Following a recent restructuring and merger with Elon Musk’s rocket maker, SpaceX, AI lab xAI is embarking on an unconventional recruitment drive. CEO Elon Musk proclaimed, “Join xAI if the idea of mass drivers on the Moon appeals to you,” signaling a departure from traditional AI aspirations like achieving AGI or disrupting software companies with deep learning models. This bold vision positions the combined company for an anticipated IPO, leveraging synergies that extend beyond initial plans for AI data centers in Earth orbit.

Musk’s vision escalates to establishing even larger computers in deep space, with the Moon serving as the manufacturing hub. He outlined plans for a city on the moon dedicated to building these space computers and launching them into the solar system using a “big maglev train,” or mass driver. This ambitious endeavor is a strategic evolution, especially as SpaceX has publicly withdrawn from its long-held goal of Mars colonization, a narrative that previously served as a powerful recruiting tool for the company.

For nearly a decade, the dream of Martian exploration and colonization, epitomized by “Occupy Mars” merchandise, effectively united SpaceX’s diverse development efforts and attracted talent. However, the Red Planet mission faced significant hurdles. Plans to repurpose Dragon spacecraft as Mars landers were abandoned due to technical challenges and escalating costs. Starship’s capabilities, initially conceived for Mars colonization, were scaled back to focus on more immediately remunerative tasks: launching Starlink satellites and securing lucrative NASA contracts to land astronauts on the Moon.

With xAI now integrated, Musk is introducing a new scientific metaphor to inspire: the Kardashev Scale. This theoretical measure of galactic civilizations, proposed by a Soviet astronomer in the 1960s, gauges a civilization's technological advancement by its energy consumption. Musk suggests that a moon base could harness “maybe even a few percent of the sun’s energy” to train and operate massive AI models. While contemplating the implications of such an intelligence, he told his staff it would be “incredibly exciting to see it happen.”

While the prospect of building AI data centers in Earth orbit, potentially in the 2030s, holds some logic given rising demand and costs on the ground, the lunar base concept presents significantly greater challenges. Realizing a self-sustaining city on the moon to mass-produce advanced computers and launch them into deep space necessitates a universe where access to space is dramatically cheaper, and the logistics of transporting raw materials are overcome. This is acknowledged as a