Elon Musk is getting serious about orbital data centers

On Friday, when SpaceX submitted plans to the FCC for a data center network of 1 million satellites, you might have thought Elon Musk was having a little fun with us. But after a week, it became clear that he was very serious.

The most obvious step, of course, is the formal merger between SpaceX and xAI that moved forward on Monday, officially bringing together Musk’s space and AI projects in a way that would make more sense if there were some sort of joint infrastructure project planned.

But even after the merger, we’re starting to see the idea of ​​orbital AI data sets — essentially networks of computers operating in space — cohering into an actual plan. On Wednesday, the FCC accepted the request and set a timeline for public comment. It’s usually a formality, but FCC Chairman Brendan Carr took the unusual step of sharing the dossier on

Meanwhile, Elon Musk has begun making the case for creating orbital data centers in public places. In a new episode of Stripe co-founder Patrick Collison’s podcast “Cheeky Pint,” which also featured guest Dwarkesh Patel, Musk laid out the base case for moving most of our AI computing power into space. Essentially, solar panels produce more energy in space, so they can cut down on one of the major operating expenses of data centers.

“It’s harder to expand on Earth than it is to expand in space,” Musk said on the podcast. “Any solar panel will give you five times more power in space than on Earth, so it’s much cheaper to do it in space.”

Close listeners will notice that there is a bit of a gap in logic here! It’s true that solar panels produce more energy in space, but since energy isn’t the only cost in running a data center and solar panels aren’t the only way to run a data center, that doesn’t mean it’s cheaper to do everything in orbit, Patel pointed out on the podcast. Patel also raised concerns about serving GPUs failing during AI model training, but you’ll have to listen to the full episode for that.