I get asked about Palantir more than any other single stock, and the question is almost always some version of “is it too late?” So I sat down and did the thing most of the bullish coverage skips: I worked out what the current price actually requires the business to do, and then asked whether that is a bet I would make on purpose. The business is real. The math is the problem.
Start with what’s working, because a lot is
Palantir is not a story stock in the way people mean it dismissively. It is GAAP profitable, it generates real free cash flow, and it has been in the S&P 500 for a year. Its “Rule of 40” score, revenue growth plus profit margin, has been running near 80, which is elite for software of any size. The US commercial business, the part that was supposed to be structurally hard for a company built on government contracts, has been reaccelerating on the back of its AI platform and the boot-camp sales motion that gets a customer from pilot to signed deal in weeks instead of quarters. Government spending on defense software and AI is going up, not down, and Palantir sits in the middle of it. Founder-led, dilution slowing, expanding into new sectors. If you were scoring the business alone, it would score well.
None of that is the argument. The argument is the multiple you pay to own it.
What 60-plus times sales has historically meant
The stock has roughly tripled over the past year. Market cap is north of $400 billion on a company doing somewhere around $4.5 billion of revenue this year. That puts it near 60 to 70 times forward sales, depending on the day, and something like 180 times forward earnings. The Palantir quote page has the current price and multiple if you want to check where it sits today.
Numbers like that are not unprecedented, but the precedents are cautionary. Zoom touched a similar multiple in late 2020. Snowflake did in 2021. So did a short list of other excellent software companies at their absolute peaks. In every case the multiple compressed hard over the following two years even as the company kept growing revenue. The stock went down because the valuation came back to earth faster than the business could grow into it. That is the specific risk here, and it has nothing to do with whether Palantir executes.
The scenario grid
Here is the exercise. Assume Palantir grows revenue at a given annual rate for the next five years, and assume that at the end of that window the market still awards it a given price-to-sales multiple. That gives you a 2031 market value, which you can compare to today’s figure, a little over $400 billion, to get an annualized return for the stock.
| Revenue growth, next 5 yrs | Exit multiple 12x sales | Exit multiple 18x sales | Exit multiple 25x sales |
|---|---|---|---|
| 25% per year | about -17% / yr | about -10% / yr | about -4% / yr |
| 30% per year | about -14% / yr | about -7% / yr | roughly flat |
| 35% per year | about -11% / yr | about -3% / yr | about +3% / yr |
Look at where the positive numbers are. There is exactly one corner of that grid that produces a return worth the risk, and it requires two things to both be true: Palantir compounds revenue at 35% a year for five straight years, which it has never done at anything close to this scale, and the market still pays 25 times sales for it in 2031, which would leave it as one of the most expensive large software stocks in the market after five more years of maturing. Miss on either and you are underwater, in some cells badly.
And the growth itself has two engines that face different risks. The government side depends on defense and intelligence budgets staying generous and on Washington’s appetite for outside software contractors, both of which have been favorable but neither of which is guaranteed through an administration change or a budget fight. The commercial side depends on the AI platform continuing to convert pilots into large, expanding contracts faster than the hyperscalers can bundle a good-enough version of the same capability into cloud deals customers are already paying for. A deceleration on either engine, even to a still-healthy 20%, is the kind of number that resets a 60-times-sales stock in a hurry. This is the single-stock version of the argument I made about whether the whole AI complex is priced for a future that arrives late.
Why I’m not shorting it either
Being expensive is not a catalyst. A stock like this can stay expensive for years, and the borrow to short it is costly, and the momentum is real. I learned that lesson the hard way betting against names that “had to” come down and then did not for eighteen months. Palantir has a durable product advantage, a customer base that does not rip out its data platform casually once it is embedded, and a retail shareholder base that treats every dip as a gift. That is not a setup I want to be on the other side of.
What I have done instead is nothing, which is its own decision. I do not own it and I am not buying it here, because the price already assumes a near-perfect decade and leaves me no margin for the ordinary things that go wrong. The level where I would get interested is closer to 25 or 30 times forward sales, which sounds absurd until you remember that is still a richer multiple than almost any other software company earns. The stock has been there before. It may get there again, either by falling or by growing into it. I would rather wait for that than pay for a forecast where everything has to break my way.
If you want to pressure-test the business quality separately from the price, run the ticker through the free quant rating tool and look at the Growth and Profitability categories on their own, then look at Valuation. The split is the whole story, and Palantir is the purest expression of it in a single stock.
Financial disclaimer: The content on StockVane is for educational and informational purposes only and should not be construed as professional financial advice. Stock market investing involves risk of loss.