AI was a focus during this week’s meeting between US President Donald Trump and Chinese President Xi Jinping.
The most advanced American models lead most benchmarks, but Chinese companies like DeepSeek, Z.ai and Alibaba have launched new models offering major performance gains in tasks such as coding.
According to usage data shared with CNBC, global adoption of Chinese models by businesses increased significantly in 2026.
Chinese AI models have grown from a relatively small share of usage to a majority on two major development platforms that provide companies with gateways to access models from different vendors. On OpenRouter, they represented 57-67% of tokens used during the week of September 14, up from 6-13% in February. On Vercel, their share increased to 55% in August compared to 11% in January.
OpenRouter data relating to companies in the The United States, Europe and what they define as the “Global South” – 82 countries in Central and South America, Africa and Asia. Vercel did not specify the geographic distribution of its data.
Concern is growing in Washington, where two House committees are studying the impact of the growing adoption of Chinese models.
The United States has sought to preserve its lead in AI by preventing Chinese AI companies from purchasing the most advanced chips through export controls.
Washington is concerned that they are accessing Nvidia chips remotely, through overseas data centers, and gaining ground by using “distillation” where new models imitate older, more established models.
Chinese AI poses “real economic and security risks to the United States,” said Daniel Remler, a senior fellow in the technology and national security program at the Center for a New American Security (CNAS), a think tank.
“The ultimate concern is that the integration of Chinese AI models draws countries into a sphere of Chinese technological influence that hardens into geopolitical alignment,” he told CNBC.
Peter Walker, head of research at OpenRouter, told CNBC that Chinese open source models released this year “can credibly work in advanced agentic use cases, particularly around coding, in a way that simply wasn’t true at the end of 2025.”
They are also “incredibly cost-effective compared to most models from American laboratories,” he added.
Earlier this week, OpenAI and Anthropic both announced new, cheaper models. Dianne Penn, head of product management, research and labs at Anthropic, told CNBC that the company is trying to make its response models “more efficient, so it uses fewer tokens based on your effort.”
Price is key to the growing adoption of Chinese models, Harpreet Arora, head of agentic infrastructure at Vercel, told CNBC. “Chinese models are becoming good enough to do more tasks at a much lower cost. Once a model hits the quality bar for the job, this price difference becomes compelling.”
But he added that companies still want to use U.S. border models for some more complex tasks.
Companies in what OpenRouter defines as the “Global South” have been the largest users of Chinese AI models on the company’s system in recent weeks.
More than two thirds – 67% – of the tokens used by these companies are based on Chinese models. About half of the tokens on OpenRouter are used by companies in the United States
“Southeast Asia, in particular, could see significant adoption of Chinese AI models given close economic and cultural ties (with China) as well as growing digital infrastructure,” said CNAS’s Remler.
“Everywhere from Lagos to São Paulo to Jakarta, where entrepreneurs and governments are looking for open and cheap models, they will turn first to Chinese AI.”
Gn bussni
