasks such as folding clothes and wiping tables, which are then used to teach AI systems. This method, while seemingly mundane, offers critical insights into the next phase of the AI race, which could determine geopolitical influence.
The US, under President Donald Trump, has emphasized the importance of winning the AI race, comparing it to the Cold War’s space race. However, the administration’s own report from Stanford’s 2026 AI Index reveals that the US remains formidable, with 59 notable American models in 2025 against China’s 35. Despite this, the performance gap between the leading US and Chinese models has effectively closed, with the US model leading by just 2.7 percent as of March 2026.
While the US has a significant advantage in private AI investment, with its spending 23 times higher than China’s, the latter is leveraging its approach to build a robust ecosystem. China’s strategy involves pushing AI into the physical economy, where cheap, open models are used in factories, logistics, and robotics. This process generates specialized real-world data, which in turn improves the models. The US, on the other hand, has built its AI boom on an internet economy that relies heavily on accumulated online content and behavioral data, raising concerns about vast surveillance and limited user control over data.
China’s approach is further bolstered by its classification of data as a factor of production. Factories are paid for access to workers wearing cameras and sensors, while others earn money performing household tasks to train robots. This creates a feedback loop where data improves AI, which in turn improves factories, potentially giving China an immense edge over the US. By 2024, China had more than two million industrial robots operating, and in 2025, it installed 354,000 new ones, accounting for 59 percent of global new installations according to the International Federation of Robotics.
The real advantage, therefore, may lie in the ecosystem itself. China’s industrial robots produce data, which is then used to improve AI, which in turn improves factories, creating a virtuous cycle. This contrasts with the US, where the AI boom grew out of an internet economy built on years of accumulated online content and behavioral data, raising concerns about privacy and data control.
While the US may have a stronger position on benchmarks, China’s approach to integrating AI into the physical economy and leveraging data could give it a significant edge. The US China Economic and Security Review Commission has warned of a reinforcing loop in China, where cheap, open models spread into factories, logistics, and robotics, generating specialized real-world data that improves the models.
The implications of this development are far-reaching. The winner of the AI race will have a much larger say in the next world order. As China continues to push its AI strategy, the US must adapt to remain competitive. The race is not just about models and benchmarks but about creating an ecosystem that can sustain and improve AI over time.
In conclusion, while the US remains a formidable player in the AI race, China’s data-driven approach and integration of AI into the physical economy could give it a significant edge. The race is far from over, and both nations will need to continue to innovate and adapt to maintain their positions.
