• Anthropic CEO Dario Amodei warned at Davos that allowing Nvidia to sell AI chips to China poses serious national security risks
• Amodei compared the decision to “selling nuclear weapons to North Korea,” despite Nvidia being a major investor in his company
• The clash highlights rising industry tension over how AI hardware should be controlled as global powers race to build advanced systems

Anthropic CEO Dario Amodei cut through the polished choreography of the World Economic Forum in Davos this week when he flatly implied that Nvidia, one of his own company’s biggest backers, was a ‘nuclear’ threat to geopolitics during an interview with Bloomberg.

The interview triggered immediate global hubbub across the tech, diplomatic, and security spheres over his response to the U.S. approval of AI chip sales to China.

The arrangement ends a ban on the sale of high-performance AI chips to China. The U.S. now allows Nvidia and AMD to resume sales of certain AI chips, including the H200 line, to pre-approved customers in China.

“I think this is crazy,” Amodei told a stunned audience during the session. “It’s a bit like selling nuclear weapons to North Korea and bragging that Boeing made the casings.” It was an especially bold reaction from the leader of Anthropic, a company that $1.5 trillion chipmaking giant Nvidia has invested over $10 billion into so far.

They’re powerful enough to dramatically accelerate Chinese AI capabilities in many ways, with military and security being one that has Amodei particularly worried. Amodei sees this as a real and immediate threat because AI models are “essentially cognition, that are essentially intelligence.”

He suggested thinking of the models powered by the chips as “100 million people smarter than any Nobel Prize winner,” all under the control of one country or another.

People were audibly shocked during the interview. Anthropic is one of the leading homes of cutting-edge AI models. The Claude assistant is often noted as a strong rival to ChatGPT in many ways, thanks in no small part to Nvidia’s GPUs.

Friction over China’s access to AI chips reflects a growing fault line inside the tech industry. Chipmakers and cloud service providers hoping to hold onto or expand their control of the AI market are tugging against companies like Anthropic with geopolitical fears over unfettered access to AI hardware by authoritarians.

Global chip war

Adding to the volatility is Nvidia’s somewhat indispensable AI‑training chips. Its architecture has become a foundation for model development, with few alternative providers, though AMD and Intel are keen to catch up. But it means that when Nvidia sells chips to China, it creates more than just new commercial rivals.

Nuclear weapons and airplane casings are unsubtle analogies, but Amodei almost certainly chose them for that reason. Davos is where CEOs talk like they’re chewing on a technical manual and marketing guidebook at the same time. A straightforward and consequential projection of the future must have thrown plenty of attendees off balance.

You might dismiss Amodei and the whole debate as high-level geopolitical drama with little relevance to everyday life. But what gets decided at the level of chip exports affects how fast the next AI-powered feature and device come out, and what they can do.

The U.S. Commerce Department has stated that any sales to China are subject to rigorous controls and that buyers are vetted for ties to military operations. But enforcement remains a murky affair, especially when front companies, joint ventures, or subcontracting relationships can blur lines.

Amodei didn’t name China explicitly, but no one needed him to. The entire discussion was a rebuke of U.S. complacency in treating AI as a neutral export rather than a lever of global influence. And while Nvidia might argue that the chips being exported are less advanced, Amodei’s counterpoint is that even slightly outdated chips can be networked at scale to produce transformative capabilities.

And as Chinese AI labs become more adept at optimizing existing hardware, the line between what’s considered to sell and what isn’t begins to erode.

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!

And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.

Eric Hal Schwartz is a freelance writer for TechRadar with more than 15 years of experience covering the intersection of the world and technology. For the last five years, he served as head writer for Voicebot.ai and was on the leading edge of reporting on generative AI and large language models. He’s since become an expert on the products of generative AI models, such as OpenAI’s ChatGPT, Anthropic’s Claude, Google Gemini, and every other synthetic media tool. His experience runs the gamut of media, including print, digital, broadcast, and live events. Now, he’s continuing to tell the stories people want and need to hear about the rapidly evolving AI space and its impact on their lives. Eric is based in New York City.