SpaceX Puts a Fuel-Free Lunar Railgun on Its Roadmap, Reviving a 1977 MIT Machine

A magnetic mass driver that flings cargo off the Moon on electricity alone landed in SpaceX's S-1 filing, and the physics behind it - 22 times cheaper than escaping Earth - has had working prototypes sitting on a shelf for nearly 50 years.

The Moon is not a destination. It is the cheapest launchpad in the solar system, and almost nobody prices it that way. That single reframe is the whole reason a magnetic catapult that hurls satellites off the lunar surface with no engine and no fuel showed up in SpaceX's pre-IPO S-1 filing as a business the company thinks will eventually make money. I know how that sounds. A catapult on the Moon reads like medieval siege equipment bolted onto a sci-fi fever dream. But run the numbers and it stops being a joke and starts looking like an inevitability nobody wants to pay for yet.

Key Takeaways

• Escaping the Moon takes about 2.38 kilometers per second versus 11.2 for Earth, and because launch energy scales with speed squared, that fifth-the-speed gap means roughly 22 times less energy per kilogram to leave the Moon.
• 0.79 kilowatt-hours is all it takes to fling one kilogram off the lunar surface - less electricity than running your microwave for an hour.
• Gerard K. O'Neill, the Princeton physicist who invented the particle storage ring in 1956, bolted together Mass Driver 1 from about $2,000 of scrap and car batteries at MIT in May 1977, shooting a bucket down a line of ~20 copper coils at 40 meters per second.
• Throwing 10,000 metric tons off the Moon a year needs only about one to two megawatts of average power - roughly a single modern wind turbine.
• A 2026 engineering study by Casey Handmer found a large lunar driver averaging around 450 megawatts but spiking to a peak demand near 16 gigawatts, the output of more than a dozen nuclear plants for a fraction of a second.
• Lunar regolith is 40 to 45% oxygen by mass, and polar craters hold an estimated 600 million to 2 billion metric tons of water ice - a propellant depot hiding in the dirt.
• NASA directed its program toward a 100-kilowatt lunar fission reactor targeting around 2030, but a serious mass driver needs hundreds of megawatts - three to four orders of magnitude more, up to 10,000 times the power.
• SpaceX already holds NASA's primary crewed lunar lander contract worth roughly $4 billion, the first domino in a five-step chain that ends with the catapult.

---

Why 2.38 Beats 11.2 By So Much More Than It Looks

To leave Earth for good you have to hit about 11.2 kilometers per second - call it 25,000 miles per hour. That is a violent, fuel-guzzling number, and it is why a Falcon 9 sitting on the pad is more than 90% propellant by weight. You are launching a giant flying gas tank with a tiny payload perched on top.

The Moon asks for 2.38 kilometers per second, less than half a mile per second. Here is the part people skip: the energy to fling something off a body does not scale with speed, it scales with speed squared. Drop the required speed by a factor of five and the energy per kilogram falls by a factor of about 22. So the electricity to give one kilogram enough velocity to escape the Moon entirely comes out to roughly 0.79 kilowatt-hours. You are escaping a whole world's gravity for less power than reheating a plate of leftovers.

The Three Reasons Earth Can't Play This Game

The energy math is reason one. Reason two is gravity: the Moon's surface pull is about one-sixth of Earth's, so the track, the structure, the whole mechanism is fighting a much gentler grip. Everything weighs a sixth of what it does here.

Reason three is the decider, and it kills Earth outright. The Moon has no atmosphere. On Earth, firing something off the ground at thousands of miles per hour means slamming it into air that behaves like a solid wall. One serious Earth-based proposal put the nose heating around 30 kilowatts per square centimeter - a blowtorch the size of your fist, thousands of times hotter, hitting the tip of your craft the instant it leaves the rail. On the Moon you fire straight into vacuum. Nothing in the way. You just go. Shallow gravity well, feather-light weight, zero air. The Moon was practically built to throw things.

The Machine Is Just a Maglev Train That Never Slows Down

You have already seen the technology. A maglev train floats above its track on magnets and rides a magnetic wave forward, no wheels, no onboard engine. A mass driver is that same idea, except instead of gliding you to the airport at 200 miles an hour, it accelerates a small carrier - engineers call it a bucket - to thousands of miles an hour and then lets go.

The bucket holds the payload, rides the magnetic wave down the track, releases the cargo to fly free at the end, then peels off, loops around, and comes back for the next load. Nothing burns. Nothing touches. The bucket levitates the whole way, so nothing wears from friction. In principle a machine like this could fire millions of times. That is the dream: a launch system with no fuel, no rocket, and almost nothing to grind down.

A Princeton Physicist Already Built This in 1977

The idea is not new, and that was the part that genuinely got me. In May 1977 a small team at an MIT lab bolted together a contraption from about $2,000 of scrap - a row of roughly 20 copper coils and a bank of ordinary car batteries. They flipped it on, the coils fired in sequence, and a little bucket shot down the line at about 40 meters per second. They called it Mass Driver 1.

The man behind it was Gerard K. O'Neill, a Princeton physicist who in 1956 invented the particle storage ring, a concept that became foundational to how particle accelerators work. This was someone operating at the absolute frontier of physics. In the early 1970s he got fixated on a single heretical question: is the surface of a planet really the right place for an expanding technological civilization? He didn't think so. He laid out his answer in a 1974 paper and a 1976 book, The High Frontier, which won a major science award the next year.

His problem is the exact one SpaceX stares at now. To build enormous structures in space you need an enormous amount of raw mass, and hauling it up from Earth's gravity well is absurdly expensive. So O'Neill asked: what if the material comes from the Moon, where launching it is 22 times cheaper, flung out by an electromagnetic catapult to a gathering point where a catcher scoops it up? Mine the Moon, throw the ore, catch it, build. His team didn't stop at one machine either. Mass Driver 1 pulled its bucket at about 33 Gs. A few years later Mass Driver 2 hit around 500 Gs, and by 1980 Mass Driver 3 pushed past 1,800 Gs.

The Trade-Off That Sets How Big Your Catapult Has To Be

There is one tension at the heart of the machine: how hard you push versus how long your track runs. The distance you need equals speed squared divided by twice the acceleration. Push harder, shorter track. Push gently, longer track.

Watch what that does. Slam a payload at 1,000 Gs and the track only needs about 289 meters to reach lunar escape speed - two football fields. At the 1,800 Gs O'Neill's third machine actually hit, you are down to about 160 meters, shorter than a city block. You could build the launch system for an entire world on a strip of ground you could walk in two minutes.

But you can only shove that hard on things that can't break. A chunk of regolith does not care if you hit it with 1,000 Gs - there is nothing fragile inside it to shatter. So for dumb mass, oxygen and construction ore, your catapult stays tiny. The moment you want to throw something delicate, the equation flips. A satellite might tolerate tens of Gs. A human passes out or dies somewhere around two or three. To get a person to lunar escape speed at a gentle 2 to 3 Gs, the track isn't 160 meters anymore - it is hundreds of kilometers, the same physics scaled up 5,000 times. The length of your catapult is dictated entirely by the most fragile thing you ever intend to throw.

Cheap To Run, Brutal To Power

Here is what turns a physics toy into something worth billions. Scale that microwave-sized 0.79 kilowatt-hours up to industrial volume - 10,000 metric tons flung off the Moon per year - and the average power comes to only one or two megawatts. A single wind turbine. An entire off-world freight operation sipping the electricity of a farm's worth of blades.

So why doesn't it exist? Because average power lies. A mass driver doesn't draw steadily, it fires in violent pulses, dumping a huge slug of energy in a fraction of a second, over and over. Casey Handmer's 2026 study on a large lunar driver found average draw around 450 megawatts but peak demand spiking near 16 gigawatts - more than a dozen large nuclear plants, needed only for an instant. The genuinely hard part isn't the track. It's the pulsed power plant, the capacitors and flywheels that trickle-charge slowly and then unload everything in a flash. Handmer likened the difficulty to building ITER, the giant fusion reactor. The catapult is the easy piece. The thing that feeds it is the monster.

What It Actually Does Is Rewrite What the Moon Is Worth

Mechanically the machine throws things. Economically it does something far bigger: it deletes the rocket. Today, getting any mass off the Moon means fuel, propellant, and the cost of hauling or manufacturing that propellant. Strip the rocket out and the cost of launching a ton off the lunar surface stops being a fuel bill and becomes an electricity bill - kilowatt-hours from a reactor or a solar array, 22 times cheaper than Earth, with no atmosphere to fight.

So the Moon stops being an expensive rock you visit and becomes the single cheapest place in the solar system to get mass moving. Cheaper than launching from Earth. Cheaper than anywhere. Once you see that, the question flips from "why build a catapult on the Moon" to "what does the lunar economy look like when lunar material is the cheapest material in space." You walked in expecting a launch story and you are suddenly holding a commodity story about who controls the cheapest mass in the solar system.

What the Moon is genuinely rich in is oxygen - 40 to 45% of regolith by mass, nearly half the ground locked up as oxidizer in rock. And oxygen is the heavy majority of any rocket's propellant load. Pull it from the dirt, fling it into orbit, and you've built a gas station in space where passing spacecraft top up on lunar oxygen instead of anything dragged up from Earth. Add the water ice in permanently shadowed polar craters, confirmed in 2009 when NASA crashed a probe into a south pole crater and read the plume - somewhere between 600 million and 2 billion metric tons. Split it into hydrogen and oxygen and you have fuel and breathable air.

Dust, Vacuum, and the Chicken-and-Egg Trap

I want to steel-man the case against this, because it is strong. Start with dust. Lunar dust has never been weathered by wind or water, so the grains are jagged and razor-sharp, like microscopic broken glass, and electrostatically charged so they cling to everything. During Apollo, astronaut gear started degrading after hours of exposure - seals bogged down, surfaces ground away like they'd been sandblasted. A mass driver is kilometers of close-tolerance moving parts and precision surfaces that must run for years in an environment that eats machinery in hours. Then layer on a surface swinging from about 121°C in daylight to -133 at night, a 14-and-a-half-day night that starves any solar array for two straight weeks, plus raw radiation, vacuum, and micrometeorites. This is the most hostile factory floor humans have ever contemplated.

The deeper problem is a chicken-and-egg trap. A catapult only pays off if there is already a mountain of mined, refined, packaged material waiting to be thrown. But the cheap transport the catapult provides is part of what would create the demand for that material in the first place. So you can't lead with the catapult. It's one of the last machines you build in a mature lunar industry, not one of the first, and the early cycle has to lean on gear hauled up from Earth the expensive way. Someone eats that cost for a long time.

And the power gap is savage. NASA directed its program toward a 100-kilowatt lunar fission reactor targeting roughly 2030, backed by a formal Department of Energy agreement signed in January 2026 - genuinely important, the first real nuclear power headed for another world, and partly a race against China's own reactor planned for the early-to-mid 2030s. But 100 kilowatts is kilowatt scale. A serious mass driver needs hundreds of megawatts. That's three to four orders of magnitude beyond anything funded today.

The Order It Has To Happen In, and Why I Think It Finally Might

Nothing in physics forbids this. That's what makes it a real story instead of a sci-fi one. The bottleneck is entirely industrial, and it has to unfold in sequence. First, Starship landing cargo and crews on the Moon routinely - SpaceX already holds NASA's primary crewed lander contract worth around $4 billion. Second, lunar surface power climbing from that 100-kilowatt reactor toward hundreds of megawatts, a thousandfold haul. Third, industrial-scale mining and refining of regolith. Fourth, manufacturing the track segments and magnets from local metal, because shipping a 160-meter precision machine from Earth defeats the entire point. Fifth, humanoid robots like Tesla's Optimus building and aligning that machine to sub-millimeter tolerance. Only then, the catapult.

The day a real lunar industry exists - mines, refineries, power, maybe people - a mass driver becomes almost inevitable, because the economics are too obvious to ignore. The only open question is when, and the honest answer is that it might be 50 more years, or it might be never. What changed is the forcing function. The most capable rocket company in history has now fused itself with an AI effort, put the concept in its filings, and is deadly serious about it. The dream of flinging mass off the Moon with no rocket has sat fully formed on a shelf since before I was born. After 50 years of gathering dust, I think this might finally be the moment somebody takes it down. For the first time, my answer is leaning yes.