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

“Join xAI if the idea of ​​team drivers on the moon appeals to you,” CEO Elon Musk announced yesterday after a restructuring that saw a group of former executives exit the AI ​​lab.

This is an interesting hiring strategy following the company’s merger with Musk’s rocket maker, SpaceX, and the combined company’s expected IPO. You might think that xAI employees must be fascinated with achieving artificial general intelligence, or using deep learning models to disrupt traditional software companies, or simply playing on bad words like “macrohard.” But instead, Elon went to the moon.

After outlining plans to build on-orbit AI data centers, a key synergy between the two companies, Musk took the idea further. “What if you want to go beyond just a terawatt per year?” – Musk asked. “To do that, you have to go to the moon… I really want to see a mass driver on the moon launching artificial intelligence satellites into deep space.”

In Musk’s telling, the step beyond Earth-orbiting data centers is larger computers in deep space. Furthermore, Musk says the best way to achieve this is to build a city on the moon to manufacture space computers and drop them into the solar system using a large maglev train.

If all this sounds like a lot, seasoned Musk watchers know there’s a clue as to where the discussion appeared in a video of an all-hands meeting shared by the AI ​​with the public. The slide describing the moon base comes at the end of the presentation, where, during SpaceX pep talks, Musk typically shares visualizations of SpaceX rockets landing on Mars and talks passionately about humanity’s multiplanetary future.

Notably, Moonbase comes right after SpaceX publicly backed away from its long-term goal of colonizing Mars. Now, with advanced artificial intelligence in the corporate fold, Musk needs a new sci-fi metaphor for the future: in this case, the Kardashev scale, a theoretical measure of galactic civilizations coined by a Soviet astronomer in the 1960s. The idea is to scale up energy use, as early civilizations figured out how to tap into all the energy sources on their planets, and then (hypothetically) go into space and build the infrastructure to capture the sun’s energy.

With the lunar base, Musk says the company can harness “perhaps even a small percentage of the sun’s energy” to train and operate artificial intelligence models. He told his staff: “It is difficult to imagine what an intelligence of this size could think, but it would be very exciting to see it happen.”

In the nine years since Musk unveiled his plan to explore and colonize Mars, the vision has been an effective recruiting tool for SpaceX: The founding story of Musk’s interest in the Red Planet provided a long-term vision that unified the company’s various development efforts, and signaled the company’s ambition among other space contractors who had settled for moonlighting over government priorities. The “Occupy Mars” T-shirts were a clear symbol of SpaceX’s aspirations.

This is where the virtual Moonbase fits in, part of a long history of Musk wrapping his companies in a powerful narrative. There are a million people living on Mars, but they are now serving a future where artificial intelligence is the most interesting thing. Mars mission creep became less apparent in Musk’s Starship update in May 2025, when the presentation ended with a now-scrapped vision of Tesla Optimus robots cruising across the Red Planet.

There was only one problem with SpaceX and Mars: no one wanted to pay them to go there. Plans announced in 2016 to reuse the company’s Dragon spacecraft as a Mars lander were abandoned the following year after the technical challenges became too costly. Since Musk unveiled the vehicle that would become Starship in 2016, its capabilities, initially intended for colonizing Mars, have been scaled back to focus on two additional lucrative missions: launching satellites for the Starlink communications network and $4 billion contracts to land astronauts on the moon for NASA.

Unlike a multi-planet civilization, there may be some sense in SpaceX buying money-generating artificial intelligence and social media to build data centers in Earth orbit, especially if expectations of higher demand and costs on Earth come true. Experts indicate that it may be possible in the 2030s.

In theory, building satellites on the moon would require fulfilling many of Musk’s other dreams first. Scientists and startups are experimenting with building chips and other microcomponents in space. But mass production of many tons of advanced computers on the Moon means that we live in a world where access to space is dramatically cheaper, which is the basic requirement for those technologies, and getting all the raw materials for such an effort to the Moon, plus everything else required for a “self-sufficient city.”

In a sense, that’s the point: this is the stretch goal. If happy retail investors buy into that argument, they could turn SpaceX stock into the next Tesla. The engineers, whether in AI or aerospace, that Musk needs to achieve his goals may find this shift disturbing. But vision is one way to explain what xAi is all about, other than the LLM which might be known as a pervert. As one departing company executive said on his way out, “All the AI ​​labs are building the exact same thing, and it’s boring.”

Producing a solar system-scale supercomputer on the moon is a lot of things (I’ll get emails for not using the word “crazy”), but it’s not quite the same, and it’s not boring.