How Palantir Was Founded After 9/11

In two thousand three, a retired rear admiral named John Poindexter gets a phone call. It's Richard Perle, one of the architects of the Iraq War, and he wants to introduce him to two guys from Silicon Valley starting a software company.

Poindexter... that's the Iran-Contra name, right? The one who resigned under Reagan?

The same. And after nine eleven he'd gone back to the Pentagon to build a data-mining program called Total Information Awareness. A conservative columnist, William Safire, called it a super-snoop's dream.

And THAT is who the startup founders wanted a meeting with?

He was exactly who they wanted. Peter Thiel and Alex Karp were building something with the same ambition, and Poindexter was by then widely seen as the godfather of modern surveillance.

So what were they actually selling? Because every company says it can connect the dots.

The pitch was literal. Consolidate the databases, and you might have prevented nine eleven. That's what they told the CIA.

Hang on. Is that a real claim or a sales line?

Both, honestly. The underlying problem was real. After the attacks, agencies realized they'd gathered plenty of indications about al-Qaeda and bin Laden's plans. Possibly enough. But it was scattered across different agencies and different databases.

Nobody could see the whole picture.

Right. And the odd part is where the fix came from. Not a defense lab. A payments company fighting Russian fraudsters.

PayPal?

PayPal. Around two thousand, their engineers found that algorithms alone couldn't keep up with how fast criminals adapted. So they built a program that flagged suspicious transactions for humans to review. They named it Igor, after a Russian criminal who'd been taunting the fraud team.

They named the software after the guy trolling them.

Fraud losses dropped tenfold after launch. And in two thousand three, with PayPal sold, Thiel goes to Alex Karp, an old Stanford Law classmate with a doctorate in German social philosophy, and asks: why not point Igor at terrorist financing?

Why Karp? A philosophy PhD seems like a strange pick for a spy software company.

They were friends, and Karp had discovered a talent for investing in Frankfurt. Thiel's framing was that the government's response to things like airport security was becoming, his word, nightmarish. He wanted a mission company: reduce terrorism while preserving civil liberties.

That's a sentence that's going to age interestingly.

Hold that thought. Incorporation was May two thousand three. Thiel, Nathan Gettings, Joe Lonsdale, Stephen Cohen, Karp. Named after the seeing-stones in Lord of the Rings.

Of course it is.

Thiel bankrolled roughly thirty million in startup costs himself, because venture capital wanted nothing to do with it.

Why not? Post nine eleven, defense money was everywhere.

Picture the demo. Karp shows a prototype tracking two hypothetical terror suspects, their travel, their money transfers. Then the questions start. Is anyone at your company cleared for classified information?

No.

Have you ever worked with intelligence agencies? No. Senior advisers who have? No. A cleared sales force? No.

Four for four. That's not a pitch meeting, that's a hazing.

But they were intrigued enough. The group was In-Q-Tel, the venture arm set up by the CIA, and they put in an estimated two million dollars.

The CIA has a venture capital fund. That still sounds made up.

It does, and Richard Perle later told New York Magazine that angel investment may have been the agency's most inspired move. More important than the money, though, In-Q-Tel put Palantir's engineers in a room with actual intelligence analysts.

So the customer built the product with them.

For years. One trick that mattered: they tagged every single piece of data separately. First name, last name, phone number. So if one line in a file is top secret, an analyst with lower clearance can still see the rest.

Oh, that's the actual unlock. The clearance stops being a wall around the whole file.

And the first big operational contract came in two thousand six, with the Joint IED Defeat Organization, for Iraq and Afghanistan. Finding the patterns in roadside bomb attacks.

Did it work? Or is that a company claim too?

Current and former US officials told the Wall Street Journal in two thousand nine that it helped root out terrorist financing networks, revealed trends in roadside bombs, and foiled a Pakistani suicide bombing plot on Western targets.

Not everyone loved it, though, surely.

Poindexter didn't. He told New York Magazine he was impressed by the interface, but there was no back end to sort the data automatically. His words: the people using it were not happy with it at all. Much too manual.

The surveillance godfather's review is "needs more automation."

Which is the part that connects to now. The company built to stop another nine eleven is reported to run the technology behind an ICE data system drawing on dozens of databases, covering twenty million people.

Twenty million. And that's the same connect-the-dots engine?

Essentially. Four oh four Media reported it lets the agency locate people in minutes rather than hours or days. Supporters say that's the capability working as designed. Critics say a tool sold for counterterrorism is now pointed at immigration enforcement.

And "reduce terrorism while preserving civil liberties" is doing a lot of work in that sentence.

That's the argument that's still live. By the way, those SEC filings said the goal out loud: become the default operating system for data across the US government.

From four noes in a conference room to the default operating system. That's a twenty-year run.

With the founders cashing out along the way. In two thousand twenty-four, insiders sold more than four billion dollars of stock. Thiel offloaded about a third of his holding.

Huh. So what do they do with the commercial side? Because that's not the CIA anymore.

That's where we're going next.