Executive Summary
This RAND report systematically characterizes the commercial AI developer ecosystems in the United States and China using a novel dataset of 1,181 firms. It finds that both ecosystems are structurally more similar than commonly assumed, with a key divergence being China's greater concentration in embodied AI and physical form factors. The study provides empirical grounding for discussions on U.S.-China AI competition.
Why It Matters
This report offers a data-driven comparison of the US and Chinese commercial AI ecosystems, providing critical insights into their strategic strengths and weaknesses, particularly regarding embodied AI and model development. Defense and geopolitical analysts should care about this as it directly informs the technological competition between two major global powers.
Key Takeaways
- US and Chinese AI developer ecosystems are structurally more similar than often believed, especially in model architecture and commercial positioning.
- A significant divergence is China's strong focus on embodied AI (robots, autonomous vehicles) and industry verticals, while the US concentrates on software-only, knowledge-intensive applications.
- Chinese open-weight models have achieved near parity with US models in token volume on platforms like OpenRouter, driven by competitive pricing and performance. This indicates a growing challenge to US dominance in AI model adoption.
Strategic Relevance
The report's findings are strategically relevant as they provide an empirical basis for understanding the technological competition in AI between the United States and China. The differing focus on embodied AI versus software-only applications highlights potential future strategic advantages and vulnerabilities for each nation, impacting military applications, industrial capabilities, and the race for artificial general intelligence.