Steelman Press
Palmer Luckey

Five Miles Down

Where do you fight when every battlefield becomes transparent?

Preface

Palmer Luckey believes the Earth's crust is the next warfighting domain. Not space. Not cyber. The dirt beneath your feet.

He has said this on stages from West Point to Joe Rogan, and he knows exactly how it sounds.

I believe [the earth's crust will be] the next major war fighting domain. I've said this many times and everyone thinks that I'm nuts.

Palmer LuckeyMoonshots with Peter Diamandis

He has been saying it since at least 2023 — first as speculation, then as aspiration, and by March 2026 as a statement about hardware that exists. To understand why, you have to follow the logic from the beginning, the way he lays it out.

The Transparency Problem

The argument starts not underground but on the surface, in the ocean, in the air — everywhere Anduril already operates. Luckey's company builds sensor towers that see across hundreds of square miles. It builds autonomous submarines. It builds AI that fuses every sensor feed into a single operational picture. He describes the mission in the plainest possible language:

It's really just trying to make the world transparent to military. We want to know where everything is. We don't want anyone to be able to hide from us.

Palmer LuckeySourcery with Molly O'Shea

Anduril's Lattice platform is designed to become a superintelligence system that connects sensors to build a comprehensive model of the world, enabling prediction of enemy actions rather than just reaction. Luckey would rather predict where the enemy will be — skate to the puck — than have a jet that flies slightly faster.

And the trend is working. The ocean is becoming more transparent. Luckey won't discuss the specific technologies, but he notes that even the open-source literature shows it is getting easier to find submarines. Anduril has even built working optical camouflage for drones — and then shelved it, because against any serious adversary equipped with infrared, thermal, LIDAR, or radar, visible-spectrum camouflage is not just useless but can actually make you easier to spot.

Every domain is converging on the same endpoint: total visibility. The air is transparent. The sea is becoming transparent. The surface has been transparent for decades. Space is transparent in principle and increasingly so in practice, with spacecraft that cannot be armored because mass constraints are too severe — meaning any successful hit results in total destruction.

If your company's defining ambition is to ensure that nothing can hide, then you are forced to confront a question. Where do your assets go when the world becomes a glass house?

The Exception

Finding mobile targets, small targets deep underground that are density matched with the surrounding terrain, nobody has any idea how to find things like that.

Palmer LuckeyUpstream with Erik Torenberg

"Density matched" is the key phrase. A vehicle designed so that its material properties approximate the surrounding rock and soil would confound the methods that detect objects in every other medium — acoustic, gravimetric, electromagnetic. Unlike the ocean, where a submarine's signatures eventually betray it to a capable enough sensor network, the crust offers no equivalent handle.

Luckey draws the comparison to submarines and then overturns it:

Someone can know exactly where a sub is, it doesn't mean they can get to it and stop it. All of those benefits are massively, massively compounded when you are in the crust of the Earth.

Palmer LuckeyInternet Marketing Association

A submarine can be found with enough sensors. Once found, it can be killed. Something five miles underground cannot be reached even if you know precisely where it is.

Very, very hard to go get something that is five miles under the surface of the Earth. You can know it's there and there's nothing you can do about it.

Palmer LuckeyInternet Marketing Association

This is a claim about physics, not technology. If it holds, the underground domain has a property no other domain shares: permanent opacity. Not opacity that degrades as sensors improve. Opacity that is a physical feature of the medium itself.

The History

The obvious objection: if this is such a good idea, why hasn't anyone done it?

Luckey's answer is that they did.

The United States and the Soviet Union during the Cold War both had subterranean programs, building vehicles that moved through the crust of the earth just like a submarine would move through the ocean.

Palmer LuckeyMoonshots with Peter Diamandis

He tells this story everywhere — at Singularity University with Peter Diamandis, on Upstream with Erik Torenberg, at West Point in front of cadets. The details are consistent. Both superpowers built nuclear-powered manned subterranean prototypes. That nuclear-powered earth-melter digger mole train was a real US military program, not just a crazy thing from one movie. The Soviets went further:

And the Soviets actually built a prototype and then lost it... in the crust of the earth. So it worked that well. It worked that it just went off and they lost track of it melting through the crust.

Palmer LuckeyMoonshots with Peter Diamandis

At West Point he added a dry observation: getting far enough to lose a subterranean in the crust of the Earth is, in its own way, a sign of significant progress. You have to get pretty far to have that problem.

So why did it stop?

The thing that killed that dream isn't that we decided that it was impossible. It was that we won the Cold War and we turned the spigot off on a lot of very, very interesting things.

Palmer LuckeyAxios

Two forces conspired. First, nuclear power fell out of political favor, and these vehicles required nuclear power. When nukes went out of fashion, everything built on them was canceled — subterrains, ramjet-powered space fighters, nuclear highway excavation through mountains, nuclear creation of new inland seas. Second, the Cold War ended. The threat evaporated. The funding dried up.

The domain was orphaned by contingency, not by physics. It sat empty for thirty years — not because it couldn't work, but because no one had a reason to revisit it.

The Unlock

The historical programs required manned vehicles. Nuclear-powered, large-diameter, manned. Luckey changes one variable: the assumption about who — or what — has to be inside.

The thing that has changed is autonomy means you don't have to have a person aboard to do intelligent tasking and navigation. And so, the diameter of what you need can go down to, instead of feet, inches, or even millimeters.

Palmer LuckeyWest Point - The U.S. Military Academy

The subterranean application is the most extreme version of a thesis Luckey applies across every domain Anduril operates in. Consider how he thinks about ships. Today, building an American warship requires extraordinary safety standards: fire compartmentalization, anti-sinking flotation, perfect welds verified by X-ray. All of this exists because humans are aboard.

I don't have to build a ship with crazy anti-sinking, long flotation properties if there's no people aboard. I don't have to dedicate weight, bulk, and cost to fire compartmentalization if there's no people on the ship when it gets hit.

Palmer LuckeyCore Memory

Remove the human and you remove every constraint the human imposed. Every pound of life support. Every dollar of safety engineering. Every limitation on maneuver that would kill a pilot. One out of 100 autonomous submarines failing due to manufacturing problems becomes a cost on a balance sheet rather than a human tragedy that paralyzes the entire system. Autonomous fighter jets can execute maneuvers from a "forbidden" tactics book that would kill human pilots. Autonomous warships can be wrapped in cellophane filled with argon gas — a "Twinkie ship" that lasts a thousand years — and warehoused until war comes. With autonomy, you could build 10,000 airplanes, put them in a warehouse, and not hire a single new pilot.

Now apply all of this to the underground. A manned subterranean vehicle needs to be large enough for a crew, powerful enough to sustain them, and connected enough for them to communicate with the surface. These requirements demanded nuclear reactors, diameters measured in feet, and communication links through miles of rock — links that do not work. Autonomy dissolves all three. The diameter drops from feet to inches. The power source shrinks to match. And the communication problem ceases to exist, because autonomy and artificial intelligence allow you to, for the first time, have payloads that can do useful things operating autonomously deep underground for long distances.

At West Point, Luckey made the economics explicit:

The cool thing about digging is that diameter is expensive and length is free. If I dig a one-foot-long, one-inch-diameter tunnel, every foot of stuff I pull through after that is basically energy-free. It takes no work.

Palmer LuckeyWest Point - The U.S. Military Academy

The Cold War programs built wide because they had to fit humans inside. The physics of tunneling rewards the opposite approach: narrow and long. Autonomy lets you build for the medium instead of for the crew.

The Machine

By May 2026, at West Point, Luckey was willing to describe what he was building:

The ideal vehicle is probably hundreds of meters long, a bunch of train cars and explosives and effects and variety of other things. And then a unique thing that I've made that is really, really good at getting rid of dirt and gasifying it, that allows it to dig through the front.

Palmer LuckeyWest Point - The U.S. Military Academy

A snake train. Hundreds of meters of linked autonomous cars boring through the crust behind a proprietary dirt-gasification system, carrying explosives and electronic warfare payloads and whatever else fits in a narrow tube. No crew. No communication link. No way to detect it — and even if you could detect it, no way to stop it.

The operational concept is not tunneling in the traditional sense. Not bunkers. Not trenches. Not the kind of SubT work the Army already does. Luckey describes using the crust of the Earth as a fully three-dimensional battle space for moving supplies, conducting electronic warfare, kinetic warfare, psychological warfare, and high-end logistics. The crust as a medium, the way the ocean is a medium. Something you move through, not something you dig into and hide.

By March 2026, on Axios, the prototypes were real:

Anduril has working prototypes of subterranean systems that can deliver a variety of kinetic, electronic, and other effects.

Palmer LuckeyAxios

The Strategic Logic

Luckey's broader defense philosophy is organized around deterrence — building things so powerful that adversaries conclude fighting is futile before they start.

The goal here is not to build the weapons that will win a war with China. It's to build the things that will make China never want to go to war with us in the first place.

Palmer LuckeyWBNS 10TV

He thinks in terms of adversary calculus. Democracies must develop overwhelming force so that when dictators look their way, they do the math and know that they can't win. He worries about magazine depth — the US would run out of munitions somewhere between one day and two weeks in a fight with China — and about the manufacturing asymmetry that underlies it: China has 300 times more naval shipbuilding capacity than the United States.

Subterranean warfare fits into this framework as a capability that changes the adversary's math. If the crust is a domain you can operate in — if you can move effects through it undetected and unintercepted — then an adversary's cost-benefit analysis for any aggressive action shifts. They can't blockade what they can't find. They can't destroy what they can't reach. They can't exhaust a magazine that resupplies itself from five miles underground.

Luckey is careful about timeline. He places subterranean warfare past the 2030 window — one of the few military developments he considers beyond that near-term horizon. He doesn't think it will matter for the current China threat. But he does think it will define what comes after.

If you're gonna ask a question like, will it be the one that defines it? I think space is gonna be very important, but I think actually subterranean is the one that will define it when people in 2100 are looking back and saying, 'What was the thing that really changed everything?'

Palmer LuckeyInternet Marketing Association

The reasoning: space technology existed in the last century; other than a handful of Soviet and US prototypes, nobody was doing subterrains. Whatever emerges in the underground domain will be genuinely new in a way that better satellites and faster rockets are not.

The Institutional Barrier

If the physics work, and the engineering is progressing, and the prototypes exist, then the remaining question is: why isn't it happening faster?

I'm trying to get people to believe that the next war-fighting domain is the crust of the Earth... I think of the way the Army might take a role is developing a belief that this is actually real and this is actually possible, and then taking a lead in actually fielding this to solve certain problems that subterranean weapons could uniquely solve today.

Palmer LuckeyWest Point - The U.S. Military Academy

Today, the people who work on "SubT" in the Army are a small group whose job is to deal with bunkers and tunnels — an institutional frame that treats the underground as a problem to be managed, not a domain to be exploited. Luckey predicts that will change:

The same way you see a Space Force, I think it's very likely you'll see some kind of subterranean corps.

Palmer LuckeyMoonshots with Peter Diamandis

In the meantime, he talks about it. Loudly. On every podcast that will have him.

I've been talking louder and louder about this because when I talk about these things, it acts as a lighthouse that attracts all of the people. They say, 'Wait, what's that over there? There's a guy who's doing this?'

Palmer LuckeyAxios

The lighthouse strategy serves double duty. It recruits engineers who want to work on the problem. And it signals to adversaries that the capability exists — or will. Traditional deterrence depends on the adversary knowing what you can build. A domain where weapons are inherently invisible requires that someone stand in public and describe them.

It's moving along slower than I would like because it seems so crazy. It isn't.

Palmer LuckeyAxios

At the Internet Marketing Association, he admitted he needs a better marketing term than "subterranean warfare", noting that Cousteau was a great marketer when he coined "inner space" for the underwater world. The underground domain still lacks its Cousteau. What it has, for now, is a man who built working prototypes and will not stop talking about them.

Nobody agrees with me, and every time I talk about it, I sound insane to people.

Palmer LuckeyInternet Marketing Association

He has built the prototypes. He keeps saying it.