Palantir CTO on The Broken Incentive Structure of How Governments Buy Defence

I’m so thrilled to be here. Thanks for having me, Harry.

father. My father, as a young man in his twenties, he left India to set up the first pharmaceutical manufacturing facility in Africa. And even before that, he was the youngest of nine children. Children. He lived in this structure where all of his siblings were working their ass off to send him to college. And this sense of, look, I have to make something of this given all this opportunity I’ve been given. He was a great man.

That there was kind of this obligation to family, to society that there’s there’s just so much value in that. And I think in in his life, in many ways, I think you could say it would be it would be kind of a trivial story. Was like, look. My dad was this guy who worked really hard. He took this risk, it all worked out. That’s not what happened. It all didn’t work out. When my dad went to Nigeria, when I was a young kid, we suffered horrible armed violence.

He was almost executed. He had to start over. We we fled Nigeria. We we resettled in The US effectively as refugees starting from scratch. I was raised, and he lived his life with this deep understanding of the counterfactual. Like, no matter how bad it is, remember what it could have been. Your alternative to this present moment is being dead in a ditch and how grateful you should be for that and how you should not squander the opportunity that you have as a consequence of that.

I was 2.5. So, fortunately, you know, I was not mentally scarred, traumatically scarred by that, but it had a profound effect on my parents.

gave me this sense of looking forward. Look. All these bad things happen. What are you supposed to do about it? What the most productive thing you can do about it is stick together as a unit and focus on what you could do next and support each other through that and support each other’s ambitions. I think accomplishing great things in the world very rarely can be done alone. At least I’ve never experienced It’s really the unit. There’s all these unsung heroes. It’s it’s better to think about it as a rock band, not a solo act.

And getting that chemistry right is, I think, crucial. And it’s as crucial at home as it is in business.

I can have a mixed feeling on it. On one end, I think I’m not sure why it’s productive to believe in luck, especially if you think your luck can vary. I think the most productive mind frame is something like, I have good luck, and the only thing I actually have control over is how hard I work. And so let me focus on that part of it. The other view of it is, is there such a thing as luck? Probably. But it’s like quantum. We often think it’s like, what matters more?

Luck or hard work? Or why not both? And only one of these variables do you really have a lot of control over. You know, we we probably somehow in this current epoch, I think we’ve undervalued or underfocused on hard work and overemphasized the role of luck in in success.

Yeah. So for the audience, this is Zoom with an x, was the, the money transfer company that subsequently ended up with PayPal. I mean, that was great. Like, I really got to the maybe backing up a little bit. I got to The Valley because in this period of time having graduated from Cornell, your opportunities as an engineer were really finance, consulting, or maybe a very large tech company where you were gonna be a cog in the machine. Like, Google didn’t even come to the campus to recruit in those days.

And so I just needed to get to Silicon Valley. I went to to Stanford for that. Within two months of getting to Stanford, I had, like, begged and broken down Kevin Hartz’s door to get a job at Zoom. That’s really how this kind of all started. So I joined Zoom as the fifth employee, and I spent three amazing and wonderful years there. But through Joe’s freshman year roommate, actually, I learned about Palantir, this this very eclectic character, Gabe Rosen, who is a poet who swims at the South End Rowing Club in freezing cold San Francisco Bay waters.

I mean, this guy, you can’t really put in a box. But he’s like, you know, you really need to meet Joe and hear about this company. And at the time, it was 12 people, and and they wanted to work on problems of national security. My uncle was a victim of the two thousand six coordinated train bombings in Bombay. I was very much and I was in Upstate New York during nine eleven. It was think that was a seminal moment for many folks of my generation. I actually tried to drop out of school to join CIA, but I think everyone had that idea at that time.

I didn’t even have a college degree, so they didn’t quite respond to my electronic application. And so, yeah, I would have much rather failed, and it certainly seemed in 2005, Palantir. It wasn’t at all clear how it was gonna work, but I would have much rather failed working on something that I thought was so important.

They had focused on only hiring engineers. They were kind of at the point where it’s like, okay. Well, who’s gonna actually interact with the customers to shape how we’re building the product? This is a very deep and complicated problem. The intersection of the domain and the software will be nuanced. You have a massive asymmetry of information that most of the software you’re deploying is into a classified environment where at the time none of us had clearances. We were all wild outsiders with no knowledge of what was happening behind these magical doors, but just almost maniacal commitment to trying to solve the problem nonetheless.

Well, I I think we were just committed to building the right product. And I don’t want that to sound naive because actually, I think it took us longer and was harder because we didn’t know how to sell to government. There was all these things that were kind of zero to one. There was no one else to really learn from. All the people who really knew it were not of the technology world and couldn’t get it to work in our context anyways. So we had to kind of flail our way through feeling it out from first principles and developing a new playbook that was gonna work.

And I think a playbook that subsequently has been really valuable to the rest of Defense tech.

If you were to follow so before I tell you what the the new playbook is, just understand the old playbook. Right? The way the government really wants to buy everything is to pay you by the hour to build it. So, like, some assessment of who should be doing it, but fundamentally, it’s a labor model. And it’s a labor model where they really want and through cost accounting and audits, we’ll look at your actual costs. They will build in overhead rates. They will then say, this is how much profit you’re allowed to have.

It’s kind of actually pretty communist for a country whose entire economic engine has been powered by capitalism as a monotonous buyer. And you could say, well, why is this? Because it’s not totally crazy. But assume you’re the only buyer for a product in the world. You’re the only buyer for the aircraft carrier. How should I compensate someone who’s building an aircraft carrier for me? What’s a fair price to pay? So the way that they solve this really starting in the last century was thinking about as cost plus.

I don’t know. It seems fair that it costs you something. You have to pay people. You have to buy materials, and I’ll give you some profit on top of that. Now this has obvious consequences in terms of what the long term incentives end up being. Why would you ever invest in innovation? And and to the extent the government wants you to innovate, they actually pay you the money to do the r and d. You’re not putting your money into it, really. You’re putting their money into it.

As the world transitioned from a place where all of the r and d was done by the government, fifties, sixties, obviously, the preponderance of r and d is done by the commercial world, this model makes no sense. You really wanna leverage the beauty of what we’ve really pioneered in the valley is is high risk, high return approach.

you’re right. I was trying to steel man it, which is like, okay. How did we end up here? Like, let’s just assume, you know, it’s Chesterton’s fats. We didn’t end up here because somehow the system is totally broken and incompetent. There was some thinking to it. But I think we’re now in a world where the second order consequences of that system are dominating, and that’s what’s leading to a structural inability to innovate. Technology is supposed to make things cheaper. Somehow that doesn’t happen in government. Why?

You know? And to the extent things do get cheaper, they get cheaper because they’re benefiting from the massive tailwinds from the commercial market’s r and d that then subsequently makes its way to government.

And so I think the the great opportunity and really, so the playbook was, look. I’m not gonna let you pay me by the hour. I’m going to refuse to accept any revenue that services based from you. And instead, I’m gonna build a product. I’m gonna keep investing in product, and that means I have to go sell a part of my company to finance that. Right? I have to go raise money to go do this. But this product’s gonna be so good, you’re gonna wanna buy it.

And when you’re gonna buy it, you’re gonna buy it on a license fee basis. You’re gonna pay some sort of subscription, software as a service. I know to the Silicon Valley, let’s say there’s this, obviously. I don’t what are you talking about? How is But there’s a lot of brain damage to getting the government to this model. Even now, we’ve made a huge amount of headway, but the the low entropy, the, like, lowest path, least resistance is to try to pay someone by the hour.

One of the other complexities of government is there’s a disconnect between who is the buyer exactly and who is the user. And so in the early days and even today, you know, the preponderance of our time is spent with the user. How do I know I’m building the right product? How do I know what I should be building next? Where is the utility? How do I get these folks who are the users to tell the buyers, no. No. Really, this is way better. If you go to the buyers, they’re gonna always wanna push you down to the conventional model of how do I buy this by the hour?

And so I think that the way to do it is to create capability that’s so compelling, so necessary that people are willing to go through a more difficult but more rewarding process in order to acquire it.

It’s closer to committees. I mean, it’s still a Byzantine process. I’m not sure even I, after eighteen years of doing this, could really fully put it on a piece of paper and and kinda tell you it out. You’re kinda feeling your way through it. But, you know, people would say that you need a requirement, you need resource, I e money, and you need a contract. And I think at a highest level, you’re trying to figure out how do you line up those three elements within the government customer you’re working with to accomplish this.

Now the pathing of doing that, I think, is specific to every opportunity you’re kind of going after.

There’s two pieces to this. The first is that the land and expand doesn’t work the way it does for commercial SaaS companies. It very much is about expanding the product surface area as opposed to the license surface area, or I’m trying to get more users. I’m trying to so it’s not sales driven expansion. It’s actually product driven expansion. And if you think about how do I connect up more of what this customer is doing, like, I’ve landed with this capability. How do I expand to the lateral capabilities that are valuable to them?

Many of those capabilities, you’re not gonna think those things up in Palo Alto eating strawberries. Like, you need to be in Djibouti or in Kandahar or in Balad or in Ramstein sitting with the users to understand what’s next because the truth is out there. It’s not in some document somewhere. It’s not in DC. The deep understanding of what’s missing that you can build true stickiness around, that’s that’s one dimension. The other dimension of this, like, the reason you have to be there to your margin optimization, it doesn’t actually come from lock in or just, like, rent collection.

It it actually comes from understanding how do I continuously productize and create the efficiency in how I deliver what I deliver. There are unique constraints in this environment. Right? You don’t have one big multi tenant or three big multi tenant geographic instances of your software that you’re running on the public Internet. You’re delivering software to air gapped environments that are tightly controlled and constrained, and people need clearances to access them. And so you need to have some sort of product methodology to invest in the software infrastructure.

You have your software, then you have the infrastructure below that that is essentially an army of robots that manages your software in production. Otherwise, you’re gonna end up deploying a team of 20 people to go manage your software in these environments, and you’re gonna have a really hard time getting to scale.

Yeah. The whole government system is geared up to buy hardware because they kind of especially in the US government, but I think you could see it’s analogous everywhere. Living in a world where government used to finance all r and d, it starts with r and d. Okay. I’m gonna pay you to do some r and d. Build a prototype. If I like this thing, I’m gonna procure it. I’m gonna buy it at scale. Once I’ve procured it, I’m I have to sustain it. So they expect costs to go like this.

Of course, that’s not how software works. First of all, software r and d costs only increase. No investor thinks Microsoft’s gonna spend less in the future on r and d than they spend in the present moment. That would be crazy. But the government somehow expects that to be true. And the second part is there is no difference in our modern technology world for between procurement and sustainment. Like, those concept even as I say the words, they may be like, what what do you mean exactly here?

You know, you’re delivering software as a service. You’re continuously innovating on it. It’s never done. I think the government has absorbed that idea that software’s never done, but they haven’t absorbed the idea that that means there’s at least a constant cost to that software.

Yeah. You could think about, like, when were we amazing at this? And I think it was really at the dawn of World War two. And there’s this great book by Arthur Herrmann called The Arsenal of Democracy. It’s probably the book I’m most obsessed with right now, and I’ve read it a couple times. It really talks about how the industrial mobilization worked. We brought Bill, a Danish immigrant, who was a very senior executive at Ford and General Motors into government. We made him from a civilian to an army three star general, and he was in charge of the industrial mobilization.

And we had this immense luxury during World War two, which was that we were at peace, Britain was at war, and we were supplying Britain. We had eighteen months when we were not getting shot at to turn on our industrial production, to build factories, and to retool them, and that took eighteen months. We needed that time. The counterfactual would have been brutal. We also needed to leverage American industry. We had just figured out mass production, and we, America, had just figured out mass production. It was our unique strength in the world.

So we applied that mass production technique not to model t’s. We turned it to to bombers and fighters and tanks and all the machinery of war you needed to actually help the Brits go forth and then subsequently to help America fight as well. And we had these crazy founders. You know, it was Henry Kaiser. We think of Lockheed Martin, but it was Glenn Martin. It wasn’t Northrop Grumman. It was Jack Northrop. All of that innovation was not driven by faceless companies run by eighth generation leaders.

They were built and run by these just maverick founders, people we would recognize their personality structure in the valley today. You know, as a consequence, after we won the Cold War, rightly, we expected a peace dividend. Why would you continue to I I think it’s it’s hard for a legitimate democracy to continue to sustain that level of spending in the absence of an adversary. So in 1993, the assistant deputy secretary of defense, Bill Perry, who was himself an entrepreneur, he started a Silicon Valley company called Electronic Systems Laboratories.

He had what was called the last supper. He invited the 51 defense primes. Today, we have five. Back then, we had 51. He invited the 51 defense primes in, and he said, look. The budget is gonna get much smaller, and it did. For every dollar they had before, they had 33¢ going forward. Not all of you are gonna survive. I I’m kind of giving you permission and a mandate to consolidate, to get lean, to get efficient. And so this is how we went from 51 to five.

And I think the off-site consequence of doing this is that the industrial base became less competitive. Oh, we only have five people to pick a month, so, you know, prices are high or something like that. And there’s some truth to that, but I think the much more profound consequence of the Last Supper is it began a deep financialization of defense. It became about dividends and buybacks and leverage ratios. It, you know, it became about m and a. New venture formation was about spinouts and mergers, not about venture backed creative risk taking ideas to go build entirely new things.

Yes. Because all the emphasis went away from creative big idea. You know, Alex Karp says this all the time. It’s like, no great company was created primarily to make money. Yes. There’s a consequence to that and it has to happen, but, you know, like, I I don’t think Larry and Sergei were like, hey. How do we make money? In fact, I think they were kinda pulled kicking and screaming into that part of it. So when suddenly it was very much first the business that became quite challenging.

And I I think it also bred a sort of dysfunction in the relationship between government and what government calls industry, a term that really we don’t use anywhere else when we talk about the partners you use to solve problems. And it’s that, like, you’re transactional, you’re money grubbing, I am doing this mission here. I’ve given up a lot to work in the government. I think we can really fix that, and we have this opportunity that’s happened right now because of how much interest there is in Defense tech.

Well, I think that’s a consequence. The core industrial problem is that we don’t have a way to reward mavericks and innovators that we’re breeding we’ve historically bred a marketplace that’s driven towards conformity. We all have the same business model. We all wanna get paid by the hour. Hey. I don’t wanna take risk because I need to know you’re gonna buy this government. There’s no way for VC to be applied here because I can’t figure out how to get to a scale of return that’s gonna matter.

I’m forgetting Churchill’s chief of the general staff. He he had a observation, which is that when you go to war, the first thing you need to do is fire all your generals. Generals that succeed in peacetime and this was a general himself. Right? The generals that succeed in peacetime are not the right generals to succeed in a war and vice versa. What I’m saying there is something a little smaller, which is that when there is war, you do have a different sort of mobilization. You have a different sort of leadership.

People are willing to make different decisions. It’s almost like the only thing that works when it really matters is breaking the process and doing what’s right, and the process never really works. It it’s kind of a digression from the main point around this, but I think you don’t have leverage in the sense that, look. You’re gonna buy whatever model works. It’s the opposite, which is the leadership says, oh, I’m willing to break the model because I see what’s right. You know, like, there wasn’t a problem big enough to solve before, so I wasn’t gonna change how we did business.

I have a problem to solve, and I, as a leader, am gonna say, yes. This is the right path.

I think there was almost a lack of political structures. You know, there was a willingness to move fast and break things. Kelly John and when I say founder, I really mean it in the broadest sense of the word. It’s almost like renegade, maverick, innovator. Kelly Johnson is is one of my favorite examples. This is the guy who founded Skunk Works at Lockheed Martin. This guy as a and we would recognize him as a 100 x engineer in the valley today. Right? 23 years old, joins Lockheed Martin, looks at a two d diagram of schematics of a plane and says, this plane is fatally flawed.

He goes to the senior engineer at Lockheed, and almost to get rid of Kelly, he’s like, yo. Why don’t you fix it? Of course, he fixed it. So this guy built 26 airframes in his career. Today, if you’re an aerospace engineer, you’re lucky to work on a third of a plane over your entire career. So you’re never even gonna ship a plane. And this guy shipped 26. And some of his planes, like the u two, we still fly. He built the first u two in nine months.

We can do this again. There’s a way in which we build up so much process, so much risk aversion. In order to constrain I think that that the greatest travesty of all of this is it’s like in order to constrain the variance and outcomes, we have constrained ourselves to mediocrity. Know? It’s like we don’t want projects to fail catastrophically, but that also means you can’t have projects succeed catastrophically. And the the amount of process we put in place just ensures mediocrity.

I think it is. I’ve been thinking about this. You know, it’s dangerous to say this because if you take me too literally, it’ll it’ll sound politically unpopular. But in an effort to have a perception of fair competition, we have created so much process that we neither have fair competition nor speed. If you you could just juxtapose the European procurement process, which I’ve The US process, and maybe alternatives. And I’d say the European process is even more slow because there’s an even greater emphasis on how do I show fairness across all of the EU states.

And so then you could say, okay. The American process is a little more streamlined, but it still takes years. Kelly Johnson could build a plane in nine months. We can’t even procure something in twice that long. That’s crazy. It’s not because we’re not competent enough to do it. It’s because we have created a structure where we do not want protests. We do not want to be blamed. We do not want to lose public trust in how these decisions are made. But I think we are retarding the growth of the human capital and government as a consequence of this.

You can make mistakes. That’s how you’re gonna learn. No one wants to make mistakes. No one wants to be wrong. I wish we could all be right all the time, but I have screwed up so many things at this company over eighteen years. If the standard of surviving here was something like you can never be wrong or you must optically be seen to be making every decision in a way that, you know, it will never blow back on you, this company would not exist, I would be screwed.

It’s not a way to actually make the world better.

I think the most honest answer to that question is America actually does have the best process when we’re serious. When there’s the political will and alignment and the problem is clear and we’re ready to go, that’s when we throw all the rules and the doctrine out the window and we go and we get it done. And this is kind of the Churchillian thing. Right? The Americans screw everything up until they’ve exhausted all the options and then they do the right thing. And so that’s the other point.

It’s frustrating is we are so clearly capable of this. I’ve seen it demonstrated in our greatest moments of need, but in the steady state, we reject it.

when you put your

Pick winners, plural, to be very clear. So I think this is one of the greatest disconnects, I’d say, between the hyper growth technology community and DOD. So we’ve had roughly a $100,000,000,000 of venture investment enter Defence tech recently, last couple years. And the VC model is not that everyone needs to be a mediocre outcome. Right? Like, I’d say the median return in a venture fund might be zero, but the average return is maybe two, three x, right, for the portfolio. And so this is the dynamics that matter, which is, like, what the peanut butter spreading as it’s kind of called.

Like, I have this much money, and I’m gonna spread it across all the companies that have come in here. That just ensures that they’re all zombies. They’re neither alive nor dead. Those firms are not gonna be able to raise their next round. There’s not enough return. And and a much better strategy is realizing, look. We need to pick a few winners, and we’re gonna go after it with scale. And these companies are gonna become large as a consequence, which then allows those funds to go bring more money in innovation in our ecosystem.

And there’s a virtuous cycle here, and the talent is gonna stay. So maybe their one company didn’t work out, but they’re gonna have many shots at the apple. We’re gonna have this ravenous new tech industrial base focused on defense problems with the capital behind it and the economics that support it. This is the part that DOD is a little uncomfortable with, which is like, well, I’d rather it’s a little bit more an aesthetic towards fair socialist returns. Like, I just wanna spread out the money across all of these participants so that no one’s upset.

But I think in reality, everyone will be upset. And if you talk to these venture capitalists, it’s o like, yes, I’d prefer that my companies win. But if that’s not possible, someone has to win because otherwise, this market’s not gonna exist. And getting the Defense Department to realize so one of the greatest lies they tell themselves is we don’t pick winners. They if you if you interact with acquisition folks, you’ll you’ll get this sort of like, oh, we don’t pick winners. But, no, literally, you’re a monopsonist.

Every contract you award, you’re picking a winner. So we have to kind of regroove how we do this to support the broader ecosystem.

I think there’s two dimensions to this. One is you don’t wanna be beholden to your adversaries. So that that’s, like, an obvious one I think everyone can agree to. The second dimension to this is we often forget that one of the roles of government is its economic role. You know, there’s an aspect of government that is a jobs program, that is welfare for the people. I think where it gets tricky with defense is where that’s the case and it’s unacknowledged. In The US, we have these CBR programs, these small business innovation kind of grants research.

To a VC, this is like the kiss of death. Study after study shows that you get these CBR grants and and your company never actually achieved scale. They don’t really wanna invest in the companies that are pursuing this. It’s not a strategy to actually become a big company. It’s a strategy to stay a zombie. I think we’d be better off asking ourselves, okay. If we want the CBR program, if we believe in small business and the aesthetic of small business, this is a type of, like, wall white collar jobs program, and that might be quite good.

It’s creating more STEM jobs. There’s that’s a policy question I’ll leave to our policymakers. But we should not think that from this program, we’re gonna get massive defense innovation. We’re gonna get new champions. We’re gonna get competition. We’re gonna get any of the transformation that we actually need to win these next fights.

I’d flag two things there. One is we are spending at historic lows relative to the cold war as a percentage of GDP. So absolute dollars, yes, are bigger. But as a percentage of GDP, like, what what do you think is the defense insurance premium the country should pay every year to protect, you know, Pax Americana and the Western international rules based order? What is

At its peak, I think it’s like, when you look at, like, Vietnam, you look at other periods of the Cold War, it’s, like, closer to six. But it’s, like, half as a percent of GDP. I’m not sure that it should double. That’s not the point. But I think, like, we somehow view this in a context of, like, we could view it as, like, what is the insurance premium we’re willing to pay on GDP to protect GDP?

I think it’s between where we are now and where it was during the Cold War. It it was absolutely right to get a even though we suffered long term consequences of the Last Supper, I don’t think you could have political legitimacy without having a peace dividend after the Cold War. So I think I think what happened was correct. But now we should analyze what needs to happen in the context of great power competition. Like, we do not live in that world anymore. It has changed radically, and we have developed some sclerotic habits of thirty years of living at, quote, unquote, the end of history that we need to really shake off and organize against.

And no place more so than Europe, where in many ways, they’ve kind of forgotten the distant past and have enjoyed a lot of what the West collectively has been able to create.

The buyers here are very serious people who wake up every day thinking, I might have to fight tonight. I’m in the position of providing capability to people who might have to fight tonight. I don’t feel a shift in tone. What I do feel is a shift in the rate of learning. What is there to learn from? You look at the tactics. You look at what is working on the battlefield. Many of these things are surprising. They’re different. They’re innovating very quickly. They’re changing the concepts of what you think you would even need to buy to have survivable capability to win the next the next conflict.

And that is evolving incredibly rapidly, and I think it continues to speak to this concept of lower cost, attributable capabilities. You know? Instead of buying something that you think, like an aircraft carrier or some space based capability that’s gonna be around for thirty or fifty years, because it’s gonna be around that long, it’s gonna be so expensive. It’s gotta just be perfect. It’s like, I’m buying something I might use for the next year. You know, you look at Ukraine. Many of these drones are single use drones.

So there is no sustainment. What are we talking about here? There is like, your whole supply chain changes. Like, you the whole theory of how you’re gonna employ the weapon is very, very different. And as a consequence, the logistics you need around it become very different. And that can liberate you to have entirely new concepts, entirely new business models that enable you to deliver the innovation, recognizing that what you’re gonna fight within two years is not what you’re gonna fight within a year, and it’s not what you’d use today.

One wish. It would be to return to a structure that we actually used to use, which is to have multiple competing programs to go after the same capability. So when we were building submarine launched ballistic missiles, we didn’t have one program to build it. I think we had four. And there’s a human factors element to this, and there’s a a practical technology element to this. Well, we weren’t sure if they should be solid fueled, liquid fueled. There was all sorts of technical questions. And so you can either have the loudest voice win or you could do a forever, never ending research project, or you can compete the ideas.

And that means, you know, you’re going much faster. You’re learning. You’re iterating. It’s it’s an empirical discussion rather than a theoretical one. You’re buying down technology risk, but I think what’s even more compelling than that is the human factor. I often think about what would motivate so many of these programs. Like, what would motivate the program manager to wanna go faster, to wanna insert new technologies more quickly? And the answer is nothing. Because the distant threat of an enemy for a conflict that you might wanna fight one day does not warrant blowing up your schedule, introducing risk, maybe having someone yell at you.

Because you’re just judged by cost, schedule, performance. That’s all you’re gonna think about. Right? I don’t think that’s healthy. So my counterfactual would be, what if there are two PMs who woke up every day? It’s human competition. You know? It’s like the government is gonna buy some percentage of my product, some percentage of their product, and I am going to wake up every day to win this. That’s like a uniquely American thing. I I think we’re very good at that, and I think it is gonna drive a lot of innovation.

It solves a lot of the problems. Like, how do we get things to be cheaper? Well, I want the government to buy more of my products, so I’m gonna build it cheaper. How do I insert new technologies? How do I go faster? Like, these problems are actually kind of a lack of internal competition. The government focuses on competition in the industrial base. I think there’s not enough competition inside of government.

There’s definitely gonna be ex post attribution. How do we get closer to something ex ante? How are you making that decision? So today, there is a pretty big disconnect between the people who are responsible for fighting, really the combatant commanders, and the people who are responsible for providing the equipment. And, really, these guys, the fighters, they don’t really have a vote. And I think that’s one of the things we need to change, where winning the competition should really be about, okay. Well, how much does the INDOPACOM commander or the CENTCOM commander want of this variant versus that competing variant?

And it doesn’t have to be winner take all. It can be something like, I want 20% of these and 80% of those, and next year, come back to me with better ideas. Let me let me taste the stakes and see which one I like, and, you know, I I might allocate resources differently next time. And I think getting that faster feedback iteration cycle would be super valuable. It also lets us be a little bit more critical of a quote, unquote requirements. You know, the requirements are the things that must you know, we’ve done this analysis.

We said these things must exist, and we’re gonna build all of them. Well, clearly, requirements change. The only thing you know for sure is requirements are wrong. And so how do we enable our program managers to take more risk and say, well, I think these requirements are more important than those requirements, and I think they’re requirements they haven’t even thought of that really matter. Because at the end of the day, you gotta taste the steak. You know, you gotta yield the product, use the product, and that’s the only way you’re gonna know if it’s good or not.

You know, can you imagine if the Valley didn’t have something like that with its buyers? That the way that every new product was gonna be evaluated was through historical list of requirements that someone else came up with and a paper PowerPoint exercise to see which products someone was gonna buy. No. We innovate because we fight with our technology in the customer’s office.

Much less than people think. There’s obviously some effect, and I think it it varies by country. In The UK system, given how involved the cabinet is with approving contracts and the oversight and integration with treasury, it’s actually more relevant in terms of what agenda people will be following. In The US, I think the bigger issue is really congress and whether there’s a budget or not. We don’t have a budget. We’ll enter a continuing resolution. In a continuing resolution, that means you can’t have any new starts.

So all the things the department wants to do, they’re gated and waiting until there is a new budget. This is a consequence of our our system. I think it’s a little facile to say, like, congress should just pass a budget on time. And, yeah, like, that would be a nice world, but this is how we exercise political power and arrive at consensus. And I think there are things that we can do to make this better, but I think that’s kind of the bigger issue in The US.

If you’d asked me ten years ago, I would have had a long list of things that I think the capital class could be doing better to help in the national interest. I don’t have those critiques. The capital class has showed up. The capital has showed up. The entrepreneurs have showed up. People are working on these problems. To the extent I had an ask from the VCs, it would be to engage with the policymakers so they understand the dynamics of your business. What is it gonna take so that the $100,000,000,000 or more continues to show up?

What does it take for these entrepreneurs to stay in this ecosystem? Why is a labor based business model so bad? Because I think we can underestimate the size of the gulf between these two worlds. And you could even say it as something as simple as, you know, in DC, there’s a, like, no conflict of interest. This idea of, like, we’re gonna manage all the conflicts of interest, and if you have any conflict of interest, it’s very bad. And, obviously, conflict of interest can be bad. On the other hand, in in the valley, it’s like, well, if you haven’t invested in something, how do you even have an opinion about it?

Like, you you you’re not even willing to put your own money behind an idea, so what do you know about it? You know, it’s like it’s almost like no conflict, no interest. There’s a middle ground between these things, but I think that’s the first place to kinda get the relationship grooved so people understand it. So it really comes down to high trust human interaction. We’re not gonna scale this process. It’s gonna come down to these two worlds coming together, and they’ve never been closer, but we still have a lot of work to do

You know, Steve Jobs has this great loss interview about how companies get confused. When they’re trying to replicate their success, they look at the process they went through to deliver their first success. Then they follow the process, and it doesn’t work. Because, actually, you know, the first and most important thing is the content. It’s the creativity. It’s the substance. And that substance rarely and as he points out, people who are good at content tend to be a pain in the ass to manage. You know, they’re a pain in the butt, I think was his his quote.

But you put up with them because the content is just so good. And so I think there’s this way in which process as a medicine is okay. You know? It’s like, can you have zero process? No. But if you hold process up as the deity, if process ends up being what is actually an opioid, it’s very bad. It’s like, look. We just need to follow the process, we’ll get to the outcome. That’s almost never correct. When the process becomes more important than the content, you’re gonna lose.

And a lot of this comes up as like, well, I wish the interaction could be less frictional. I wish there was more predictability. I wish there was all sorts of things which are just not true in the real world. Like, yes, we all wish the world could be different. But when you honor what the world is, you kind of recognize how important it is to have the right rock band. Right? There’s no, hey, follow this process and we’re gonna get there.

Absolutely. So, you know, I think you need some way of having both, but you can’t ever lose the primacy of content. Put it this way, you can get really far with incredible content and bad process. You’re gonna get nowhere with incredible process and bad content. So And the primacy of it, you need to keep in mind. And this is what I mean by process as medicine versus opioid. And so there’s a rule for this here, but especially when you’re and I feel you know, Palantir is this company that keeps reinventing itself.

In those moments of reinvention, there’s just content. There’s just the crazy content and your commitment to it and making it better, and then you figure out how to get it to scale. And sometimes maybe getting things to scale is itself not scalable, which I don’t mean as a critique. It’s more like, first, you gotta figure out the substance, and then you work really hard every day to get it to scale. There’s an old Silicon Valley saying, there’s the first 80% and the second 80%. And so maybe there’s a little bit of this and this, where it’s like, yeah, the first part of the content is 80% of the work, and then getting it to scale through process is 80% of the work.

Yeah. I think this is so important. I think there’s a way in which we’ve kind of constrained our cultures and our companies because we’ve leaned into the opioid of structure, essentially. Like, look, I’m gonna structure my company as a factory. Here are all the roles and the levels I have, and here’s the org chart, and it’s all very clean and laid out. And if you play by these rules, you can figure out how to navigate it. I think one of the challenges is that, like, really talented people are highly asymmetric.

You’re much better off designing a role around who they are rather than saying what role can I slot you into? None of those roles are gonna fit them. It’s gonna put them in position of relative weakness. They’re gonna blow something up. They’re gonna feel like they have to be good at something they’re never gonna be good at. And so it’s it’s a total evisceration of their potential. And if you said, instead of being a factory, we’re gonna be an artist calling. I don’t get Dali to paint better by yelling at him to paint more like Monet.

Each one of these individuals is an individual, and I’m gonna go on a journey to craft a role around who they are, and I’m gonna make it okay for them to, like, flex their role to meet their capabilities. Even more importantly, artists’ work don’t monotonically get better. This idea that you’re gonna climb levels in this sort of strictly increasing way, I think, is deeply in inauthentic. Some of your works are gonna be hits, and then you’re gonna go through a lull. And, you know, your work might not be great.

The community might not appreciate it. And then your next piece is gonna be a masterpiece. And so how do you get into a place where you can be committed to your talent over a long duration? You know there’s gonna be ups and downs, but they’re so fundamentally talented that you’re gonna stick with them, and you’re not gonna be judging them work to work. It’s not like a daily mark to market. It’s like, look. This is you are special. This is a home for you, and we value you over this long arc of what we’re building together.

That means that when they’re not performing, you don’t suddenly have this responsibility to sideline them and put them in a hole or fire them.

Well, we could keep leaning into the Hollywood metaphor here of of the talent agent and the talent. How do you manage talent is a very open ended question, but I think that’s roughly how you would think about it, which is like, look. I’m really here to support. I need to create the environment that allows your art to flourish. I’m not the one managing your art. So if you think about that, okay, they need a huge amount of autonomy. They need the resources. They need some protections from the daily distractions of certain aspects of the business.

If you’re always inhaling the full visceral scary reality of everything going on, I think that distracts you from doing the art. So how do you have complete autonomy over an area that you’re very, very good at and the ability then to to execute against it? And then I think, Maxim, I always think about is smart people cannot be taught anything. They can learn, and that’s the environment you’re trying to set up. It’s like, how do I create this environment of maximum learning for these incredibly talented artists?

Yeah. I mean, I look think the artist might even blow up at that point. Artists want their art to be really good too. If they’re really going through a rough patch for a long time, that might be part of your I think that’s gonna be a judgment call as the talent manager. No. No. I really believe. Like, I understand this is a long dry spell here. I’m gonna keep you motivated. I’m gonna keep you managing. I’m gonna figure it out. But, of course, you could just be wrong.

Maybe maybe the the moment has passed. The art and the inspiration and the muse doesn’t exist.

There are sometimes there’s just external realities. You may have to do that because that’s reality. This is one of those things you have to think about how you shield or when you choose to shield the artist from, like budgets. Budgets are anticreative. Constraints are important. Don’t get me wrong. Budgets are just anticreative. So you have to think about what’s the right mix of, like, hey. At this moment, we really need budgets. It matters for these reasons, and here’s why. And and so I and I think you tend to find that, like, you kind of oscillate between two ends of of a continuum.

Right? Like, there’s no budgets. It’s free for all. It’s max creative. It’s like, oh, this is completely ungoverned, we need, you know, this thing, and you’re you’re just trying to maintain the stable unstable equilibrium rather

I definitely agree that constraints drive creativity. Now I think you can be overly constrained. So I already have a time constraint. I have a customer constraint. I have a problem constraint, and, you know, maybe the budget constraint blows it up. I mean, that’s the job of the talent manager. It’s to manage to the budget without exposing the brutal reality of the budget to the talent. You know, how can I get the talent to feel a few less constraints while actually being constrained?

My biggest hiring mistakes were hiring for a factory. Hiring what I thought I would need. Hiring what the world told me, you know, the way to solve this problem is to hire this profile. And, you know, I didn’t earn it from first principles. Principles. The most successful ones have always been when I took the talent and grew them into the roles that actually reflected their talent. And the worst mistakes were it’s where I tried to import a solution to a problem because that that’s the version that’s my version of the opioid.

Wouldn’t it be nice if I could just magically import this solution, and it will solve all my problems in this area? I think I am, like, zero for 10. It’s never worked for me. So what do you do now? First principles. Who do I have? How can I give people more responsibility? How can I stretch them? How do I bet on just raw talent? And when I’m out there meeting people, and I still spend a fair amount of time doing this, I’m not really even thinking about the role.

I’m thinking about this human. And it’s like, would I be excited to fight shoulder to shoulder with this person over the next decade? And can I imagine continuing to invest and grow in them to take on more problems? And if they’re talented, I’ll figure out the role. They’ll figure out the role.

The the form of the mistake is thinking that outside knowledge will translate here without a massive retranslation function. And so the sort of personality that’s like, look. I’ve solved this problem before. I know the answer versus I’m betting on my talent to be able to solve the answer. That’s the shape of challenge.

I honestly think all of them have been challenging. And I I the reason they’re challenging, there’s the work, which is more obvious. The less obvious part of this sort of reinvention is reinventing yourself. It’s the toll on the individual where you’re like, look, But I got this thing, and it’s at this place, and it’s it’s successful. And where is my, like, earned credibility from having done that? And in the reinvention, what you’re saying is no. Like, we’re all every single person, we’re starting from zero. And it might be true that some person who joined this company two months ago is going to be more sophisticated at this new problem than you who joined five years ago.

And that’s you know, there’s a, like, a a commitment to pushing through that. And I think I think it’s actually only survivable because the whole thing is set up as an artist.

That’s one aspect that gets harder. Let me give you one aspect that gets easier. The first few times we’re doing it, I felt this deep personal responsibility to, like, handhold everyone through their own reinvention, and that is exhausting. And I felt their pain of of it. And it’s not that I don’t feel it now, but I now recognize, like, that’s the medicine. Like, I was trying to minimize the pain of the reinvention, but, actually, their greatness is gonna come from going through this this crucible and being proud of how they reinvented themselves.

So now I I kind of view it as I view it proudly rather than with trepidation.

Sounds great.

I think it biases towards the incumbents who own applications in the app layer.

I think the models are gonna be commoditized because the models are not going to really give you enough of the value. It turns out that, you know, like, you have the model, then getting to a business outcome with that model happens in an application. And so if you own those pixels, you’re gonna be able to apply it more quickly. I also think that very few problems are a 100% AI. They’re like a third AI, a third traditional deterministic code, a third human thought. The only place you can drive that elegant integration is at the app layer.

I think so. I’m not saying some won’t be better than others, but I’m saying most of them are gonna be good enough.

I think they’ll get a lot of the compute for sure. And I should say, I think the value is going to accrue in those two directions, infrastructure and app layer, and that the models are gonna be squeezed.

I was a model company, I would be pushing into the app layer. I’d be trying to figure out how to own more of the value I’m creating with the model rather than just selling an endpoint. And this is usually the opposite advice you’re gonna get from traditional venture where they’re like, build something simple that scales with an API.

Well, I think a lot of their criticisms are right that essentially if you’re just building a chatbot, it’s a pretty thin layer. So you could what makes it a thick layer is what’s the use of deterministic code or algorithmic reasoning in addition to the LLM? What’s the integration into workflow? Are we still chatting here? Are prompts for developers or for users? Why do LLMs need tools? Because otherwise, you’re completely constrained to the parametric knowledge of the model. And I think that’s a Sisyphean task. Your people keep trying to jam more and more knowledge into the model, but what they fail to recognize is that these models are actually very good at speaking regular grammars.

JSON, code, it’s a regular grammar. How do you use these models in their current state to manage and manipulate application state?

Some of the folks who have been created are definitely gonna be core winners. I think there will be more winners to come.

Unquestionably, OpenAI has has done quite well. I think we have some time to see where does Google really shake up in in all of this. And I think Meta has I mean, it’s amazing to see how much they’ve reinvented themselves around this.

The fear I really have here is how much are we modeling the appropriate use, and how fixated are we on proof of value instead of proof of concept. We had a self driving car demo in 2005. It drove a 132 miles through the desert. That’s really impressive. That’s not a toy like demo. You could argue only in 2023 did we have a proof of a self driving car that could drive around the city reliably as a commercial offering. Nobody wants to be in that business. No enterprise wants to adopt this thing where the demos are amazing.

In twenty years from now, they might be able to put this in production. And I think as technologists, in order for us to not be in a bubble, we need to be really focused on proof, not proof of concepts, and the speed by which we can affect these enterprises.

Probably the biggest one is content and process. I mean, look, content has always mattered, but I always had this view that, like, it’s gotta scale, and I’m now willing to fight incrementally, you know, almost like trench warfare to get things to scale as long as the content is gorgeous.

I’d probably, Begin, former prime minister of Israel, one of kind of the founding fathers. I’m I’m so obsessed with the conditions around founding. You could also pick a a US founding father. I think we don’t study, like, that the mindset of 1776, the unlikelihood of this, the unlikelihood of founding the modern Israel state, how difficult it was. I I just wanna relish those founder stories.

Probably the most important one is this idea of the artist colony, that humans are unique individuals and that you maximize value by staying true to that and never letting yourself aggregate or think about them as a class.

I think the world is likely to change much more in the next five years than most people are modeling. The geopolitics are highly unstable. That maybe is a more obvious statement now to say after October 7. And I think we’re entering a a very challenging time here, and I think that has consequences for the commercial world as well as governments. But the other aspect of this is as we approach that sort of instability, what is the role for real transfer? Like, we have this unique opportunity with AI to actually change things, not just to build businesses, but to fix many of the problems that drive at the fundamental legitimacy of the institutions we have in the world.

Why don’t these institutions work? Well, I am not someone who thinks you’re gonna fix them by tearing them down. You’re gonna fix them by partnering with them and helping them, helping the humans who wake up every day trying to make their institutions work work better.

Thank you, Harry, for having me. It’s been fun.