Ep 153: Palantir CTO Shyam Sankar on Heretics, AI Weapons, and Rebuilding the Arsenal of Democracy

21:0894 % of the money for major weapon systems went to dual purpose companies, companies that were invested in our national economy, but also our national security. And we've kind of forgotten the stories of their involvement. Like Chrysler, yes, they make minivans. They also make missiles. They are the prime contractor on the Minuteman, the original Minuteman missile. Ford made satellites until 1990. General Mills, the cereal company, made torpedoes, an inertial guidance system. That sounds like crazy today. It sounds And you can say, why? Why is that even rational? Well, General Mills, the cereal company, had a mechanics division to make machinery to process grain.

21:42Everything they learned about building that type of machinery, they were able to repurpose it into machine tools and factory production that subsidize our national security. We lived in a world where every car, camera, and cereal box that an American bought was actually subsidizing the R &D against our national security. And national security is not an end unto itself, it's a means to an end. And that that end is the economic prosperity of America. And so getting those things to be aligned, I think, is really critical. And in particular in a world where so much of the R &D spending from the 60s was driven by the government, now the vast majority of the spending is driven by the commercial sector.

22:18You wrote that the Pentagon started breaking in 1961. What do you mean by that? We brought in a brilliant businessman, Robert McNamara. But I think one of the things we really got wrong is that McNamara ran Ford at an era when Ford was entirely supply constrained. The demand for cars was essentially unlimited. Any car that he made, they could sell. That is a distorting school to go to, right? It teaches you actually that the only thing you need to do is manage your supply chain, grind your suppliers, focus on efficiency, not on effectiveness. There was not enough competition. So actually the auto industry that McNamara learned all his lessons in, when it met Japanese competition in the seventies, got destroyed and had to reinvent itself.

23:01But the Pentagon didn't reinvent itself based on all those lessons of Japanese competition. And so it kind of started this steady process. It would be intellectually dishonest to place all the blame on one person like McNamara. I think even just stepping back structurally, defense is a monopsony. It's the opposite of a monopoly. A monopoly, you have one seller of a thing. A monopsony, you have one buyer of a thing. Yeah, but it's the only buyer. And it's really hard to be a buyer if you really believe in the free market. it's the fact that you have so many disparate customers with so much signal, you get to aggregate it, you get to work against it.

23:32If one customer's low vision doesn't want to buy your thing, there might be a hundred other customers who have the vision and do want to buy your thing, right? But in a world where you have one buyer, now you have this single point of veto. And I think our history would suggest as part of the fun part of writing the book was documenting it. I talk about the, I call them heretics and heroes. If you really look at the history of defense innovation, almost none of it happened because of the system. All of it happened. It's always the heretics who break the rules who do it, right? We can go all the way back to Winston Churchill.

23:59So Churchill, as the head of the Royal Navy. Yeah, World War I, before World War I. Yeah, exactly. Built the tank. How crazy that the British Army was like, I got horses. I don't need this tank thing. Go away. And he built landships as a rogue renegade effort. I love that he called them. It had to be landships. Of course. It wouldn't be in his job jar. It's one of my favorite books, actually, is when he writes about before World War I. And it's like, he just like fired all these bureaucrats and pushed everyone out of the way. I feel like it'd be harder today to have someone that strong, able to push through, but it's needed, huh?

24:28Yeah. You build up, you know, the entropy of the universe is against you. You, you build up these bad practices. And I think part of his understanding is like, it's not an ad hominem critique. It's a structural critique. The other example I love is Andrew Higgins, this foul mouth, Scots Irishman down in Louisiana. He was a huge, anyone from Louisiana knows Higgins, you know, he, he, he was a boat builder. And in the, in the twenties, he observed how Chinese bootleggers were able to build these shallow draft boats to kind of outrun the policing. And he used that to inspire his own boat building.

24:59So Higgins, he built the boat. The Navy wouldn't let him compete. Then finally, there was an enterprising Marine first lieutenant who got him in. He won the competition. Then the Navy tried to steal his IP and build the boats themselves. They couldn't even steal the IP. In the end, 92 % of all boats in World War II were Higgins boats. 92%. The boys would not have landed at Normandy or Iwo Jima without the Higgins boats. I mean, you see this even in the Civil War. You had like, I have upstairs a Gatling gun with the patents from the 1860s, but they weren't used because people were ignoring them because they didn't think it made sense.

25:30Like the renegades would figure things out that don't always get in, I guess, until later. Yeah. And it's equally important to have the renegades inside of government. Like some of the fun stories are like Higgins on the outside, but you had Admiral Rickover who built the nuclear Navy. When Rickover wanted to build a nuclear powered submarine, Oppenheimer himself told him, I don't think this is going to work. Imagine the chutzpah, like in the face of Oppie telling you this, you're like, no, I think you're wrong. I'm going to do it. And he did it in six years. I love it. I love it. What are some of the stories?

25:57Is there anyone with Palantir itself got to be a renegade and push something through? Well, the fun part was documenting for the first time Colonel Drew Cukor's story. So here you have this guy who grew up in Southern California, single mother, Mormon. uh he was didn't have a lot of money so he went to rotc to go to college joined the marines and he became the founding father of really maven the intel the the ai enterprise kind of marquee project and kind of looking at the amount of bullshit this guy had to deal with you know his nickname was the iron dome of pentagon bullshit this whole project started it in the basement of the pentagon which is kind of the equivalent of a garage in palo alto i guess for for these sorts of things and now it is one of the most successful programs.

26:42But he suffered personal and professional slight, IG investigations, all completely made up horseshit because the bureaucracy was threatened by the amount of success that he was really having here. And it's a great lesson of like what it actually takes to be a heretic, how much you have to be willing to sacrifice for the ultimate outcome at the expense of your career. It feels like there's so many stories of heretics, like with the airplanes from the 30s and 40s with Lockheed's special groups early on. And like, how come they don't just like stop the bureaucracy from trying to kill heretics anymore?

27:13Like, why is it always like almost win and almost stop all these things? Well, that question you just asked is why I wrote the book. We have to remember that we need the heretics. We have them, by the way. It's not like we're missing them. It's like heretics require two things, the space for their heresy to exist and a little bit, a modicum of protection from the top to enable that to happen right here. And so you see a little bit of protection. Yeah, exactly. I feel, I feel like this is like every company needs to do this for their most talented people. Cause the, cause the organizations always reject the talent and your job is leader to keep them.

27:44Like we were, I mean, Palantir was a bunch of heretics early on, right? It's basically like, like, I felt like I was like the leader of a bunch of heretics and we were just like beaten by the mainstream people for years and years and years. And so it's really fun. We're now, it's very funny now to be attacked is like this, like in military industrial, like leaders now, cause we were just like beaten by the bureaucracy forever as these outsiders. Like, do you still feel like an outsider? Are you an insider now? Because you're a lieutenant colonel and Palantir is winning. Like, is your mindset changed?

28:11Constitutionally, I'm going to always feel like an outsider. I always feel like I'm the heretic looking in. Maybe I relish that role a little bit, which is why it appeals to me. But absolutely, in the beginning of Palantir, really, like, we tried to go through the front door and very much got, you know, the front door was not an accessible door. So disrespected. So disrespected. Like, people just wanted us to die. There's a positive piece of that. You know, you carry that chip on your shoulder to like earn your place every day. And you recognize your product has to be 10 times better than the alternatives.

28:39And, you know, there's like a fire in your belly from it. But it was, yeah, it's, I think it's, it's motivating. It's motivating to know, look, why should it be easy for us? It wasn't easy for any of these guys. And so chapter 11, you said factory is the weapon. What's your framework for reviving manufacturing? I'm always, I'm working on this in a few ways. We're starting to scale up some things. We still have a lot of work to do. How are you thinking about this? You can kind of break down like what are the laws of physics and what are the laws of man? And ideally our factories are marginally if, how do we break down the laws of man that are constrained in the factory?

29:11How do we qualify this part? How do we do this? This is just the way we've done it. You know, reimagination of these processes. I do think, you know, I'm a one trick pony in the sense that I think everything starts with the people. And if you look at, I don't think it's fair to say that America is not great at manufacturing. Like the, look at SpaceX, look at Tesla. Like we have modern manufacturing marbles. They just think about manufacturing differently. So how do we bring a kind of software continuous iteration, co-locating R and D with production mindset, a bias towards vertical integration to really get after these things, squeezing out the deadweight loss that happens through what I think is the legacy of globalization of like, Hey, we're going to just really like push out our supply chains.

29:50We're going to optimize on cost instead of innovation. That's kind of the precursor starts with that. Then recognizing the Ukraine example is really important because what we got wrong to summarize it would be we thought the stockpile was the deterrent. It's not the stockpile. It's the ability to generate the stockpile. And this is what the frog boil that happened, which is like, well, how many of these weapons do we really need? We're not using them. Maybe we can have a little less, a little less. Let's make it a little more fancy. Let's make it more exquisite. Well, now that it's more expensive, we can afford even less.

30:19and you kind of got into this doom loop. And a recognition that, frankly, in World War II, the Germans were superior engineers. We just could produce more. And that counted for a lot. And they produced small, exquisite quantities. So today we look like the Germans and the Chinese look like we did. Which is why we got to fix that in 100X it. Well, we're working on it with Sironic and others. Are you still getting any companies in particular you're working on, you're excited about for scaling up manufacturing? Well, I think a lot of the new entrants are going to have a huge role to play. Hadrian, Saronic, Andrel.

30:51But I also think reinvigorating the existing industrial base is a huge part of this. We have a huge number of smaller, very small, they're not even medium-sized enterprises. They're small businesses that make parts. They're coming up on a succession challenge where the owner, it's a sole proprietor type thing or limited partnership, and it's going to need to be transitioned. You need kind of outside capital to reinvest in the machine toolings and modern processes. So I think we're going to have to get both of these things to work. Certainly, we're going to need a high-low mix. Our exquisite stuff is exquisite.

31:23As we see with Epic Fury, this stuff really works. The real problem is that we don't have a hundred times more of it. And we're going to need to invest in scaling the low end of this as well to kind of fight the future war. A few more questions on defense. If you were in charge of fixing the culture at the Pentagon to protect heretics and to make sure we're scaling all this stuff up, what are the number one or two things you'd focus on? You know, I think culture follows from execution. And, you know, Bill Perry, who was a deputy secretary of defense during the last supper and subsequently the secretary of defense, like he had this fork in the road where he could think about, I could try to either fix the system or I can make sure that stealth and GPS happen by going around the system.

32:02And he chose the latter. And I think that's the wise move, which is like, how do you deliver the incredible successes that motivate the next generation of heretics that set the path for those who are willing to see it. Because if you try to fix the system, you're going to lose, you're going to get bogged down in trench warfare. And, you know, I think you got to fix it through inspiration. Isn't this always a problem though, that it's never worth fixing these really hard problems. The government just as we just always just go around them. And so this thing just gets nastier and nastier and more and more broken.

32:29I think that is a problem. I think that what I would offer you though, is like, can you fix the system by literally trying to fix the system or do you have to fix the system by getting parts of it to work? Kind of like eat your way through the elephant. I always thought we should just fire a ton of the core and just build new things around it, but that's apparently not appropriate. There's a Chesterton fence question there too, though. What do we need that we don't know we need? And so I think a lot of it comes down to people. There's so many tens of thousands of people doing things that don't seem necessary to me.

32:56Am I crazy? We're not as efficient as we could be. The only pushback I'd have on that is we built 154 different airframes in World War II. I think probably 10 of them really mattered. Yeah. So part of, part of what's gotten us to this situation is a fixation on an efficiency over effectiveness. Like you, it's a pathing question. You have to get to effectiveness first, then you can work on. Just build the most effective things around it and don't worry about trying to take it, take it apart. Yeah. It's very much like we manage everything as if it's a one to N thing when sometimes there's a zero to one thing and those things have to be managed differently.

33:32I guess I'm probably better at zero to one anyway. I should just do more zero to one things around it and not worry about this giant mess in the middle. Oh, is the, is the right strategy. Yeah. Well, I think the zero to one things as they scale end up replacing and fixing it, right? Like it is easier to fix and lead with momentum than, yeah. I have to ask you about some of these like magical AI weapons that we're using right now. I know you can't tell us everything, but this has been interesting to watch the war from afar. And I think Israeli fighter jets supposedly in the first 20 minutes hit 500 targets.

34:00And obviously Palantir is involved very much in helping take all the data and decide things there. What are you allowed to tell us about how these things work? Unfortunately, not a huge amount, but I think what you could say, there's a way in which there's nothing new under the sun. It's just the OODA loop. John Boyd's Observe, Orient, Decide, Act. And whichever side is able to go through the OODA loop fastest is the side that's going to win. And so you can think about the application of AI to the doctrinal targeting process as about providing an advantage to cut through the fog of war and to be able to apply effects to your adversary before they understand even what's happening and respond to it.

34:38A way of tying national intent to the targeting and fires process. And I would say Epic Fury is a leapfrog in capability, but I think it also shows that there's another 10X, 100X to go that's illuminated by what's already been done here. So as AI improves and as the infrastructure improves, we can actually do an even much better job next couple of years. Oh, yes. Sorry, enemies better not screw with us. That's deterrence. And that's the whole point. That is the whole point. Let's look ahead. The clash between Anthropic and the Pentagon over the cutting edge AI is obviously a preview of some of these kind of tough questions.

35:13You can imagine AI is a hundred times better and there's people working on it. And it's, and it's interesting questions around like what you want the government to do or not. You have obviously some of the very best minds in the private sector these days working on this. It used to be those best minds were in the government. So it's like, it's kind of more aligned. Like how's this dynamic going to function where the best minds are outside of the government, but then having to work with it and might have different opinions about things? Well, one of the things I really hoped helping stand up Detachment 201, you know, how do we build more of a bridge between Silicon Valley and the department?

35:43So Detachment 201 is what you're serving as a lieutenant colonel. That's correct. Yeah. And I think part of it is like, we have to walk many miles in each other's shoes to understand these things. You know, I think there's a lot of historical analogs that suggest if you, you know, if you really believe in democracy, you need people making these decisions who are accountable to the American people. They're elected or appointed and accountable to elected officials. If you go back, the analog I like historically is thinking about Theodore Hall. Theodore Hall was one of the youngest scientists in the Manhattan Project.

36:14He joined the project at 18 years old, right out of Harvard, genius physicist. He had an older brother, Edward Hall, who was actually the inventor of the Minuteman missile. But there were two types of traitors in the Manhattan Project. There were those who were committed communists, like Klaus Fuchs, always a spy, got him to the program. Theodore was not that sort of traitor. He was naive. Theodore thought, look, I'm a genius at physics, so I must be a genius at geopolitics. And in his conception of the world, the only way to have world peace was for two countries to have the bomb. So without any penchant for communism or even an ill will towards America, he thought this was his moral duty, was to walk into the Soviet trade mission in New York in 1944 and give them secrets to the bomb.

36:57But what happened was not world peace. Every death since 1949 from communism, which is the year the Soviets got the bomb, you could say that blood is at least partially on Theodore Hall's hands. And so I think we live in a world where many smart people struggle with epistemic humility. What are the limits of my knowledge? How do I have some humility around that? Because the alternative, you could say Theodore Hall was essentially tyranny by tech bro. He decided global policy and it had huge consequences. On the other end, there's a true deep commitment to democracy, even when it feels uncomfortable.

37:29How do you think about the future of warfare with AI and like prosecuting war without humans in the loop? I mean, obviously there's things where right now you're like, if something's incoming, you're going to want the AI to target it and destroy it for you. And I'm learning a little bit about this from Sironic and other companies I'm involved in. If you have hundreds of, you know, ships having a war and China has hundreds of ships and they're reacting in real time with AI, you're probably going to want to react in real time with AI too. And you're going to probably I want to give more leeway in response in order to, in order to beat them.

37:56Like there's ways you have to use AI there. And then it goes kind of higher and higher up like for targeting, for example, and for, you know, at what point should human be in a loop or not in the loop? And it's, it's, it's, it's, it's interesting as Congress probably has some opinions on this and they're trying to figure out how to regulate it. And, and like, obviously, you know, we're probably not ready for AI to be making certain high level decisions, but there's other ones that have to make or we lose. And like, like, how do you, how do you see this? Yeah. Well, I think there's a lot of complexity and nuance to the issue and you have to kind of slice it in layers.

38:24First of all, we've had autonomous weapons since the seventies and eighties. You know, you think about like the Aegis system that defends a carrier, uh, when it senses an incoming threat coming at the speed of a missile, it's firing off to protect the ship. And you don't have the time for a human to do many of these things. Wouldn't make sense for a person to have to say, yes, the ship's already hit. Yeah, exactly. Then you can kind of even look at something like a weapon that you fire from a fighter jet. Well, it's your radar system that's telling you that it's detecting an adversary. And you as a human have weapons release, but then it's also the guidance system that's going and closing that kill chain.

39:02So there's a lot of this that's actually happening already by computer. So I would posit that this is more of a difference of degree than kind. And because we're miscategorizing it as a total difference of kind, we kind of have some deranged conversations around it. And so we need to have a risk-based decision of like, what is the risk to our force? You know, nobody is trying to jump the shark and have killer robots. It's really like, how do we have something that doesn't advantage the adversary's OODA loop over ours and mean that we're showing up to a gunfight with a knife? and so you have to measure the risk as you go and i think uh you're gonna have to feel your way through that right and i think much to the department's credit they want the frontier labs they want everyone who's really smart about this to have a seat at the table and to help work through it together but recognizing that like policy has to be set by policymakers to play devil's advocate because i 100 i agree with what you said and i 100 agree we have to stay ahead of our adversaries um and use ai for these things to play devil's advocate there's some people on the else.

39:59I don't know this stuff as well. They see maybe there's some Karatek inside of the department of war who builds a program that does create some kind of thing that you might call a killer robot and maybe breaks the rules. And then all of a sudden you have something that is very scary and causing problems. I mean, it sounds ridiculous when I say it, but people are worried about this. Like what should we be talking about it? Like how are we going to protect people for this not to happen? Well, if you're concerned about that, like why, why aren't you worried that someone in a garage is going to create a killer robot?

40:25You know, I think the probability of that happening is really no different. So then that becomes almost like a technical question, not a matter of will. It's just not how weapon systems are developed. You have to believe that there's a conspiracy where a rogue actor can go do something like this in an organization the size of the department. And I think it fundamentally underestimates the moral fiber of people who are in our uniformed services. These are people who take the law incredibly seriously, who work very diligently, who have put their lives on the line to defend the nation, to defend all of us.

40:57And I think we owe them the benefit of the doubt in this capacity here. We trust your bureaucracy to stop crazy things from happening in some ways, which is ironic, but it's true. I guess, yeah. I trust people. It's not the faceless bureaucracy that I'm trusting. It's like, these are Americans. These are exceptional human beings. They care about what they're doing. This is their craft. They're not YOLOing it. If you have some cartoon caricature of who they are, okay, fine. Maybe from the degeneracy of the last 30 years of media that always paints the ultimate bad guy as the corrupt bureaucrat or the corrupt military official, but that's not reality.

41:32And I think they deserve more from us. Yeah. I think like a random PhD student in a garage is probably more danger with AI than anyone in these that we were serving with in the department of war. I don't disagree. Yeah. And it's actually interesting taking the other side of that. It does seem like recent wars, thanks to AI, thanks to new technologies are actually a lot safer, like a lot less death. Is that a trend you'd hope? Are we going to see that going forward where hopefully Americans lose a lot less lives based on this technology? Yes and no. So yes, in the sense that like, yes, this is an empirical trend.

42:03I think 70 million people died in World War II from all causes and effects. That's not even a quantity we can contemplate today. Hundreds we struggle with. right? The part that's known is like the part, I don't buy the idea that these sorts of robotic weapons are going to somehow mean that humans don't have to do the fighting. You certainly want the robots to draw first blood. You want them to serve as a deterrent, but the will of the nation to actually win and prevail in these consequences is always going to come down to brave humans willing to put themselves in harm's way when the moment calls for it in order to prevail.

42:43So it shouldn't let us off the hook of the moral weight and the consequence of what it means to actually fight a war, which is why we're a peace-loving nation who does not want to fight a war. We want to prevent the wars from happening. We want to prevent adversaries from having so much leverage over us, like a nuclear bomb, that actually peace is no longer possible. Well said. So Sean, we started this podcast to push back against cynics and the doomers. What's the best case for an optimistic future for America? What do we need to get right over the next decade? Well, you know, I've always said, I think America's greatest risk is not homicide, it's suicide.

43:15And so I think reminding ourselves of what we have in common and reigniting the national spirit for innovation and excellence. We are a great nation. There's no nation who has more to be proud of than we do. And that is ultimately the case for optimism. I'd say, you know, one more maybe off the wall comment is in the end, the reason we're going to win in whatever domain it is, is that we're crazy. You know, even we don't know what we're going to do when we get punched in the face. And I think that's actually an exceptional quality. It speaks to the plasticity and adaptability of the American mind and how we innovate.

43:51Alex always told me I'd be very successful if I didn't get arrested from a bar fight. I guess that's a good note to end that on. Thanks, Sean. Thank you, Joe.