Sixty-Seven Times

Let us begin by giving Palantir its due, because the usual way of writing about expensive stocks — sneering at them as hype, as emperors with no clothes — does not fit here and would mislead you. Palantir is a real, formidable, deeply impressive company. It makes software that governments and corporations use to fuse oceans of messy data into decisions, it has become one of the most important defense-technology contractors in the Western world, and unlike most of the companies riding the AI wave, it actually makes money — a great deal of it, at extraordinary margins. In its most recent quarter revenue grew 85% year over year; its US commercial business is compounding at well over 100%; it earns gross margins north of 84% and is solidly, GAAP-profitably in the black. Management has raised full-year guidance toward $7.66 billion in revenue, implying growth around 70%. These are not the numbers of a fraud. They are the numbers of one of the best software businesses of its generation.

That is the setup for the only question that matters, and it is not "is Palantir a good company." It is: how much is a good company worth? Because the price the market has put on this particular good company is one of the most extreme valuations ever attached to a large, real business — and the people in the best possible position to judge whether it is justified have answered, with their own money, in a way that should give every shareholder pause.

The number

Palantir trades at roughly sixty-seven times its sales. Not its earnings — its sales, the top line, before a single cost is subtracted. Stretch to earnings and the multiple runs past 150 times, and on some measures well beyond that.

To understand how far outside normal experience this is, you need a sense of the gravity that usually governs software valuations. A fast-growing, profitable software company is typically considered richly valued at, say, ten or fifteen times sales. Twenty is aggressive. The legendary investor Scott McNealy, whose Sun Microsystems traded at ten times sales at the peak of the dot-com bubble, once delivered a famous lecture to anyone who would buy a stock at that multiple: at ten times revenues, he pointed out, to return your money over ten years you would need to pay zero costs, zero taxes, retain every customer, and pay out all revenue as dividends — assumptions he called "ridiculous." He was talking about ten times. Palantir trades at nearly seven times that.

A multiple of sixty-seven times sales is not a valuation in the ordinary sense. It is a statement that the company will not merely grow but grow enormously, for many years, almost without stumbling, and then keep most of what it earns — and that nothing will go wrong along the way: no recession in government budgets, no competitor, no slowdown in the AI spending that powers the story, no disappointing quarter. The valuation does not price success. It prices perfection, sustained for the better part of a decade. And perfection, as an investment thesis, has a poor historical record, not because great companies fail, but because even great companies occasionally merely do well — and a stock priced for perfection has nowhere to go when "well" arrives instead of "flawless."

The market has already offered a preview of what that looks like. Even amid all the triumphant growth, Palantir's shares are down more than a third from their late-2025 high near $207, and off around 20% in 2026. The business kept compounding; the multiple simply began, gently, to exhale. There is a great deal more exhaling available.

The sellers

Here is where the story stops being about valuation theory and becomes about something more concrete, and more telling: what the insiders are doing.

In 2024 alone, Palantir's executives and directors cashed out more than $4 billion of stock, and the selling has continued through 2025 and into 2026. This is not a couple of executives trimming around the edges; it is the entire inner circle, in concert. The chief executive, Alex Karp — brilliant, combative, the public face and intellectual engine of the company — sold around $2 billion of Palantir shares in 2024, much of it around the election, and has filed to sell hundreds of millions more. Peter Thiel, the company's co-founder and most famous backer, the man whose conviction predates and underwrites the entire enterprise, sold roughly $1.5 billion of stock. Co-founder Stephen Cohen sold about $310 million; the chief technology officer, Shyam Sankar, more than $420 million since 2024. The people with the deepest possible understanding of Palantir's prospects, its pipeline, its contracts, and its true competitive position have spent the past two years converting billions of dollars of their belief into cash.

The standard defenses apply, and they are not nothing. Much of this selling is conducted through pre-arranged 10b5-1 plans; founders and executives have every right to diversify after a stock rises 2,000%; selling shares is not the same as disbelieving in the company. All true. But the same logic that applied to CoreWeave's founders applies here: a 10b5-1 plan governs the timing of a sale, not the decision to schedule billions of dollars of selling in the first place. When the founders and officers who know a company best cash out more than four billion dollars in a single year while the stock trades at sixty-seven times sales and the marketing department tells retail investors the dips are a gift, the most informed money and the least informed money are doing precisely opposite things. The tell is not that they are selling. It is the size, against that price.

The trap of the great company

The reason this matters more than an ordinary overvaluation is psychological, and it is the trap Palantir lays for exactly the most thoughtful investors.

It is easy to refuse to buy an obvious junk stock at an absurd price. It is very hard to refuse to buy a wonderful company at an absurd price, because every instinct you have been taught — buy quality, hold winners, don't bet against greatness — pulls you toward the purchase. The bulls are not wrong that Palantir is special. Rosenblatt and other analysts who call every dip a buying opportunity are not lying about the earnings momentum or the defense-AI synergies; those are real. The danger is subtler than hype. It is that a genuinely great company can be a genuinely terrible investment if you pay a price that already contains a decade of its greatness — and that the very quality that makes the company easy to admire makes the stock easy to overpay for.

The history of markets is littered with superb businesses that were ruinous purchases at the wrong multiple. The "Nifty Fifty" of the early 1970s — Coca-Cola, Xerox, Polaroid — were the unquestioned great companies of their age, "one-decision" stocks you were supposed to buy and never sell. Many of them were great, and kept growing for decades. Investors who bought them at fifty and eighty times earnings still spent years underwater, because the price had borrowed the future and the future takes time to arrive. Palantir, at sixty-seven times sales, is a one-decision stock for a new generation, sold with the same certainty.

None of this is a prediction that Palantir's business stumbles. It may well grow into a substantial fraction of its valuation; it is good enough that it might. But "grow into" is doing brutal work in that sentence. At sixty-seven times sales, even spectacular execution can leave the stock flat for years as the company's reality slowly catches up to its price, and any genuine disappointment — a budget cycle, a missed quarter, a cooling of the AI fever — can compress the multiple violently, exactly as the recent third-off-the-high drawdown hinted. The downside is not that Palantir is bad. The downside is that it is wonderful, and you paid as if it were wonderful and certain and immortal, and only one of those three things is true.

The company is real. The growth is real. The profits, refreshingly, are real. It is only the price that is a fiction — a fiction the insiders are quietly, methodically, billion by billion, selling into. When the people who built the cathedral are filing out the back while the ushers wave you toward the front pews, it is worth asking what they can see from the altar that you cannot see from the door.