Palantir just had a nearly flawless quarter. At forty times sales, flawless is the minimum the price requires.

There are two Palantirs, and the mistake almost everyone made this week was to talk about one while pricing the other. The first is a software company that just grew revenue 93 percent, raised its full-year outlook by the largest margin in its history, and posted a set of numbers that, on the operating merits, are about as good as you will see from a company its size. That Palantir is real, and it is very good. The second Palantir is a stock trading at roughly forty times the revenue it has guided to for the entire year — not its profit, its revenue — and that one is a bet, with a specific and demanding set of assumptions baked into the entry price. The company had a wonderful quarter. Whether you should own the shares here is a different question, and running the two together is how people lose money in both directions at once.

Start with the quarter, because I have no interest in pretending it was anything other than excellent. Revenue came in at $1.94 billion, up 93 percent from about a billion a year earlier, ahead of the $1.8 billion analysts expected. Adjusted earnings were $0.41 a share against a $0.34 estimate. U.S. commercial revenue grew 149 percent to $764 million; U.S. government revenue grew 90 percent to $809 million; the two together carried domestic revenue past $1.5 billion. Management then raised full-year guidance to roughly $8.15 billion, up from a prior range near $7.65 billion — the biggest guidance raise the company has ever produced — and guided the current quarter to about $2.16 billion. You do not short a company on numbers like these. The demand is not imaginary, the acceleration is not a one-off, and anyone still calling Palantir a story stock with no business underneath it stopped reading the filings a year ago.

The company is not the argument

So let me be exact about what I am not saying. I am not saying the AI-software trade is fake, that Palantir cannot grow into a much larger company, or that the government and commercial demand it is booking will evaporate. It is genuinely rare to watch a company this size put up triple-digit commercial growth and simultaneously accelerate its government book, and rarer still to do it while throwing off real operating margin. On the company, the bears have been wrong, and I include in that the version of me that thought the commercial ramp would take longer than it has. The company is not the argument. The price is.

And the first place the price shows up is in the metric everyone quoted approvingly: a Rule of 40 score of 155. The Rule of 40 is a rough sanity check — a healthy software company's revenue growth plus its profit margin should clear 40. Palantir cleared it by nearly four times. That is being presented as evidence the stock is cheap for its quality, and it is worth pausing on the logic, because it runs backwards. A Rule of 40 of 155 is not a durable operating state that a spreadsheet can extend for a decade. It is a photograph of a peak — the specific moment when a growth rate and a margin are both unusually high at the same time. Distrust the round number, and distrust the very round number even more. The healthiest reading of 155 is not that the company will stay there. It is that the market is being invited to capitalize a peak as if it were a baseline.

A great technology and a sane price are different claims

A record quarter tells you the company is winning. It tells you nothing about whether the price already assumed it would.

Here is the precedent the enthusiasm has agreed to forget. In 2002, after the dot-com peak had finished humiliating everyone, Scott McNealy — then the chief executive of Sun Microsystems, a genuinely great company that had traded at around ten times revenue — was asked why investors had been so happy to pay it. His answer has aged into one of the most useful paragraphs in the literature. At ten times revenues, he said, to give you a ten-year payback I have to pay you one hundred percent of revenue for ten straight years in dividends. That assumes zero cost of goods sold, which is hard for a company that makes things. It assumes zero expenses, which is hard with tens of thousands of employees. It assumes I pay no taxes, and that you pay no taxes on the dividends, which is illegal. And it assumes I can hold revenue flat for a decade with no research spending. "What were you thinking?" he asked. He was talking about ten times sales. Palantir trades at roughly forty.

The point of dragging McNealy back into the room is not that Palantir is Sun, or that the business is weak. Sun's business was not weak either; that was the whole point of his complaint. The point is that a great technology and a sane price are two different claims, and a bull market's favorite trick is to let the obvious truth of the first smuggle in the unexamined assumption of the second. Palantir being a very good company is not in dispute. Palantir being a good price at forty times sales is a completely separate proposition, and the quarter — however strong — is evidence for the first claim and silent on the second.

What forty times sales actually asks

It helps to say out loud what a multiple like this requires, because the number is abstract until you translate it into obligations. At roughly forty times forward revenue and something near ninety times forward earnings, the price is not paying for the business Palantir is; it is prepaying for a business several years larger, at margins it has not yet demonstrated at scale, and it is assuming the multiple itself does not compress along the way. All three of those have to go right. The company can execute flawlessly — keep compounding revenue at a high rate, keep expanding margin, keep landing government contracts — and the stock can still deliver a poor return for years, because a market that pays forty times sales for perfection has priced perfection, and perfection merely delivered is not an upside surprise. It is the terms of the trade.

We have watched exactly this movie with a company nobody would call a fraud. Cisco in early 2000 was the most valuable company on earth, growing fast, dominant in a real and expanding market, and priced at a multiple in the same neighborhood as Palantir's today. The business went on to do fine for the next decade. The stock did not; buyers at the peak waited many years just to get back to even, not because Cisco failed but because the price had already assumed a future that then had to be paid off in real time. The company was never the problem. The entry price was. That is the difference between a bad company and a bad price, and it is the single most expensive distinction that investors refuse to make, because in a good story the two feel like the same sentence.

• What the quarter proves: revenue up 93 percent, U.S. commercial up 149 percent, the largest guidance raise in company history, real margin. The business is winning.
• What the quarter does not touch: whether roughly 40x forward sales and ~90x forward earnings is a sane price to pay for that winning.
• What the price already assumes: years of high compounding, further margin expansion, and no multiple compression — all three at once, with flawless execution priced as the base case, not the bull case.

Compared to when

Now the part I am obligated to say against my own thesis, because a columnist who only remembers her hits is running a marketing operation, not an argument. I have thought this valuation was stretched before, and the stock kept going, and being early is a way of being wrong that I refuse to launder into being right. It is worth keeping the ratio honest. But notice what the tape already did while everyone was arguing: the shares were up near $190 in the winter, fell roughly 40 percent to the $130s by spring, and are now bouncing back toward $145 on this print. "Compared to when?" is the only question that keeps a round number honest, and the answer here is that the stock has already made a full emotional round-trip inside a single year, on a business that never stopped growing. That is not the signature of a market calmly compounding a great company. It is the signature of a market repricing the same set of facts violently, in both directions, depending on the mood — which is exactly what you would expect when the price has floated far enough from the fundamentals that sentiment, not results, sets the daily number.

So here is the whole thing in one breath, since it keeps getting split into two arguments that pretend to disagree. Palantir is a very good company that just had a very good quarter, and Palantir is a very expensive stock that has priced a very good future as a certainty. Both are true. The bull case does not require the company to fail for you to lose money; it only requires the company to be merely excellent instead of flawless, or the multiple to drift back toward something a spreadsheet can defend. The growth is real. The growth was never the question. The question, as it always is under every story confident enough to make you forget it, is what you paid.