Anthropic CEO Opposes AI Chip Sales to China

Anthropic CEO Dario Amodei has once again urged the United States government to avoid selling advanced artificial intelligence (AI) chips to China, warning that such exports could pose national security risks.

Speaking on the Dwarkesh Patel podcast, Amodei said the US should maintain control over its computing advantage. His remarks come at a time when Nvidia CEO Jensen Huang has been advocating for broader access to American-made AI chips in international markets, including China.

Amodei argued that advanced AI chips should not be treated as ordinary trade goods. According to him, these chips power AI systems that represent “intelligence” and could significantly influence global power balances. He stated that the US should avoid building data centres in China or supplying high-end chips that could strengthen China’s AI capabilities.

He framed the issue as one of long-term security rather than short-term economic benefit. While he supported sharing medical advances and economic growth opportunities globally, he suggested that sensitive AI infrastructure should remain out of the control of authoritarian governments.

Amodei also discussed broader risks related to AI competition between the US and China. He warned that if both nations develop equally powerful AI systems, it could create a dangerous rivalry. Unlike nuclear deterrence, he suggested, AI competition might not offer stable safeguards, as both sides could believe they hold a strategic advantage.

He raised concerns about cyber dominance, noting that highly advanced AI could potentially give one side deep visibility into digital systems unless countered by equally strong defences.

On the economic front, Amodei said AI-driven growth is likely to expand rapidly. However, he emphasized that distributing the benefits fairly and protecting political freedoms would be more challenging than generating economic value itself.

Instead of excluding developing countries from AI progress, Amodei proposed expanding data centres and AI-related industries in regions such as Africa. According to him, the goal should be to spread AI’s benefits globally without compromising security.

He also suggested that AI technology could eventually reshape governance models, though he acknowledged that such outcomes remain uncertain.

The debate highlights a growing policy divide between leaders in the AI industry. Nvidia, which has invested heavily in AI development and maintains business interests in China, has advocated for allowing chip sales under certain regulatory frameworks.

Jensen Huang has previously questioned whether strict export controls could limit global innovation and economic growth. He has also expressed concerns that excessive restrictions may affect competitiveness.

Interestingly, Nvidia has invested billions of dollars in Anthropic, yet the two companies appear to differ sharply on export policy. The disagreement reflects a wider discussion within the US about balancing economic opportunity with national security concerns in the AI sector.

As AI technology becomes more powerful and influential, policymakers face increasing pressure to define clear rules around exports, regulation and global competition.

The issue is expected to remain central to US technology and trade policy discussions in the months ahead.