Episode Details

This episode looks at the news that Anthropic suspended access to two of its latest models outside the United States — not just for foreign nationals, but for all users, because filtering by citizenship proved technically difficult.

It draws a comparison to rocket engine export controls, specifically Japan's long effort to develop the H3 and LE-9 independently — a project driven by the need to avoid a scenario where a single foreign policy decision could halt an entire program. AI, the episode suggests, may be entering that same territory.

There's also a brief look at why Mars makes this question especially sharp: with communication delays of up to twenty minutes each way, any settlement would need rockets, power, communications, and AI that no outside party could switch off.

The episode traces the phrase "AI sovereignty" through recent national efforts — DeepSeek and Alibaba in China, Mistral in France, Preferred Networks and Sakana AI in Japan — as countries quietly begin preparing for a world where access to frontier AI is no longer guaranteed.

A quiet look at how a routine service suspension, read from a certain distance, looks less like a technical inconvenience and more like the moment borders began to appear around intelligence itself.