Musk needed a new vision for SpaceX and xAI. He landed on Moonbase Alpha.

“Join xAI if the thought of wide drivers connected the Moon appeals to you,” CEO Elon Musk proclaimed yesterday pursuing a restructuring that saw a watercourse of erstwhile executives exit the AI lab.

This is an absorbing recruitment strategy aft the company’s merger with Musk’s rocket maker, SpaceX, and the combined company’s anticipated IPO. You mightiness deliberation that xAI employees ought to beryllium fascinated with achieving AGI, utilizing heavy learning models to disrupt accepted bundle companies, oregon simply atrocious wordplay similar “Macrohard.” But instead, Elon is going to the moon.

After outlining plans to build AI information centers successful orbit, the superior synergy betwixt the 2 companies, Musk took the thought further. “What if you privation to spell beyond a specified terawatt per year?” Musk asked. “To bash that, you person to spell to the moon…I truly privation to spot a wide operator connected the satellite that is shooting AI satellites into heavy space.”

New year, caller dream.Image Credits:SpaceX

In Musk’s telling, the measurement beyond information centers orbiting Earth is adjacent larger computers successful heavy space. And furthermore, Musk says the champion mode to execute that is to physique a metropolis connected the Moon to manufacture abstraction computers and hurl them into the star strategy utilizing a large maglev train.

If that each feels a spot much, seasoned Musk watchers cognize there’s a hint astir wherever the treatment appears successful a video of an all-hands meeting xAI shared with the public. The descent describing the Moon basal comes astatine the extremity of the presumption deck, where, during SpaceX pep talks, Musk typically shares renderings of SpaceX rockets landing connected Mars and waxes rhapsodic astir the aboriginal of multi-planetary humanity.

Notably, the Moonbase comes conscionable aft SpaceX has publically backed distant from its long-held extremity of colonizing Mars. Now, with xAI successful the firm fold, Musk needs a caller subject fabrication metaphor for the future: In this case, the Kardashev Scale, a theoretical measurement of galactic civilizations coined by the eponymous Soviet astronomer successful the 1960s. The thought is climbing the standard of vigor usage — aboriginal civilizations fig retired however to leverage each the powerfulness sources connected their planets, and past (hypothetically) spell to abstraction and physique infrastructure to seizure the vigor of the sun.

With the Moon base, Musk says the institution could harness “maybe adjacent a fewer percent of the sun’s energy” to bid and run AI models. “It’s hard to ideate what an quality of that standard would deliberation about,” helium told his staff, “but it’s going to beryllium incredibly breathtaking to spot it happen.”

In the 9 years since Musk unveiled his program for Martian exploration and colonization, the imaginativeness has been an effectual hiring instrumentality for SpaceX: The founding communicative of Musk’s involvement successful the Red Planet offered a semipermanent imaginativeness that agreed the company’s assorted improvement efforts, and signaled the company’s ambition among different abstraction contractors that settled for incremental enactment connected authorities priorities. “Occupy Mars” t-shirts offered a disposable awesome of SpaceX’s aspirations.

That’s wherever the hypothetical Moonbase fits in—part of a agelong past of Musk wrapping his companies successful a almighty narrative. It’s 1 cardinal radical surviving connected Mars, but present catering to a aboriginal wherever AI is the astir absorbing thing. Martian ngo creep became evident little successful Musk’s May 2025 Starship update, erstwhile the presumption ended with a now-cancelled imaginativeness of Tesla Optimus robots clomping crossed the Red Planet.

Poor robot. Image Credits:SpaceX

There was conscionable 1 occupation with SpaceX and Mars: No 1 wanted to wage them to spell there. Plans announced successful 2016 to repurpose the company’s Dragon spacecraft arsenic a Mars lander were abandoned the adjacent twelvemonth aft the method challenges became excessively costly. And since Musk unveiled the conveyance that would go Starship successful 2016, its capabilities, initially intended for Mars colonization, person been scaled backmost to absorption connected 2 much remunerative tasks: Launching satellites for the Starlink comms web and $4 cardinal worthy of contracts to onshore astronauts connected the Moon for NASA.

Unlike a multi-planetary civilization, determination whitethorn beryllium immoderate logic successful having SpaceX acquisition a money-burning AI and societal media to physique information centers successful Earth orbit, peculiarly if forecasts of rising request and costs connected the crushed travel true. Experts suggest that it mightiness beryllium imaginable successful the 2030s.

Hypothetically, gathering satellites connected the Moon would necessitate a batch much of Musk’s different dreams coming existent first. Scientists and start-ups are experimenting with building chips and different precision components successful space. But mass-producing galore tons of precocious computers connected the Moon means we’re surviving successful a beingness wherever it is dramatically cheaper to get to space, which is the cardinal request for those technologies, getting each the earthy materials for specified an effort to the Moon, positive immoderate is required for a “self-sustaining city.”

In a sense, that’s the point: This is the, uh, agelong goal. If meme-happy retail investors bargain the argument, they could crook SpaceX shares into the adjacent Tesla. The engineers, AI oregon aerospace, who Musk needs to execute his goals whitethorn find the displacement jarring. But the imaginativeness is 1 mode to explicate what xAi is about, different than an LLM possibly champion known as a pervert. As 1 of the company’s departing executives said connected his mode retired the door, “all AI labs are gathering the nonstop aforesaid thing, and it’s boring.”

Mass-producing a star system-scale supercomputer connected the Moon is galore things (I’m going to get emails for not utilizing the connection “insane”), but it is not the nonstop aforesaid thing, and it is not boring.