US AI Race with China Hinges on Global Adoption, Not Just Superiority

The global Artificial Intelligence race between the United States and China is not solely a contest of technological prowess but a critical battle for global adoption, according to analysis. While the U.S. currently holds an edge in cutting-edge AI development, securing its leadership depends on ensuring its technology becomes the ubiquitous standard worldwide.

This competition is described as a triathlon with three simultaneous legs: innovation, cost, and trust. In the innovation leg, experts estimate the U.S. has a lead of at least two years in chip technology and eight to 12 months in frontier AI models, driven by advancements from companies like Nvidia and OpenAI. However, Chinese AI models are rapidly improving and may operate with fewer restrictions.

The cost factor is crucial for widespread adoption. Frontier AI models are expensive to develop and operate. While Chinese models may not match the capabilities of the most advanced U.S. offerings, they are often "good enough" for many tasks and are significantly cheaper. This cost advantage makes them attractive to budget-conscious businesses, startups, and governments globally. According to OpenRouter, a model marketplace, approximately 50% of token usage in the past year involved Chinese models. Research from Booz Allen indicates that Chinese models can introduce more vulnerabilities into American applications than into Chinese ones, potentially impacting the U.S. software supply chain.

The third leg is trust, which is essential for widespread acceptance. America's democratic values and free-market system provide a foundation for trust, contrasting with China's state-controlled internet. However, both nations face challenges in building and maintaining trust. China's AI models are trained to adhere to Communist Party dogma, while in the U.S., concerns about disinformation and the rapid advancement of AI are leading to public apprehension.

To secure victory in this "adoption triathlon," the U.S. must simultaneously advance on all fronts. This includes continued investment in technological innovation, the development of lower-cost AI alternatives, and rebuilding public trust. The President's America's AI Action Plan offers a framework, but further steps are recommended.

These recommendations include designating the AI stack as critical infrastructure, similar to other vital sectors like banking and energy, with a combination of voluntary and mandatory regulations. The framework should also support lower-cost, open-weight models, not just frontier technologies. Greater transparency regarding AI advancements and challenges is also advised, drawing parallels to the unifying effect of the space race. A key objective is to establish this critical infrastructure framework by the end of 2026, before AI models become too advanced or China gains insurmountable global adoption.

Winning the AI race is framed as essential for U.S. national security, economic security, and global standing. The author, Horacio Rozanski, is chairman, CEO, and president of Booz Allen.