theft. confrontation. limit. spy. and statement. The race for artificial intelligence between the United States and China has all the hallmarks of a thriller—and it now appears to be coming to a head.
Beijing-based Z.ai’s public announcement of its artificial intelligence model GLM-5.2 last month drew admiration from Silicon Valley. The buzz around it is that the model is capable of completing complex tasks with minimal prompting and can rival its American counterparts at a fraction of the cost.
Some are calling it a “mini DeepSeek moment,” referring to January 2025, when China surprised the U.S.-dominated artificial intelligence industry and global markets by launching a chatbot. It turns out that China can develop cutting-edge artificial intelligence without advanced Nvidia chips – the chips on which ChatGPT and Claude in the United States rely – but which have been deliberately kept out of Beijing’s control.
Also read: Negotiating the AI Paradigm
Is China closing the gap?
According to Reuters, GLM-5.2 currently ranks fifth on Artificial Analysis’ intelligence rankings and second on Code Arena’s front-end coding rankings.
Z.ai founder Tang Jie told xAI’s Elon Musk on social media that GLM-5.2 is equivalent to Claude Opus 4.8 and aims to reach the capabilities of Claude Fable 5 (the most powerful model Anthropic offers to the public) in the first quarter of 2027.
The incident puts GLM-5.2 at the center of a growing debate over whether China is finally catching up to the United States in the artificial intelligence race.
“We now have a Chinese open weight model that is as good as the models currently available from OpenAI and Anthropic,” David Sacks, a former AI czar in U.S. President Donald Trump’s administration, said last week.
The timing of this debate is no accident. The release of GLM-5.2 came the same month that Washington briefly restricted Anthropic and sparked other flashpoints, all of which raised concerns about unstable U.S. regulation of the burgeoning artificial intelligence industry.
Also read: India must ride the next wave of AI, not chase past waves
Myths and Fables Close
In early June, Anthropic released Fable 5 – a public version of its most advanced model, Mythos, which is not more widely used because it is powerful enough to easily detect vulnerabilities in software and help bad actors carry out hacking and other illegal activities.
Three days later, the U.S. government ordered the company to cut off access to both models to all non-U.S. citizens. Anthropic shut down these models globally due to the inability to easily verify the nationality of users.
The Trump administration has said the directive was triggered by the discovery of a “jailbreak” that could induce Fable 5 to provide users with dangerous information.
Washington state partially eased restrictions last week and allowed Anthropic to resume access to Fable starting July 1.
For some observers, the brief closure had less to do with safety concerns than with a feud between humans and the Trump administration. In February, Anthropic refused to allow the Pentagon unfettered use of its artificial intelligence for surveillance and weapons use, prompting Trump to order government agencies to stop using all of the company’s products and describe it in social media posts as a “left-wing nut job” company.
Compounding the situation, as the midterm elections approach, support for the Trump administration is declining, and the Trump administration is leaning toward regulating the artificial intelligence industry, which many Americans believe could affect their jobs and employment opportunities.
Machines ‘missing’, Anthropic blames Alibaba
The same weeks brought different flashpoints.
U.S. Commerce Secretary Howard Lutnick told senior leaders at Dutch chip equipment maker ASML that Washington suspects one of its extreme ultraviolet (EUV) lithography machines used to make state-of-the-art processors may have arrived in China, violating U.S. export controls, Bloomberg reported.
ASML has never been allowed to ship these systems to China, and the company denies any violations and said in an internal document that all 314 EUV machines in operation worldwide originate from outside China.
Senior U.S. officials told Bloomberg that they have evidence that ASML has shipped specialized equipment related to EUV systems to China, but they declined to make the evidence public. ASML’s access restrictions are important because the lack of EUV tools is the most stringent restriction for Huawei Technologies Co., Nvidia’s main competitor in China’s artificial intelligence chip design.
The third dispute broke out after Anthropic accused Alibaba Group of illegally accessing its artificial intelligence models. The news sent Alibaba’s Hong Kong-listed shares to a 16-month low, and shares of other Chinese artificial intelligence developers such as Xiaomi and Baidu also fell more than 3% on the news.
Bloomberg Intelligence analyst Robert Lea said the incident, combined with Anthropic’s export restrictions, showed “an increased risk of Chinese AI models facing a U.S. ban.”
Games and their bets
The race for artificial intelligence has been compared to the nuclear race between the United States and the Soviet Union in the 20th century. But this has a clear endgame: nuclear weapons. Artificial intelligence doesn’t have that, at least not yet.
Experts say both countries currently have unique advantages.
Kyle Chan, a fellow at the Brookings Institution who specializes in Chinese technology and industrial policy, said in a podcast last month that the United States’ advantage lies in the “virtual world”—advancing research and developing artificial intelligence models and the software and business ecosystems that surround them.
China’s strength, he said, is the “physical world” of manufacturing and supply chain innovation. Chen said few countries can make great progress in areas where they don’t have access to the best inputs.
