It's great to be here again at the World Economic Forum in Davos. I'm very honored to introduce Alex Karp. I want to start with something a bit more personal between him and me. I'm quite proud of what I've built at BlackRock, but my total return since becoming CEO has only been a compound rate of 21%.
Since Alex and Palantir went public, his compound return has been 73%. So congratulations, Alex.
But more importantly, we are in the midst of a profound technological shift. I think we're all hearing about it, reading about it, feeling it, and being part of it. And everyone is asking: what can AI do for me, or how can I apply it? How does it help growth? How does it help workers? What can it do for countries and national security?
We're discussing its potential to grow capacity, modernize industries, and expand opportunities. It will also transform how we work, where we work, and the way we work. And the question is whether governments are ready for that genuine transformation of society.
So we need to ensure that when it is deployed, it empowers people, empowers organizations, and builds a more resilient global economy. However, very few leaders are truly at this intersection. I'm certainly not that leader. But at the intersection of technology, national security, and the real economy, Alex Karp stands as co-founder and CEO of Palantir, which has worked closely with defense organizations and governments, private sector, to apply how AI can be used in many more critical areas, and that truly matters. And I must say that my conversations with Alex over the past year have really enlightened me a great deal. So I'm very much looking forward to this. Alex,
So let me start. Sovereign nations are often the earliest adopters of advanced technology, and I think we're seeing that very clearly in the United States. But from your perspective, how is AI supporting decision-making in defense and security?
First of all, it's great to be here, and with that introduction, perhaps I should just stop right there. Everything will be worse from here. You want to talk more about the introduction, and I'll... it's time. I've been working publicly for 26 years. No, no. I think one of the things to remember, like the context for your question, I think because you have
in America, and I would also say in Europe, in the history of industrial and military technology development, they are obviously linked together.
And in a general but more right than wrong way, you develop a product for the military, then it has a dual function and raises the standard of living of the country.
And then, for many reasons, I'll set that aside for now, that is not the way, at least, technology has been built so far. There are all these defense technology startups and Palantir. But the problem is you have something you're about to create that has to work in the harshest conditions, which is probably better, considerably better than anyone else, to the point that it gives you an advantage, even a slight one, on the battlefield, especially if combined with the way you fight.
And you've seen this. There's a very famous German sociologist historian who said one of the problems Germany had was that its war machine was so good that they just said, 'We'll decide on the battlefield who is right.'
That obviously led to a lot of disconnect and real problems in the American context. And I would say there you have a shift between what has happened recently in the US and what is happening in Europe, and what I think.
So I think the US and China are quite successful. I've lived a long time in Europe and I'm very supportive of it. But I think any honest assessment would show that it's not looking very good.
You've built things that can be used in adverse conditions, that is, harsh, dirty, morally gray conditions. Changing the ethics to fit the way the West fights is a big factor. So ethics is a difficult thing. The conditions under which you use technology are very difficult. In the software context, you're not directly connected to the network.
In many cases, you have constraints that you have to fight with, even though that's not the optimal way of fighting, and you have very specific ways of fighting in each country. But the positive side of that is that you're also building something that can be deployed and bring clear value to ordinary people.
And so now you have this. The AI question has a lot of really interesting things. But if you start with the banality that I think, until very recently, all kinds of Western adversaries, broadly understood, believed that defense investments based on software were just some crazy thing Americans did to advertise. The company got rich and then blew up, you're on a beach in the Bahamas. But the shareholders were happy, and that was that.
Something like that. And what you've seen is, first of all, you could say that has changed.
But directly related to your question, their learning process in building, how to build this for sovereign governments, is also a learning process about how they adopt these technologies. So it's not just the technology, because if you're building a tank, like the British, then the French, then the Germans optimized tank technology, it's very obvious how you would deploy it. How do you deploy a system that has value? It's by deploying parts on the battlefield without seeing the parts on the battlefield and examining whether it works, how effective it is, whether it's better than what we had, whether we can do things we couldn't do before.
And then there's a hidden thing about AI software where most of the really interesting value lies. People always assume the value is where you are versus where you should be.
That's obvious, but in most sovereign countries in the world, and we deal with almost every country in one form or another, broadly defined, the ability to deal with a company with an American character, the actual technological rigor of enterprises has significant gaps. So it's like, it's very... I just find it very messy. There are large parts of the enterprise that only exist on PowerPoint, and when you go to battle, you discover they don't exist at all.
And this, regardless of which country you're from in this room, if you're from the West, this is the problem you're facing. One day on the battlefield, you'll discover this is one of the advantages Ukraine has. They basically started from zero, and so there's no question of having to rediscover that your enterprise doesn't work after you've gone to war. And one of the biggest advantages America has is simply because, for better or worse, whether I believe it or not, mostly for worse, I've always opposed interventions, I'm not a neoconservative, is that we have all this battlefield experience so you can see what works and what doesn't work. But back to that issue. The first thing that sovereign nations really struggle with is: can I objectively identify which technology is better? Can I even evaluate it? Can I evaluate what I'm looking at to say, 'Yes.' But don't you need to know where you want to go to ask the right question? That is, you really... let me rephrase. You need to know where you are to know where you want to go.
So the point I want to make with this explanation is that one of the most important things Palantir has done on the battlefield is to compensate for the reality that half of your enterprise doesn't work on the battlefield. It works in the lab, on a PowerPoint, built by a nation. It doesn't work because of machines or people, because battlefield conditions change. And as an example, if you take Ukraine as an example, just to make it concrete, because everyone reads about this and how hard it is to move a drone from A to B. Well, actually, first of all, you would need to know where you want to put the drone. That would require synchronizing all your data. You would need to do that without transferring that data to your adversary. That means you would need to know every single person who has touched it.
You would have to conceal it until the last step. Then you would want to do that in a way that fits your strategy or your ethics. So it can't go anywhere. You need the ability to maneuver. Do you want to put the drone on your target? No one. Only two Ukrainians can know that one of the generals is your target. You can't say that's the target. How do you make your soldiers see that you're actually destroying the enemy and you just missed your target? And then the war progresses. So the Russians, who are often underestimated for reasons I don't understand, but mathematically they may be the best in the world.
And the things they may not have had at first, they can patch together. They started causing a lot of disruption. So now you face a completely different problem. But your enterprise probably still has to hold or grow, because now the problem is not going from A to B, but going through a completely jammed environment where you have no connection while in reality still collecting data. When you go through that environment, and each of those is like a dynamic challenge. None of that was anticipated, even before the war in Ukraine, and every battlefield in the world has it. By the way, another thing is that everyone fights differently. So if you look at the major battlefields, I'm sure some people like our work here and some people hate it. By the way, we welcome all opinions at Palantir. Palantir has people who like and don't like this work.
We embrace all of it. That's the spirit of dialogue with a bit of a leader's attitude.
But if you look at the way...
it's a small group of people and very brave soldiers who are very technically knowledgeable.
They have highly technically skilled people building on top of our products, things we don't understand, and proprietary uses of the product. In Israel, according to rumors, they use their intelligence. Most people fight militarily. This is intelligence. From what I can gather from the articles, and then in the US, you just have a massive force that no other country has. But they have to be integrated, and the integration capability. So each of those is different. For sure. And so the dual role of enterprise software on the battlefield is, one, to ensure that all the foundational elements actually work. And
two, to elevate it to a level that no one else in the world has.
First of all, in reality, this is a completely raw, bare environment. Bare means you actually know the truth about what can work independently of what enterprises think can work.
And at a high level, it can transfer almost one-to-one.
But precisely because, just to take enterprises in general, not every enterprise tends to want to become like every other enterprise over time. So if you take enterprise A and enterprise B and enterprise B and technology enterprise C, they're in the same market, their technology infrastructure is trying to make them the same enterprise. They have the same organizational chart, certainly. They probably have about the same... they don't have the same data and infrastructure. And what you learn on the battlefield and in life is that that has no special value.
An enterprise that does what no one else does is very valuable.
And that is the goal of every intelligence and military agency.
Intelligence and military agencies have their own expertise. When it enters commerce, how does an insurance underwriting company assess risk?
How do you turn it into internal knowledge?
How do we transform knowledge that everyone has into knowledge that only you have, with effects that no one else can achieve? So, for example, on the battlefield, one of the most important problems is how you collect data, process it, and then put it into a framework where it can be acted upon from an intelligence perspective. Yes. So what does an enterprise do?
In the end, what is an enterprise doing? Like in underwriting and banking.
Admitting patients in a hospital. That's information. That's organizing information in a way that you have a clear advantage over others in a similar position, and that advantage cannot be easily wiped out. And so, in fact, what you're doing on the battlefield with our products, without playing the game, I'm very happy to say, like what you're doing with Ontology and Foundry in some cases on the battlefield, but certainly in commercial organizations around the world, especially in the US, is when you approach an underwriting unit or a hospital, we provide energy to many hospitals.
They have admission problems. All of them have admission problems. They all lack doctors and nurses. They operate on thin margins. But every facility has a different way of handling their patients according to their specialty, and what kind of patients they don't do well with, and how you manage that. And so the admission flow into your enterprise in a way that you can actually handle these things 10 to 15 times faster than before.
It saves lives. It saves many lives. It's also really interesting because you're processing AI in a... you have a structure that, despite what everyone might want to believe, also strengthens civil liberties. Because now you can ask simple questions: Is a person being treated based on economic considerations? Or are they being treated based on their foundation? Those things are invisible unless you have a very large civil liberties improvement here, which normally people don't believe we care about, but in reality it's the exact opposite. We really do care, and what we show is that we care. Like we can display in detail why someone came, why they were admitted, why they were rejected, and we can do that in a way that makes business sense for that enterprise itself. And then it leads to safety outcomes and perhaps cost reduction.
If you want to do a slightly shorter financial version of this, to do what we can do in the daylight of the public market, you need to take the company private.
And then you would remove the cost structure and perhaps sell it again. Now you can reduce costs and improve labor. So real workers, not the fat in the middle. Then change how products are delivered, hospitals process information faster. Nurses and doctors are happier too.
