The United States and China are pursuing structurally different artificial intelligence strategies, according to Harvard Business School Professor Meg Rithmire. The U.S. prioritizes a "frontier-first" approach, focusing on increasingly powerful foundation models and infrastructure to achieve global dominance. China employs a "diffusion-forward" strategy, embedding AI technology throughout its economy with focus on industrial efficiency in manufacturing and robotics.
The divergence reflects competing theories of AI dominance. Washington and Beijing both believe that developing artificial general intelligence—systems capable of outperforming the median human in economically valuable work—could grant enormous capabilities, enabling one country to dominate competition and potentially penetrate critical infrastructure. The competition extends beyond AGI to which country's technology stack the world will adopt.
China's state-directed approach is quantifiable. In 2025, nearly one in four Chinese AI services credited some degree of state participation, up from less than five percent in 2023, according to "China's Diffusion-Forward AI Strategy: The 'AI Race' in Political Economic Context," published in August in Asian Economic Policy Review. The research, co-authored by Rithmire and Harvard Kennedy School postdoctoral fellow Hao Chen, draws from Chinese policy documents and AI services registration data.
UBTECH, a humanoid robotics firm, exemplifies Beijing's industrial integration approach. While Silicon Valley bets heavily on frontier models, China is industrializing AI at scale—AI generation is already built into major video platforms, reflecting state control and widespread adoption.
The distinction carries material implications. The country that translates AI most effectively into productive economic activity may gain more than the one that develops the most powerful model.
