Anthropic CEO Dario Amodei delivered a blistering critique of the U.S. administration and leading chipmakers at the World Economic Forum in Davos on Tuesday, denouncing the decision to approve the sale of high-performance AI chips to China as a grave strategic error. The comments, particularly sharp given Anthropic’s close financial ties with Nvidia, underscored the intensifying anxiety among AI leaders about the geopolitical stakes of cutting-edge technology.
Last week, the U.S. reversed an earlier ban and officially approved the export of Nvidia’s H200 chips and a high-performance line from AMD to approved Chinese customers. While these chips are not the most advanced in the companies’ portfolios, they remain powerful AI processors. Amodei’s response was scathing, reflecting deep unease among some industry leaders about the consequences of enabling China’s AI development. His critique was especially notable because Nvidia is not only a major partner of Anthropic but also an investor, having announced a potential investment of up to $10 billion in the AI company just two months earlier.
“The CEOs of these companies say, ‘It’s the embargo on chips that’s holding us back,’” Amodei said, responding to a question about the newly relaxed export rules. He warned that the decision would “come back to bite the U.S.” and argued that America’s technological lead in chip manufacturing remains strong. “We are many years ahead of China in terms of our ability to make chips,” he said, framing the export as a strategic misstep. His comments painted a dystopian picture of future AI capabilities, describing them as “essentially cognition” and “intelligence” that could be controlled by a single nation. He compared such advanced AI to a “country of geniuses in a data center,” suggesting that an entire population of highly intelligent agents could be harnessed by state actors.
Amodei then delivered his most provocative analogy, calling the export decision “crazy” and likening it to “selling nuclear weapons to North Korea” while praising the manufacturer. The comment signaled the depth of his concern and the urgency he attributes to AI’s national security implications. The statement also highlighted a growing rift within the industry: even those closely aligned with leading chipmakers are willing to publicly challenge their strategic decisions.
The criticism may have struck a nerve within Nvidia’s ranks. The company is not merely a chip supplier but a central infrastructure provider for AI worldwide, powering the servers of major cloud providers and supporting Anthropic’s own AI models. The partnership between Nvidia and Anthropic, announced with high-profile optimism, now sits under a more contentious spotlight after Amodei’s remarks. His comparison of Nvidia to an arms dealer may have been an unguarded moment, but it also reflects the heightened anxiety around AI competition and the perceived dangers of empowering rival nations.
The exchange also reveals a broader shift in how AI executives view the geopolitical landscape. Amodei’s willingness to speak so bluntly on a global stage suggests that AI leaders increasingly see their technology as a matter of national security rather than just business strategy. The industry’s race has taken on existential overtones, with leaders apparently feeling freer to prioritize public warning over diplomatic caution.
At Davos, Amodei’s remarks were not just a critique of policy but a signal of how deeply the AI arms race is influencing corporate behavior. His comments implied that the usual constraints of investor relations, strategic partnerships, and diplomatic etiquette are losing weight against what many see as the urgent need to shape the direction of global AI development. The fearlessness with which he spoke may be the most telling aspect of the exchange, suggesting that AI executives are increasingly willing to challenge both governments and corporate allies in pursuit of what they view as a larger strategic imperative.
