Foreign Media: If the U.S. bans Chinese open-source AI models, it could increase costs for American companies by $12 billion annually

Foreign Media: If the U.S. Bans Chinese Open-Source AI Models, It Could Cost American Companies $12 Billion Annually

As technology companies increasingly shift toward more cost-effective Chinese AI solutions, Assistant Professor Daniel Yue from the Scheller College of Business at Georgia Tech points out that while the exact economic cost of such a ban is difficult to quantify precisely, usage data from New York-based large model aggregation platform OpenRouter can offer useful reference.

He estimates that if OpenRouter users were forced to switch from Chinese open-source models to top-tier closed-source alternatives, these users alone would incur an additional annual cost of approximately $2 billion—based on token usage during the period from July 21 to 27 and the price differences between open-source and closed-source models. If this ratio is extrapolated across the entire U.S. economy, the cost increase could range between $3 billion and $12 billion annually, depending on the overall degree of U.S. reliance on Chinese open-source AI models.

Researchers emphasize that due to difficulties in tracking usage outside decentralized platforms, the above figures are merely "order-of-magnitude" estimates rather than precise predictions. While OpenRouter enables developers to switch among different AI models via a unified interface, its data covers only a small fraction of the global large model inference market.

Original article: toutiao.com/article/1872515494938700/

Disclaimer: The views expressed in this article are solely those of the author.