After Monday’s closing bell in New York, Palantir Technologies delivered a quarter of a kind that rarely appears in the history of listed software companies. Revenue rose 93 percent to 1.935 billion dollars. Net income came in at 1.062 billion dollars — more than the entire revenue the same company booked in the year-ago quarter. Full-year guidance was raised for the second time this year. Chief executive Alex Karp reached for adjectives that do not normally survive a legal review of an earnings release.
And the stock is still down for 2026. Even including the roughly 16 percent jump in Tuesday’s pre-market session, Palantir trades below the price at which the year began. Before the numbers landed, the gap was almost 30 percent. That is not a contradiction. It is an arithmetic problem — and it is the single most important arithmetic problem of this market year.
What Palantir Actually Reported
The numbers first, because everything else rests on them. Quarterly revenue of 1.935 billion dollars came in well above the consensus estimate of roughly 1.80 billion. Diluted earnings per share were 0.41 dollars against expectations of 0.34 to 0.35. GAAP operating income reached 912 million dollars, a 47 percent margin; on an adjusted basis it was 1.19 billion dollars and 62 percent.
Adjusted free cash flow hit 1.220 billion dollars — 63 percent of revenue, and the first quarter in the company’s history above one billion. Operating cash flow of 1.216 billion sits almost exactly alongside it, which matters: this is not a one-off adjustment or an accounting artifact but money arriving.
The composition is more remarkable than the total. US commercial revenue grew 149 percent year over year to 764 million dollars, and 28 percent sequentially. US government revenue rose 90 percent to 809 million. Together that is 1.573 billion dollars, or 81 percent of group revenue, from a single country, growing at 115 percent. Total remaining deal value stands at 13.1 billion dollars, up 83 percent. Customer count rose 24 percent to 1,049, and in US commercial by 35 percent to 653. Net dollar retention is 157 percent, meaning last year’s customers spent roughly half again as much this year. The industry’s standard growth-plus-margin gauge, the Rule of 40, scored 155. Forty is considered good.
For the full year the company now expects 8.150 to 8.158 billion dollars in revenue, growth of about 82 percent, and adjusted free cash flow of 4.5 to 4.7 billion. Third-quarter guidance of 2.160 to 2.164 billion compares with a street estimate near 2.0 billion.
The Calculation Nobody Runs
Palantir entered 2026 at 177.75 dollars. On Monday evening, immediately before the release, it closed at 125.65. That is a 29.3 percent decline over seven months. Across the same stretch, quarterly revenue rose 93 percent.
It follows necessarily that the price-to-sales multiple must have contracted by roughly 63 percent. When the numerator falls 29 percent and the denominator nearly doubles, a little over a third of the ratio survives. The calculation ignores dilution from stock-based compensation, which at Palantir is not trivial; it does not change the order of magnitude.
The picture sharpens if you start from the valuation at the beginning of the year. Palantir went into 2026 at roughly 62 times trailing sales and 155 times trailing earnings. Against 2025 revenue of about 4.48 billion dollars and the 8.15 billion now guided for 2026, that means something specific: even if the share price had stood perfectly still all year, the price-to-sales multiple would have fallen from 62 to about 34 through growth alone. Because the price also fell 29 percent, it now sits nearer 24.
There is the core of it. Most of the total de-rating came from the business growing into its valuation — and the smaller remainder came from the market letting out additional air. That smaller remainder was the shareholder’s return, and it was negative. At a company compounding at 93 percent, the trailing and the forward multiple describe what are effectively two different companies. Confuse the two and neither the price chart nor the valuation will make sense.
Why a High Multiple Is a Duration Bet
A high valuation multiple means nothing other than this: most of the enterprise value consists of cash flows that do not yet exist. At 62 times sales, the center of gravity of expected earnings sits a long way into the future. In the language of the bond market, such a stock is a very long-dated instrument — and long-dated instruments react most violently to changes in the discount rate.
The discount rate has moved considerably in 2026. The ten-year Treasury yield closed July at 4.75 percent and stood near 4.69 percent on Monday, after touching its highest level since January 2025 the week before. The thirty-year is trading around levels last seen before the financial crisis. The Federal Reserve under Kevin Warsh held its policy rate at 3.50 to 3.75 percent in late July, but over three dissents that wanted an increase.
When the long end rises half a percentage point, a cash-flow stream whose center of gravity sits a decade out loses meaningful present value — entirely regardless of what the current quarter reported. That is the cleanest technical explanation for how a company can compound at 93 percent and still lose money for its owners. It also explains why the same stock fell back in May, when the first quarter had delivered 85 percent growth, 133 percent in US commercial, and the largest full-year guidance raise in company history. The shares went from 144.45 to 133.79 dollars the following day. It was not a verdict on the business then either.
The Counter-Test Arrives Tonight
On Tuesday evening SpaceX reports the first quarterly numbers of its life as a public company. The group, which acquired chatbot developer xAI in an all-stock transaction in February and has since reported across three segments — space, connectivity and artificial intelligence — listed on the Nasdaq on June 12 at 135 dollars per share in the largest initial public offering in history. It first closed at 160.95 dollars and peaked at 225.64 on June 16. It last traded at 114.53. That is down roughly 49 percent from the high and about 15 percent below the offer price.
Second-quarter revenue is expected around 6.82 billion dollars. On earnings, analyst estimates run from a loss of 1.26 dollars per share to a profit of 0.33 — a range that is less a forecast than an admission. For comparison, the first quarter produced 4.69 billion dollars of revenue and a 4.28 billion dollar loss. The satellite business earned 1.19 billion in operating income on 3.26 billion of revenue. Space produced 619 million of revenue against a 662 million operating loss; artificial intelligence produced 818 million of revenue against a 2.47 billion operating loss. Starlink counted 10.3 million subscribers across 164 countries at the end of March, roughly double a year earlier.
So within 24 hours the market gets two companies suffering the same condition in opposite forms. Palantir delivers a 62 percent adjusted operating margin and more than a billion dollars of quarterly free cash flow, and gets cheaper anyway. SpaceX trades at roughly 49 times expected revenue, burns billions a quarter, and has to demonstrate tonight that the second number is moving toward the first. Both share prices are being driven in 2026 by the same variable: not by what gets reported, but by what was already in the price.
Why This Is Not a Bear Market
An objection belongs here, without which the whole story hangs crooked. The broad market is not falling. The Dow Jones Industrial Average closed Monday up 693.38 points, or 1.32 percent, at a record 53,178.41. The S&P 500 gained 1.48 percent to 7,600.50 and the Nasdaq Composite 2.1 percent to 25,913.9. Tuesday morning futures were higher again, with the volatility index at 15.63. The driver was geopolitics rather than earnings: President Trump called off a planned military strike on Iran and announced fresh negotiations, sending Brent down almost five percent to around 84 dollars and West Texas Intermediate toward 81. Tehran denies direct talks but points to progress through the Omani channel. OPEC and its partners have also approved another production increase.
What is happening, then, is not a devaluation of equities but a devaluation of valuation premiums. Caterpillar rose nine percent pre-market on Tuesday after quarterly revenue topped 20 billion dollars for the first time — an industrial with short earnings duration and a moderate multiple. On the same morning, a company compounding in triple digits is down on the year. Capital is not leaving the market; it is moving within the market from long earnings duration to short. Mistake that process for pessimism and you sell the wrong securities.
The scale of the confusion shows up in price targets. Morgan Stanley’s bull case for Palantir sits at 382 dollars; Jefferies carries a sell rating with a 70 dollar target. What separates those two houses is not a different view of the business — both expect rapid growth. What separates them is purely an assumption about what multiple that growth deserves in three years. The dispersion itself is the message.
Where American Investors Meet the Same Pattern
Tonight’s other report makes the point domestically. Advanced Micro Devices is expected to post roughly 11.3 billion dollars of revenue and 1.61 dollars of adjusted earnings per share, with data center revenue near 6.5 billion — around 4 billion from server processors and 2.5 billion from AI accelerators. AMD has beaten revenue estimates in each of its last eight quarters and earnings estimates in seven. That record is precisely the problem: a company that never misses is priced as though it never will, which leaves the multiple to do the work in either direction. The same logic governs Nvidia, CrowdStrike and every large-cap name whose valuation is a statement about 2030.
There is a tax dimension American investors should think through before acting on any of this. Long-term capital gains, on positions held more than a year, are taxed at preferential rates topping out at 20 percent; short-term gains are ordinary income at up to 37 percent. A de-rating that plays out over eighteen months therefore rewards patience twice — once through the arithmetic of growing into a multiple, once through the rate schedule. Conversely, if a position is being sold at a loss to offset gains elsewhere, the wash-sale rule disallows the deduction if a substantially identical security is repurchased within 30 days. High-multiple growth names are exactly where investors are most tempted to sell and immediately buy back, and exactly where that instinct is most expensive.
The Case Against This Reading
Four objections deserve to be taken seriously. First, the link between the long end of the curve and Palantir’s share price is plausible, not proven. It is equally possible that all we are watching is the unwinding of a 2025 excess that has little to do with bond yields. Second, there is a more mundane explanation requiring no interest-rate theory at all: company insiders, the chief executive included, have sold shares steadily for months with essentially no offsetting purchases.
Third — and this is the most important objection — the arithmetic cuts both ways. If the multiple determines the return, then a falling multiple does not make the stock worse but mathematically better. The same arithmetic that hurt in 2026 works for the holder in 2027, provided growth persists. Fourth, and here is the genuine weak point: by the company’s own guidance, growth does not persist at this level. Set the full-year number against the first half and the implied second-half growth rate is about 77 percent, after roughly 89 percent in the first half. That is still extraordinary, but it is a deceleration — and growing into a multiple only works as long as the growth rate exceeds the rate at which the valuation is deflating.
What Investors Should Take From This
A stock’s return decomposes into three parts: growth in earnings, change in the valuation multiple, and dividends. For Palantir in 2026 the first term has been 93 percent, the second roughly minus 63 percent, and the third zero. The result was negative. Most investors watch only the first term, because that is the one in the headlines — and the second is the one that did the work this year.
From that follows a rule applicable to any growth stock, in any sector, on any exchange. When you buy, write down two numbers: the multiple you are paying today, and the multiple you are still assuming when you sell. The difference between them is the part of your return the company cannot influence — only the market can. A chief executive can double revenue. No chief executive can decide that the shares will still change hands at 62 times it.
One thing more. The entry multiple is the only variable in this equation entirely within the buyer’s control. Palantir shareholders did not lose money in 2026 because the company did something wrong this year — it did very nearly everything right. They lost money because the price left behind by the enthusiasm of 2025 already contained a better year than the one that actually arrived. The business owed the valuation more than even 93 percent growth could repay. Tonight, when SpaceX reports for the first time, the same test runs again — this time on a company that has yet to produce any evidence for its multiple at all.
Try TradingView Free for 30 Days
Plus get a $15 discount on your first subscription through this link.
Read more in our topic hub: Topic Hub: AI Stocks Investing 2026
