JPMorgan: Low-Cost Advantage Spurs China AI Models' Global Share Surge

JPMorgan strategist Michael Cembalest's latest research points out that low-cost open-source systems are driving China's artificial intelligence models to rapidly rise in global usage rankings.

The data shows that the recent weekly token consumption through OpenRouter API calls is significantly shifting towards Chinese AI models. Among the nine mainstream large language models being tracked, the token consumption share of Chinese models has been continuously expanding since the beginning of 2026.

The acceleration of Chinese model usage during February to May 2026 is particularly evident, while the utilization of American models has maintained a relatively steady growth pace. Michael Cembalest believes that China is leveraging its vast domestic market size to accelerate the application of artificial intelligence in consumer and enterprise sectors through low-cost open-source models and integrated platforms. Data as of May 11th indicates that with the expansion of global enterprise and consumer AI usage, the overall token consumption of leading AI models is still surging worldwide.

Enterprises across Asia are currently accelerating the deployment of AI applications related to productivity tools, e-commerce, search engines, and mobile ecosystems, which has given Chinese open-source models a very high exposure recently. Influenced by this trend, including Chinese internet exchange-traded funds and artificial intelligence-themed funds are being closely watched by the market. Driven by the widespread application of low-cost models, Chinese AI models are reshaping the market share structure of the global large language model market.