AI Race: Anthropic CEO Warns Against Nvidia Chip Exports to China
San Francisco, Thursday, 22 January 2026.
Dario Amodei likens selling advanced AI chips to China to ‘selling nuclear weapons to North Korea’. He voiced concerns at Davos on January 20, 2026, about the national security implications of the Trump administration’s policy.
Amodei’s Concerns
Speaking at the World Economic Forum in Davos, Switzerland, on January 20, 2026, Anthropic CEO Dario Amodei voiced strong opposition to the U.S. administration’s decision to permit Nvidia to sell its advanced H200 AI chips to China [2][4][7]. Amodei equated this policy to ‘selling nuclear weapons to North Korea,’ emphasizing the potential dangers of providing a geopolitical rival with access to cutting-edge AI hardware [1][2][8]. He believes that the U.S. holds a significant lead in chip manufacturing and that exporting these chips would be a ‘big mistake’ [1][5].
The Policy Context
The Trump administration’s decision involves allowing Nvidia to ship H200 accelerators to Chinese customers, with the stipulation that the U.S. would receive 25% of the revenue [5][6]. This move is a reversal of an earlier ban and comes amid an ongoing technology battle between the U.S. and China for AI dominance [2][5]. Advanced Micro Devices Inc. (AMD) is also seeking approval to sell its MI325X chips, an alternative to the H200, to China [5].
Impact on AI Development
Amodei argues that the embargo on U.S. chips is holding back Chinese AI companies [1][6]. He acknowledges that while there was excitement around DeepSeek’s AI model, R1, it was primarily optimized for benchmarks and not necessarily a threat to top U.S. LLMs [1][5][6]. Despite this, he warns that providing China with more sophisticated silicon could enable them to compete more effectively with the West [6].
Industry Reactions and Partnerships
Amodei’s criticism is particularly noteworthy given that Nvidia is a major partner and investor in Anthropic, with investments potentially reaching $10 billion [2][3][4]. Nvidia’s GPUs are crucial for powering Anthropic’s AI models [2][4]. This situation highlights the tension between the tech industry’s financial interests and concerns over national security and the potential risks associated with China’s access to advanced AI technology [1].
Sources
- www.capitalaidaily.com
- techcrunch.com
- finance.yahoo.com
- www.thestreet.com
- siliconangle.com
- www.theregister.com
- archax.com
- www.bloomberg.com