#190 Shyam Sankar - Chief Technology Officer of Palantir: The Future of Warfare

With AI and the AI revolution, people think it will make the median person better. That's true, but it will make the very best person superhuman.

When Rocket Man was rattling his saber in 2017, the Army wanted to know how many tanks there are. That is a 3-week data call. Operation Paperclip brought Nazi scientists to the US for our defense program. How worried should we be about China?

Great to be here. Thanks for having me, Sean.

Thank you. Everybody starts with an introduction here.

I love that quote, man.

Thank you. I'm grateful for everything this country has given us.

Absolutely. Quantum surprise, staying ahead. It's about the OODA loop. Quantum has exponential implications on how quickly you go through observe, orient, decide, act. What is quantum computing? Is it just really fast?

It's millions of times faster. It uses quantum mechanics to compute enormous permutations. It breaks legacy encryption. Anyone storing historical encrypted communications can decrypt them instantly. We need quantum-proof encryption. Beyond that, how do we employ this for decision advantage?

Smart people say we're close, but they said the same about AI in the 60s. We'll see.

The Chinese.

What?

My father was the last of nine children, born in a mud hut in South India. His siblings helped him get to college. He couldn't go to med school due to DEI, so he became a pharmacist. He slept under his brother's kitchen table. His sister-in-law found him a job in Nigeria building the first pharmaceutical plant in Africa. He did well. I was born seven years later in India, then returned to Nigeria. We had an armed robbery. They decapitated the dog, pistol-whipped my dad, threatened my mom. I was excited to have visitors. One robber was a regular at my mom's open kitchen and spared us.

We left everything and resettled in Orlando selling knick-knacks at theme parks. My after-school care was going to SeaWorld with my mom. I thought it was normal. I wore misprinted t-shirts. But it was a privilege to grow up near the space coast, watching shuttle launches. It gave me a sense of scientific progress and American prosperity.

He is my hero.

He passed away in 2021.

Early. My dad was an entrepreneur. I worked at his pharmacy, started programming in 11th grade, dropped out of Cornell to build a company. I went to Stanford to be on the West Coast with builders. Within two months, I was the fifth employee at Zoom with an X, a money transfer startup funded by Peter Thiel. My interviews were with PayPal Mafia members. That's how I got started.

I was at Zoom for three years. A colleague said there's a company perfect for me. My freshman dorm mate at Stanford was a co-founder, Joe. He put me in touch, and Joe hired me.

It was smaller but ambitious. We wanted to help institutions with counterterrorism. The political debate was security vs. privacy. As engineers, we knew both matter. We built technology to protect data and maximize sharing. One operation defeated an ISIS cell that planned to blow up a hospital with a downed drone. It required intel from Denmark, the US, and the UK. Our technology enabled machine-to-machine sharing, and they were gone before the drone left the garage.

We do much more now. The business is 50/50 government and commercial. Commercial covers 50 industries. Government is mostly defense and intelligence, but also health, IRS, justice. Our technology is an operating system for organizations. It lets you see yourself. For example, the Army took three weeks to count tanks because there's no canonical record. We built systems to integrate data. Data is not the new oil; it's snake oil. Value comes from using it to make decisions. It's about decision advantage.

So the

The decision is leading up to realizing what they're trying to do. How do you integrate human reporting from different countries? How do you automate sharing? Combine SIGINT, FMV footage, and intercepts to piece together the conclusion and timeline.

We provide software to the government. They have the data but it's in the dark. We spotlight what matters. Can you answer the next 10 questions in 10 seconds? If it takes 10 weeks, you won't bother. We automate information sharing with software that enforces rules, creating a hive mind. The government can operate competently without human gatekeeping.

It was hard. None of us had clearances. We carried our own projector to DC. We coded based on customer feedback overnight. We used synthetic data until we got conviction. It started with a renegade analyst, Sarah Adams. She brought CIA analysts to see our demo. They revolted against IT trying to shut it down. That broke the floodgates. I did 300 demos, and she made it happen in one day.

It was always about defense. The commercial business powers innovation we bring to government. We started because we cared about the government mission. We only built a commercial business because the government was too slow and we would have gone broke.

They wanted to build it themselves. Analogy: SpaceX next to Boeing. Boeing looked sophisticated, SpaceX looked like a joke. But the first derivative matters. SpaceX improved fast. In our early days, we seemed like kids in shorts. I once got dressed down for not wearing a tie in the Pentagon. But the substance is the code, not the tie.

A lot of pushback. Fake partnerships. But the government was our real competitor. Program offices have long-term plans. We disrupted them. Our adoption came from the field. First deployment was with 10th Group in Balad. The colonel called us three days before deploying. He palletized our servers and gave us four seats. We earned our keep. Matt Grimm and Greg Barbaccia were there.

What was that like for you?

Nice place. I've been there.

What year?

I was there around that time.

Tell me about it. We had these experiences that were like a water tank of motivation. People came back from missions with feature requests. I coded in a plywood setup and had something better the next day. I was the first forward deployed engineer. Investors hated it, but it was right for users. You need to be in the tent with the people doing the work. That builds deep understanding and addiction to purpose.

Credit goes to our customers. That colonel took risks. More people put in urgent operational needs to get Palantir. Some suffered career consequences. Good ideas don't come from Palo Alto; they come from Djibouti or Detroit. We haven't held ourselves accountable to winning. Does the software work? What can you change to make it work better tomorrow? Are you impatient enough to drive that change?

Institutionalizing rebellion. There was nothing that headquarters was going to tell our engineers in the field that would make them better at coming up with the next feature. That was going to come out of their own crazy creative ideas, being right next to the problem. So my job is to support them in doing that to the greatest extent possible as we continue to build the company, so that the machine doesn't crush them, the machine encourages them. That's led to this philosophy at Palantir that we are not a factory. We're an artist colony. My job is to find Delali and Monae and not do stupid things like yell at Monae to paint more like Deli. Each of these people are unique artists, and my job is to get out of their way, make sure other people stay out of their way, provide as much room as possible, and help them. Just like an artist, it's not like every piece of art is better than the last one. You go through waves and cycles and just create the environment that allows them to do their best work.

Yeah. In 2016, we sued the army because we were unable to compete on their army intel program. The story is, all these units as they would deploy, they would request Palantir. The reason they had to request Palantir is there was an army program of record called D6A. And D6A was the blessed system that was supposed to do army intel, except it didn't really work. $26 billion, many years of development, and you would honestly find it, when the generals came, people would turn it on and pretend like it worked. But it was a paperweight to them. It was just sitting there collecting dust. We were there on the ground, highly responsive, and all we wanted was the opportunity. This program was coming up for recompete and we wanted to compete to win it. The way the army structured it precluded us from competing. It basically said you have to custom build a product and turn over all your IP to us or you're not even allowed to compete, which is illegal. It turns out, which is why we sued, there's a law from 1994 called the Commercial Item Preference, that if a commercial product exists, the government is not allowed to go custom develop something because that's risky and probably going to cost a lot more. As a philosophy, we'd rather tie our investments to what the commercial world has adopted because it means it probably works, it's been derisked, hypercompetitive free markets are voting on it and providing the stimulus to keep making it better. It's going to be cheaper. So we sued them. It went to appeal. We won on appeal. So it's a very strong precedent. But the key lesson from that is how hard it is to change the momentum in the bureaucracy. One of my favorite anecdotes is there was a contractor working on this legacy D6A system in his civilian job, but he was a reservist. He got flipped to active duty and deployed to Afghanistan. The first thing he did when he hit the ground was put in an arms for Palantir. So when his life was on the line, he knew what he wanted to back.

Constantly working with the warfighters. Yeah, constantly. We were in Iraq and Afghanistan throughout the conflict supporting users. We were working with coalition as well, with UK special forces, other places, Philippines, Djibouti, supporting ODAs. We get on the conveyor belt of flights that would go visit the forward operating bases and make sure folks had what they needed. Make sure that their forward lock, they would take their laptops to do key leader engagements where all the intel was really coming back. When you were alone and unafraid and disconnected, this was your brain, this is how you were coordinating things. A lot of the value was doing our part to change the nature of the conflict from being 17 one-year-long wars to really creating continuity. Some Marines will tell you there are Marines alive today because of the simple fact that we could look back at past deployments and realize every time prior teams landed at this HLZ, they got bombed to shit. It was always an ambush. Just being able to see things that we couldn't know because we weren't there on the ground, but being able to learn from the history and the experience of everyone else who's been here before us made us smarter and more lethal and more survivable.

Not for me personally. Some of our teams definitely were under mortar attack and a few convoys that took some fire, but fortunately for us no one got hurt.

Yeah. I think we could start by what does it not mean? Why do doors fall off airplanes? Why do we have this legitimation crisis for our institutions? This sense that the institutions we have aren't working, and I think it's because they're stuck in manager mode. There's this idea that there's a playbook and if you just follow the process, everything will work. My experience is it's all well-managed into the ground. Whether it's a company or a country, there is no playbook. It really comes down to human discretion, human decision-making, bold leadership, people with the right ideas. In our history, it comes from one of our founders, Peter. It's called the Founders Fund. The whole idea is that the value creation engine comes from the founder personality. The most dynamic parts of our economy are led by founders. Europe has created zero companies in the last 50 years from scratch that are worth more than a hundred billion dollars. Zero. We have created all of our trillion-dollar companies in America from scratch in the last 50 years because of founders. There's something special about founders. There's a reason we call them the founding fathers. This is a unique legacy of America that we should really be leaning into and embracing. It's more than just did you found a company. It's a mentality, a personality. When you think about Bill Knudsen and how he mobilized us for World War II and built the arsenal of democracy, he's got a founder personality. Or Admiral Rickover. This guy born in a Polish shtetl, came here when he was six years old, short man, feisty. He was so hated at the US Naval Academy they tore his picture out of the yearbook. Zumwalt said when he was CNO, the Navy has three enemies: Soviet Union, the Air Force, and Hyman Rickover. But he built the nuclear navy. After World War II, he went to visit Oak Ridge and realized, 'Holy crap, this could power an entirely different regime of sea power and give us the ability to keep our subs underwater more than an hour.' Those diesel subs back then, it's laughable when we think about it. He was a four-star admiral for 30 years and we kept him in place and he built one of our last remaining asymmetric advantages. That's a founder personality. Gene Kranz who built the Apollo program, Kelly Johnson who built 41 airframes in his life, including the U2 which we still fly and the Oxcart and the SR71. These people did incredible things. That's the promise of America. We have these people who do incredible things. Don't squash them. Don't crush them. Let their innovation thrive. Those people are in government. I think the reason JSOC was one of the first places to adopt us is you got a lot of founder personalities in JSOC. That's how that institution works. That's what happens on the ground. They are entrepreneurs in a different domain. For us, it's great. When you have someone who has the creativity and the vision, I can build a software Iron Man suit around them. I know how to make them a hundred times more lethal or effective in the business context than otherwise. When you have a manager who's just trying to follow the rules, maybe you can make them 10% better. You're just rearranging the deck chairs on the Titanic. There's this constant need for reinvention. We can't rest on our laurels. It's these founder personalities that push that reinvention, who see what could be and see what is broken and close that gap.

Yeah.

Operation Paperclip was our covert action to bring the very best scientists from Nazi Germany to the US to enable our defense program, in particular our rocket and space program. Think about how hard that would be. These were our sworn enemies. They were horrible people. They did horrible things. But they were without exception the world leaders in this capability. They could either fall into the hands of the Soviets, which would only create more problems, or we could bring them here, hold our nose, and figure out how to rehabilitate them and use that for our own economic prosperity and national advantage. We did that. What's really instructive about Operation Paperclip is there were two competing programs. One was Operation Paperclip and the other was called Fiat. Fiat was this idea that we don't need these dirty scientists, let's just steal the scientific papers, take the technical documents, and we'll be able to learn from the documents alone. It was a catastrophic failure. It turns out there is something special about the human separate and apart from just what they wrote down on a piece of paper. There's more to replicating these things than math. There is something about the human mind that's really relevant here, and it underscores this point around the founders and the primacy of people, which I think is one of our greatest national resources. The reason it was called Paperclip is that's how we told them to come in through the front door through State Department, which obviously would screen out Nazis. So you had to have some sort of cover. We would have these files and we would resettle them. This is how Huntsville got started. Where can we put these people where it'll be low viz, low profile? We built a whole space industry in Huntsville, Alabama around von Braun and his colleagues.

We don't do predictive policing. It's widely believed publicly we do. We don't. And it has probably the most scary name possible.

I think a better way of thinking about it is the projects I've seen and been close to, it's about resource allocation. Where should I preposition police to deter violence? We look at the historic data. We see there's a lot of stabbings at this Jack in the Box at 2 a.m. on weekends when the temperature is pleasant enough for people to be outside roaming around. It might be nice to have a cop on that block, make the presence known, deter any bad actors. That's been effective where it's been possible. The branding has been really detrimental, but there is a clear correlation between the presence of policing and deterring violence. The other aspect is understanding that with gang violence, it's usually a cycle of retribution. This person got killed. The family of the person that got killed, their gang knows who did it. There's going to be retaliation. If you can figure out ahead of time these relationships, you can have presence there to suppress the ability for retaliation to happen without consequence, which then hopefully gives time and space to quiet down the violence.

Yeah.

What I think is wrong writ large with the world is that, let's steel man it. The executives of these agencies or companies have a steering wheel they're trying to steer the ship, they're trying really hard, but what they don't know is that steering wheel is like a prop from the Jungle Cruise ride in Disneyland. It's not connected to anything. The people on the factory floor look up and say, 'How can my leadership be so dumb? How can they not see what's actually happening here?' Then you have all the layers in between. Part of their job is to make sure the steering wheel doesn't actually work. They're controlling and massaging the information as it goes from one place to another. So you can't see yourself if you're at the top. This breeds a nihilism. Nothing works. We should tear it all down. Sometimes that's the answer. But you have to wake up. There has to be some force trying to make these institutions actually functional. That's what we see our software doing. It's creating a steering wheel for the executive that is connected directly to the people who do the work. That gives you a basis to understand what is happening. How do I know when I provide steering input, is it actually going to result in changes? How do I iterate on my policy and strategy? When I have a question like how many tanks are in the army, I should know that in three seconds. That lets me know, do I need more or not? I can move on to the next problem. If I have to wait three weeks to answer that, the whole system gets gummed up. Nothing actually works. You don't understand that the quality on assembly of doors is not working, and therefore small problems that can be solved become big problems. This breeds a legitimation crisis. The legitimacy of our institutions is based on the fact that they work. How do you spend $11 billion on high-speed rail in California and have 1,600 feet of track that goes nowhere? With a government institution that's really proud about how many jobs they created. Compare that to SpaceX. They've only spent $9.6 billion and they put more than 300 rockets into orbit. We spent $40 billion on rural broadband and have no rural broadband. We spent $80 billion on electric chargers and I think we had like eight. There's something wrong. Part of the disease is this idea that all you need to do is cut a check. Money will just solve the problem. Money enables the founder personalities. Money enables people to do the work. It's the front end of enabling the work, and it's the back end, the reward for having done the work. But someone's got to do the work, and you got to be really good at doing the work. That's art, that's craft. It's not a commodity. It's not mindless. You don't just outsource it. We need people to be proud of that and our efficacy in doing that. The counterfactual: if we didn't have Rickover, if we rotated people in and out of driving the nuclear navy every two or three years, we wouldn't have it. It wouldn't have worked. Why does the F-35 take 30 years when Kelly Johnson could build 41 airframes in his career, build the U2 in 13 months?

I do. The reason I've been writing about it is that I'm wildly optimistic that we're on the verge of changing it. It's a little bit like a ketchup bottle. You shake it, nothing comes out, and then it comes out all at once. This has been building. People can feel this subconsciously. They can see it in the parts of the American economy that are working, which frankly are the only parts of the economy. Look at Europe, nothing's actually working over there. The parts of the world economy that are working look like this. The parts that don't, you see it with when DOGE goes to DC. There's a lot of noise from the other side around DOGE, but it's been amazing. You have people who are the NBA athletes of technology walking around realizing, 'Oh, that's a seventh grade basketball team over there.' People who can call balls and strikes. Conversely, that person in government is a genius, they're not empowered. We need to empower this person who actually knows the right answers. DOGE has the ability to make those calls and to do so with taste. That realignment is profoundly needed.

It's not supposed to work this way. We were following the playbook. There's a system. It's this belief in process over outcomes. It's profane to have to blow up all the process and to believe that everything we believed in the past didn't actually work, or that there would be another way of organizing yourself that would be superior and would work better. That's the subconscious undertone of the pushback. People, it's like the best engineers. Like I said with Hyman Rickover, he was hated by the Navy, but he was a national treasure. Can you hold that contradiction? The engineer said something that hurt your feelings, but they're right. Can you hold that? If you care about the outcome, that's easy to get over.

What a shame. But I think the results will speak for themselves. I've seen this so many times. Once the bureaucracy gets out of the way, you can often solve these problems inside of 30 days. I know those guys, that's what they're focused on every day, delivering results, keeping their head down. They're a national treasure.

From a technical perspective, we could have done it 30 years ago. From a political will perspective, it took President Trump's political will and the mandate that he got with the election to make it happen.

Yeah, it's insane. I just don't understand it. It seems like it's better for everybody if we do this. People are screaming on the rooftops. To tie it back to founder mode, my own support for President Trump is, I had a three-hour dinner with him in the presence of a few other folks. What I felt profoundly was a founder personality. This is a person who was going to throw away the manager playbook, who actually had an intuitive understanding of the things that were broken and what to do about them, was going to make contrarian calls, and actually cares about the outcome.

I'm not. I think there will be some volatility. I've been looking at the markets, thinking about how much of this is the market reacting to the playbook. It's prophylaxis before the pain is even felt. They take as an axiomatic given from economic theory that this is a bad thing, versus empirically it's already resulted in badness. There's going to be some chop between here and there. But the ultimate thing about the founder playbook is not that founders are always right, but that they are first intuitionally driven and often can only fully explain their intuition over time. You have to decide: is the rebuttable presumption going to be we lean in and try what the founder is saying, or are we going to resist every step? Is the burden of proof such that I will not take any action until I perfectly understand everything? No great things are accomplished that way. It's too slow. The OODA loop is too long. So I'm in a suspension of disbelief mode.

I would say Scott Bessent is a very smart human, very smart macro trader, part of an elite world. There's probably on the order of 10 people in the world like him as a current Treasury Secretary. It comes down to confidence in the team. I have a lot of confidence in the team.