外星人、AI武器、中国与全球冲突：Palmer Luckey 拉响警报｜EP #169

Over the last 2 years, there have been an unprecedented number of congressional hearings about NHIs—non-human intelligence—and alien craft. What are your thoughts?

I’ve probably seen things that I can’t talk about. It’s hard for me to have an open discussion about it.

I started Anduril because I wanted to work in the national security space for a variety of reasons. GPT o3 just came in at a 135 IQ. I’m not betting my company on superintelligence, but I do believe it will happen.

It’s not the US versus China. It’s the US and China versus the rogue actor. China is not going to purposely build a tailored bioweapon that wipes out all the Jews, for example. But at the same time, China has made real material threats and said it is going to reunify with Taiwan by force if necessary within this generation.

Now, that’s a moonshot, ladies and gentlemen.

Palmer, you’ve been on Moonshots about 4 times in the last 2 years. As a friend, there are a bunch of questions I’d love to ask you that I truly, deeply want to know the answer to.

Let’s do it.

All right. Here’s the first one.

Over the last 2 years, there have been an unprecedented number of congressional hearings about NHIs—non-human intelligence—and alien craft, by some of the highest-level generals, admirals, and Air Force officials. It’s crazy. What are your thoughts?

I want to believe.

You want to believe?

This is a tough topic for me because I’ve probably seen things that I can’t talk about. If that was the case, it’s hard for me to have an open discussion about it.

I think I can plainly say I have not seen evidence of anything conclusive or obvious. I have not seen recovered craft. I have not seen the programs that are analyzing alien wreckage, but I have seen things that are not necessarily public and are very hard to explain.

What makes it difficult in the public eye is that there are dozens of examples of very strange things that can be explained in one way or another. There’s really only a small handful that, even when you dig deep, have no explanation for the combination of human eyes, sensor data, behavior, activity, and timing.

The brain makes a lot of things up. The brain is very, very trickable. It’s not just people; different animals see the world in different ways. Our perception of reality is constantly lagging behind what we perceive. What feels instantaneous to you is actually up to a second in the past.

It’s amazing. When you clap, it feels instantaneous. In reality, your brain is basically filtering out anything it perceives before you should be perceiving it because it knows that it’s not real.

I once wrote a science-fiction speculative short story about an alien species that’s like that, but taken to the extreme. What if your perception was actually hours behind what you observed?

If you had something that reacts to things well enough, sees a snapshot, and reacts to it by predicting what’s going to happen next, how long would it take a person to figure out that the person on the other side—or an alien spacecraft, for example—is operating under completely different principles of consciousness, awareness, perception, and reality?

It’s like having a conversation with someone on the Moon and having a 2.5-second time delay. But what if the alien was so smart that it could predict what you would say in response to it, and then vice versa, and then back again, such that it seemed instantaneous?

What would it look like for a being to exist whose consciousness doesn’t lag? We have a lag of about a second. What if the lag were 100 times greater? Is that really conceptually impossible to imagine? I don’t think so. It just makes for a boring conversation, though.

Well, maybe so. In the story that I wrote, people don’t figure this out until bad things start happening. They don’t understand that the perception of these alien beings is much slower, but the beings are so smart at reasoning forward that a person can’t tell they’re not responding to what you’re doing. They’re responding to what they predicted you were going to do 5 steps ahead.

Once you understand that gimmick, you can do things that take advantage of it—things that are completely unpredictable, outside of what they would expect. When you live outside of social norms, you can do things that are very unpredictable.

From time to time, I write these short stories just to entertain myself. I’ve never published any of them. But getting back to the topic, I’ve not seen anything. There’s some weird stuff. There are at least a handful of examples that are very difficult to explain.

I think we talked about this at Abundance a few years ago, but I suspect that in the end, it’s going to turn out to be something different from what we all expect. It’s probably less likely that they’re aliens from a nearby planet. I suspect it’ll be something like a natural phenomenon we have not yet begun to understand.

But you hope it would be super cool.

I do think it is going to be something beyond our current understanding. I think it is more likely, for example—not saying that this is what it is—that some of these craft are somehow traveling through time than coming from a nearby galaxy, if you look at what’s more possible.

By the way, I don’t mean traveling through time necessarily backward. People say, “Palmer, how could they go forward?” Perhaps they’re coming from the distant past. There are a lot of ways to look at this. I don’t know. I wonder.

Wait, I’ve got to ask: What do you think it is?

I believe life is ubiquitous in the universe. I truly believe it. Probabilistically, it seems likely, and in one sense, almost thermodynamically, I think life is the end result of a series of processes.

Where do you fall on Dark Forest theory?

The proud nail gets hammered. Life doesn’t make it. I don’t have an opinion. I haven’t figured out my opinion. I’m agnostic on Dark Forest theory.

The question I have is, if it proves out to be true, what would the public reaction be? What I find fascinating isn’t the UFO sightings from the 1940s to the 1980s or the blurry photographs. It’s all of the testimonies that have been given over and over again before Congress by seemingly extraordinarily credentialed individuals who have a lot to lose and very little to gain.

Yep.

I’m curious about decorated war veterans and politicians whose careers are everything and whose credibility is all they have.

Yeah, it’s extraordinary.

And so the question is, what would be the public reaction if it actually played out to be true?

You’re in the warfare business, and I’ve always thought the only thing that could bring some sort of unified peace to the planet—besides a massively dominant force—is an external threat. An external threat is like an asteroid coming toward us, with 10 years to organize a response to—a planet-killer that we have 10 years to respond to—or an alien saying, “We’re going to come and eat you.”

It brings all of our differences to the surface and vaporizes them in the process.

Yeah, I do think that is probably the case. Historically, it seems to bear out, too. It’s not just that we’re unified with people we don’t care about. Even bitter enemies or ideological enemies can be unified by a common threat.

Look at the Japanese and the Germans during World War II. Culturally, they couldn’t have been more different. Ideologically, the Japanese were subhuman to the Germans. The hilarious thing is that the Japanese believed exactly the same thing about the Aryans. Yet they allied and worked together, smuggling controlled materials and chemicals back and forth—things that they uniquely had—because they had a common enemy.

I think that if they were an enemy, we would unite. Assuming that they weren’t necessarily an enemy, though, maybe I’m crazy, but I often feel like people wouldn’t respond the way you expect.

I almost feel like culturally, we’re so inoculated. You believe that life is proliferated throughout the universe. I think the average person does, too. If we found out that, in Alpha Centauri, there were some guys living over there who were kind of like us, I think a lot of people would say, “Wow, that’s really interesting, really fascinating.”

I’m not sure it would even be the top-trending topic on Twitter by day 3.

I think you’d get a day or 2, and then it would retreat.

They’d get back to the Housewives of Hollywood, right?

I mean, people are focused on the things that are in front of them. Can I get food on the table? What’s the price of gas? What’s the price of eggs? Raising kids?

I think the existence of aliens is probably going to be as important as the context of those aliens. Are they coming to burn us all down? Okay, then that’s going to threaten my way of life. Are they just out there in the world? I think it would prompt a lot of navel-gazing from the media class and the academic class.

I mean, forget about selecting a new pope.

I want to see what the Catholic Church does if intelligent life is proven to exist elsewhere. I suspect they would probably be one of the faster-moving entities to say what the Catholic Church has been pretty clear about. I’m not Catholic by any means, but I do appreciate that, at least for the last couple of centuries, they’ve said, “Look, anything new that comes to light that violates our previous understanding is proof of God’s plan being further revealed to us and needs to be incorporated into the doctrine.” I would love to see how they would deal with that. That would be very interesting.

Yeah, for sure. A lot more unprecedented than a pope. There have been a lot of popes, but not a lot of doctrinal changes on the order of a new species of sentient beings. And, of course, the current folklore—and it’s all that can be said—is: Has all our microelectronics technology emanated from UFOs? The question is, if the UFOs are real, are China, Russia, India, and the US all vying for advanced technology there?

I will take a stand there. I don’t think our current microelectronics technology came from alien wreckage. I think that’s one area where we deserve the credit. We made it happen. We figured it out. Gordon Moore worked hard for his I think we really did make that happen from scratch. Semiconductors and microprocessors—I think we can take credit for that.

If there is technology that’s been derived from alien wreckage, I suspect it’s more likely to be related to fission or fusion, or advanced metallurgical or ceramic compounds. Though gravity shielding would be awfully convenient for your vehicles.

Well, yes, that is true. But, on the other hand, I have not seen any antigravitic drives, and I’ve kept my eye out. I think it’s one of those things where people say, “Well, we’re holding it in reserve for the right moment.” Looking at how the government operates, it’s just hard to believe that they’re capable of having made it through the last half-century of conflict without ever feeling like that was the right moment. Maybe I’m wrong, but I also think people understand that you go to war with the tools you have, not the tools you want.

If you haven’t started a program to implement gravitic drives in an aircraft, look at how long the F-35 has taken to get across the line. The F-35 was conceived during the Cold War. People think of it as a much more modern thing because of how long it was delayed. Remember that the Cold War ended on December 25, 1992. The F-35 program had already started. If you were going to get a gravitic drive into a bunch of fighters, apparently it takes 30 years to make it happen. I don’t really buy into this idea that there’s a secret vault of technology that’s going to be busted out the moment the threat level gets high, because reality suggests it’ll take us decades to make use of it.

Yeah. Well, at least for the traditional defense contractor here at Anduril.

Yeah, well, we’re doing things differently. I mean, the government has been betting that companies like Anduril would exist. I think that was a crazy bet 8 years ago. It’s maybe a crazy bet today.

So there’s a different alien race that’s landed on the planet and is emerging right now, and that’s the whole AI world.

Sure.

Let’s jump into that. I saw Eric Schmidt recently saying that AI is being underhyped, and that if you truly understood the power we have today and what’s going to emerge from the recursive self-programming of AI models, we’re in the midst of this intelligence explosion and it’s about to get really crazy, really fast.

Obviously, Lattice and everything you’ve built have been a beautiful platform for AI. How much are you thinking about digital superintelligence? Just to define that, I mean orders of magnitude more intelligent than human systems.

Anduril was an AI company back when it wasn’t cool to be an AI company. The name of the company is Anduril Industries; the acronym is literally AI. But back in 2017, AI was kind of like how VR used to be: “Oh, it’s always in the future, never in the present.” It was the thing for wacky, crazy people to waste their lives on, not something for a serious doer to build a company on top of.

I knew that AI was imminent because the smartest people I knew were telling me that and illustrating it in ways that were very believable, showing how a bunch of schemes that had been improbable for decades were clearly scalable. One of those people was John Carmack.

I love John. John is brilliant. He’s one of the smartest people in the world, definitely the smartest I know.

Yeah. I remember John. Years and years ago—

For the listeners, just so they know who John is, he basically invented 3D gaming. He also started a rocket company and later became the CTO of Oculus. He created Doom and Quake and basically the modern 3D game engine, and he deeply understands hardware and software.

Yeah, he’s amazing.

He had one of our teams in the original Ansari XPRIZE, Armadillo Aerospace. They were doing vertical-takeoff-and-landing rockets way before SpaceX, like a decade before SpaceX. We had these lunar lander challenges where you had to launch, hover, translate 100 meters, do a soft landing, and come back.

I remember back then he had this thing on the side of his rocket. I could never figure out how to pronounce it: NVIDIA. NVIDIA was a sponsor back in 2005 or 2006. Had I only known.

Well, I mean, NVIDIA—there was a time when Oculus was acquired by Facebook, and there was a point where we were considering acquiring NVIDIA. I know that sounds crazy, but remember that when you go back to that point in time, NVIDIA was worth about $4 billion. It’s not that crazy. And you don’t have to buy the whole company. They were publicly traded, so you just need to take a dominant position. You don’t necessarily have to buy out every share.

You’re looking at a low single-digit-million-dollar investment to have control of it. People have often looked back and said, “Oh my God, imagine if we would have done that. Imagine what a big deal that would have been.” My point to them is, if we had bought NVIDIA, they never would have turned into what they are today. They wouldn’t have focused more on AR and VR processing. They wouldn’t have focused on cluster computing or crypto, and they wouldn’t have gotten extremely lucky in the fact that their GPU architecture happened to be exactly what you need to scale large language models.

Extraordinary. We’ve talked about this with your kids playing video games at home, but in terms of superintelligence—

John ended up leaving Oculus long after I was fired, actually, because he wanted to work on AGI. He told me, and has now said publicly, that even though he thought it was a relatively low probability that he would crack the nut, the impact on humanity would be so fundamental that even a 1% chance of succeeding made it, from a risk-cost-benefit-weighted analysis, obviously the right thing to spend his time doing.

That was one of the reasons I had such confidence starting Anduril and saying, “I’m going to build a company where the whole premise is: Take as a given that AI is finally going to work. Take as a given that autonomy is here. What would that mean for the military?” We went about building all the things that assumed it was true. That was really our early advantage.

Other companies were not running their programs and their research-and-development programs as if AI were a real thing. It’s not that the people in the Air Force were dumb or the people in the Navy were stupid. It’s that they were making decisions assuming that AI wasn’t going to be real. Once it becomes real, that changes everything.

And there’s a vast difference between an AI-native startup and an old-school company trying to retrofit it, trying to ram it in.

Yeah, right. There’s also a difference between a company that’s premised on AI versus one that’s just helped by it, and a founder-led, AI-premise company. I mean, the advantages that you have—and Zuckerberg has, and Elon has—as visionary leaders able to say, “No, no, no. I know this is the way we used to do it. We’re changing it. We’re going this way now because it’s the right way to go.”

Yeah, it’s impossible for a large-scale public company in the defense industry to make any kinds of shifts like that.

Well, I mean, Zuck had an AI research lab of significant size back when all this stuff was considered crazy.

So when they opened up their AI research lab, where they were doing integrated AI and robotics, we're talking about 2014 or 2015. They were doing this, and a lot of people, including people in the public markets, saw it as folly. They saw Mark working on this ridiculous thing, burning money—totally a waste of time.

What it really is is what you're saying: these founder-led companies can make bets that a hired executive would never make. You would never have a hired CEO from the outside, who's also thinking about what his next job is going to be, say, “You know what? I think I'm going to burn billions of dollars on this technology that everyone thinks is a total waste of time, and I'm going to be punished for it quarter after quarter after quarter.” Eventually, someday, it's going to come to bear, and everyone's getting rewarded, because none of those guys are usually even around long enough to see the fruits of that labor.

Even if they were, there are probably safer bets they could make. Founder-led companies can afford to do that. They can afford to say, “You know what? I'm going to do it anyway because this is my company, and I care more about it than anybody, and I'm going to do the right thing for it in the long run.” It's a powerful thing.

It is powerful, especially when the founder really is so technically literate and has his teams revere him. I'm not saying you ever said that about yourself, but your teams do, as they do for Zuck, as they do for Elon and others.

I think that's true because the reality is, I'm not sure “revere” is quite the right word, but if they didn't believe in the company, they probably wouldn't have joined. If they didn't like what they see, they probably won't stay. And so you end up building—if you're a founder-led company, you're going to attract people who, to a certain extent, reflect the vision of the founder, because you want to attract people who believe in that vision and then, equally important, repel people who do not like it.

Yes, equally important. You saw that ad campaign we did: “Don't Work at Anduril.” The whole point was, “Hey, here, we work hard. This is a real job. You're going to be in the shit. This is 8-hour workweeks, guys.”

We had a lot of people coming in and saying, from the outside, “Oh, this is a bizarre campaign. This seems like—why would you do a campaign about how hard it is to work at Anduril?” The point is, guys, this is going to attract exactly the right type of person and, most importantly, repel anyone who wouldn't enjoy working here once they get here. By the way, our applications went up 3 times the week after that campaign, and they're all exactly the type of people.

I believe that's beautiful. One of the challenges a company has as it matures is deciding whether to go public or not. I don't want to get into whether Anduril is going public or not, but eventually, my guess is yes.

We actually are. We are going public. We have to at some point. You can't win an F-35-scale program without doing so.

So you've got Elon saying, “I will never take SpaceX public. I don't want to have shareholders telling me whether I can spend money to go to Mars.” And I don't want to disclose all of my secrets in a 10-Q and so forth.

Then you've got folks like Bezos. I've known Jeff for 40-plus—almost 45 years. Jeff famously says, “Don't invest in Amazon if you're looking for me to maximize near-term returns or shareholder returns.” It's like, “I'm going to build.”

How do you balance the benefits of going public, so you can pursue these large contracts, with maintaining the agility that has allowed you to survive and thrive?

I've never run a publicly traded company, so take this with a grain of salt. It's as valuable as what you're paying for my advice, which is nothing.

Now you have a CEO.

We do. We do. Our CEO is Brian Schimpf.

He's an amazing individual.

He is, and I think we're totally agreed. Everyone's in alignment on this. When we become a public company, we have to keep doing what we've done in the private markets. It's really no different from hiring. We need to attract people who believe in our vision and repel people who don't believe in our vision.

A lot of people imagine that if Anduril goes public, we'll become like other public defense companies because we'd take less risk, be unwilling to invest in the future, pay out dividends rather than investing in R&D, and maximize quarterly returns. But that doesn't necessarily have to be the case.

Even when you transition to the public markets, you can take action through your communications, your filings, and your decisions that scare away people who want you to be like a traditional company. You want to attract investors who believe in your vision of the world, and if that's your whole investor base, they're not going to force you to be something different.

I think Elon's other company, Tesla, is probably the strongest example here. Tesla has an extremely high price-to-earnings ratio. Why? Because their investors believe that they are going to win across the board on a multidecade time scale. They think they're going to win at robotics, they're going to win at energy, and they're going to win at—and I think they have a very good shot at winning on all or most of those items.

You could ask yourself, “Wait, what? They're a publicly traded company. Why aren't they like one of the more traditional automotive companies? Why weren't they forced to be more like a traditional company?” The answer is simple: they've cultivated an investor base that believes in what Tesla is, and they've repelled everybody else. Many of the people they've repelled are now shorting Tesla because they don't believe in it.

I think it's the same way for us. We're going to need to attract people who believe in what we are and repel everybody else. If we ever start getting enough of an investor base that is pressuring us to do the wrong thing, I'm going to need to go in the press and say some crazy shit to scare them all away, because I don't want those guys voting at my quarterlies. I don't want them picking my board members. I want all the Anduril haters and people who want Anduril to be what defense companies used to be running for the hills.

Nice. Nice. All right, let's get back to AI. You are an AI-first company. I mean, software makes up the majority of your workforce and the majority of the products you're developing.

Digital superintelligence—let's define that. Very famously, we saw AI reaching IQs just above human levels of 101. This was Claude 3, and then GPT-o3 just came in at an IQ of 135. The prediction is that we'll reach intelligence as great as all humans combined by 2029 or 2030. Elon's prediction, when he was on our stage at the Abundance Summit a couple of years ago, was that AI would be as smart as all humans combined by 2029 or 2030. So that's a vastly accelerated curve.

Are you skating to where that curve is going to be? How are you thinking about it?

Am I skating to where that curve is going to be? I'm actually probably running my company a lot more pessimistically than that. To bet that those most optimistic predictions will come true is probably not a responsible way for me to run my company.

But a billionfold, an 8-billionfold—let's just say a millionfold smarter than a human—I think we're operating under the assumption that will happen. You could pick almost any point in even the last, let's say, 2 years, and people say, “Well, AI might be getting smarter, but it'll never be able to do this.” Then within weeks or maybe months, it's doing it.

They say, “Well, but this video has this problem. You see, the man has 6 fingers, and so that proves that it will never be able to replace a real illustrator.” Then, of course, you wait a couple of weeks, and all of a sudden, that's no longer the case. I'm not going to be one of those people who bets that we're not going to get there.

At the same time, I would say Anduril's thesis makes sense even if AI doesn't get smarter than a person. I need it to have fast reaction times. I need it to be able to think much faster. For example, processing what would have taken a person a month to process: if I can do as good as a person would have done in a month, but do it in a minute, that's a superpower in and of itself for military operations.

I don't actually need things to be superintelligent. I just need them to be better than people at speed and latency. If I've got a truck and I need a truck to drive itself around, I don't need it to have a 135 IQ. A 100 IQ is more than enough to drive a truck sufficiently well, especially if it's a 100 IQ that's not distracted, not sleep-deprived, and never going to be abusing substances like that.

That's actually pretty great, especially when you can duplicate it 100,000 times for free, unlike a trained truck driver. I'm not betting my company on superintelligence, but I do believe it will happen.

I have to imagine that superintelligence in the warfare game is not just a small advantage; it is often a huge one. I often tell people, “What would you rather do? Would you have an airplane that is twice as fast, or an airplane that makes decisions that are twice as good?” Or, put another way, would you rather have an airplane that can carry twice as many weapons, or be twice as smart about which targets to use them on and when?

Would you rather be able to predict the next 5 minutes of combat better, or would you rather have sensors to see more? Would you rather predict the battlefield, or actually sense it? In most cases, it's the software advantage that you'd rather have, which, by the way, scales much faster than the hardware advantage ever will. Every copy of software you duplicate is free once you've invented it, and it can often be applied to many hardware systems.

If I make one aircraft better by investing in that one airframe design, it’s not nearly as useful an advantage as a piece of software I can deploy to 10 different kinds of aircraft. That’s really the core of Anduril. Our core product is Lattice, the AI engine that powers everything we do.

The reason we’ve been able to pivot into so many different industries is because we invest so much in that AI platform that runs all of our products.

I do have to tell you, actually, I haven’t thought about this in a while. Have we ever talked about the philosophical origins of Lattice?

I don’t know if we have.

No, I’m going to guess Skynet.

Skynet is the fictional example everyone thinks of, but there was a French mathematician and philosopher, Pierre-Simon Laplace. He was best known for the Laplace transform, but he had a thought experiment known as Laplace’s demon.

It was a thought experiment around whether free will exists or not, and this was before anyone was talking about simulation theory. We didn’t even have computers. He posited, “Think about whether free will exists. Suppose that there was a supernatural being, this demon, that was so perceptive of the world that it could perceive every particle of matter in the entire world, and the energy and motion contained therein, throughout the whole universe.”

He could perceive it all at once. Also suppose that this being were so intelligent that he could, in an instant, reason about the reactions that would occur as those particles collided, as physics and chemistry occurred, and reason all the way until the end of time. Suppose that such a being could derive an equation that described the actions of every person in the universe until the end of time.

His question was this: If that being can even exist theoretically, doesn’t that mean free will isn’t real? Isn’t everything deterministic, just physics playing itself out?

The question was also whether there are things that change this idea. Are there nondeterministic elements in our universe? Are there supernatural effects? Are there spiritual effects? Are there things we cannot observe, that nothing can observe? Could it be that the act of observing something, in fact, changes the outcome, as in quantum mechanics?

This was long before quantum theory was in play, but he was asking these questions. Could it be that such a being is not possible? I think he posited that if such a being is even theoretically possible, free will definitionally does not exist. If such a being is impossible, then at least free will is a possibility.

Most people get deeply into the philosophical side of this question. When I became familiar with Laplace’s demon as a thought experiment, my first thought was, “Who’s going to build Laplace’s demon?” What would it look like to build something that is as close as you could get? It perceives as much of the world as you can.

Omniscience.

Omniscience. And not just the present. What if you could gather enough information to be smart enough to reason about where it’s going to lead? It’s the same as seeing the future or even traveling into the future.

What if I could predict what my enemy was going to do 10 seconds into the future with a high degree of certainty? What if I could predict what he’s going to do for the next week with reasonable certainty? It won’t be right every time, but you spread it across enough bets. You take 10,000 guesses and 9,000 of them are right. That’s a superpower. That would seem superhuman.

In terms of superintelligence, that is actually what Lattice is supposed to eventually become. By tying enough sensors together, you can build a model of the world where you can react not to what the enemy is doing, but to what they will be doing.

That’s the type of capability where I’d rather be able to predict where my enemy is going to be and what my best response is than be able to have a jet that goes ever so slightly faster. What if I can start going to where I need to be—skate to the puck—because I’ve predicted that’s where I need to be? I’d rather have that.

In the commercial world, I talk about this as a trillion-sensor future, where you can know anything you want because the sensors are there. You can predict a man’s blazer color on Madison Avenue because you can ask your AI to look at the camera feeds and so forth.

If you can know anything you want, then what’s interesting is that it’s important to ask amazing questions. The questions you ask are more important than what you know.

Sure.

At that point, the world pits the United States versus China in the AI space. I had a conversation with Eric Schmidt about this, where the concern in my mind, and I think in Eric’s as well, is not the United States versus China. It’s the United States and China versus the rogue actor—the individual out there who’s using digital superintelligence to code up the next virus or the next hack, whatever the case might be. How do you think about that?

I would probably take a bit of a counter position. I’m very worried about rogue actors because rogue actors don’t necessarily act as rational actors. Nation-states typically exist on a rational basis. You could take an extremist theocracy like Iran and argue that they’re not acting rationally, but those are few and far between.

In general, nation-states follow game theory. They act in their rational self-interest. They don’t want to destroy themselves in the process of destroying you. Whereas when you bring that down to the level of an individual or a small group, you can have people who believe they win by losing. They could think that them dying is the victory. They could believe that bringing about an apocalypse is their destiny.

I’m terrified of, for example, tailored bioweapons built by rogue groups. The idea that it’s the United States and China versus these rogue groups, I’m not so sure. I think that China on its own poses its own unique type of threat.

It doesn’t terrify me as much. China is not going to purposely build a tailored bioweapon that wipes out all the Jews, for example. I don’t worry about China trying to do that. But at the same time, China has made real, material threats and said it is going to reunify with Taiwan by force if necessary within this generation.

I watched your most excellent TED Talk on wargaming the war, and it’s chilling to see how that plays out.

Yep. One, it’s not just Taiwan. In living memory, China tried to invade Vietnam.

A lot of people want to pretend to whitewash China, largely because they’re often working with China, so they have to carry water for them. They say, “China and Taiwan is a special case. There’s this long history.” I say, “Okay, well, what about when, in living memory, they invaded Vietnam?”

What about the fact that they are currently occupying huge swaths of territory in the Philippines? What about the fact that they are illegally building artificial islands in the sovereign territory of other nations? What about the fact that you now have Xi Jinping going to conferences and saying that Okinawa, Japan, is actually a territorial vassal holding of China?

He’s pretended to have this awakening. He says, “The Okinawan people used to be a tributary state to China. They gave us tribute, and we failed them by failing to protect them from being taken over by the Japanese imperialists.”

He’s laying the groundwork for revisionist history because he knows he can’t motivate a bunch of young Chinese men to go and take over a territory they have nothing to do with, with absolutely no way of convincing anyone that it’s theirs. He has to tell them a story where this is actually part of the great Chinese Empire.

That is where I think China is its own unique threat. They are willing to reinvent history with their own population, much like Russia has with Ukraine, to justify death and violence at mass scale.

I think they want to take elements of Japan. They want to take elements of the Philippines. They want Korea. They want Vietnam. And certainly, they want Taiwan.

Imagine what a world looks like where China achieves even half of that. A rogue actor is not going to do that. A rogue actor might take a virus, but they’re not going to take over a democratic nation and seize control of the semiconductor supply.

Having said all that, what you said about being rational actors and being able to take actions politically and militarily to prevent that from occurring gives you a game plan.

The question I have is: Are Anduril systems, in the idea that Lattice is giving us an omniscient level of knowledge to prevent rogue actors, going to be critical to having enough data about what’s out there and being able to track it? Do you imagine that as part of your future?

I think it’s a part. Honestly, I’d probably have to give more credit to companies like Palantir. I think they’re building more of these non-tactical-edge, real-time tools that allow you to find these bad actors.

They’ve been involved in apprehending and killing a lot of really dangerous people, terror cells, and repeat violent criminals. I think Palantir and companies like them are probably doing that.

If I had to split it, I’d say a company like Anduril is much more relevant to a traditional hard-power deterrence theory that stops a rational actor like China, and less so to a rogue nation-state group.

Let me flip to the positive side of ASI, of artificial superintelligence. There are a lot of breakthroughs on the precipice. We just saw the first Nobel Prize given to Demis Hassabis and John Jumper for AlphaFold. What are you hoping for out of advances in physics, math, and science?

Medicine.

I think there are so many elements of low-hanging fruit that we have not been able to seize, partly because of the regulatory climate, but also because the cost of developing and testing new drugs is so high. And not just drugs, but new therapies—think therapies that require continuous intervention and monitoring. We have not had the resources to try everything. You have to pick very, very tightly what you are going to do, and even then, it mostly does not work.

Business, I know.

And so, automation at scale of those—I mean, what if instead of one lab, you could run 10,000? What if instead of running 10,000, you ran 1,000,000 simulations?

I am very optimistic about medicine. I think energy is another area where AI-assisted design of fission and fusion energy-generating systems is going to be a massive, massive change in the way that we use energy. Energy is such a huge part of our way of life. It drives food costs, it drives the cost of materials, and it drives the cost of a country.

That’s right.

There are really no examples of high-GDP countries that do not consume lots of energy—not necessarily produce it. There are countries that buy their energy from elsewhere, but it takes energy to build the future. I am very excited about that.

Do you ever see Anduril getting into the adjacency of the energy space or the biotech space?

I think maybe we are already a little bit adjacent, but we are really focused on our mission of trying to modernize military capability and force, and doing it well. With energy, we partner with a lot of these companies. There are a lot of companies doing interesting things in the nuclear power space, and we are partnering with them. I do not see any reason for me to try to compete with them. I want to be a customer of theirs, and I want to use the DoD as an early customer that can help accelerate the deployment of these new ideas in how to split atoms and how to fuse atoms.

I want to talk about the speed of defense-system innovation. I have a few metrics for comparison here, looking at the glorious days of World War II in manufacturing and innovation.

Kelly Johnson brings on the first U.S. jet in 143 days, from a clean sheet of paper to a jet flying. The Liberty ships got cut from 230 days per production to 4.5 days. The P-51 Mustang fighter goes from concept to flight in 102 days. One of the references you had was the B-24 Liberator bomber—one every 63 minutes by a single company. That’s insane. These were not small planes. These were flying fortresses.

What happened? It was a war footing, but I know you studied Kelly Johnson and Lockheed Skunk Works. His philosophy was—I mean, if I remember correctly—he had a single blueprint in the center of the workspace, and any of his engineers could go and make a change on it, but they had to sign their name to it because they knew that if they made a mistake, someone’s life was at stake. The rate of iteration was so rapid. What happened that killed that level of innovation and iteration?

Look, you can blame a lot of things. I think it was probably the end of the Cold War. I am not saying that we should have continued the Cold War; it is just that the end of the Cold War is what caused the change to happen.

The United States government came in, and you may be familiar with the Last Supper. They brought together the heads of all of the major defense companies for one dinner, and they said, “There will be consolidation. Half of you should not exist by the end of next year. Consolidate, consolidate, consolidate. The party is over. We are going to decide who the winners are.” Musical chairs.

If you do not get with this program, then you are out and you are done. It was very much a top-down-driven thing. So you ask, why did the innovation go away? Why did the speed go away? It was because there was no longer a drive to move quickly. There was no longer a government directive to move rapidly against threats, and we moved into a peacetime posture that was willing to accept a high level of inefficiency because they felt like that was okay.

I think it went worse than they expected. I think they expected some level of inefficiency. They did not expect that reduced industry to then capture the political side and maintain that inefficiency for decades. It was one of those ideas that seemed okay but did not turn out so well.

By the way, the argument that the people who architected the Last Supper would make is that we made the right decision. They would say, “Look, we reaped a peace dividend. Look at what we did through the ’80s and the ’90s and the early 2000s. Look at the economic growth in the United States.” It is hard to argue with the results. They argue that we did reap a peace dividend on the back of this, but we can still recognize that it was a problem for our military prowess.

We had that huge explosion of economic development and technological investment elsewhere, to the detriment of our military. One last thing I will say is that a lot of the smart people left. A lot of those people who helped build critical things like GPS for the military did not stay in government labs. They went into the private sector.

Look at the microprocessor industry. It was the same thing. These same people who built microprocessors for the DoD instead went into the private sector. We had the explosion of Silicon Valley. That is where the smart people went, and that was definitely also to the detriment of organizations like Skunk Works.

That’s been your philosophy—to pull that talent back.

Exactly. We are just bringing them back to where they were. There is a long tradition of the smartest people in the country wanting to work on national-security problems, and there was a time when that was not the case. I think that is finally reversing.

I want to dive into your design philosophy here at Anduril. I have spent a lot of time with Elon talking about his design philosophy at SpaceX. It seems to be a very similar parallel: simplify the parts count, simplify the designs, but do not make them overly simple. How do you think about your design philosophy in the systems that you are building?

Boy, this is a huge question. I am trying to think about some of the common threads. One of the common threads that I think is different about Anduril than people would expect is that we generally do not vertically integrate.

SpaceX, Tesla, and others have really fetishized vertical integration, and it makes sense for some of them. It really does. But when I get pressure, usually from people who do not know my business that well, they say, “When are you going to bring this all in-house?” I get asked that as well.

The thing you have to remember is that when you are building, let’s say, space-launch systems, your customer base is pretty well known months or really years in advance. You know what your schedule is, and you know how many rockets you are going to need. You can plan all of this very predictably.

That is not necessarily the case for weapons production. You need to be ready for shit to hit the fan and to 10x or 100x your production.

Got it. It’s actually pretty irresponsible, ramping up and ramping down.

Yeah. If I build a vertically integrated capability where I build every wiring harness, I do not work with any partners on my fasteners, on my composites, or on my casting, and I can only do that in-house, what happens when the DoD suddenly needs 100 times more of that system every single day?

That means I have to build 100 times more factory space. I need to hire 100 times more people. How the hell am I going to do that? What is much more responsible is for my engineers to design a part that can be made by any machine shop in the country, anywhere. Pick an adhesive where there are 10 suppliers in the country, not just one, and certainly not something we only make ourselves.

That means that if I need to ramp up, I can multisource these things. I can outsource them to lots of other places, or I can do what we did during World War II. I can go to the industrial capacity that exists in, let’s say, the American automotive industry or the American commercial-aviation space, and take over.

You say, “Hey, good news. Our submarine can be manufactured by the same robot arms, the same plasma cutters, and the same assembly lines and people that were cranking out cars yesterday.” That is how you build a resilient defense infrastructure. I mentioned this in my TED Talk: We have to design for mass production using existing infrastructure.

You can't assume that you're going to have the time to build an alien dreadnought to build your thing. And that's again, Tesla with the Model 3 wanted to build this hyper-optimized capability. But Elon is never going to have 1 year where he needs 100 times more Model 3s and then the next year 100 times less. It's just not quite like that.

Makes sense. It's crystal clear now. Going beyond that, in terms of how rapidly you iterate a product, how you focus on parts count, materials, and so forth, are there other design elements that you have as a basis for the company that you learned perhaps when you were at Oculus?

There's so much that I've brought from my Oculus days. That's what makes this company different. What's interesting is that so many of these things are—I almost don't want to belabor them, because if I belabor them, it sounds like I think I'm a genius for doing things that are already done everywhere. What we're doing is taking the same approaches to design, design review, and velocity of manufacturing from the consumer electronics world and just bringing them to defense.

At Oculus, we were launching a new product every single year. We had to manufacture millions of virtual reality systems. It's a totally different mentality than you see in the defense space. I'd say the main thing we brought here is just: do it like that. Do it the way that you do it in industries where you have to move fast, where you can't afford to fail.

Imagine if the iPhone was delayed by 4 years. iPhones get delayed from time to time, right? You've seen this happen. But it's usually, “Oh, you missed by a month, but they'll be over here soon.” Or manufacturing was behind, so they couldn't send it over on a boat. They had to air-freight it, and they lost a little bit of money. Have you ever heard of an iPhone being delayed by 4 years? How about 20 years? Have you ever heard of that? It's just unthinkable.

A lot of what I do is just doing things the way that they're done in industries that aren't subsidized by taxpayer dollars, that can't afford to fail. When you skin your knees after you fall, you're a lot more careful not to trip. I think that's really what has helped Anduril in an overarching way. We hire people from consumer electronics, the automotive industry, and the maritime industry who are used to working in those kinds of conditions.

Do you ever expect the tech you develop at Anduril to go back into the consumer space?

Not really. Maybe it's even a breach of fiduciary duty, but I just don't have a big interest in it. I started this company for a reason: to fix national security.

Early in our company's history, we had the opportunity to do quite a bit of commercial work that I think would have actually grown faster than our Department of Defense work. That would have been a problem. Imagine a world where Anduril has a product line where half of the team is dedicated to the military and half is dedicated to, let's say, commercial oil and gas security or critical infrastructure security. Imagine a world where the commercial side is growing 3 times as fast. What investor is going to allow me to continue to spend half the team on the thing growing at a third of the rate?

I was terrified early on that that could become a reality. It was actually similar to our border security work. I was worried that part of the business would put us in a position where we weren't able to invest in the military side. There were times when we said, “You know what? We think we could make money there. That is not our mission. We need to stay laser-focused on our mission.”

That's how we're going to get to where we want to be, which is being the next-generation defense product company that really—our first page of our first pitch deck said, “Anduril will save taxpayers hundreds of billions of dollars a year by making tens of billions.”

I love that line. Love that line.

And that was the mission. So will it come to consumers? I don't know.

I'll finish off this bit by also noting that anything that we sell to consumers is, at the end of the day, going to end up in the hands of our adversaries. People have asked me over and over again, “Palmer, you're building EagleEye, this new integrated visual augmentation system that's giving soldiers superhuman thermal vision, night vision, and augmented views of the world. When are you going to build a consumer version of that?”

I would love to, except it will end up on the heads of Chinese commandos, and they'll say, “Oh, but Palmer, there's export restrictions.” Yeah, but if you're selling something to civilians, eventually you will sell to a trader, and that traitor will get that gear into the hands of your enemies.

The Russian special forces aren't wearing Russian gear. They're wearing American night vision, American helmets, and American armor. They're using the best. That's because they prioritize getting these things smuggled out of the United States and into Russia.

Sure, maybe you can stop it from being on every Russian soldier or every Chinese soldier, but how do you think I would feel if I built advanced capabilities that we sold to civilians and then, in an invasion-of-Taiwan scenario, a bunch of Chinese commandos dropped out of helicopters and killed all the top political leadership of Taiwan using Anduril gear?

That would be the worst reversal of intent in my life that I can imagine, in terms of intent versus effect. That's my biggest problem with selling back to civilians. I would only sell tech that I don't worry about getting into the hands of our adversaries.

Makes a lot of sense. And that's not what you're passionate about.

If I figured out how to do—I'm not doing this—but if I figured out how to do, let's say, better biological defense, I would be absolutely passionate about it. I've long been interested in long incubation antibiotics, things like antibiotics that are encapsulated, live in your body for long periods of time, and are only released when you have some biological trigger that causes them to be released and become active.

Or biological antibiotics—same idea. Sort of a loitering defense system, but one that is only active in the bloodstream when there's a threat. If you just loaded people up with antibiotics all the time, you would create superbugs because they would continuously be active in people.

Suppose I figured out how to do that and there was crossover to the civilian side. I would absolutely be a fan of that, but I would have to make sure that I'm not inadvertently giving a tool of great power to an adversary.

I want to jump 5 years out. It's 2030. What does warfare look like in 2030? You've got AI far more advanced, humanoid robotics—and I know your position on humanoid robotics—but the ability to enhance super soldiers takes on a brand-new meaning.

Yep.

Drones have gone from zero to infinity in record speed. It's extraordinary. What are you thinking?

I hate to be a cynic here, Peter, but I actually think warfare in 2030 is going to look more or less the same as it does today, with a few very small exceptions where breakthrough capabilities get in. I said earlier: you go to war with the tools you have, not the tools you want.

The reality is that the vast bulk of our arsenal was built a decade or 2 or 3 ago. And so, even as companies like Anduril move very, very quickly—we're trying to build things that are relevant to a fight with a great power, whether it's Iran, Russia, or particularly China—even if we move at breakneck speed, as fast as we can, we're going to end up being 1% of the fight, 2% of the fight.

We can try our very best, but it's going to take years and years to replace these legacy capabilities with new things.

What will the battlefield look like? You're going to have a weird, anachronistic mash of things that were built in the Reagan era: tracked vehicles built in the Reagan era, operated by humanoid robots that just rolled off the line a few weeks ago—but only 1 column of them. All the rest are going to be crewed by people.

You're going to have things like AI fighter jets flying alongside aircraft that were built under Bill Clinton, and they're going to be flying together in formation. Unfortunately, there's probably going to be a lot more manned aircraft, and the AI aircraft are going to be a valuable tip-of-the-spear component. They'll be the tip of the spear, making first contact, and they're probably all going to be blown up. We're going to say, “Shit, I wish we would have been building those for another couple of years.”

It's just 2030. It's close. It's there. There's just so little time to build it, deploy it, and then train people on it. Remember, you can't just deliver these things on day 1. People have to train for years to become proficient in something.

Imagine if you showed up with a new alien weapon system pulled straight out of the Roswell wreck today, and you handed it to a soldier and said, “You have to go to war with this tomorrow.” That won't work. You need to develop tactics. You need to develop doctrine. You need to have him train with his squad for years, potentially.

Let's take it slightly differently. Let's talk about the 0.01%. Let's talk about the elite Navy SEAL team, or equivalent, out there that will have the most advanced technology.

Well, you're going to see lower fatality rates. You're going to see people who are acting as omniscient technomancers, kind of acting as a central hub in an AI surround. They know where the good guys are. They know where the bad guys are. I think, to a certain extent, the future of warfare is going to look a lot more like chess than dodgeball.

If you understand what's happening and you know exactly what you're up against—where it is and when it is—you can know when you can win and also know when you need to retreat. You don't necessarily get to the point where you win or lose the Battle of Midway. You know well ahead of winning or losing what the likely outcome is. That probably drives better decisions.

I think you're going to see a lot fewer casualties, a lot fewer fatalities. You're not going to allow yourself to will your way into a scenario where everyone gets wiped out. There's good and bad there. When you give people better visibility into what's going to happen, imagine this: Imagine a world where we get into a fight that we can't really afford to lose, and then we find out that to stay in that fight, we're going to have to send 50,000 sailors to the bottom of the sea.

I don't think the United States has the political will to do that. We just don't, especially knowing that it will happen. It's a double-edged sword, but I think in general I'm on the side of having the information to make that decision. It's going to make decisions a lot harder for these guys.

I'll end with this: In current warfare, the fog of war allows for enough indeterminism that someone can make hard decisions without really knowing what the impact would be. You believe, “Hey, this might work. Everyone might be fine.” It's interesting to ponder what happens when that uncertainty is removed.

What happens when you order someone to do something and you're no longer sending them into a non-determinate, liminal space—“Oh, well, they might live, they might not”? What happens when you know that they are, with a high degree of certainty, going to die? That will be a change in the nature of warfare at a very high level.

Now, of course, the flip side is, like I said, I think there will be lower casualty rates and better decisions will be made, but it's going to make for a very hard set of ethical quandaries. I don't think anyone would argue that it's better to not know. I don't think you'd find anybody saying it's better to not have that information in your decision-making process.

So, this Navy SEAL is omniscient. They've got enhanced imagery, enhanced knowledge, probably a 100-to-1 ratio of autonomous systems to men. Every person who's going to be out there is going to be working in a highly networked fashion, commanding drones and robots that are basically their extensions.

Some will be commanding, and I think a lot of them are going to be autonomously doing their jobs. Suppose that you have that Navy SEAL. He might be aided continuously by 10 drones that are sensing the world around him, looking for things that are a threat. He's not so much commanding them as consuming the information that comes in.

He's not watching 10 drone feeds. He's just seeing, in his augmented view of the world, where those threats are. As things become a critical threat, the system is able to highlight that to him. He doesn't have to look at 10 drone feeds and say, “That guy's running. I think he might be going over there.”

The system is going to say, “Hey, this is the top threat. It's the only thing that might kill you in the next minute. You need to deal with this. What do you want me to do?” It's going to be a little different than commanding them. He won't be commanding the drones so much as having them feed him a view of the world.

I act like this is the future, but of course, this is what we're doing with our customers right now. People are doing these things in exercises and in small-level conflicts all over the world right now. It's just going to be a different thing when it's deployed at scale.

Have you taken the time to dream 5 years out beyond?

I mean, you're building with 5 years out. I know exactly what I'm doing. 5 years is easy. The things that are going to be relevant 5 years out, we're starting to build them today. We just started construction on a $900 million factory in Columbus, Ohio, to build our autonomous fighter jets. Those are going to be in combat before 2030.

So, 2030 is easy for me. I know exactly what we're doing.

The question is, what are you starting to design and build in 2030?

Yeah, that's the interesting one. It's actually hardly anything. In general, Anduril is very focused on building weapons for the immediate near term. It's leaked out through the press that we have certain teams working under a mandate called China 2027. If the feature you're building or the capability you're working on is not going to be ready for a fight with China before the end of 2027, you can't be working on it. You need to find something that is relevant to that.

I don't want to say that I'm not even thinking about 2030 and beyond. I probably dedicate 1% of my time to it. I'll tell you one thing I think: I think you're going to see subterranean warfare become a much bigger part of the future.

Really?

I believe it's the next major warfighting domain. I've said this many times, and everyone thinks that.

What does that look like—drilling machines?

So, yeah, more or less. Have you seen the movie The Core? Oh, my God. It was a while ago—2006, I think. It's about a group of guys who have to drill to the center of the Earth to use nuclear bombs to restart the Earth's core spinning, to protect us from cosmic rays. It's not a scientifically sound movie, but something like that.

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. The Soviets actually built a prototype and then lost it in the crust of the Earth, so it worked that well. It just went off, and they lost track of it, melting through the crust.

I think that's going to become a very powerful part of the future of warfare. I'm not talking about just tunnels or bunkers. I mean using the crust of the Earth as a fully three-dimensional battlespace that you will be moving supplies through.

You'll be doing electronic warfare, kinetic warfare, psychological warfare, high-end logistics, and more. That isn't relevant to China. The technology is just not quite there. I can't really build things at scale that are relevant by then.

That's one of maybe the few things I think about past the 2030 timeline. I think it's going to become a huge deal. At some point, in the same way you see a Space Force, I think it's very likely you'll see some kind of Subterranean Corps.

I don't know exactly what that's going to look like, but right now, the people who work on subterranean bunkers and tunnels—I think it'll become a large enough part of warfare that you're going to need a dedicated group that focuses on the unique challenges of the subterranean domain.

Amazing. I saw the video you posted on Pulsar-L recently. That was a bit of magic. It looked like a bunch of mosquitoes flying out of—dropping out of the sky. I don't know if you described that for folks.

Well, yeah, we just launched this video that shows Pulsar-L. It's basically a thing the size of a small cooler. You can carry it in the back of a truck, and it is an AI-powered electronic warfare system capable of jamming, spoofing, hacking, targeted cyber effects, general cyber effects, and doing things that make the motors of drones want to stop working and make their navigation not do what it's supposed to do.

It does things that don't just work against remotely piloted drones. It even works against autonomous attack drones. It looks like an EMP being triggered, and everything just falls down. We released this video where—and by the way, that was a real test event—we had 25 autonomous attack drones flying toward the target. You turn on Pulsar-L, and they all fall out of the sky, fall to the ground.

This is a real capability. We've been selling it to real military customers. They're using it in combat right now, and we finally were able to start showing it publicly. We released this video, and people were all saying, “There's no way this is real. This is all totally fake. It's all CGI.” They wish it were this easy.

What they don't understand and don't see is that we've been investing in electronic warfare at Anduril for the last 5 years. This is a culmination of all of that work. This is a real capability, and in fact, the video is literally an actual live test event.

I actually tweeted about it. I said, “Okay, fine. We'll release all of the behind-the-scenes footage. We'll just take all the video footage that we showed of it actually working.” Now, we're not going to be able to talk about the specifics of exactly the way that we're being clever with the electrons, because that stuff falls into the classified domain.

But I will note, too, things like Pulsar-L—they're not the solution, because it's possible to make a drone that can survive that type of attack.

They’re a very useful part of a layered approach, right? You need to have directed energy, you need to have EW, and you need to have kinetics. You need all these things working together, and it’s very hard to make a drone that makes it past all of those things. It’s very hard to make a drone that can survive all of the ways that Anduril has for stopping a drone.

Yeah. It looked like the kind of device that I’d want in every Jeep on the battlefield. I’d love to see it at every airport. I’d love to see it at every sports stadium.

The biggest obstacle is actually regulatory. Pulsar L is completely illegal in the United States for nonmilitary use. There’s nobody in the United States who’s allowed to use something like Pulsar L. The only guy who’s allowed to push that button is someone with very special authorizations via the military.

I think that’s going to change. I’ve been spending years now talking with members of Congress who understand that we can’t afford to have our airports shut down by drones. We can’t afford to have our military bases surveilled by drones. We can’t afford—I suspect inevitably there will be someone who commits a large-scale terror attack, or a series of terror attacks, using drones. It’s cheap enough to get ahead of these threats that we should at least try.

You mentioned recently that we need to look at the ethics of using AI in warfare on a case-by-case basis. Absolutely. The examples you gave were compelling, and I agree with you. I’d like to scratch that a little bit: where does that case-by-case ethical review happen? Do you guys have that kind of conversation inside of Anduril? Is this something that’s happening with your DoD customers? How do you think about this?

The good news is that the DoD already has these processes in place, and they have for decades. The reason that so many people are freaking out about autonomous weapons is because they think that it’s a new thing. I mentioned in my TED Talk that people think they’re keeping Pandora’s box from being opened. What they don’t realize is that every US military base and aircraft carrier is protected by autonomous weapons that shoot down incoming boats, incoming missiles, and incoming aircraft.

They don’t realize that destroyers are all capable of operating in a fully autonomous mode, even if the bridge is completely destroyed and not a single person is living topside. You go back to World War I and World War II. All the landmines that were used were autonomous weapons triggered on their own. We go back even further. I’ve given a couple of talks where I argue that this idea of building weapons that execute the intent of the designer, even when the person is not immediately physically present, goes back thousands of years. Spike traps, pit traps, poisoned wires—all of these are autonomous weapons.

Now AI allows you to do new things, but in Vietnam we were also using missiles that would be fired from a jet, fly into an area, look for, for example, surface-to-air missile launchers, and then destroy them. Those are fully autonomous weapons. They’re deciding which targets to hit and which to destroy, and they’re discriminating between one type of target and another.

What I mentioned is that you have to look at these on a case-by-case basis and not have a blanket prohibition on AI autonomy. Imagine if you could say, “Hey, I can take this landmine—it’s an antivehicle landmine. It’s not set off by people; it’s set off by vehicles.” Right now, it can’t tell the difference between a school bus and a tank.

Yeah. Why would you want that? Why would you want that?

There are people who are fighting for that. They want a UN-level resolution to condemn the use of AI in weapons, to make it illegal for a robot to pull the trigger. My point to them is: if you’re going to use landmines, shouldn’t they be able to make that difference? Shouldn’t you be able to use every tool to achieve the most precise, most surgical, least-civilian-casualty-attached outcome?

They’ll say, “Oh, here’s why I don’t believe that.” My point to them is: if you have a problem with landmines, ban landmines. Don’t ban landmines from being as good as they can be at not killing civilians. It’s the same thing with a bomb. If I can make a bomb that, using autonomy, does not kill the person who’s 100 yards over to the side of the guy that I need to get rid of—if I’m taking out the head of al-Qaeda, isn’t it better to have something that kills that guy and doesn’t blow up the building next door?

There are people who would argue no. It’s such an ethically fraught problem that they can’t deal with how icky it feels to have a robot decide who lives or dies. My point to them is, guys, the DoD has a process for this that they’ve been applying for decades. The key is to never abdicate human responsibility.

A person always needs to be responsible for how force is used. When that AI weapon kills the wrong person, there needs to be human accountability, as if there were a person pulling the trigger. That is the thing we cannot afford to compromise on. Banning AI wholesale is just going to ensure, one, that we lose, and two, that we’re fighting with our hands behind our backs, and a lot of civilians are going to die as a result. That is not a moral outcome, in my opinion.

I have to admit, you’ve been on my stage at the Abundance Summit twice now, and the first time you were on stage I was a bit nervous about how the audience was going to react.

Sure.

It was a standing ovation. People were completely won over by that argument: if we’re going to get into a war, if we’re going to aim to kill somebody, let’s make sure that the collateral damage is completely minimized. Let’s focus on the intention. What arguments have you gotten against that? Because I can’t imagine one that would win.

It’s differences in philosophy. There are perhaps two arguments people will usually make. Either they’ll make a purely philosophical argument, like, “It is not the place of tools to rebel against man. We cannot outsource certain things, no matter the cost in lives.” They would rather civilians die today than outsource these decisions to AI models, and it’s just a difference in philosophy.

I think that minimizing civilian deaths is really important. There are other people who take a more existential-risk approach, like the x-risk people in the AI community. They say, “I have no problem with the landmine—the landmine that doesn’t blow up the school bus full of kids, I have no beef with. But first it’s the landmine, then it’s the gun, then it’s the nuke, then it’s Skynet, and it wipes out everybody.”

My point to them is, look, that just isn’t how the DoD looks at these things. These are usually people who are not familiar with how the DoD actually makes decisions. It’s hard enough for me to get AI into that landmine. That’s actually hard. There is such an extraordinarily stringent review before they deploy new weapons.

Here’s a great example: there was a new landmine capable of a fully autonomous mode that was developed during the ’90s. It was developed by the United States Army, and it was capable of basically being a sensor that could trigger remote mines around it. It would detect what kind of vehicle it was and blow up if it detected it.

They actually disabled that capability in the final version because they couldn't figure out how to attach responsibility for malfunctions. They couldn't figure out how they were going to say who was responsible. Is it the person who ordered the landmine deployed? Is it that, any time the instructions to it are updated or the categorization is updated, it's that person's responsibility? Is it the contractor who develops the differentiation model? What's their liability for civilian casualties? The military is fundamentally very conservative. They don't take these crazy risks.

People imagine there's a slippery slope to Skynet. Remember that our nuclear arsenal, until a few years ago, ran off floppy disks. They were so conservative that they didn't even want to move to digital circuits controlling these things; they kept it all analog. Given that, I'm just not that worried about the slippery slope. I think the people who are in charge of these problems are very sharp, and if you don't believe in the process that puts these people in positions of power, then you just don't really believe in the democratic process, period.

I mean, look, the alternative is that you have people making all of these decisions, flawed as they are. If you trust a 19-year-old kid not to nuke the wrong people—I'm just kidding; that's a little ridiculous. Nineteen-year-olds don't get the nuclear keys. It's people who are a little more senior, but you get my point. If we are trusting young men with decisions of great life-and-death import, it seems a bit strange to say, “I think the system as a whole is just going to trend toward irresponsible use of force, and the machines are going to kill us all.”

I understand the existential-risk people, but it's similar to you. I don't know how you feel about the idea that people say this about AI that has nothing to do with weapons. They say we shouldn't develop AI to help us with physics because what if it develops new physics and then uses it to exterminate humanity? I'm a lot more worried about evil people with existing AI. It's not artificial intelligence; it's human stupidity I'm worried about.

Yes. The AI part is the part I'm least worried about. I'm worried about bad people using good AI, not superintelligence turning against everybody.

A lot of people don't realize that the technology you've been developing has some significant nonwarfare applications. Here I'm pointing directly at the prevention of perilous wildfires. Two years ago—two and a half years ago—we were in Washington, D.C., together, and you were the first registrant for our $11 million Wildfire XPRIZE. You really pulled together a lot of people. You had the lieutenant governor of California there, the head of CAL FIRE, and a lot of people from the U.S. Forest Service.

Yeah, it was a great event. It was valuable to have you step up as our first registrant, and even more valuable because of what you said. Six months ago, we all had front-row seats to the Palisades Fire. That shouldn't ever happen again. It shouldn't have happened then.

The really crazy thing is—and you know this, but not all your listeners might—we started working on firefighting technology right at the start of the company. We built the Century firefighting tank. It was a tracked, autonomous firefighting vehicle that could continue to fight fires even after a fire had overwhelmed an area, long after all the people had shipped out.

The problem we ran into was purely political. People were afraid it was going to replace jobs and automate jobs. They were saying, “If you fund this, then we're going to come out against you politically in the upcoming elections.” That was a really big problem.

I think we should have been working on this, and I think that you have similar things. The whole point of that, for people who don't know, the Wildfire XPRIZE, is to end destructive wildfires using autonomous technology: to build things that can detect and react to fires instantly.

Yeah, and the problem is that not everybody wants things to work that way. There are people who don't necessarily want to stop wildfires because their job is tied to fires continuing to exist at scale. That's the hardest part of this.

Well, look, if your whole job is to do, let's say, large-scale firefighting tanker operations, you're not going to be excited about giving money up in your budget to build something that stops that from ever happening. It's a perverse set of incentives.

I don't think any of these people are waking up in the morning saying, “I can't wait for there to be more fires and more death so that I can get my budget.” But they're never going to want to take a risk if even a successful outcome is probably bad for them.

I said it at the event, and I'll say it again: We can do this. This is not a distant future. This is not a superintelligence problem. This is a matter of product execution. The technology to detect and extinguish destructive wildfires—maybe not all of them, but let's say 95% of them—exists today. We just have to put the pieces together and demonstrate it.

The evaluations for the prize are coming up in October, so that's not that far away. A whole bunch of companies are participating, and you've seen the companies. You're a registrant right now, and we're teamed up with some of them.

The coolest part about this prize is that, unlike some other XPRIZE competitions where a bunch of people were trying to figure out if something was even possible and pushing the limits, in this case I think there are lots of companies proving that it is very much possible. It's now just a matter of cost and effectiveness: How much will it cost to do this? Everyone is trying to drive down the cost and increase the effectiveness. I think all the teams are in a place where they've proven it can work.

Have you publicly said what technology you're going to use?

I don't think we've publicly gotten into too much detail. The plan right now is to demonstrate multiple things, and this is how I think it will be in the real world. In the real world, you're not going to have one type of solution.

The aircraft that will respond to a fire on a hill, a hundred miles out in the brush, is very different from one that is going to respond to a fire starting in a power substation right next to a bunch of trees. You'll need different tools for each job.

What you need to do is detect fires, classify what kind of fire it is, determine what kind of firefighting agent you need, and understand the environmental conditions, including the wind and how you can approach it. How far away is it? What's the closest asset? What type of compounds do you need to fight that fire?

You need a system that autonomously decides which of the available assets is best suited to stop this particular problem, deploys it, and then sees whether it worked. Did it slow the fire? Did it stop it? Do you need to continue deploying assets? Do you need to bring out the big guns while you try to just tamp it down?

Our plan is to have multiple autonomous assets, different types of assets with different capabilities, and then demonstrate how different types of fires trigger different responses from the system. I also think that's how it will work in the real world.

If Anduril were to deploy, let's say, a Lattice instance with sensors, fire-watch towers, and space-based layers, people would ask, “Why not just do it all from space?” The answer is that sometimes you have weather that makes it impossible to see what's going on. You might have a lot of fog or a lot of clouds, so you probably need a terrestrial layer as well.

Anduril probably isn't going to be building all the vehicles that respond. I think you're actually going to see a lot of different companies focusing on their niche. Some people will build a vehicle for more urban environments, while others will build for long-range operations.

What I love about XPRIZE is that it's a Darwinian evolution, where you have hundreds of different approaches all competing. At the end of the day, as you said, you'll probably collaborate with a number of them.

I think that, in the end, probably half a dozen of the teams that are competing will be involved. If CAL FIRE were to award a multibillion-dollar contract to deploy these systems at scale and stop wildfires, I'm very sure it would not be any one of us getting all that money. It's going to end up being a distribution of money to a lot of different people for a lot of different things.

The thing that is insane is when you have these Malibu fires, these Palisades fires, over and over again, and then it's impossible to get insurance for your home.

Well, and the cost is so immense, and the requirements California has put on these insurance companies are also so immense.

What blows my mind is when we talk to people in Cal OES, the Office of Emergency Services—and I'm not putting them down.

They've got a lot of constraints they have to work under. But what's so fascinating is when we talk about the cost to deploy a system that would detect every fire and put out many of the fires, they say, “Oh my God, that's billions of dollars. Where are we going to get billions?” My point is, if you stop even 1 fire, you've already made the money back. You've already saved lives, prevented property loss, and saved time. It's crazy.

The lives are irreplaceable. But even if you look at it just in dollars and cents, stopping 1 of these fires would pay for the whole system in terms of the economic damage. It's one of those things where it seems like a lot of people are being pennywise and pound foolish, and we have to fix that. I'm going to ask you a rapid-fire set of AMA questions from my Twitter audience.

Let's do it. I'll be efficient in my answers so that we can hit as many of them as possible.

Can a drone fly in formation, land, transform into a robot with 2 or 4 legs or wheels, reconnoiter, and attack on land? Do you imagine a mixed-mode set of drones?

Such a thing is definitely possible, and I've actually seen companies that are building exactly this. I've seen companies building quadcopters with legs, for example, where they land on their legs so they can loiter for long periods as a watch capability. I've seen people building robot dogs that basically have jetpacks. I've seen the gamut.

The thing is, I'm not going to say any of these don't make sense. It's really a matter of how many situations need both. Why not just keep flying? Why not get there just walking? Or here's another example: why not put the robot dog onto a flying vehicle that drops the vehicle in place, and then you don't have to carry all of that extra parasitic weight?

It's always possible to come up with some niche scenario where you need it to fly, land, walk into a cave, jump over a hole, and fly out the other side. Those do exist. But here's the good news: making these different niche robotics, I think, is going to be a big part of the future. There's not going to be 1 form factor that dominates everything, right? You're not going to see C-3PO-style humanoid robots doing literally everything. There are going to be 100s of different form factors, and I bet some of them will have wings and legs, just like in nature.

All right, next one. A serious one: between U.S. military tech and Chinese military tech, is America behind?

There are places where America is behind, and there are places where we're ahead. It's hard to give a universal answer. In general, I think the United States has a strong lead in a lot of the areas that people would consider critical, but at the same time, you have to look at the fact that China has about 300 times more shipbuilding capacity than we do.

I mean, people can't visualize it.

I saw your tweet. And by the way, that's not during wartime. They say, “Well, America would just scale it up during wartime.”

America would just scale it up during wartime, and so will China. They've proven that they can do it. I'm not saying we should do this, and I'm not saying we should copy every move of an authoritarian, centrally planned state. But China has made it a law that many types of boats—for example, passenger ferries and car ferries—can be commandeered, and they have to build every passenger ferry to military specifications so that it can be used for a Taiwan invasion.

All of their car ferries have to meet a certain deck-plate load standard so that they can move armored vehicles onto them and move them to Taiwan. That's an advantage that they have. Are they ahead of us on amphibious landing capability? Immeasurably so, by orders of magnitude.

I had a tweet that I found particularly interesting about the importance of a navy.

Yep—in projecting global domination.

Yep. Protecting freedom of movement and freedom of trade. Our navy is kind of the backbone that allows global commerce.

Yeah.

All right. Here's one you may or may not want to answer: What are your honest thoughts about Mark Zuckerberg?

What are my honest thoughts about Mark Zuckerberg? The subtext there that people may not be picking up is that Facebook acquired my company in 2014. I worked there for a few years on VR. My company was Oculus VR, and then I was fired after giving money to the wrong political group: the libertarians.

But I think people need to realize that—look, I'll put it this way: I took every single liquid dollar that I had and bought into Meta stock the day that they announced they were changing their name to Meta. Mark is the number 1 VR fan in the world. It's a title I wish I could have.

Yeah, I wish I could be the world's number 1 VR boy, but I'm not.

Mark spent $60 billion on AR and VR. He beats me handily, and he's done so through immense pressure from people who don't understand his vision or where he's going. Whatever beef I might have with Mark over other items, in general, I think he understands the future.

I think he's been very practical and pragmatic in his engagements with the government, even to the detriment of his press coverage and the attitude that people show toward him. The thing that I like, and that I've had to come to terms with, is that it wasn't Mark who fired me. It was the apparatus that was under Mark.

One of the things I've had to come to terms with is that the people who ousted me, the people who orchestrated my destruction, who seized my baby from me, aren't even at Meta anymore. It's been 8 years. The people who conspired to stab me in the back are gone.

Can I really be upset at the corporate structure that remains behind those people? Am I mad at their ghost? Am I mad at the ghost of the people who once walked the halls of Meta? My opinions have varied over the years, and this is probably more than I should even be saying about it, but in general, I have a lot of respect for Mark.

There have been times when I've been a lot more upset with him than I am at present. A lot of that came down to the fact that, through a series of unrelated litigation, it became very clear in the discovery process that it was not Mark who had stabbed me in the back. It was people who were much closer to me.

Well, some could say that Anduril exists now because of that action, and the world is a better place—or at least the United States is a better place—because of that.

That's an argument that's been made. The point that I make to those people is this: if a guy got shot in the head by a burglar and then gained superpowers—he became supernaturally intelligent—the guy still shot you in the head, right? I'm not going to say, “Oh, but he gave me superpowers.”

I understand that, but the point that I would make is that Mark was in charge of the company at the time. Imagine this: you're the executive of a major company worth hundreds of billions of dollars. The people you trust come to you and say that the people they trust have come and said, “We have to fire Palmer. There's no other way around it. This is the only way to handle the situation.”

What are the odds that you're going to go and say, “You know what? I think the people that I trust are being lied to by the people they trust. The entire thing is a farce, and they're doing it for purely political reasons. I reject you, and I override this decision 2 levels down”? That's not how the real world works.

At the same time, you've got 1,000 other problems going on that you're dealing with. I hate to say it, but that's probably the decision I would make in my company. If I had people coming up through multiple levels and they said, “This is the only way that this is going to work. Here's what's going on,” my first instinct isn't going to be, “I think you are all lying to me.”

Maybe I'm not saying the people who talked to Mark are lying. I'm saying that when you go far enough down the chain, it's hard to say. I think everyone is actually engaging in an orchestrated coup based on false information to run Palmer out of his company so that we can seize power and get more money out of the performance-bonus fund that he won't get access to if I blow him out. That would be a crazy thing for you to perceive from the top.

As someone who's now running an organization with 4,000 people in it—almost 5,000 people—I'm very sympathetic to the realities of large companies.

Bitcoin: How much do you love it? Do you own it? What are your thoughts on it?

I'm a big-time Bitcoin guy. I've been one from the beginning. I've been mining my own Bitcoin.

People have often asked you, “When did you buy in?”

I didn't buy in. I mined in. I've been doing that since before there were any exchanges. I was on the Bitcointalk.org forums. I sold a banner ad on one of my websites for 700 Bitcoin. I remember very vividly that my website was a little crappy internet forum, and I still did that.

I remember very vividly going to an online Bitcoin slot machine and betting 60 Bitcoin on 1 pull. Nice. It didn't work. I was caught up in the Mt. Gox hack. I lost all of my coins that were in Mt. Gox, and then 10 years later, I got about 13% of them back through the recovery process.

So, I’ve been in Bitcoin since the very beginning. I’m a huge fan of Bitcoin relative to other cryptocurrencies. I’ve often said there are two kinds of crypto: Bitcoin and shitcoin. It’s a long discussion as to why I believe that, but I’m a big fan.

I originally became interested in cryptocurrency because of an essay by Jim Bell on his website, The Outpost of Freedom, called “Assassination Politics.” It was about how he believed cryptocurrency would reshape world politics, the insurance industry, the military, and governments across the world. Jim Bell was later arrested and sent to prison for being a terrorist, and he also didn’t pay his taxes, so he was a very interesting guy. I’m not saying he’s my hero, but I am saying he did predict Bitcoin and many of its impacts back in 1996. That’s when he wrote “Assassination Politics.” I highly recommend it to anyone who wants to read what someone who was very ahead of his time, although on the fringes of society, was thinking about crypto before anyone else was.

Fascinating. All right, here’s a fun one. Would you ever consider—or would you ever buy—a defense prime, like Northrop Grumman or Lockheed Martin?

I won’t rule anything out, but I suspect it won’t make sense. We’re in the same industry, but we’re very different businesses, and our investors are very different. I talked earlier about how you have to attract a certain type of investor and repel another. Their type of investors compared with our type of investors, and in terms of what they want us to be, I think it’s a bit like oil and water.

We’ll team up with those companies. We frequently do. We’re selling rocket motors to some of those companies, and they’re supplying payloads into some of our systems. We’ll work together, but I think to actually bind our fates in that sort of way, it would have to be exactly the right mix. I think that only happens if the world changes a lot.

All right, I’m going to put you on the spot here. It’s a conversation we’ve had over dinner on a couple of occasions. We just awarded a $100 million prize that Elon funded for carbon removal, which was amazing. The winning team had a brilliant approach.

I didn’t see that. I was at TIME100 last week. Is the carbon that’s recovered just stored, or is it turned into some kind of synthetic long-chain hydrocarbon?

Yeah. So, we had 1,300 teams enter that competition from 88 countries. We awarded 6 of them part of the $100 million. One team called Mati Carbon got $50 million, and I handed the guy a $50 million check on stage.

Now, this guy’s amazing. He’s living in Houston, was born in India, and spent much of his life there. They’re actually using a technology for rock weathering.

Okay. So, it turns out that basalt—

Yeah, yeah, yeah.

—and then it absorbs the carbon.

It’s crazy. You just bust up the rocks and they become a fine powder. Are they doing it with an atmospheric process or an in-water process?

They’re basically spreading it on farmland.

Oh, fascinating. I was mostly familiar with marine weathering projects that use the ocean as the carrier for the carbon, but atmospheric weathering is interesting. It increases crop yields by 20% to 30%.

Oh, because you’re pulling in all of that carbon, which also improves water retention.

He’s been building it out in a number of nations, and he’s just going to spend the money. I introduced him to an incredible philanthropist who’s going to 100x what he’s doing right now, so it’s a beautiful one-two punch.

The 100th anniversary of Lindbergh’s flight is coming up, from 1927 to 2027. I’m a huge fan of Charles Lindbergh. I have a signed portrait of him that my grandfather gave me before my grandfather passed away. He was an airline pilot for 40 years, and Charles Lindbergh was his hero. I’ve been to Lindbergh’s grave out in Hawaii.

Well, Eric Lindbergh, who’s one of my trustees at the XPRIZE, is amazing. I’d love to introduce you to him.

I had no idea.

Yeah, he’s the grandson. When I announced the original XPRIZE for spaceflight, it came out of the book The Spirit of St. Louis, right? I was reading about this $25,000 prize, and it sparked the aviation explosion. Lindbergh, the most unlikely guy, pulled it off.

Long story short, Eric Lindbergh and the Lindbergh Foundation want to fund or put together a massive XPRIZE again. So, we’re looking for a big, bold idea—a big, bold idea that we should build an XPRIZE around.

It’s interesting. You asked me the same question, I think, 4 years ago, and I’m trying to remember what I said then. I think tuna farming was one of them.

Tuna farming was one. Basically, large-scale aquaculture of species that are on the precipice. What else did you say?

You talked about upleveling animals.

Uplift.

Uplift. Yeah, I’m still a fan. Can you bring a nonhuman species to human-level sentience? I’m not sure what the right bar is. It’s probably not the Turing test, because even an intelligent species is probably going to think so differently.

Well, you and I are both fans of Ben.

But what if you could get an octopus to an IQ of 100, which is the human intelligence measurement? It’s not far off, probably. Maybe it’s not. I stopped eating octopus because of that.

There’s a question: you don’t have to do this naturally. the octopus is actually one of the hardest because we understand them so little. For birds, we understand what the common traits of the smartest birds, even in a local population, are. You see more brain folding, higher surface area on the brain, and we also know—I mean, you mentioned Colossal—they know how to modify animals to produce exactly those effects. It’s not like we need to come up with this from scratch. If we just take the things that we know make animals smarter—

Another example is dolphins. They have very high-glucose brains, very similar structurally to humans. If you were to make a few choice modifications, you could probably massively increase their intelligence with just a few edits.

People say, “Well, how come they didn’t evolve that way, then, Palmer?” The answer is because that’s not how evolution works, right? You need to reproduce to be fit, not necessarily be smart.

Yeah, right. There were no natural environments that would favor them dedicating even more calories to being more intelligent. Humans developed in a very complex environment where using tools and working as social animals is critical. The ocean is a relatively sparse environment.

One of my favorite ideas is: what would happen if you took even existing marine mammals—if you took a whale and put it in a VR headset and trained it to teleoperate a humanoid robot? Could you train a mammal to interact in a much richer environment that requires tool use and collaboration at a manual, physical level? I’m not so sure it wouldn’t work.

There were some NASA projects back in the 1950s and 1960s where they tried to have various animals interact with people. They raised them from birth around people to see how smart they could get them and whether that would be relevant for spaceflight. I would not rule it out. There’s quite a bit of science fiction that suggests this fun idea that maybe humans are not the optimal Earth-bound species for spaceflight.

It’s not a crazy thought. Well, your homework assignment is to keep thinking about this challenge.

Keep thinking. Keep thinking.

What is a challenge? Come up with another challenge that would spark people to take risks but isn’t so—

One of the people might say, “How about New York to London in 60 minutes?”—hypersonic flight. But the cost for a team to take that on as an XPRIZE challenge is just a fundraising competition.

I mean, another one is probably—maybe you guys haven’t done an interspecies communication prize, have you?

We’ve talked about it, and we’ve been trying to raise the capital for it, but I think that’s a great prize.

Well, what’s interesting is you’re getting to that team idea. The money to tackle something like that 10 years ago would have been just unthinkable. Now it’s 2 guys in a garage with some clever ideas and a GPU and an AI model. You might have seen Google just released—I think it’s DolphinGemma.

Yeah.

They’ve adapted their dolphin translation, and they’re working with the Wild Dolphin Project. I’ve actually given a lot of money to those guys over the years.

If you want to do that XPRIZE, we’re ready to run with that one. Now, here’s the question: are you guys going to prohibit me from using the dolphins we translate? Is there going to be a prohibition from inducting them into the United States Navy?

Absolutely not. The Navy has a large dolphin program.

Not mega-large. They’ve spent a lot of money over the years and have a small dolphin program that consumes a lot of money.

Probably the best experts in the world in terms of dolphin psychology are actually in the United States Navy. I do think an interspecies prize—for dogs, for birds—could you imagine the commercial potential of being able to talk with your dog? It would be a trillion-dollar company just like that.

So, another AMA question here is: Will aging warriors be able to keep fighting using robotic technologies?

Absolutely. One of the interesting things about special forces is that they actually tend to be much older than conventional forces. People often don’t understand that. They imagine that these must be the youngest guys at the peak of their athletic prowess.

It turns out that what you more often need in special forces operations is people who have unique experience—a lot of hard-fought, hard-won lessons implanted in their brains. The thing that takes them out is that you still need a certain level of high physical competence and excellence to survive on the battlefield.

I think it’s almost inevitable that, as you have more and more resources shift to robotic systems, remote systems, and exoskeleton systems—well, I think it’s exoskeleton systems, but I’m not even sure that it’s about putting old guys into exoskeletons. I think you might have more of a wizard approach.

A wizard doesn’t fight through strength. He doesn’t imbue his limbs with force so that he can use a sword. He fights through other means. He fights at a distance. He perceives the battlefield well enough that he can act in other ways.

I suspect that, if we do our job right at Anduril, we should make physical prowess maybe not irrelevant. You still have to be able to walk around. You still have to be able to get in and out of your car. But I don’t see a reason you couldn’t have someone who’s much older, or who has a missing limb or missing limbs—people who today can’t operate effectively. I would not at all be surprised to see them be able to stay in service much longer.

The question then becomes: How do we keep them in? Right now, it’s really hard to keep people, especially those who have had life-changing injuries; people who are getting much older; and people who maybe want to focus on raising families.

It’s two things. If we’re going to keep that experience in the military, keep those guys, and maximize utility, we need to make tools that allow them to safely keep operating into later life. We also need to figure out how we can pay these guys and give them good enough benefits that they don’t depart the armed forces for very practical, pro-family reasons.

At the end of the day, most people want to do well by their families. We can do a much better job of keeping and retaining those people. We just have to give them better benefits. We have to pay them more. You do that, you’ll keep them.

All right. Next question from X: Could a neural-linked trigger finger fire faster than your nervous system?

Absolutely. It’s without question. There’s an enormous amount of latency in the link from your brain, all the way through your peripheral nervous system, out to your finger.

I’ve talked about this several times, but years ago I actually built a peripheral nervous system bypass to test exactly this. I wasn’t going directly to the brain, which would be fastest. I was just triggering it off of a muscle that was basically a jaw muscle.

It turns out that your jaw and tongue muscles have much lower latency than your fingers, all the way out at the end of your hand. The nerve transmission velocity is much higher to here, and the physical length of the link is much shorter.

You actually need very good control of your tongue and your mouth not to bite your tongue. Try chewing sometime. Notice how crazy it is that you’re basically opening your mouth, shoving food into the hole with your tongue, and then, as you bite down, your tongue pulls out just so. You do it hundreds of times in a meal without even thinking about it. That coordination is crazy.

What I did was make a system that would let me click my mouse on my computer as a proxy for trigger-finger movement by flexing a muscle in my mouth. In doing so, I greatly reduced latency in playing first-person shooters. It turns out that it totally works. You can trim a lot of your reaction time right off by just using different muscles, and that’s not even directly to the brain.

It’s also worth noting that you don’t just have to go to the brain. Nervous signals are a mix of chemical and electrical signals. Those of you who’ve been in high school chemistry might remember that most chemical reactions happen at an accelerated rate when you raise the temperature.

One thing you can do to reduce peripheral nervous system latency is just heat up your arm—the whole thing. If you soak your arm in hot water to a very uncomfortably high temperature, it will be very uncomfortable, and your reaction time will actually improve for clicking a mouse or pulling a trigger.

I’ve actually pondered the idea for years of a product like the Magma Sleeve or something. You’re playing your game, it’s down to the last round, and you’re thinking, “Oh no, I really have to pump it up.” It would superheat your arm to the point of causing first-degree burns if you did it for more than a minute, but it would give you that little last bit of extra edge. I think that would be a really interesting product for somebody to make.

Another question related to video games: A lot of kids are playing a lot of video games today. How do you feel about that? How do you think about it for the next generations coming up? Is it a good thing or a bad thing? Was it valuable for you, or is it distracting from education?

For parents who are like me, with teenagers who love their video games and are focused and obsessed with them, what’s your advice?

I struggle with this because I love video games. I started the ModRetro forums, the game-modification and game-console-modification community, when I was a teenager. We just launched our first product after 17 years, which is a clone of the Nintendo Game Boy Color. I love games. I spent a lot of time as a teenager playing games—thousands of hours.

On the one hand, it makes me worried when I see the current generation spending all this time playing Fortnite, Minecraft, and Roblox when they could be doing more productive things. But then I remember that I did the exact same thing.

It’s hard to know where the value is. There are connections being created that are valuable in other contexts. Are they more valuable than the things they could be doing if they were playing sports or reading books?

I feel like I’m just becoming the old person. On the one hand, I say, “I’m not going to let my kid play games like I did. All these kids today, they’re all iPad babies.” But then Socrates supposedly said, “What are they to become? You look at the children of today. They have no respect for their elders or their society. They riot in the streets, inflamed with wild notions, pursuing their own desires. What is to become of them?”

I realize that people have been saying this for 2,000 years. They say, “This new generation—what’s going to become of them?” They seem to generally turn out fine. I don’t feel like I’m in a place to speculate beyond saying it’s probably going to be fine. It’s probably going to be fine.

You’re up on the TED stage giving your most excellent talk, and for those who haven’t seen it, they should take a look at it. You’re wearing these glasses, and there’s a lot of speculation about whether or not your speech is being fed to you in the glasses. Were you cheating?

I was cheating. I had my notes up on my glasses. The hardware in particular is made by a company called Even Realities. I was wearing a pair of Even Realities G1 glasses, and it’s a really remarkable product. They’ve done a great job of making smart glasses that really do look like normal glasses.

The arms are very, very thin. It kind of hides the battery and interface back in your hair, which I have huge hair, so I hide most of that bulk. The lenses look like normal lenses, and it’s giving you—not just in one eye—a full stereoscopic little window that’s green only. It’s not full color, but it has a really great function that can show you your notes and your script.

You can pull up critical information in messages. If someone were to tell you, “Palmer, slow the fuck down,” you could have that message pop up and just see it. I’m not saying I got a message like that, but if I had, I’d be able to.

For people who don’t know, TED doesn’t allow any teleprompters. They want you to memorize your whole talk. They want you to just memorize the whole thing.

I’m pretty good at this stuff when I prepare, but I have to admit, especially when you talk about specifics—like the number of bombers that were built in a specific year per minute, or the specifics of some technical item—it’s really nice to have your notes up so that you can refer to them and make sure you don’t say something that has everyone making fun of you for a year.

People pointed out, “Oh, what about Palmer, where he said that China has 200 times instead of 300? This guy barely knows what’s going on.”

I’m a huge fan of augmenting human capability. When you expand your capabilities beyond what you were born with, you can kind of extend yourself out into your phone and your wearable glasses. You augment your vision. You augment your haptic perception. I’m living what I preach.

It was especially funny when there were people saying, “Why would Palmer wear something like that when it makes him look so dumb?” All I could think when I was reading these criticisms was, “Have you seen me? Have you seen me? Do I look like the kind of guy who wakes up in the morning and says, ‘Okay, first things first: What will people think of my outfit today?’”

I’m glad you’re wearing flip-flop shorts and a Hawaiian shirt on stage.

I mean, look, here’s what I’ve realized: I’ve been doing it for so long, and I get away with it now. I think the point that I usually make to people is, when you achieve success, you earn a certain level of eccentricity that is allowed.

If I want to be eccentric, that’s okay to a certain level. I’ve decided that I’m going to put all of my allowed eccentricity points into my mullet and into my clothes so that I can focus on other things and look how I want.

I’ve always wanted a mullet my whole life. How long had I wanted one? My whole life.

When did you start?

A few years ago. My mom would never let me get one. Then I started dating my wife, Nicole, when I was 15, so we’ve been together for a very long time. We met at a debate camp at a law school in Maryland or Virginia, as teenagers.

She didn’t want me to get a mullet. Then, when we got married, I realized that there was nothing she could do. I thought, “Oh my, I can get the mullet. She can’t leave.” So I got the mullet, and luckily, she actually likes it now.

Although, I’m way overdue for a haircut. It’s out of control. I’ve transcended mullet man to homeless man. But again, that’s the eccentricity that I’ve earned.

I love that. Is there a favorite principle or mental model that you live by, a guiding principle?

There are so many. Marcus Aurelius had a whole bunch of stuff in Meditations that speaks to me. I wouldn’t say I’m fully in his philosophical camp, but you can pick and choose a few.

Probably the one that I like most recently is what Roosevelt said: “It is not the critic who counts; it is the man in the arena,” the one who actually bets it all, the one who actually sacrifices and gives, and inevitably will fail. He will get beaten up and bruised, but it’s his contribution that counts, not the people who stand on the sidelines, not risking anything, picking apart everything he might have done wrong, whether it’s wrong or not.

I like to remember that, especially when things do go bad, because they don’t always go well. Sometimes you flip a coin and you get tails.

Along that line, one of my pet peeves has been that so many incredible people who have created tens and hundreds of billions of dollars of wealth are sitting on the sidelines and not betting it to make the world a better place.

You talk about the people who are cruising the Mediterranean on their yachts. They’re investing. I’ve always wanted to take out a New York Times page—I hate the New York Times and all traditional media—but these are the people who are working to make the world a better place. These are the people with the biggest yachts.

Brutal. It is brutal. I’ve asked myself a similar question.

I’m in a group chat called the B Boys Club, and it’s all boys who have sold a company for at least $1 billion. That’s the membership criteria. It doesn’t exclude women; women would be allowed. It’s just that, thus far, only boys have applied.

For years—I got invited to the B Boys Club when I sold my company—I’ve tried to shame people and said, “Guys, you have so much money. Why are you not doing what you know is the right thing to do with this capital? Why are you not acting according to whatever your system of values is? I’m not even saying, ‘Do what I want you to do.’ Why are you not doing what you know you should be doing?”

Some people have told me, “This really changed my thinking. I should be doing what I know is right.” There are other people who have said very clearly, “You know what? Racing old vintage race cars is extremely fun, Palmer. I’ve paid my dues. I’m not in it to stress like I used to when I started my company.”

My point to them is: Are you going to be Batman, or are you going to be you? I’m not saying everyone should become a vigilante crime fighter, but why wouldn’t you use your resources to do the thing? It’s your intelligence, right? Because money—giving away money—is the easier part of that.

That’s a great point. Money, intelligence, network, and reputation—money was a small part of why I was able to start it. It was because I was able to raise further money. It was because people believed in me, because I was a proven founder. I had started a multibillion-dollar company, and I had successfully exited it. That makes it easier to recruit people.

You’re right. These people are in a position that no amount of money alone could replicate. You could take some guy off the street, give him $10 billion, and he won’t be able to accomplish half of what some of these people would be able to accomplish.

I don’t want to pick anybody in the B Boys Club, but imagine what would happen if—I don’t know—who would maybe be a good fit? Who’s a tech founder who’s not really doing anything anymore?

I’ve got examples. Imagine that guy with $100-plus billion. What if they announced they were starting a new company to solve some problem, hiring the founding team, and building another multibillion-dollar company? They would attract the greatest players on the planet instantly.

Instantly.

Yeah. That’s almost a free resource. You don’t even have to spend your money. You’re just betting your reputation.

What do you do? Do you leave your money to your kids to ruin their lives? What I love even more is the irony of the Giving Pledge. I’ve had this conversation with Gates; I haven’t had it with others, but you’re pledging to give half of your money to a nonprofit before you die that could sit on the money and do nothing with it.

You’re basically betting that other people can do more good with your money than you can. You’re saying, “I think my view of the world will be more competently executed by others than by myself.”

Maybe there are people who are really like that. Maybe some of them are truly mentally or physically not what they used to be. But for many of these people, I think they absolutely could achieve their goals better than handing it to a non-founder NGO and a bunch of lawyers to go and correct things.

To be fair, I think Bill Gates has done that. Bill Gates has done that for sure. Despite signing the Giving Pledge, he has also gone and said, “Here’s my stake. We’re going to eliminate this disease. We are going to achieve this level of carbon capture per dollar.”

That’s actually pretty cool. It’s very real.

For everybody else, for God’s sakes, commit yourself to using the wealth.

That’s why I haven’t done it. My thinking is, look, I’ve got a view as to how the world should be, and I don’t think there’s anyone else who’s going to more faithfully execute on that than me.

My proposal is that, when I’m old and used up, maybe I’ll change my mind. Hopefully, we’ll have some good longevity products by then. But my view of the world is that, instead of the Giving Pledge, I want people to do an impact pledge.

I pledge to eliminate child slavery, or hunger in this country, and then—or die, or go bankrupt trying.

There are some amazing people, like Tony Robbins and Marc Benioff, who do this. You call your shot, and then you invite others to come and join you.

Imagine a list of all of the impacts around the world being worked on. You can measure the results by the quality of the outcome rather than the number of dollars put in. Right now, it’s, “Oh, well, he put $1 billion into this thing.” It’s really hard to track that outcome versus saying, “I will end this disease.”

I mean, because you can only spend so much money in your lifetime, right? We have the ability to do such extraordinary things as humans and solve so many problems, especially in the time we live in. I know you do. It's so great to wake up and realize, wow, there are things I can do today almost trivially easily that would have been the work of a lifetime just a few decades ago. Isn't that extraordinary? The stuff you did between breakfast and lunch would qualify you as a god 100 years ago.

Yep.

It's so easy to lose sight of that when you fall into the human routine and say, "I have these problems. I'm struggling to deal with this situation. I'm having this family problem." Those are all very true. They're real, but there are some people who feel like, "Oh, man, I was born in the wrong generation. I can't imagine being born any other time." This is the only time. More exciting today is tomorrow.

Yeah, yeah. Of course. I like to imagine what it must have been like to be born in the age of exploration. That's the only other time. But even then, I don't think I would make the trade.

I could see you with an eye patch and a sword.

I could. Look, what guy hasn't watched Master and Commander with his bros and said, "Oh, man, that is something else"? Just imagine us on the high seas having adventures. How great would that be? True discovery, not knowing what culture would be on that land over there.

Well, my plan is to die on one of the moons of Jupiter right now. I've reserved the right to change my plan. It's not my life drive. I'm not like Elon, where I must get to Mars. But the thing I would like to do, all things being equal, right now, would be not to die on Earth, but to die on a reasonably colonized, reasonably terraformed moon of Jupiter or elsewhere in our solar system.

Nice. I feel like setting my sights on another solar system is a bit much. The Jovian moons have really high radiation belts.

Yeah. Look, if you have enough nukes, it's not a problem. Just generate a synthetic magnetic field. Bam, you're all set.

All right. I don't know if it's that easy, but I'm just making it up on the fly. I'm sure people a lot smarter than me are going to figure it out. But I'm going to need to make a lot of money if I'm going to buy Jovian real estate. I think it's going to be in high demand in our lifetime if everything goes well.

I love it.

Oh, no, no, no. I want to live in a nice gated Jovian community with a nice HOA making sure the oxygen stays on and making sure our plutonium prices aren't through the roof.

Yeah, that's the type of world I hope for.

Oh, no. I'm a 1G guy. I want 1G. I want my full bone density. I want my normal metabolic process. Maybe 0.9G; that could be kind of fun. But having read up on the impacts of low G, I'm—do you know Gerard K. O'Neill and the work that he did at Princeton?

So he designed these large rotating space colonies that were basically cylinders. I'm familiar with most of the weird designs. You've got tin cans on strings, you've got the big rings. The big thing, of course, is that at the center of rotation there's zero gravity; on the outside there's 1G, so as you get older, you could sort of move up a mountainside. That's a fun idea.

Yeah. I had a thought for a reality television program at one point. It's called Fat Flight. You've seen My 600-lb Life. It's a show about people who are 600 pounds or more, and they try to motivate them to lose weight, exercise, get gastric bypass surgery, and regain control of their lives. I had an idea for a show called Fat Flight: You would take really, really obese people—these are people who are immobile, who can't even walk around—and you would put them on one of those zero-gravity flights.

I did this. I founded Zero-G, right? The commercial operator here. There was a television show called The Biggest Loser.

I'm familiar. I don't watch it frequently, but I'm familiar with it.

It was for people losing weight. At the beginning of their season, we took, I think, 8 Biggest Loser candidates who were 300 pounds or higher into zero G.

Ah, so—see, here's what I wanted to do. I wanted to put them in a zero-G plane, build a living room in the plane, and then fly at 0.1G or 0.2G so they could stand up, walk around, and experience normal life. Imagine what it would be like if you could walk. Give them 1/6G.

Oh, we gave them lunar gravity.

You already did this, then.

You can go find the episode someplace, man.

See, this is very much aligning with what I usually say, which is that none of the ideas I've ever come up with are something that someone hasn't already done before. You've already done Fat Flight. Bummer.

Yeah. Well, my favorite flight—I did hundreds of flights. My favorite flight was taking Stephen Hawking up into zero G.

That's got to be fun. How did it end up being a 727 that you guys bought? That's always my favorite airplane. I want to fly around myself in one someday, but I haven't figured it out yet.

It was interesting. We looked at 737s. We looked at 767s, but the wide-bodies were too expensive for us. The problem with the 737 was that the fuel lines between the wing tanks and the engines were very short. The 727 had 3 advantages. It had the rear airstairs, which I loved because we could load and unload there, and you didn't need any infrastructure wherever you're taking off, wherever we go.

The second advantage was centerline thrust. There are 3 engines in a 727: engines 1, 2, and 3. When we were going into a parabolic flight, we would take engines 1 and 3 back to neutral, and just throttle engine 2 to have the perfect amount of thrust overcoming drag. The third thing was the fuel lines. When you're descending, especially, you need to match 2 737 engines exactly, and the fuel lines between the tanks in the wings and the engines were long enough that during the zero-G portion, the fuel in the lines fed the engines and you didn't run out.

Our original business model was that we were going to be using cargo airplanes and palletized interiors. Cargo airplanes fly at night. FedEx and DHL were flying at night, and the airplanes were sitting on the ground during the day. In fact, I think UPS has a whole fleet of 727s that sit at Ontario Airport, not used for most of the year, or they only use them during the holiday season, so they're sitting unused. I tried to negotiate with all of the players out there and finally said, "No, we've got to buy our own plane." So we fell in love with the 727 the same way you did. It's a beautiful plane.

It's a brick shithouse. It was made by a Boeing that knew how to really make airplanes for pilots. It's fast, too. You can cruise at Mach 0.97.

Yes. What an incredible plane.

It's my dream to own a 727 and put Volvo RM8s on it, which was a license-built derivative in Europe of the Pratt & Whitney JT8D.

Yeah, the JT8D.

There was a version of the engine in the 727—one of the later engines—and the difference was that it was built by Volvo with an afterburner on it. So it would be a direct one-to-one swap, and I could have a 727 with triple afterburning engines. I think that would be the fire plume out the back. Wouldn't that be the coolest plane ever?

The only runner-up would be—Aeroméxico had 727s with real FAA-certified rocket boosters. Have you seen these?

No.

When they would take off hot and heavy out of Mexico City and an engine went out, they wouldn't be able to maintain altitude. So they actually built in solid rocket boosters that could be activated only by an emergency pull in the cockpit. They could pull it, and I think 4 boosters would allow it to maintain altitude. The FAA certified this, which is so cool.

One of my dreams is to find one of those old Aeroméxico 727s and restore it. They would never certify it today, but all of those STCs would be grandfathered in, and I would then have a nice rocket-assisted 727. Small aside: What are you flying these days?

I'm mostly a rotary-wing pilot. I own a few helicopters. I've got a UH-60 Black Hawk, an Agusta, and a Robinson.

When can Lattice become operational for pilots?

Already.

No, no, no—commercial pilots. I want my heads-up display. I want to be able to see fields in the distance. The entire ATC system is so ridiculously broken.

I have to admit, one of the nonmilitary problems I would love to work on would be modernizing ATC. It's unclear exactly what's going to happen.

President Trump has said he's going to build a beautiful system like nothing anyone's ever seen. I would love to be part of that. I'm not sure if it will end up making sense for us to do so. But, to your point, we should be giving pilots full synthetic vision and full awareness. They should know everything that's going on.

You should never have, for example, military helicopters running into aircraft. No, that's just—it shouldn't even be within the realm of possibility of that happening. It's kind of crazy that, with all the tech that we have, things like that are still happening.

We're building things for the military that I think are oriented around avoiding those types of situations: day, night, weather, you name it. I would love to see that tech flow into this. That is one place where I think tech could flow back into the commercial world. That's one where I think it could.

The way I would justify it is that these things are operating in close proximity. Making manned aviation safer and commercial aviation safer does actually make military aviation safer. The two are so closely intertwined in terms of operating out of the same airfields, operating out of the same airports, and using a lot of the same support infrastructure. Safer airports make for a safer military.

So that's maybe how I'll justify doing something I wanted to do anyway.

Anyway, buddy, listen, thank you for your time today.

No, this has been a lot of fun, and thanks to everybody who sent in questions. It's fun to go through a long list, but I figure it's a good place to break.

Just outside of this media room that we're in are our beautiful devices connected by Lattice.

Well, their form follows their function.

Yeah. It's very interesting how a lot of the things that we build end up following natural form.

People say, “Oh, wow, this looks like some sleek predator.” It's like, well, it turns out that predators have certain characteristics inherent, and it doesn't matter whether they're biological or technological. It turns out a lot of them share the same characteristics, whether they're moving through air, water, or land.

I remember I was with Burt Rutan at Scaled Composites, and he was putting up an equation for drag, C_D,B, and I said, “What is that?” He goes, “It's the coefficient of drag due to beauty.” Hey, good-looking airplanes are good-flying airplanes.

Anyway, always a pleasure.

Always a pleasure, pal.

Thank you, everybody.