Musk wanted a brand new imaginative and prescient for SpaceX and xAI. He landed on Moonbase Alpha.

“Be part of xAI if the thought of mass drivers on the Moon appeals to you,” CEO Elon Musk proclaimed yesterday following a restructuring that noticed a stream of former executives exit the AI lab.

That is an fascinating recruitment technique after the corporate’s merger with Musk’s rocket maker, SpaceX, and the mixed firm’s anticipated IPO. You would possibly assume that xAI staff must be fascinated with attaining AGI, utilizing deep studying fashions to disrupt conventional software program corporations, or just dangerous wordplay like “Macrohard.” However as a substitute, Elon goes to the moon.

After outlining plans to construct AI knowledge facilities in orbit, the first synergy between the 2 corporations, Musk took the thought additional. “What if you wish to transcend a mere terawatt per yr?” Musk requested. “To do this, it’s important to go to the moon…I actually wish to see a mass driver on the moon that’s taking pictures AI satellites into deep house.”

In Musk’s telling, the step past knowledge facilities orbiting Earth is even bigger computer systems in deep house. And moreover, Musk says the easiest way to realize that’s to construct a metropolis on the moon to fabricate house computer systems and hurl them into the photo voltaic system utilizing an enormous maglev practice.

If that every one feels a bit a lot, veteran Musk watchers know there’s a clue about the place the dialogue seems in a video of an all-hands meeting xAI shared with the general public. The slide describing the moon base comes on the finish of the presentation deck, the place, throughout SpaceX pep talks, Musk sometimes shares renderings of SpaceX rockets touchdown on Mars and waxes rhapsodic about the way forward for multi-planetary humanity.

Notably, the moon base comes simply after SpaceX has publicly backed away from its long-held objective of colonizing Mars. Now, with xAI within the company fold, Musk wants a brand new science fiction metaphor for the long run: On this case, the Kardashev Scale, a theoretical measure of galactic civilizations coined by the eponymous Soviet astronomer within the Sixties. The concept is climbing the dimensions of vitality utilization — early civilizations work out how one can leverage all the facility sources on their planets, after which (hypothetically) go to house and construct infrastructure to seize the vitality of the solar.

With the moon base, Musk says the corporate may harness “perhaps even a couple of p.c of the solar’s vitality” to coach and function AI fashions. “It’s troublesome to think about what an intelligence of that scale would take into consideration,” he advised his employees, “nevertheless it’s going to be extremely thrilling to see it occur.”

Within the 9 years since Musk unveiled his plan for Martian exploration and colonization, the imaginative and prescient has been an efficient hiring device for SpaceX: The founding story of Musk’s curiosity within the Pink Planet supplied a long-term imaginative and prescient that united the corporate’s varied growth efforts, and signaled the corporate’s ambition amongst different house contractors that settled for incremental work on authorities priorities. “Occupy Mars” t-shirts supplied a visual image of SpaceX’s aspirations.

That’s the place the hypothetical moon base suits in — a part of a protracted historical past of Musk wrapping his corporations in a robust narrative. It’s a million individuals residing on Mars, however now catering to a future the place AI is essentially the most fascinating factor. Martian mission creep grew to become obvious much less in Musk’s May 2025 Starship update, when the presentation ended with a now-cancelled imaginative and prescient of Tesla Optimus robots clomping throughout the Pink Planet.

There was only one downside with SpaceX and Mars: Nobody needed to pay them to go there. Plans introduced in 2016 to repurpose the corporate’s Dragon spacecraft as a Mars lander have been deserted the following yr after the technical challenges grew to become too pricey. And since Musk unveiled the car that will change into Starship in 2016, its capabilities, initially supposed for Mars colonization, have been scaled again to deal with two extra remunerative duties: launching satellites for the Starlink comms community and $4 billion value of contracts to land astronauts on the moon for NASA.

In contrast to a multi-planetary civilization, there could also be some logic in having SpaceX buy a money-burning AI and social media to construct knowledge facilities in Earth orbit, notably if forecasts of rising demand and prices on the bottom come true. Specialists counsel that it could be attainable within the 2030s.

Hypothetically, constructing satellites on the moon would require much more of Musk’s different desires coming true first. Scientists and startups are experimenting with building chips and different precision parts in house. However mass-producing many tons of superior computer systems on the moon means we’re residing in a universe the place it’s dramatically cheaper to get to house, which is the central requirement for these applied sciences, getting all of the uncooked supplies for such an effort to the moon, plus no matter is required for a “self-sustaining metropolis.”

In a way, that’s the purpose: That is the, uh, stretch objective. If meme-happy retail traders purchase the argument, they may flip SpaceX shares into the following Tesla. The engineers, AI or aerospace, who Musk wants to realize his objectives might discover the shift jarring. However the imaginative and prescient is one method to clarify what xAI is about, aside from an LLM maybe greatest generally known as a pervert. As one of many firm’s departing executives said on his approach out the door, “all AI labs are constructing the very same factor, and it’s boring.”

Mass-producing a photo voltaic system-scale supercomputer on the moon is many issues (I’m going to get emails for not utilizing the phrase “insane”), however it isn’t the very same factor, and it isn’t boring.