Palantir's US commercial business overtakes government for the first time

The quarter in numbers

Palantir reported third-quarter 2026 results after the close on Monday 2 November. Revenue came in at $2.313 billion, up 96% year over year, against Wall Street estimates of about $2.17 billion according to Bloomberg data cited by Yahoo Finance. CNBC put the consensus at $2.18 billion. Either way, the company blew past $2 billion in a quarter for the first time.

Adjusted earnings per share were $0.44, beating the $0.40-$0.41 consensus depending on the data provider. Net income more than doubled to $1.13 billion, or 44 cents per share, from roughly $476 million, or 18 cents per share, a year earlier, per CNBC.

The number that matters for anyone tracking AI adoption is this one: US commercial revenue reached $984 million, up 148% year over year, and for the first time exceeded revenue from Palantir's US government business. In the shareholder letter published 2 November, the company says sequential US commercial growth was 29% for the quarter.

Why commercial crossed government

Palantir spent two decades as a government contractor with a commercial side project. That has inverted. The company attributes the shift to demand for what it calls sovereign AI: the layer that sits under a model and turns an organisation's own data and operations into something an AI system can act on.

In the shareholder letter, the company attributes the surge to demand for the sovereign AI systems it began building before large language models existed, describing that appetite in the United States as far from satisfied and its window as a matter of quarters rather than years.

That framing is a direct attack on the way most enterprises have bought AI so far. Palantir's public position, repeated by its chief executive, is that the market is splitting into companies paying millions for model tokens and getting what it calls slop, and companies buying infrastructure that constrains what the model can do and records what it did.

Whether you buy that pitch or not, the revenue line is real. A 148% year-over-year jump in US commercial sales is not a pilot programme. It is enterprise budget moving.

AI Lookout has tracked the same enterprise pattern in [ServiceNow's expanded ChatGPT and Codex integration](/blog/openai-and-servicenow-expand-partnership-to-bring-chatgpt-and-codex-into-enterprise-workflows/) and in [Cohere's alliance with PwC](/blog/cohere-pwc-enterprise-ai-alliance/).

Karp, the pitch, and the valuation problem

Chief executive Alex Karp gave CNBC the quote of the quarter: "I don't believe anyone's had consistent results like this in the history of software AI, ever." He also claimed the business is succeeding without salespeople, a line that is either an insight into product-led enterprise adoption or a piece of theatre, depending on your patience.

On the earnings call he used a different metaphor, according to Yahoo Finance: "We've gone from being the most attractive person in the physics department to being the captain of the football team."

The market was less moved. Shares dipped about 3% in extended trading after the release, CNBC reported, then fell around 4% during Tuesday's session before trimming losses as investors weighed the shape of the guidance rather than the size of the beat. Yahoo Finance notes the stock was on pace to turn negative for the year.

That reaction is the interesting part. A company growing 96% and doubling net income is being judged on whether the next quarter grows 15% or 20% sequentially. When a stock is priced for perfection, arithmetic becomes the enemy.

The guidance and the deceleration question

Palantir guided fourth-quarter revenue to $2.65 billion, above analyst estimates. The problem is what that implies: 15% quarter-over-quarter growth, down from 20% in the same period last year, per Yahoo Finance.

For the full year, the company now expects revenue of $8.530 billion to $8.534 billion, ahead of the roughly $8.2 billion consensus, and raised its US commercial revenue target to more than $3.559 billion.

A single point of sequential deceleration is not a collapse, and any company lapping a 96% growth quarter is going to face harder comparisons. But the AI trade has become intolerant of second derivatives. AI Lookout covered the wider mood in [Capital Economics' warning that the AI boom is turning](/blog/capital-economics-says-hide-in-defensive-sectors-as-the-ai-boom-turns-to-bust/), and this quarter shows how quickly good numbers get graded against expectations instead of against last year.

The other structural fact in the numbers is that Palantir is no longer a government story. If US commercial keeps compounding at anything near this rate, the government business becomes the ballast rather than the engine, which changes both the growth profile and the political risk profile of the company.

What enterprise AI buyers are actually paying for

Strip out the earnings theatre and the quarter says something concrete about how large organisations are buying AI in late 2026. They are not buying tokens. They are buying integration: software that touches their own data, their permissions, their audit trail, and their existing workflows.

That is a harder sale than a chatbot and a slower one, which is why the sales cycle commentary matters. It is also stickier. Once a Palantir deployment is wired into an insurer's claims process or a hospital's operations, ripping it out is a programme, not a procurement decision.

The risk sits on the other side of the ledger. Palantir's revenue is concentrated in a relatively small number of very large customers, and its growth depends on those customers expanding usage rather than on a broad base of new logos. A 148% commercial growth rate flatters that model while it works. It also means a handful of paused expansions can move the number quickly.

The company also carries a defence and intelligence footprint that most software peers do not, including its Maven Smart System becoming an official Pentagon programme earlier this year, per Yahoo Finance. That is a moat in Washington and a complication everywhere else.

What to watch

Three markers for the next quarter. First, whether US commercial growth holds above 100% year over year on a much larger base; that is the difference between a genuine platform shift and a burst of catch-up spending.

Second, the sequential guide. If the fourth quarter lands at or above the implied 15%, the deceleration story dies. If it undershoots, expect the valuation argument to harden regardless of the headline beat.

Third, how the rest of enterprise software reports over the next three weeks. Palantir's pitch is that it is taking budget that would otherwise go to model providers. If other enterprise vendors show the same commercial acceleration, the AI spending story is broadening. If Palantir is alone, it is winning a category, not riding a wave.