How Palantir's Software Is Used in War

A sensor feed comes in. Satellite, drone, whatever's overhead. A unit finds a target, passes it up the chain. That used to take hours. With Palantir's software running, that same unit cut it to minutes.

Minutes. From hours. That's not an upgrade, that's a different kind of war.

Hold one question for the next few minutes. The program that made that possible? Google had it first. Google walked away. So why did Palantir say yes to the thing Silicon Valley's biggest company decided it couldn't live with?

Start with what's new. This story keeps moving.

Mid-September, this year. The Army moves a program called TITAN out of prototype and into real production. A hundred and ninety-two million dollars, split between Palantir and Anduril. Palantir's share is a hundred and twenty-seven million... as prime contractor.

TITAN being what, exactly?

Tactical Intelligence Targeting Access Node. Ground stations. They pull in satellite and sensor data and turn it into targets for the artillery. And here's the striking part. Palantir is the prime. A software company running a hardware program. That had basically never happened before.

So the software company isn't the supplier anymore. It's the one in charge.

Now rewind. Twenty seventeen. The Pentagon launches Project Maven. Simple idea, enormous consequences. There is more drone and satellite footage than humans can possibly watch. So let computer vision scan it and flag what matters.

And Google builds it.

Google builds it. Then in twenty eighteen, thousands of Google employees sign an internal letter. The line everyone remembers... we believe that Google should not be in the business of war. About a dozen people resign. By that spring, Google says it won't renew.

A dozen engineers quit and the whole thing collapses?

The work doesn't die. It moves. Palantir picks it up, and never apologizes for it. This is a company started in two thousand three with money from In-Q-Tel, the CIA's venture arm. Defense wasn't a detour. It was the premise.

Can't betray a principle you never had.

Alex Karp says it louder than that. He's told investors Palantir helps its partners, his words, on occasion kill. And in February this year he said the company's weapons software is in, quote, every combat situation I'm aware of.

He's not even hedging.

He thinks hedging is the problem. His pitch is that the West needs hard power built on software, and that engineers pretending otherwise is a kind of cowardice.

And the money followed him.

May twenty-first, last year. The Defense Department raises the ceiling on Palantir's Maven Smart System by seven hundred and ninety-five million dollars. Takes it to roughly one point three billion, running through twenty twenty-nine. A year before, that ceiling was four hundred and eighty million.

That's nearly triple in twelve months.

Demand, officials told DefenseScoop. Combatant commands running live operations with it. More than twenty thousand active users... four times what it had been that March.

Hang on. Twenty thousand people inside an AI targeting tool. Who's checking the machine?

That is the fight, right there. Palantir's answer, and the Pentagon's, is that the software proposes and a human disposes. It surfaces options, a person decides. Critics say when your cycle compresses from hours to minutes, the human review becomes a formality. You're clicking, not deliberating.

And nobody outside can audit which one it actually is.

No. That's the honest state of it. Then Ukraine. Karp was the first global CEO to meet Zelensky after the twenty twenty-two invasion. Palantir opened an office there. And Karp said on the record that the company is responsible for most of the targeting in Ukraine.

He said that out loud? Tanks, artillery?

A spokesperson confirmed it helps target things like tanks and artillery. Though lately his framing has softened. He told United Twenty-Four Media the Ukrainians built one of the best targeting systems in the world, that Palantir is part of it... but that it was largely shaped by Ukrainian engineers and soldiers.

And for a lot of people, that's the sympathetic version. Software helping a country defend itself.

Which is why the next part splits the room. January twenty twenty-four. Palantir announces a strategic partnership with Israel's Ministry of Defence, to support what it called war-related missions. The board held a meeting in Tel Aviv. Months earlier, the company ran a full-page New York Times ad saying Palantir stands with Israel.

And that's where the criticism gets loudest.

The U.N. Special Rapporteur Francesca Albanese named Palantir in her reporting on companies profiting from the war in Gaza. Palantir disputes that characterisation, and has never detailed what its tools do there. So let's be precise. The partnership is confirmed. What the software actually does inside Israeli operations is not public.

That's the whole pattern. We know the contracts. We don't know the consequences.

We can tell you the Maven ceiling is one point three billion through twenty twenty-nine. We can tell you TITAN went into production in September. We cannot tell you, from outside, whether a human truly paused before a strike.

So... back to my loop. Why did Palantir say yes?

Because Karp believes the alternative is worse. If the people who build the best software refuse to build it for their own democracies, someone else builds it for someone else. That's his argument. And it isn't a stupid one.

The Google engineers' argument wasn't stupid either.

No. They just lost. And here's why it lands on you. The debate those engineers started in twenty eighteen never got resolved. It got outsourced... to a company that was never going to blink.

So the hand-raising moved somewhere nobody can see it.

The hard thing about a machine that thinks in minutes... is that conscience still works in hours.