How to Lose a Global AI Monopoly in One Afternoon

Summary

The video recounts how the US Commerce Department forced Anthropic to shut off foreign-national access to its most advanced AI models, and how Anthropic disabled them globally rather than verify citizenship, days before a confidential IPO filing at an expected ~$965 billion valuation. Patrick Boyle explains the deemed-export rules and the disputed jailbreak that triggered the order, reportedly flagged by Amazon, which is both Anthropic's largest backer and a competitor. He argues the episode undercuts the winner-takes-all assumptions behind US frontier AI valuations, highlights Europe's sovereignty response, and points to cheap, self-hostable Chinese open-source models as the main beneficiaries.

• A Commerce Department is-informed letter gave Anthropic roughly 90 minutes to suspend foreign-national access to its Fable 5 and Mythos 5 models.
• Anthropic disabled the models globally shortly after confidentially filing for an IPO at an expected ~$965 billion valuation.
• Deemed-export rules also locked Anthropic's own non-US engineers out of the systems they had built.
• Amazon CEO Andy Jassy reportedly flagged the jailbreak to the White House; Amazon is a major Anthropic backer and a competing AI developer.
• Chinese open-source models such as Jipu and DeepSeek are self-hostable, roughly 60x cheaper, and gaining usage share on OpenRouter.
• Europe reacted with panic and a proposed 422 billion euro technology sovereignty package; JPMorgan and Goldman Sachs cut Anthropic access in Hong Kong.
• Boyle concludes there is no monopoly on math, so competitive pressure threatens frontier-lab margins, valuations, and IPO economics.