Palantir Beats on Every Metric, and the Stock Wobbled Anyway

Palantir grew revenue 96% year over year to $2.31 billion in the third quarter, crossing the $2 billion mark for the first time in its history — and the stock still traded slightly down the next day. The reason: enterprise contract value, the clearest forward indicator of future sales, fell quarter over quarter. That tension between a blowout earnings report and a warning sign buried inside it is the real story of Palantir's Q3.

The company itself set the tone. Its press release bragged that the numbers were "crushing consensus expectations." Wall Street agreed on the surface. Revenue of $2.31 billion beat analyst estimates of $2.18 billion. Adjusted earnings per share came in at $0.44 against an expected $0.41. Revenue also grew 20% from the previous quarter. Management raised full-year revenue guidance, which under normal circumstances would send a stock soaring.

Retail investors were lukewarm on Tuesday instead. The explanation sits in the segment Palantir has spent years building as its next growth engine: enterprise sales.

What did the quarter actually deliver?

The headline numbers were unambiguous. Revenue grew 96% year over year to $2.31 billion, the first quarter above $2 billion in company history. Sequential growth was 20%. Adjusted EPS of $0.44 beat the $0.41 consensus. Guidance went up, not down.

The structural milestone matters more than the beats. Palantir reported $984 million in U.S. commercial revenue against $953 million in U.S. government revenue — the first time commercial contracts have surpassed government contracts in its home market. Commercial revenue grew 148% from the year-ago quarter. For a company long stereotyped as a defense-intelligence contractor, that crossover is a genuine pivot point in its business model.

Takeaway: Palantir is now, by revenue mix, an enterprise software company in the United States rather than a government vendor.

Why are enterprises signing up so fast?

The pull, according to the company, is AI infrastructure. Enterprises are buying Palantir's software to run frontier models (the largest, most capable AI systems from labs like OpenAI or Anthropic) and to custom-train open-weight models (AI models whose parameters anyone can download and modify) securely. The selling point is isolation: customers can deploy advanced AI without sharing their data or business processes with outside AI labs.

Palantir pairs that with what the industry now calls "forward-deployed" software engineering — engineers embedded with customers to customize the product on site. That services-heavy model, once criticized as unscalable, has become a differentiator as corporations struggle to turn AI models into working systems.

CEO Alex Karp, on the quarterly call with Wall Street analysts, described the demand in characteristically blunt terms: "What Ryan and I are seeing on the front line and commercial is this much quicker interaction, much quicker to deployment, shorter duration of the contract but then an ability to expand that contract dramatically."

Takeaway: companies want AI behind their own firewalls, and Palantir is currently one of the easiest places to buy it.

What is the warning signal?

Karp's sentence contains the caveat. Contracts are quicker to sign — and shorter in duration. Palantir is experiencing a phenomenon now common across enterprise AI: corporations will trial AI software fast, skipping their usual procurement bureaucracy, but they will also cancel just as fast if the software does not prove value.

The hard number behind that shift: Palantir closed $1.46 billion in total contract value (TCV, the combined worth of all deals signed in the period) from U.S. enterprises in the quarter. That is a decline from the record $2.13 billion in enterprise TCV the company booked in the previous quarter. A sequential drop of that size, in the segment management identifies as the growth engine, suggests decelerating sales — at least for one quarter.

Management did not dwell on that figure. The press release led with beats and raised guidance; the TCV decline required digging into the comparison with the prior quarter. One quarter is not a trend, but it is the metric most sensitive to the short-trial behavior Karp described on the call.

Takeaway: enterprises are experimenting eagerly, and experiments are smaller and easier to walk away from than traditional software deals.

What did Karp say about the risk?

Karp showed no concern about holding onto the enterprise business, deploying one off-color metaphor after another on the call.

"We've gone from being the most attractive person in the physics department to being the captain of the football team, kind of thing," he said, describing the influx of prospective corporate customers.

He was equally blunt about churn — in the opposite direction: "Conversely, you know, once you go Palantir, you don't go back, and that's just that's how it is, and you see it in our numbers."

The second claim is where a skeptic should press. The numbers show record revenue and record commercial growth; they do not yet show long-term retention on short-term AI trials, because those trials have not existed long enough to mature. Expansion across existing contracts — the "ability to expand that contract dramatically" Karp cites — will only be visible in future quarters. Karp's confidence is a forecast, not a reported fact.

Takeaway: the CEO asserts lock-in; the financials so far prove demand, not permanence.

Can Palantir sustain a 143% growth rate?

Management is betting that it can. Palantir raised its guidance for U.S. commercial revenue to "in excess of $3.559 billion, representing a growth rate of at least 143%" for the full year. That is an aggressive target for a segment that was a fraction of this size two years ago, and it assumes the short-trial behavior converts into expanded, longer contracts rather than cancellations.

The next earnings report carries unusual weight. Many enterprises historically sign their largest technology purchases in the final quarter of the year, so Q4 will show whether the $1.46 billion TCV dip was seasonal noise or the start of a plateau. It will also test Karp's "you don't go back" claim against the first cohort of expanded AI contracts.

The stakes extend past one company. Palantir has become the most visible proxy for whether enterprise AI spending translates into durable software revenue or a cycle of fast trials and faster cancellations. If short contracts keep expanding dramatically, the enterprise AI market is real and compounding. If they do not, a 96% growth rate built on brief engagements will be hard to defend — and Palantir's guidance will have been the first thing to outrun its numbers.