At an internal xAI all‑hands on 10 February, Elon Musk sketched a bold, unconventional plan: build a factory on the Moon to manufacture artificial‑intelligence satellites and use large space ejectors to launch them into orbit. "We must go to the Moon," he told employees, arguing that lunar production would give xAI an edge in the scale of compute and satellite deployment available to its AI business.
The proposal folds together long‑standing themes in Musk’s portfolio: the ambition to industrialize space, the economics of mass launch through SpaceX’s Starship family, and the drive to keep AI development vertically integrated. Musk framed the Moon not as a novelty but as a practical way to secure vast extra capacity for running large models and for producing satellites without Earth‑bound constraints on energy, cooling and launch cadence.
Turning that rhetoric into reality would require surmounting major engineering and logistical hurdles. Lunar manufacturing faces harsh radiation, extreme thermal swings, limited local infrastructure and high transport costs for equipment and personnel. It would demand either substantial in‑situ resource utilization to exploit lunar regolith or repeated and cheaper heavy‑lift to ferry materials from Earth — both capabilities that are nascent at best.
There are, however, plausible strategic advantages to space‑native AI hardware. Low gravity could reduce the propellant cost of sending satellites into high orbits, continuous solar illumination in selected lunar locations can supply power, and operating outside Earth’s atmosphere can ease large‑scale heat rejection for power‑hungry processors. For a company that controls launch capability, satellite buses and AI software, the integration could unlock new architectures for distributed, low‑latency intelligence in orbit.
The announcement also raises geopolitical and regulatory questions. Large‑scale deployment of AI satellites touches on national security concerns, dual‑use technologies and norms governing the militarization of space. The Outer Space Treaty prohibits national appropriation of celestial bodies but is vague about commercial manufacturing; cooperation and competition among states and firms will shape whether such projects are treated as civil, commercial or strategic.
From a business perspective, the plan is an aggressive bet. It leans on SpaceX’s ability to deliver routine, cheap heavy launch and on xAI’s capacity to scale compute workloads in non‑terrestrial environments. The timing and costs are uncertain, and attention will fall on whether Musk can marshal investment, secure partnerships, and keep technical teams intact. The vision may also serve as positioning — signaling technological leadership and challenging rivals who are focused on terrestrial data‑centres and commercial cloud providers.
Whether or not a lunar factory is realistic in the near term, the announcement is consequential. It reframes the debate about where future AI compute will live, accelerates discussion of space as a domain for infrastructure competition, and will prompt regulators and competitors to rethink the intersection of AI strategy and space policy. For investors, engineers and governments, Musk’s pitch underlines that the next phase of competition for AI supremacy may extend beyond Earth’s surface.