Anthropic CEO Dario Amodei sharply criticized the policy of exporting AI chips during the World Economic Forum in Davos. He questioned the US administration’s recent decision to allow the supply of high-performance processors to China.
The statement was prompted by the approval of sales of Nvidia H200 chips and numerous AMD solutions to a limited circle of Chinese customers. Although these processors are not flagship models, they are widely used for training and running artificial intelligence systems.
Amodei said that exporting such technologies could have “incredible implications for national security” because AI models are a form of scalable intelligence. According to him, such systems are comparable to “an entire country of geniuses in a data center.”
The most controversial part of his speech was the comparison of AI chip exports to the proliferation of strategic weapons.
The businessman believes that the sale of such technologies is reminiscent of “supplies of nuclear weapons to North Korea.” In his opinion, the US has a significant technological advantage that should not be eroded.
It is noteworthy that these statements were made against the backdrop of Anthropic’s close commercial ties with Nvidia. The graphics processor manufacturer not only supplies equipment to the company’s cloud partners, but has also previously announced multi-billion dollar investments in the AI giant, as well as deep technological cooperation.
Experts believe that the contrast between public criticism and strategic partnership highlights the growing tension in the AI industry. National security issues are increasingly coming into conflict with commercial interests and global supply chains.
In addition, in December, information emerged that Chinese startup DeepSeek is using Nvidia Blackwell chips to develop a new AI model. The company became known in early 2025 after presenting a model comparable to the solutions of leading Silicon Valley companies.


