The 'Silent Cold War': How AI Chip Restrictions and Safety Warnings Fuel US-China Tensions

China has warned against “fear-mongering” after the CEO of a top artificial intelligence (AI) company in the United States sounded the alarm against AI advancement and called for restrictions on the sale of AI chips to China. Anthropic CEO Dario Amodei on Saturday called for the development of the most powerful AI systems to be slowed down, warning that they could soon pose potentially catastrophic risks to humanity itself. His proposal also calls for Washington to restrict China’s access to advanced technology – prompting Beijing to accuse the US of using AI safety as a pretext to preserve its technological dominance. The dispute comes as Washington and Beijing compete over increasingly capable AI models, with US President Donald Trump declaring, “Whoever wins AI wins.”

Anthropic’s Amodei warned in an essay published over the weekend that AI capabilities have been advancing rapidly, and he proposed pacing the development of frontier models to give governments and researchers more time to manage their risks. “Given the accelerating rate of AI capability development, it’s my worry that in 6–12 months such a swarm could be capable of taking over the entire internet with a persistent botnet (potentially causing hundreds of billions of dollars in damage), and that the scale of damage would continue to increase from there if AI becomes more powerful without the necessary guardrails,” he wrote.

His assertions were endorsed by his rivals – OpenAI CEO Sam Altman and SpaceX founder and Tesla CEO Elon Musk. In addition to AI safety, Amodei warned that a “Chinese lead in AI would pose grave danger for the US and the world” and urged Washington to maintain restrictions on cutting-edge AI chips and chipmaking equipment going to China. He also called for action against alleged “distillation” of US models by Chinese AI laboratories. On Sunday, Trump rejected calls for a slowdown, describing some concerns as exaggerated. “We’re leading China in AI. We’re the most sophisticated country in the world, and frankly, I want to keep it that way because whoever wins AI wins,” he said on Sunday.

China’s Foreign Ministry on Monday urged countries to pursue an “open, inclusive and benevolent approach to AI”. “Fomenting various threats, engaging in confrontation and malicious competition will only disrupt the process of global governance of artificial intelligence and is not in the interests of any party,” said ministry spokesperson Guo Jiakun. The state-backed Global Times went further, describing Amodei’s proposal as a “Cold War playbook” whose real purpose was to curb Chinese AI development and preserve US technological dominance. “This ‘silent AI Cold War’ is hypocritical and short-sighted,” it said.

The US remains ahead in several of the most important features of the AI race, but China has been closing the gap, experts say. American companies continue to dominate private investment and production of leading AI models. According to Stanford University’s 2026 AI Index Report, US companies invested $285.9bn in AI in 2025, compared with $12.4bn in China. The US also has an advantage in the most advanced semiconductors needed to train cutting-edge models. However, Chinese companies have demonstrated that they can produce resilient AI models despite US restrictions on their access to the most advanced AI chips. In January, the release of Chinese AI startup DeepSeek made headlines after it caused share prices in US tech companies to plummet, claiming that its model would cost only a fraction of the money its competitors had spent on their own AI programmes to build.

DeepSeek sent shock waves throughout AI circles when the company published a paper in December stating that “training” the latest model of DeepSeek would require less than $6m worth of computing power from Nvidia H800 chips. This is only a small fraction of the multibillion-dollar AI budgets enjoyed by US tech giants such as OpenAI for ChatGPT and US-owned Google for Gemini. OpenAI CEO Altman said that his firm’s latest programme, GPT-4, cost more than $100m to train, Wired reported in April 2023.