America is arguing over the wrong AI obstacle

Last month, two Chinese companies unveiled artificial intelligence models unfamiliar to most outside the industry – Kimi K3 by Moonshot AI and Qwen3.8-Max by Alibaba Group Holding. Despite unverified performance claims, the market reaction resulted in a $3 trillion decrease in global chip stock value. This reaction, rather than the models themselves, is the central issue.

The focus should be on the scarcity of powerful AI systems, not their potential dangers. The assumption that exclusive, expensive, and controllable American systems would remain dominant has been challenged.

Nvidia CEO Jensen Huang argues that America's real danger lies in a failure to diffuse its technology globally, as opposed to the creation of advanced machines. Restrictionists believe in limiting access to maintain American AI technology dominance, while accelerationists argue that regulation is smothering innovation. However, both camps are arguing about the wrong obstacle.

The United States lacks a comprehensive federal AI statute, and American institutions face challenges such as inefficient procurement cycles, strict compliance teams, and hesitant middle managers. These management issues, rather than technological limitations, prevent the successful deployment of AI systems. There is little regulation to blame, as American hospitals, insurers, or car plants can already deploy these systems.

The real question is not which model to license but whose model to build on. Chinese open models enable most institutions worldwide to run them on their servers, keeping data within their control. Inheriting Chinese design choices and standards is a strategic concern for many organizations. Targeted controls on sensitive chips may be necessary, but blanket containment is not effective, as it detracts from more pressing issues.

To maintain national advantage, societies must reorganize themselves around AI faster than their competitors. The United States should focus on targeted controls for sensitive chips and address the challenges that hinder AI adoption in various sectors.