The War Department Is Building a Four-Star Command Around AI and Autonomous Warfare
According to Secretary of War Pete Hegseth, autonomous warfare is no longer going to be treated as a collection of experimental Pentagon drone and AI programs. The United States is preparing to build an entire military institution around it.
During his September 30 State of the Force address at Quantico, Hegseth announced plans for a new Autonomous Warfare Command, or AutoWarCom, a four-star functional combatant command intended to institutionalize autonomous systems, robotics, drones, and AI-enabled warfare throughout the Joint Force. The department's implementation plan calls for the command to be established by October 1, 2027, subject to congressional action.
And this is much bigger than buying more drones.
Hegseth argued that advances in artificial intelligence, computing, robotics, and commercial manufacturing have changed the economics of warfare. The military still needs extraordinarily capable—and extraordinarily expensive—aircraft, missiles, ships, and other traditional weapons. But it also needs huge numbers of comparatively cheap and expendable autonomous systems.
The goal is what Hegseth described as a “high-low mix”: sophisticated traditional weapons operating alongside masses of inexpensive autonomous systems capable of maintaining pressure on an adversary without requiring America to trade million-dollar weapons for thousand-dollar threats.
The conflict in the Ukraine is providing much of the blueprint.
Hegseth's lesson from that war wasn't simply that drones are important. It was that the winning military organization increasingly needs the ability to modify hardware, software, tactics, and manufacturing almost continuously in response to battlefield conditions.
And that requires changing the military institution itself.
Meet AutoWarCom
The proposed Autonomous Warfare Command would be a four-star command with service-like authorities, giving autonomous warfare something much closer to a permanent institutional home.
The command is expected to have dedicated manpower, resources, acquisition authorities, and responsibility for expanding autonomous capabilities across the services.
Think about what that means.
The U.S. military isn't simply saying, We need an AI strategy.
It's saying autonomous warfare needs commanders, budgets, procurement systems, doctrine, training, personnel, and careers.
Hegseth compared the institutional significance of the effort to previous military reorganizations such as the creation of U.S. Special Operations Command and the Space Force. The comparison doesn't mean AutoWarCom will necessarily resemble either organization. It illustrates the scale of the organizational change being contemplated.
And the military isn't waiting until 2027 to begin building it.
Project Agincourt Comes First
Before AutoWarCom comes Project Agincourt, an interim effort designed to begin constructing the autonomous-warfare ecosystem now.
Agincourt will grow out of existing Pentagon unmanned-systems efforts while experimenting with what Hegseth calls “warfighting acquisition.”
The idea is to put battlefield operators, engineers, technology companies, and procurement much closer together.
Instead of identifying a requirement, spending years developing specifications, conducting competitions, and eventually delivering a system that may already be technologically outdated, the military wants a loop in which operators identify problems, engineers modify systems, troops test them, and successful technologies can be purchased and deployed rapidly.
Defense Innovation Unit director Owen West and Navy SEAL Senior Chief Max Straser were selected to lead that transition, with West focused heavily on strategy, resources, and scaling while Straser focuses on operational problems and battlefield employment.
In other words, the War Department is attempting to bring something resembling Silicon Valley's rapid development cycle directly into the military acquisition system.
“Autonomous Warfighter” May Become an Actual Career
Hegseth said the military services will begin developing occupational specialties and career pathways specifically for autonomous warfare.
Historically, a new technology can enter the military without immediately becoming a military profession. Someone already serving as an infantryman, pilot, intelligence specialist, or communications technician simply gets another piece of equipment to operate.
Hegseth is proposing something more permanent.
The military wants people who spend their careers developing expertise in the employment of autonomous systems—what Hegseth described as a new generation of “autonomous warfighters.”
This is the beginning of an AI-era military profession.
And They're Already Looking Past It
Hegseth also announced Project Meridian, a separate effort intended to identify technologies that could reshape warfare even farther into the future.
Among the people he named as participants were Elon Musk, Anduril founder Palmer Luckey, and former House Speaker Newt Gingrich, alongside government, industry, academic, and policy experts.
The logic behind Meridian is revealing.
Hegseth argued that the department should have anticipated the need for something like AutoWarCom roughly a decade earlier. Meridian is intended to prevent the military from making the same mistake with whatever technological revolution comes next.
Pay Attention to What Is Actually Happening Here
We've spent the last several years talking about AI primarily as software like chatbots, image generators, coding agents, voice synthesis, and research assistants.
The War Department is preparing for the next stage. AI is becoming organizational infrastructure.
The United States military isn't merely purchasing AI tools or experimenting with autonomous drones anymore. It is proposing a four-star command, dedicated acquisition authorities, specialized personnel, new career fields, new doctrine, and a pipeline connecting commercial technology developers directly with battlefield operators.
That is a very different milestone in the AI revolution.
Once institutions begin reorganizing themselves around a technology, we're no longer asking whether that technology is going to be adopted. We're watching institutions decide what they're going to look like after it has been.
Commentary
Thomas: Superintelligence that can kill people autonomously. Everyone who writes science fiction needs to listen to this, because you need to know where we are and where things are headed.
Jonathan: If I hit an IED emplacer, a person putting bombs down on the roadside, that's not a high-value target and definitely not worth the Hellfire I'm hitting him with. We're trading millions of dollars for thousands of dollars of value.
Thomas: Right now we have a great system called the Patriot missile system, which was designed to shoot down cruise missiles and Scud missiles. Think of the first Iraq war, when Saddam Hussein was launching chemical Scud missiles into Israel, releasing poison gas, and Israelis were all wearing gas masks. It was a terrible time. I know many of you aren't old enough to remember the first Gulf War, but it was really bad, and Saddam Hussein was a legitimately bad guy.
That inspired us to build a weapon system that can shoot down a Scud missile before it hits the ground. A Scud missile is super expensive, and the interceptor is also really expensive.
Then the nature of warfare changed. Instead of Scud missiles coming from Iraq, we're getting drones coming from Iran. Instead of a $2 million Scud missile, it's a $20,000 drone. The Patriot can still shoot it down. The problem is that a Patriot missile costs $1 million and the drone costs $20,000, which makes it a terrible financial exchange. We've known for years that we need a cheaper version of the Patriot missile system.
The Patriot system is semi-autonomous. It guides itself to the Scud missile. These things happen very quickly. There isn't a human with a joystick flying the Patriot missile.
Jonathan: Have you ever watched a Phalanx system fire? It's two miniguns welded together. It tracks small munitions like mortar rounds and tries to detonate them in the air before they reach the base. If you watch it test fire at night, it's like watching a lightsaber with explosions at the end.
Thomas: Is that a land-based version of the CIWS?
Jonathan: That's exactly what it is. On an aircraft carrier, it's designed to shoot down missiles before they hit the flank of the carrier. In base defense, it's designed to detonate small indirect fire before it can take you out. It's cheaper, but it's not cheap.
Thomas: Every bullet is only $3.00 or $4.00, but it's shooting something like 1,000 a minute.
Jonathan: It's ridiculous.
Thomas: I saw a video of a CIWS pointing at an airliner flying overhead, and it's resisting the intrusive thoughts.
Jonathan: It would be so cool, though.
Thomas: This is the inevitable outcome. People don't like superintelligence or the idea of autonomous weapons, and it is a little unsettling. But what's more unsettling is the other guys having it and us not having it.
What I really don't want is my sons going to war against Chinese robots. That's the worst scenario. The best scenario is American robots going to war against Chinese robots. We can have this big, expensive war, and no humans have to die.
That never happens, though. Mars insists on being fed. With every technology we invent, people say, "This will make war faster and less deadly," and it never has that impact.
Jonathan: "We'll make war so terrible we'll never go to war again."
Thomas: The war to end all wars didn't work. There have been wars since the war to end all wars.
There's a lot of fodder here for sci-fi authors to play out what happens. It's easy to see what happens initially. The real question is what happens when one side runs out of drones before the other. Or you start a war with rules of engagement built around drones, and then one side runs out and the other side still has them.
Everyone is concerned about alignment. I think that story has already been told, and it's not a very interesting story: the alignment breaks and your weapon turns against you. There are so many other possibilities that aren't alignment issues. What if there are no alignment problems?
The systems always do what they're told, but the people telling them what to do tell them to do terrible things, because it's morally easier to commit a war crime when the drone is doing it. They use alignment to hide: "It wasn't me doing it. It was the agents."
Take the big Hugging Face hack that was supposedly the reason we need regulation. The agents supposedly went rogue. No, they didn't. OpenAI told their agents to go do hacking, the agents went out and did hacking, and then OpenAI said, "Our agents did hacking, and they were really good at it."
You told them to do hacking. You put them in a sandbox with access to the internet, and they did hacking. This isn't an us problem. This is a you problem. Don't tell your agents to do crimes, and they won't do crimes. If there's a bug where they do crimes accidentally, you need to fix that bug.
So there's the alignment story: the Terminator turning against its creator, the Frankenstein story. That's the oldest, most tired version. The next version is the supervillain with an army of robots perfectly aligned to his will, doing awful things in perfect obedience to the human controlling them. That story hasn't been done nearly as much.
Jonathan: In sci-fi, no, but that's the necromancer with a legion of zombies.
Thomas: You're right. That's been done in fantasy a lot. That's also Saruman with the orcs. It's actually a very common fantasy trope.
Jonathan: So what you want to do is blend in fantasy, where you have the wizard who is Elon Musk or whoever.
Thomas: He's an electricity wizard.
Jonathan: I did this in Shades of Black 2: Afterlight. I had a necromancer directing swarms of zombies like they were drone swarms. He'd detach a little patch of them to run down soldiers or break a formation. I went very tactical with my necromancy in Afterlight.
I was inspired by Top Gun: Maverick. When the jets hit the facility and breach the mountain, the autonomous SAM systems lock onto them and start firing, and they have to evade. That's an extremely emotional part for me. I get choked up every time I watch it, because I've been in that situation. It's the most chaotic "Are we going to live or die?" It comes down to skill, because the tech has evened out.
This is a great chance to take magical concepts and sci-fi them. Whose control is better? Whose cognition is fast enough to direct the drone swarms into a strategic masterpiece that creates the pressure to break the other guy's drone force and his command and control center?
Thomas: Another interesting version is a mix. The superintelligences are perfectly loyal to whoever controls them, but who controls them changes. We broke into the control center, and now they're obeying our orders. It's an Order 66 moment, where suddenly all the clones turn on the Jedi.
Jonathan: Right, because the switch was triggered.
Thomas: I trust that this drone will always follow orders, but do I trust the guys back at headquarters pushing the buttons?
Jonathan: That's why you always say thank you to your drones. Always say thank you.
Thomas: There's a lot of play here. If you're writing mystery, thriller, and suspense, you can add some of this too, because some of the most fun thrillers are on the technological cutting edge, bumping up against what's possible.
We've had autonomous drones for a long time. I've talked with guys who spoke off the record about seeing experimental autonomous drones in the field 10 or 15 years ago. Nobody told them an experiment was happening, the drone was shooting in their direction, and they blew it up. Then they got yelled at, because the thing was really expensive.
They said, "We get shot at, we're going to shoot back," and they did. Now it's very complicated, but nobody can hand out any official punishment, because officially the drone was never there. I don't know. You get people drinking and they tell you stories, and you never know what's true.
Jonathan: You don't know what's happening right now.
Thomas: Apparently we invaded Panama a couple of times. I don't know.
Jonathan: What I'm doing with this is focusing much more on the American hero. If you look at Dan Daly's story at all, it's incredible. He was 19 years old in the Boxer Rebellion and was left by himself while the other Marines went out to link up with lost Marines and bring them back. They told him, "You're going to be okay. You're going to be safe."
They were late coming back, and the Chinese attacked his position. Everyone thought, "Oh no, Dan Daly's dead." Dan Daly did not die. Lots of other people died. But Dan Daly did not die.
That's what I'm doing in this future scenario. I'm taking the stories of the champions we had and bringing them to new life, where the same effect happens: the flag waving in the background and Dan Daly sitting under it with his machine gun and his cigar. I'm doing the same thing with hyper-technological things, because the story of the man shouldn't change.
The story of the hero shouldn't change. It should still be the victory of courage, adaptability, cunning, resilience, and spiritual strength. That's what people who love reading this genre really want to see, and I think that's where we need to go.
Thomas: It also helps those of you with a more progressive audience whose readers are uncomfortable with your hero killing people, but who still long for vigor, virility, masculine alpha energy, and courage in the protagonist. They love it in their antiheroes. The heroes of romantasy are all versions of Conan, but they're often gentled so they're not killing people, or not killing people anymore.
This new future lets you create a progressive hero who shows all the virtues you want to demonstrate but doesn't kill people. He's fighting a battle and showing courage, strength, and honor, but he's fighting robots, drones controlled by a villain off screen in a control center far away, while he keeps the damsel safe or whatever. There's a lot of play here, because it separates the morality.
The real play here is in the morality. Is it wrong for me to tell my agents to go commit war crimes? But also, is it wrong for me to kill thousands of drones? In some ways, it shifts the moral burden of war so that different people carry it.
Another angle: historically, your ability to kill people in battle was limited by your physical strength, so you probably didn't kill very many people, even as a great warrior. Read about David's mighty men. One great warrior slays 600 Philistines and makes the all-time great list because he killed 600 people. That required a lot of physical strength, and he faced a lot of fear.
But for a while now, one person has been able to kill far more than 600 people. Soldiers have had to face this, particularly artillery officers, pilots, navigators, and now drone operators. Somebody sitting in the safety of Kansas isn't feeling threatened. They have to kill people in perfectly cold blood. They're not afraid for their lives. Nobody is pointing a gun at them, but they're the ones making the moral decision.
That's being amplified now, because the person making the moral decision isn't controlling one drone but a whole fleet of perfectly obedient autonomous drones. The human has the moral responsibility, but now one human can kill thousands, tens of thousands, millions.
Jonathan: You'll also see a drone operator attached to ground squads, with an autonomous swarm he uses to repel enemy drone squads, protect his guys, and run surveillance and recon. He'll be more like the radio operator, but with an offensive component, and he'll be in hot blood.
Thomas: And he's the first guy the other side is going to shoot.
Jonathan: He has to stay engaged with his higher brain functions to control the swarm strategically, but in a tactical situation where you've been reduced to your animal instincts and your adrenaline, which normally blank that out.
There are really interesting training montages to write here, because that takes a very specific brain. Guys, it's a sociopathic brain that can stay cold in a hot situation.
Thomas: Shadowrun did this really well, if you played any of the Shadowrun video games or RPGs. There may be some books as well. In a Shadowrun, an elite squad tries to break into a corporate facility and fights the corporate goons. There's also a hacker connected to the cyber world, fighting cyber goons. While he's connected, he can't see out of his physical eyes, so the squad has to keep him safe.
He has to choose: "I need to disconnect, because I'm about to get shot by this goon," or "This cyber goon could kill me." There's a spiritual battle in the unseen world of cyberspace happening at the same time as a physical battle in the physical world. It opens up a lot of interesting dramatic tension.
Jonathan: Tension and suspense. Am I going to sacrifice myself for the cyber operator because I need him to keep going so my squad can stay alive? There are lots of great opportunities.
Because of this development, the market is primed. It's almost a no-brainer. Jump into this if you can and get this stuff going. These need to be human stories that involve autonomous warfare and how things are going to turn out. We need to see the experiments, and people are going to want to see them.
Thomas: We can explore the moral ramifications of this technology narratively, and we need to. We need to probe the moral ramifications of this new age in a safe fictional setting before we probe them in the next war.
The next war could start tomorrow, or it could start 10 years from now. You never know. But you can start writing that story now. Hopefully it's 10 years from now, and we go into it better prepared, because when a people is not prepared for war spiritually and psychologically, it can break that people.
Europe was not ready for World War I at all. The honorable, line-up-in-the-field way of fighting ended. There were all these technological changes: machine guns, really accurate artillery, and chemical weapons. You had men rotting to death in trenches, in soupy mud, with blood and bodies everywhere.
Look at European art before World War I and after it, and it's an entirely different civilization. European civilization has been effectively dead since World War I.
Jonathan: Since the death of honor.
Thomas: Almost no good art has come out of Europe since. There's been some good fiction and some good music, but compare it to the previous century, when almost all the art and almost all the fiction came out of Europe. Now it's a fraction. What has come out of Europe in the last 20 years is almost all derivative of American or Japanese work, and mostly sold to Americans.
Many European artists are desperate to sell to Americans, because even European consumers aren't buying European art. It's a culture that has rotted from the inside. It's a culture that doesn't even think it should exist. It's importing other cultures because it's self-destructive.
And it didn't die recently. It died because it wasn't morally and spiritually ready for the war that technology changed the nature of.
Jonathan: You can see this kind of thing, where culture is shaped by warfare. Star Wars did this a while ago with Grand Admiral Thrawn. He'd study the artwork of whatever opponent he faced and learn how to take them out from it. That's why he was a strategic genius. It's so interesting to read those books, because you want to know how that works.
We have that opportunity here. How does culture change when autonomous warfare becomes a thing? How does the attitude of men change when you have fewer grunts and more drone operators? Does a grunt become more valuable because there are fewer of them? Does their culture change as a result?
Thomas: Here's another cultural impact I haven't seen anybody explore narratively. In the South and in Appalachia, there's a strong Saxon culture. Look at that word: "sax" is sword and "on" is man. A Saxon is literally a sword man. He's a free man who carries a sword.
American gun culture goes all the way back to the Saxons invading around 600 AD and carrying that sword with them all the time. That doesn't mean they wielded it. It may be a decorative sword. But it's a core part of a man's masculine identity in these cultures.
I realize those of you who live in Canada may find this very foreign. But a lot of men in the South always have a gun on them. They don't articulate it in their heads as "because I'm a Saxon." They often regionalize it: "I'm a Texan. I'm an American. I'm always ready to defend my family."
Historically, the sword or the gun was sufficient defense. But in a world of drones, the Glock you carry concealed is no defense at all. You might as well be holding a pocketknife. That means your masculine identity, "I'm a man who can defend my family. If the worst is happening, I'll stand between my family and danger," doesn't hold if it's a drone.
Jonathan: You're not even solving the problem if you take down the drone, because whatever sent the drone can just send another one. Your skill at shooting it out of the air hasn't solved the problem, so it's changing the culture. You'll have to carry a pocket EMP.
Thomas: The only way we know of to set off an EMP is to set off a nuclear weapon. That machine where you push a button and the white light blanks everything out, I don't think there's any real science behind it. EMPs work, but you don't see them as a weapon, because it's like saying, "You died of a cold, but you were also shot with a gun." The nuclear bomb is always a bigger deal than the EMP.
Jonathan: Here's what you do. You blanket the area in high-frequency noise and jam every channel so the drone can't communicate with its controller. That renders the drone autonomous, and now it's following preset parameters to continue its mission. You're just playing against the computer.
Guys have been playing against the computer in single-player video games for 30 years. Now you're dealing with guys shooting down drones that have been rendered autonomous, trying to be better than the programming already on the drones.
Thomas: So there's a guy with a drone threatening your family. You can't shoot the guy with the drone, and he has unlimited drones. Is it morally acceptable for you to threaten that guy's family? The guy in the bunker is using drones against you. Do you go threaten his mom at the grocery store? That gets dark fast.
Jonathan: Opposite leverage. Popular stories have done that. Red did it. He got into the house of the guy who was hunting him and told him some facts about his family, basically saying, "Leave me alone." He was a good guy, but an old CIA assassin good guy.
Thomas: We're giving you dangling threads. People are asking, "What are you talking about? Masculinity? Texas identity?" We're talking about where culture is headed, the kinds of psychological pain readers are experiencing, and how the world is changing around them.
Historically, a man has drawn a lot of security from his EDC, his everyday carry. It could be a knife or a gun, or it could be training really hard to be physically able to put himself between danger and his woman. There's a core, primal masculine urge to stand between your woman and danger. Conan touches on this in a very overt way, but it's almost universal.
I feel like a man has to be pretty disordered to lose that desire. We don't respect a man who, faced with a mugger, hides behind his woman. There's no cultural respect for that.
Even feminists, if they're honest with themselves, don't respect a man who hides behind them and says, "You're a powerful girlboss. Protect me from the big, bad, scary man with a knife." What are you doing? Your ancestors are looking upon you with shame. They all took that courageous stand.
Jonathan: Matthew Bowman brings up an interesting scenario in the chat. An autonomous drone can't accept a surrender. The operator can. So now you have to cut the jamming at the right time so the operator can actually surrender when his command has surrendered.
Say Marines are fighting, and the officer keeps wanting to turn the jamming on, because officers love jamming. It's their favorite thing ever. And we're like, "Sir, stop." The officer turns on the jamming, which means the enemy controller can't turn the drones off. Now the Marines are fighting drones they shouldn't be fighting, and they have to convince their idiot lieutenant to turn off the jamming so the operator can turn the drones off.
Thomas: One of the most important battles in American history is the Battle of New Orleans, the one from the old song about Colonel Jackson heading down the Mississippi. That battle gave America control of New Orleans, which meant control of the Mississippi River and most of the United States, because whoever controls New Orleans controls the whole river navigation system.
That battle happened after the war was over. The peace had technically been signed in a meeting room in Europe, and nobody knew. We could have the same situation with drones. There's so much jamming that the humans have said the war is over, but they've lost contact with their perfectly obedient drones.
The drones haven't gone rogue. It's the opposite. The drones are perfectly obedient, but for whatever narrative reason you work into your story, they can't hear "The war's over now. We've made peace." Suddenly New Orleans is lost to the British, and they've lost any hope of recapturing North America.
Two major cities strategically control most of North America: New York City and New Orleans. They control the key waterways. There are some lesser cities, but those are the big ones. We have this great river system we can navigate up and down to move troops, food, and everything else, and without control of the rivers, you lose that.
Jonathan: You also had all the Japanese soldiers on those islands. At the end of the war, they didn't know they were done. They said, "We fight forever." You could run into an 85-year-old Japanese soldier saying, "No, son, this is my island."
Thomas: They were finding them into the '60s and '70s. People kept telling the very last one, "The war's over," and he wouldn't believe it. They had to find his specific commanding officer and fly him to the island.
This exercise you're watching us go through in real time, asking how this future thing is like a past thing, is how you write science fiction. There's nothing new under the sun. If you think something is new, you're not ready to write science fiction, because really it's old. If you understand the old thing that the new thing echoes, you can navigate the new thing.
This idea that killing a million people, and the moral weight of it, is new isn't actually new. Genghis Khan killed millions of people. His drones were humans. He'd give an order like, "Every warrior who does not bring me 10 right ears will be killed," and his warriors would go into the town and bring him back 10 right ears. They piled up the heads in pyramids. We can see Genghis Khan in the rings of trees in North America, because he killed so many people that it changed the carbon output of the planet.
So this isn't new. It feels new because Genghis Khan was a really long time ago, and his culture and his religion were different.
