Chinese AI Models Rapidly Gain US Market Share, Sparking National Security Concerns

A surge of powerful and inexpensive artificial intelligence models originating from China is rapidly attracting users in the United States, prompting concerns that America's leading position in AI development could be threatened. Experts are also raising alarms about potential intellectual property theft and the broader economic implications for the U.S.

One notable example is the open-source model GLM-5.2, released by China's z.AI on June 16. This model specializes in coding tasks and Z.AI claims it is comparable in performance to leading models from U.S. companies like Anthropic, OpenAI, and Google. Guillermo Rauch, CEO of U.S. AI firm Vercel, described GLM-5.2 as "genuinely impressed, almost shocked, at how good" it is, stating on X that "This changes things."

Mat Velloso, an AI executive with prior experience at Meta and Google DeepMind, echoed these sentiments, calling GLM-5.2 the "first open model that passes the bar as a daily driver" and predicting that "Things are not going to be the same."

The Trump administration has previously expressed apprehension regarding China's rapid advancements in AI, with officials warning in April that the country was engaged in "industrial-scale" efforts to replicate AI technology. OpenAI and Anthropic have accused Chinese firms of utilizing a technique known as "distillation" to extract data from their U.S.-developed models.

Several Chinese AI models are already making a significant impact on the U.S. market. Six out of the ten most popular models on the AI marketplace OpenRouter were developed by Chinese tech companies, including DeepSeek, Tencent, Xiaomi, and MiniMax. These Chinese firms are reportedly leveraging open-source models, which are free to download and customize, to accelerate their growth.

This open-source strategy mirrors an approach previously adopted by Meta, though the company has since shifted towards proprietary, paid models. Some AI experts now worry that Chinese open-source models could eventually become the global standard.

Z.AI's founder, Jie Tang, has publicly stated that it would not take long for Chinese firms to rival frontier AI models, responding to a prediction by SpaceX CEO Elon Musk that such a development might occur by the first quarter of 2027.

Unlike subscription-based U.S. models, many leading Chinese AI models are open-source, making them more accessible and cost-effective for extensive projects. The high cost of AI tokens, a unit of usage, has already led major U.S. companies like Meta, Uber, and Walmart to implement or consider limits on employee AI usage.

Cursor, an AI coding firm acquired by SpaceX, acknowledged in March that its "Composer 2" model was developed using an open-source model from China-based Moonshot AI. Separately, Microsoft faced criticism earlier in June after reports surfaced that it was considering offering a version of China's DeepSeek model on its new "Copilot Cowork" tool. This tool allows users to select from various AI models for long-term projects.

Senator Rick Scott (R-Fla.) reacted strongly to the Microsoft report, stating that "Communist China wants to destroy our way of life" and that "American companies have no business selling out our national security by partnering with CCP tech companies like this."

Microsoft CEO Satya Nadella indicated that offering access to a variety of more affordable AI models, not just the high-cost leading U.S. models, is part of his company's strategy. Nadella also offered a different perspective on AI's economic impact, contrasting with warnings from figures like Anthropic CEO Dario Amodei, who has suggested AI could lead to national unemployment as high as 25%, particularly affecting white-collar jobs. Nadella, in an interview with The Wall Street Journal, pushed back against what he characterized as alarmist rhetoric, stating, "You can’t say, hey, all white-collar jobs are gone and this could even be a weapon and we will use all the power to build data centers."