Palantir Posts Record 93% Revenue Growth as US Commercial Business Surges 149%, Raises Full-Year Guidance by Largest Margin Ever
Q2 2026 earnings call, August 3, 2026
Palantir Technologies delivered its strongest quarter on record, with total revenue growing 93% year-over-year to $1.935 billion, and used the results to push through its largest-ever full-year guidance raise, lifting the 2026 revenue outlook to a midpoint of $8.154 billion, up 82% year-over-year and 11 points above the prior quarter's guidance. The company's US business, now over 81% of total revenue, grew 115% year-over-year and 23% sequentially, with US commercial revenue accelerating to 149% growth. Chief Revenue Officer Ryan Taylor called the results "unprecedented, but entirely unsurprising," framing them as confirmation of a shift in enterprise AI spending that management has been forecasting for years.
US Commercial Bookings Nearly Triple Prior Record
The standout metric was US commercial bookings. Palantir closed $2.132 billion in US commercial total contract value in the quarter, up 153% year-over-year and 81% sequentially, and nearly $800 million above its previous high-water mark. On a dollar-weighted duration basis, US commercial TCV growth was 271% year-over-year. Trailing 12-month US commercial bookings reached $5.964 billion, up 117% from the prior 12-month period. CFO Dave Glazer raised the full-year US commercial revenue guide to "in excess of $3.424 billion," implying growth of at least 134%. Net dollar retention rose to 157%, up 700 basis points sequentially, and total remaining deal value hit $13.1 billion, up 83% year-over-year. Palantir also reported record deal volume: 220 deals worth $1 million or more, 98 deals above $5 million, and 73 deals above $10 million, all all-time highs.
The "Sovereign AI" Pitch and the Case Against Token-Based Models
Much of the call was devoted to Palantir's positioning against what it characterizes as an unsustainable enterprise AI model built around token consumption from frontier labs. Taylor argued that "enterprises that are not using Palantir are seeing their token meters spinning endlessly just to get slop without any correlation to value," and warned that companies interacting directly with frontier models are effectively handing over their competitive advantage as training data. CTO Shyam Sankar extended the argument technically, describing AIP as infrastructure that allows customers to run "an automated model factory that runs inside the customer security boundary and accumulates intelligence in weights they control." The framing matters for the investment case because it positions Palantir not as a systems integrator on top of third-party models, but as the layer that lets enterprises fine-tune and own their own models — a higher-margin, stickier proposition than simple API reselling if the pitch holds up with customers.
Benchmarks Are Broken, According to Palantir's Own Testing
Sankar disclosed a specific and unusually concrete technical claim: within 24 hours of integrating Nvidia's Nemotron Ultra model into its stack without any post-training, Palantir found five production tasks where the vanilla model outperformed frontier models. "This underscores that a handful of common benchmarks can be gamed," Sankar said, arguing that the industry has moved from an era of "benchmaxing" — optimizing for popular public benchmarks — to what he called "benchmaking," where customers build proprietary benchmarks tied to their own operational reality and Palantir's platform continuously optimizes and fine-tunes against them. This is a meaningful data point for investors trying to gauge whether frontier-model scale advantages actually translate into enterprise value, and Palantir's answer is an emphatic no.
A Competitive Bake-Off Palantir Says It Won Outright
Sankar recounted an anecdote in which a major Silicon Valley technology company ran a head-to-head test pitting a frontier lab's deployment team against Palantir's forward-deployed engineers on a ticketing automation problem. According to Sankar, "the lab picked a ticketing automation problem and failed to deliver anything of value against it," while Palantir's team built agent swarms that generated a $10 million annual contract. "The lab was shown the door," he said, adding "same customer, same time line, same models. The only difference was AIP and Palantir's unique FDE tradecraft, and it was determinative." The claim is unverifiable from the outside and self-serving by nature, but it is consistent with management's broader argument that deployment methodology, not raw model capability, is now the binding constraint on converting AI spend into economic value.
Government Business Still Underpenetrated
US government revenue grew 90% year-over-year and 18% sequentially to $809 million, and Sankar noted that Maven secured its first program of record this quarter, with a government program choosing Maven as its operating platform. The Maven builder base has grown to over 25,000 users across uniformed service members, civilians, contractors and companies. Despite the growth, management pointed out that trailing 12-month Department of War revenue remains under 25 basis points of the Pentagon's total budget, an explicit signal that the company sees itself early in penetrating its core government market rather than approaching saturation.
Margins, Cash Flow and the Guidance Raise
Adjusted free cash flow hit $1.22 billion, a 63% margin and 115% year-over-year growth, and the company surpassed $1 billion in each of GAAP net income, adjusted free cash flow and adjusted operating income for the quarter. GAAP net income was $1.062 billion, a 55% margin, with GAAP EPS of $0.41. The combined revenue growth and adjusted operating margin — Palantir's "Rule of 40" metric — reached 155%, marking the twelfth consecutive quarter of expansion. Adjusted gross margin came in at 86%, slightly compressed by Palantir taking on cloud hosting costs for a government customer, which management said would drive faster time-to-value and greater long-term efficiency rather than signal margin pressure. The company raised full-year adjusted operating income guidance to $4.889 billion-$4.897 billion and adjusted free cash flow guidance to $4.5 billion-$4.7 billion, while guiding Q3 revenue to $2.16 billion-$2.164 billion. Palantir ended the quarter with $9.2 billion in cash and short-term treasuries.
Karp's Framing: Sovereignty as Strategy and as Ideology
CEO Alex Karp used his remarks to draw a sharp line between Palantir's model and what he described as a "parasitic" software model common in Silicon Valley, arguing that being "fully aligned" with clients — as opposed to extracting value through lock-in — is the structural reason Palantir is winning share now. He described a recent "sovereignty boot camp" that drew unexpected demand from both new and long-tenured Foundry customers, some of whom are now migrating to the full sovereign AI stack. "The NDR number is anomalously strong... and it's also going to become even more positive because some of our older partners we haven't really interacted with, they also showed up," Karp said. He also set an explicit growth target, stating he is "driving the business to grow at a rate equal or above what we have in US commercial for the next 18 months," a notably aggressive bar given commercial growth of 149% this quarter. Investors should treat this as a stated management aspiration rather than formal guidance, but it signals confidence that the current growth rate is not viewed internally as a peak.
Karp's commentary veered well beyond standard earnings-call territory into ideological territory about AI, labor and sovereignty, including extended remarks framing frontier labs as seeking to "colonize" enterprise clients. While colorful, the rhetoric is consistent with the company's long-standing brand positioning rather than new financial information, and investors parsing the quarter for signal should weight the hard numbers — the guidance raise, the bookings acceleration, and the Nemotron benchmarking disclosure — more heavily than the philosophical framing.