A potential US ban on Chinese AI models could cost American businesses as much as US$12 billion per year, according to calculations by Daniel Yue, an assistant professor at the Georgia Institute of Technology’s Scheller College of Business. Yue said the estimate reflects growing reliance by technology firms on cost-efficient Chinese solutions.
Potential impact of banning Chinese AI models
Using usage data from New York-based OpenRouter, an LLM aggregator, Yue estimated that if OpenRouter users were forced to migrate from Chinese open-weight models to leading proprietary alternatives, they would incur an additional annual bill of about US$2 billion. That figure was derived from token usage and observed price gaps between open and closed models recorded from July 21 to 27.
Extrapolating those OpenRouter-based calculations to the broader US economy, Yue said the resulting cost increase could range between US$3 billion and US$12 billion per year, depending on the country’s overall reliance on open-weight models from China. He emphasized the numbers are an ‘order of magnitude’ approximation rather than a definitive projection, citing the difficulty of tracking usage outside centralized platforms.
New York-based OpenRouter, which allows developers to switch between various AI models through a unified application programming interface (API), captures only a fraction of the global LLM inference market, Yue noted. The limited scope of the data and the challenges of measuring decentralized usage underline the uncertainty around any precise economic toll from a potential ban.
