On Tuesday night, Elon Musk called an all-hands meeting for xAI employees. Apparently he wanted to talk about the future of his AI company, especially how it relates to the moon.
Musk told employees that xAI would need a manufacturing facility on the moon, a factory that would manufacture AI satellites and send them into space with giant catapults, according to The New York Times, which heard the meeting. “I have to go to the moon,” he said, according to the Times. The move will help xAI leverage more computing power than any of its rivals, he explained. “It’s hard to imagine what an intelligence of that scale would think, but it would be incredibly exciting to see it happen,” he added.
What Musk didn’t seem to explicitly mention was how these would be structured, or how he intends to reorganize the newly combined xAI-SpaceX entity, which is simultaneously moving towards a potentially historic IPO. He proudly acknowledged that the company is in flux. “If you’re moving faster than everyone else in any technology field, you can be a leader,” he told employees, according to the Times. “And xAI is moving faster than anyone else, and no one can come close to it,” he added.
It’s unclear what prompted everyone to participate, but whatever the cause, the timing is at least interesting. On Monday night, xAI co-founder Tony Wu announced his retirement. Less than a day later, xAI’s other co-founder, Jimmy Barr, who reports to Musk, said he too was jumping. This means a total of six of xAI’s 12 founding members have left the young company. Any split is said to be amicable, and with an IPO reportedly scheduled for this summer that would value SpaceX at $1.5 trillion, everyone involved should do very well financially.
The moon itself has been of recent interest. For most of SpaceX’s 24-year existence, Mars has been the ultimate goal. Last Sunday, just before the Super Bowl, Musk surprised many by posting that SpaceX had “shifted its focus to building self-growing cities on the moon” and claiming that colonizing Mars would take “20+ years.” The moon could get there in half the time, he said.
This is quite a change of direction for a company that has never sent a mission to the moon before.
Rational or not, investors seem to be considerably more excited about data centers in orbit than colonies on other planets. (That’s a long timeline, even with the most patient capital in the room.) But for at least one xAI venture backer who spoke with this editor last year, the moon’s ambitions have nothing to do with Wall Street or distract from xAI’s core mission. They are inseparable from it.
tech crunch event
boston, massachusetts
|
June 23, 2026
The theory put forward by VCs at the time was that Musk had been building towards a single goal from the beginning. It’s an AI trained on the world’s most powerful global model, not just text and images, but unique real-world data that no competitor can replicate. Tesla contributes to energy systems and road topology. Neuralink provides a window into the brain. SpaceX provides physics and orbital mechanics. The Boring Company adds some underground data. Add a lunar factory to this and you start to see the outlines of something very powerful.
The big question is whether that vision is achievable. The other thing is whether it’s legal. Under the 1967 Outer Space Treaty, no country, or by extension any company, can claim sovereignty over the moon. However, a 2015 US law opened up a significant loophole. You can’t own the moon, but you can own everything you extract from it.
As Mary Jane Rubenstein, a science and technology studies professor at Wesleyan University, explained to TechCrunch last month, this distinction is somewhat illusory. “That’s like saying you can’t own the house, but you can own the floorboards and beams,” she said. “What’s on the moon is on the moon.”
While not everyone has agreed to abide by its rules (China and Russia certainly haven’t), the legal framework clearly provides a scaffolding for Mr. Musk’s moon ambitions. Meanwhile, the team that helps him get there continues to get smaller, at least for now.
