
Dario Amodei, the CEO of Anthropic AI, criticised the recent approval by the US for NVIDIA to export its H200 AI Chips to China. Amodei likened the approval to 'selling nuclear weapons to North Korea' while claiming that the US seems to allow China to get ahead in the AI race.
Speaking with Bloomberg Television, Amodei addressed the Trump administration's decision to allow NVIDIA to export its H200 AI chips to China. The question was part of a wider debate amongst experts on whether NVIDIA should enter the Chinese AI market. Amodei made the nuclear weapons comparison regarding the decision, warning that there would be 'grave' consequences for the US in the AI race. 'So I think this is crazy. I think it's a bit like, I don't know, like selling nuclear weapons to North Korea and bragging, oh yeah, Boeing made the case. No, I wouldn't refer to any particular people, but I would say that this particular policy,' said Amodei.
We're partnering with @TeachForAll to bring AI training to educators in 63 countries. Teachers serving over 1.5m students can now use Claude to plan curricula, customize assignments, and build tools—plus provide feedback to shape how Claude evolves.https://t.co/UfBohQ6BCl
— Anthropic (@AnthropicAI) January 20, 2026
Not the First Time
This is not the first time Amodei has been critical of Washington's export policies, especially regarding NVIDIA. In May 2025, Amodei's company Anthropic AI accused NVIDIA of 'telling tales' to the Trump administration so they can re-enter the Chinese AI market.
Anthropic has also highlighted how chip smuggling is rife in China, citing techniques such as hiding them in 'prosthetic baby bumps.' Amodei has also criticised NVIDIA CEO Jensen Huang on other aspects linked to the supposed 'AI race.'
According to Amodei, the US is providing Chinese AI firms with the technology, which has now placed them at a disadvantage in developing frontier models. He said that companies like DeepSeek have publicly stated they are falling behind because they do not have enough NVIDIA chips.
The Anthropic CEO also claimed that even if the US has allowed the exporting of chips that are a 'generation behind' such as the Hopper-class, they are still competent compared to what China has produced in its own markets. Huang has repeatedly called out the 'China hawks' working in the Trump administration, whilst Anthropic has expressed its support for policies like AI diffusion.
We’re heading to Davos tomorrow!
— NVIDIA (@nvidia) January 20, 2026
NVIDIA CEO Jensen Huang joins @BlackRock CEO Larry Fink at the @WEF Annual Meeting to discuss how AI is driving economic growth — creating new jobs and boosting productivity across the global economy. #wef26
January 21, 2026 at 11:30–12:00CET… pic.twitter.com/Dwr3OQ6F8O
US Conditions on the Export
While the US has given NVIDIA the green light on exporting its H200 AI Chips to China, it is not without some conditions. According to their approval, the chips will be subject to testing by a third-party testing lab to confirm the AI capabilities of the chips before being shipped off to China. China cannot receive over 50 per cent of the total amount of chips that are sold to US customers.
NVIDIA will also be required to certify that there are enough H200 AI Chips in the US and Chinese customers must carry out 'sufficient security procedures.' The chips are also not allowed for military use.
NVIDIA welcomed the approval and in a statement said it 'strikes a thoughtful balance that is great for America' and can help the company compete in the global chip market. 'The administration's critics are unintentionally promoting the interests of foreign competitors on US entry lists – America should always want its industry to compete for vetted and approved commercial business supporting real jobs for real Americans,' said the company in the statement.
The clash between Amodei and NVIDIA signals deepening divisions within the tech industry over how aggressively the US should restrict China's access to advanced AI hardware, a debate likely to intensify as both nations race to dominate artificial intelligence development.