Third installment (here's part one and part two) of my breakdown of Elon Musk's interview with The Economist. This one covers AI geopolitics and why Musk doesn't buy into classic government-style regulation.
🔹 Electricity is AI's new oil: the real bottleneck on neural network progress today isn't chips — it's power. China already has more electricity than the US, Europe, and India combined [[18:10]](https://www.youtube.com/watch?v=XuoqKYxDHVc&t=1090). Musk's fix? Orbital data centers — straight up in space [[19:55]](https://www.youtube.com/watch?v=XuoqKYxDHVc&t=1195).
🔹 Companies should be checking each other: governments simply aren't equipped to judge how dangerous a new model is. Musk's proposal: market leaders — Anthropic, xAI, OpenAI, and the Chinese tech giants — should swap access to their models a couple of weeks before release to test each other for safety [[12:50]](https://www.youtube.com/watch?v=XuoqKYxDHVc&t=770). Only if someone ignores a real red flag should regulators get pulled in [[22:01]](https://www.youtube.com/watch?v=XuoqKYxDHVc&t=1321).
🔹 Respect for Dario, a diss for Sam: Musk spoke warmly about Dario Amodei (CEO of Anthropic) and his team, calling them people of principle [[26:43]](https://www.youtube.com/watch?v=XuoqKYxDHVc&t=1603). Sam Altman (CEO of OpenAI) got the opposite treatment — Musk went after him hard, pointing out how OpenAI under his leadership morphed from an open nonprofit into an $800 billion closed corporation [[26:00]](https://www.youtube.com/watch?v=XuoqKYxDHVc&t=1560).
The road to hell is mostly paved with bad intentions, but a few good stones end up in there too. The point is — don't let your guard down.