How Palantir Went Public in 2020

September thirtieth, twenty twenty. Palantir's first afternoon on the New York Stock Exchange... and some of its existing investors literally cannot sell their shares.

Cannot, as in the button doesn't work?

The Wall Street Journal reported technical issues with Morgan Stanley software that blocked some shareholders from unloading stock for much of the afternoon. Which raised a nasty question nobody could answer that day: were there more sellers waiting in the wings?

So on day one you can't even tell what the real demand is.

Right. And this is a company that had stayed private for seventeen years. Born in two thousand three with CIA seed money, as the AP put it.

Seventeen years is forever. Why so long? Most startups are desperate to list.

Because they didn't need the money in the normal way, and going public means opening the books. And the books were ugly. Net loss of five hundred and eighty million dollars in twenty eighteen. Almost exactly the same again in twenty nineteen.

Two years, over half a billion lost each time? On what revenue?

Seven hundred and forty-two point six million in twenty nineteen. So they lost roughly eighty cents for every dollar they brought in.

That's the kind of number you hide forever if you can.

They stopped hiding it on August twenty-fifth, twenty twenty, when the S-1 hit the SEC. And here's what actually changed their mind. Chief operating officer Shyam Sankar told Reuters they decided in mid twenty nineteen to go public, originally aiming for the second half of twenty twenty-one.

And then the pandemic hit and they... delayed, surely?

Opposite. They sped it up. Sankar said, quote, broadly speaking COVID has been a tailwind for our business. They started eighty-three new engagements.

Huh. Governments suddenly needing to track a virus across a hundred databases. That's exactly their thing.

And first half twenty twenty revenue was four hundred and eighty-one million, up forty-nine percent year on year. The growth story finally looked better than the loss story.

Okay, but why a direct listing and not a normal IPO? What's the difference for someone who doesn't follow this stuff?

In an IPO the company prints new shares and sells them to raise cash, with banks setting the price. A direct listing raises nothing. You just open the doors and let existing shares trade.

So the founders keep control and skip the banker circus.

Karp told the New York Times exactly that. Quote, with a listing, you can still kind of keep your culture. We didn't bring in the super-experienced but culturally foreign A players.

That is such a Palantir sentence.

He also defended his grip on the company through special share classes. He said the control structure has to be tethered to a philosophical or mission bent, and that just saying we're the founders, take it or leave it, wasn't satisfactory to him.

Which is still, functionally, we're the founders, take it or leave it.

Pretty much. So the exchange set a reference price of seven dollars twenty-five. First trade came in at ten dollars. Before I tell you where it closed... where would you guess?

Pandemic, loss-making, weird listing... I'll say it sagged. Nine something.

Nine fifty, by Reuters' count. Below the ten dollar open, above the reference price. Valued around twenty point six billion.

So a decent day that still feels like a shrug.

And notice: at that ten dollar open, CNBC pegs the valuation at sixteen and a half billion. Below its private peak of twenty point four billion back in twenty fifteen.

Five years private and the public market says, yeah, you're worth less than you were. Brutal.

Which makes what followed genuinely hard to believe. Five years later, on the anniversary, CNBC ran the numbers. The stock was up more than seventeen hundred percent, closing at a hundred and eighty-two dollars and change. Market cap over four hundred and thirty-two billion.

From sixteen billion to four hundred and thirty-two. What on earth happened in between?

Government contracts and then the AI boom. Customers went from a hundred and twenty-five in the first half of twenty twenty to eight hundred and forty-nine. Quarterly revenue passed a billion dollars for the first time. It joined the S&P five hundred, replacing American Airlines.

And was it actually earning that, or was it just a story people liked?

That was the fight. At that point it traded at two hundred and twenty-six times forward earnings. The short seller Andrew Left of Citron called it detached from fundamentals, and argued it should be priced around forty dollars.

Forty, against a hundred and eighty-two. That's not a quibble, that's a different company.

And this year the doubters got their innings. As of late June twenty twenty-six, CNBC reported Palantir down nearly thirty-three percent year to date. The fear is that Anthropic and OpenAI's models just do the data work themselves.

Wait, the AI boom that lifted it is now the thing threatening it?

One analyst described investors imagining someone saying, create me a Palantir, and it magically appears. Karp called that idea a farce. He says the models matter, but quote, the implementation is where the value is.

And the market's verdict?

Unsettled. Michael Burry's been shorting it, though he said he covered half his short at a hundred and seven fifteen.

So the guy betting against it is also taking money off the table. Nobody's sure.

Nobody's sure. Seventeen years of refusing to show anyone the books, and now every quarter is an argument.

Honestly, I'd take the broken trading screen over that.