A lunar factory vision
Serving as CEO of SpaceX and xAI, Elon Musk told employees during an all-hands meeting that the company needs to build a factory on the moon, a facility that would manufacture AI satellites using lunar materials.
A massive electromagnetic rail system, known as a mass driver, would then launch those satellites into deep space. Musk said he wants to see a mass driver on the moon shooting AI satellites outward, tying the plan to long-term growth in computing and to energy access beyond Earth's limits.
Admit it, if anybody else made such declarations, you'd call them ridiculous. Elon Musk, though?
"My estimate is that, within two to three years, the lowest-cost way to generate AI compute will be in space," Elon Musk wrote in a Feb. 2 update that announced SpaceX's acquisition of xAI.
He reinforced that belief on Feb. 11 in an all-hands meeting with xAI staff, video of which the company posted on X. Musk said that, while launching AI satellites from Earth is the immediate focus, SpaceX's new Starship megarocket will also enable operations on other worlds.
Musk argued that mining lunar soil for silicon, aluminum, and iron avoids lifting every component out of Earth's gravity well, claiming a lunar factory paired with a mass driver could generate between 500 and 1,000 terawatts per year of AI compute capacity operating in deep space.
He framed the concept as a step toward capturing more solar energy and advancing humanity along the Kardashev scale. Musk also linked the plan to a self-sustaining lunar settlement that eventually supports Mars missions.
Not a new idea
A professor at Princeton University first outlined the mass driver concept in 1974. Physicist Gerard K. O'Neill envisioned mining the moon for raw materials and launching them without chemical rockets. He believed electromagnetic acceleration could send lunar cargo toward construction sites for large space habitats positioned between Earth and the moon.
O'Neill worked with Henry H. Kolm, a professor of physics at the Massachusetts Institute of Technology, and built early working prototypes, called Mass Drive One, in 1976 and 1977.
Those models demonstrated that a sequence of electromagnetic coils could accelerate payloads smoothly along a track. Later, O'Neill founded the Space Studies Institute to continue research into space manufacturing and the use of lunar resources.
How the mass driver works
A lunar mass driver relies on superconducting coils arranged along a long track. Electrical pulses move in sequence, creating a traveling magnetic field that pushes a sled carrying cargo, increasing speed along the rail.
Escape velocity on the moon is close to 2.4 kilometers per second. Not having an atmosphere removes drag and eliminates most aerodynamic heating concerns. Engineers would power the system with solar arrays or compact nuclear reactors.
Satellites would need structural reinforcement to survive high acceleration forces, and once launched, they would enter precise trajectories that limit the need for heavy onboard fuel. Since the launchers are fixed on the lunar surface, operators could reuse them indefinitely.
Over time, lunar manufacturing could expand the rail network using locally sourced materials.
Why now
AI data centers use vast amounts of electricity and occupy large areas. Musk sees space-based compute clusters as a way to break those limits.
SpaceX Starship development suggests heavy cargo can land on the moon at a lower cost than in previous decades. Musk merged xAI with SpaceX operations to align rocket engineering with AI expansion goals, publicly stating that building a lunar mass driver represents a decisive step toward long-term growth.
Starship testing continues to advance cargo capacity and reusability; several governments plan sustained lunar exploration through programs such as NASA's Artemis initiative, led by Administrator Bill Nelson.
International interest in lunar infrastructure has grown, increasing the strategic appeal of industrial operations beyond Earth.
The engineering obstacles
A serious threat to any manufacturing process on the Moon is lunar dust; fine particles cling to surfaces and degrade mechanical systems. Temperatures swing from about -280 ºF during the long lunar night to over 200 ºF in daylight.
Power storage must bridge two-week periods without sunlight. Anchoring a long electromagnetic rail in low gravity demands careful structural design to counter recoil forces.
Also, this isn't cheap.
Financial risk also looms large. Early construction would require billions in investment before any payload generates a return. Navigation precision must remain exact at launch because course corrections in deep space add cost and mass.
Could it happen
O'Neill's prototypes proved that electromagnetic launch works in principle. Modern advances in superconductors, power electronics, and maglev technology strengthen the concept. Musk has already demonstrated reusable rockets and a global satellite network through Starlink.
Whether a lunar mass driver works within a decade isn't known. Hardware progress, funding stability, and sustained political support determine the pace. The physics holds, the materials exist, but the challenge rests in scale, cost, and execution.
If built, a lunar mass driver would mark the first large-scale launch system operating from another world. Such a step would reshape satellite deployment and space manufacturing.
Musk revived a 1970s vision, pairing it with modern ambition. The next several years will show whether that ambition turns steel, silicon, and lunar dust into a functioning rail aimed at the stars.
