Palantir Grew 96% and Signed 11% More — Both Are True
The Sovereign AI label is one quarter old; the bookings line is older, and quieter, and it is the number that will decide whether 45x revenue is cheap
PLTR · Earnings Call · 2026-11-02
The gap between the revenue line and the signature line
On the numbers management chose to headline, this was a historic quarter. Revenue of $2.313 billion, up 96% year-over-year and 20% sequentially — “the highest ever reported year-over-year revenue growth rate” — David Glazer, Chief Financial Officer · 2026-11-02 in the company's public life. U.S. revenue of $1.937 billion, up 119%, now 84% of the whole. And for the first time, commercial revenue eclipsed government revenue: $984 million of U.S. commercial against $953 million of U.S. government, growing 148% and 96% respectively. GAAP operating income crossed a billion. Rule of 40 hit 158%, the thirteenth consecutive expanding quarter.
Then, three paragraphs into the CFO's script, the other number. Commercial TCV bookings of $1.652 billion, up 17%. U.S. commercial TCV of $1.463 billion, up 11%. In a quarter where U.S. commercial revenue grew 148%.
We closed $1.652 billion in commercial TCV bookings in the third quarter, representing 17% growth year-over-year. ... We signed multiple 8-figure deals in the quarter that were shorter in duration than usual, as customers rapidly adopted initial Sovereign AI capabilities in their enterprise.
Nothing here is fraudulent, and the backlog is genuinely strong — total remaining deal value up 55% to $13.4 billion, RPO up 88% to $4.9 billion, trailing-twelve-month U.S. commercial TCV of $6.114 billion up 62%. But read together, the two lines say something specific: revenue is being harvested from expansions of contracts already signed faster than new paper is being written. Net dollar retention of 165%, up 800 basis points sequentially, is the confirmation — expansion, not new logos, is carrying the quarter. Management's explanation is bandwidth, not demand: “we don't have even the bandwidth to do that” — Alexander Karp, Chief Executive Officer · 2026-11-02. And Karp is explicit that the unpersuaded are not a priority: “this is a market that is growing so rapid. We don't need the people who aren't convinced.” — Alexander Karp, Chief Executive Officer · 2026-11-02 That is either the most honest thing said on the call or the most convenient.
"One quarter old" — on whose clock?
Karp and Ryan Taylor each insist the sovereign revolution is a single quarter old. The keyword record gives them partial credit: Sovereign AI appeared as Palantir's top-ranked term only in the August call, and remains near the top now as Sovereign AI. The label is new. The machinery underneath it is not. In August, Shyam Sankar was already arguing that “the benchmarks are wrong. I mean the benchmarks are right for what the benchmark is measuring, but that's not my business” — Shyam Sankar, Chief Technology Officer · 2026-08-03. In August, Karp was already describing “a sovereign boot camp after this kind of revolution exploded” — Alexander Karp, Chief Executive Officer · 2026-08-03. Dan Ives asked about boot camps in August; he asked about boot camps again, and got the same answer with a hotter metaphor. What actually changed is the intellectual framing, crystallized in a genuinely fresh keyword this quarter: atomic unit. The claim that the unit of the enterprise is the workflow, not the knowledge worker, is the load-bearing idea, because it explains how revenue can compound without headcount compounding. It is also the same argument Sankar made in August with less compression.
The pivot-to-ACV deal motion is real and consistent: a $3 million pilot converting to $29 million of ACV, a $1.6 million pilot to roughly $21 million, a $1 million pilot to $15 million across three quarters. And there is a new vocabulary of scale — agent swarm per wellhead, per ICU patient, per customer-of-customer.
What vanished matters as much. American worker was the company's single top keyword as recently as 20254, part of a blue-collar, populist register that dominated half a dozen calls. It is gone. So, largely, is frontier model, which ranked third in August and now sits among the quarter's decliners. Beating the frontier has been demoted from thesis to anecdote — a 30-billion-parameter model outperforming a 3-trillion-parameter one in a constrained PACOM exercise. The pitch has moved from model performance to enterprise economics. Harder to falsify, and harder to verify.
The tape already voted — on the other side of the stack
Over ninety days, global thematic money has been leaving AI infrastructure: AI data center down roughly 22% with more than fifty decliners in the cluster, high-bandwidth memory down 29%, critical IT load down 47%. It has been arriving in the application and retention layer: net retention rate up about 20% across a cohort including CrowdStrike and Varonis, software-development life cycle up 32%, platform-of-choice up 29%. Palantir is squarely on the winning side of that rotation, +57.9% over ninety days.
But the company's own tape is less triumphant than its income statement. The stock peaked at $209.05 on October 9 and closed the pre-print session 14.7% below it. More telling, the multiple has been paying down even as the business accelerated: price-to-revenue has fallen from 111.6x a year ago to 45.5x, and price-to-free-cash-flow sits at 111x after 424x. A 96% growth quarter against a 45x sales multiple is a different conversation than the same quarter against 111x. The business earned the de-rating down; the question is whether the bookings line lets it keep earning.
What to watch
Guidance was raised across the board — full-year revenue of $8.530–8.534 billion for 91% growth, a nine-point raise, U.S. commercial above $3.559 billion at 143%, $4.8–5.0 billion of adjusted free cash flow. Q4 revenue of roughly $2.65 billion implies about 15% sequential growth, a visible step down from the 20% just delivered, which is the arithmetic of comps rather than a forecast of decay.
One geographic oddity deserves a line: international commercial revenue was $184 million, up 21% year-over-year but just 1% sequentially. The sovereign revolution, by the numbers, is an American event. Karp told us in February he was admonishing Germany in German; nothing on this call suggests Germany listened.
The discipline is intact — gross margin of 84.7%, up 3.9 points year-over-year, and stock-based compensation holding near 14% of revenue while growth compounds. Karp's February line — “inexplicable growth in revenue. But not inexplicable growth in customers” — Alexander Karp, Chief Executive Officer · 2026-02-02 — has aged into the central tension. The customers are explicable. The bookings are explicable. The revenue is the mystery, and it is being financed by expansion of an installed base whose net retention is now doing the heaviest lifting in the model. If U.S. commercial TCV reaccelerates next quarter, the bandwidth story wins outright. If it prints another low-teens number against triple-digit revenue growth, the short-duration eight-figure deals are the tell, and 165% NDR becomes the single line item the entire $410 billion market cap is standing on.