Revenue +93%, U.S. Commercial +149%, FY26 Raised by $498M: the Best Print in Company History Meets a 29.5% Single-Session Re-Rating — Maintaining Outperform

Results vs. Consensus

This was the strongest quarter Palantir has reported as a public company on every dimension that matters, and it was not close. Revenue of $1,935M grew 93% year over year, the highest rate the company has ever printed, at a revenue base nearly double the one that produced the prior record. The beat over the Street was $123M (+6.8%) and the beat over Palantir's own May guide was $136M (+7.6%), a magnitude squarely in line with the 6–8% pattern that has now held for three consecutive quarters. What separates this print from the last two is not the size of the beat but the fact that growth accelerated again at $1.9B of quarterly revenue: +70% in Q4 2025, +85% in Q1 2026, +93% in Q2 2026. The law of large numbers has been deferred, not repealed, but it has now been deferred for twelve consecutive quarters.

Year-over-Year Comparison

Sequential Comparison

Quality of the Beat

• Revenue. No acquisitions, no FX distortion, no accounting change. The entire $303M sequential add came from the United States, which grew from $1,282M to $1,573M while the international book added $12M. Within the U.S., commercial contributed $169M of the sequential add and government $122M. There is no single mega-contract explanation available either: the largest deal disclosed on the call was a three-year, nearly $370M total-contract-value expansion at a multinational technology company, which spread across three years is roughly $31M per quarter against a $303M sequential step. The growth is broad, and the deal-count data (220 deals above $1M, 98 above $5M, 73 above $10M, all records) says the same thing.
• Margins. This is the one place the print is genuinely mixed, and it is worth being precise about it. Adjusted gross margin fell to 86.3% from 87.9% in Q1 because Palantir took on cloud-hosting costs for a government customer, moving spend that the customer previously carried onto Palantir's own cost of revenue. Management framed the trade as buying faster time-to-value, cost certainty for the customer, and a wider future workflow footprint. That is a credible rationale, and it is also a structural change in the shape of the P&L: a business that hosts its government customers is a lower-gross-margin business than one that does not. Adjusted operating margin still expanded 150 bps sequentially to 61.7% because adjusted expense grew 14.2% against 18.6% revenue growth, so the operating leverage story is intact. But the composition shifted from "gross margin plus opex leverage" to "opex leverage carrying the load alone," and that is a less durable configuration.
• EPS. GAAP and adjusted diluted EPS both printed $0.41, and management disclosed that unrealized gains on the company's SpaceX holdings contributed $0.03 to the GAAP figure and $0.02 to the adjusted figure. That flows through the "other income, net" line, which went from $6.6M a year ago to $91.8M this quarter. Adjusting for it, the operating EPS beat is roughly 13% rather than 18%. Two further points on EPS quality cut in opposite directions. Against: the effective tax rate was 1.4%, so essentially none of the $1,081M of pre-tax income was taxed, and interest income of $77.5M on a $9.2B cash pile is another $0.03 of below-the-line contribution. In favor: adjusted net income of $1,047M is lower than GAAP net income to common of $1,062M, because Palantir's non-GAAP reconciliation applies a normalized tax charge of $297M that GAAP does not. That is an unusually conservative construction of a non-GAAP number and it deserves credit.
• Free cash flow. $1,220M of adjusted free cash flow at a 63.0% margin is the highest the company has produced in both absolute and percentage terms, up 115% year over year. Adjusted FCF exceeding GAAP operating income by $308M is a function of the stock-compensation add-back and working capital, not of financial engineering: the cash flow statement explicitly reverses $62M of realized and unrealized securities gains out of operating cash flow for the half. Contract liabilities added $257M of cash across the first half against $434M of receivables build, so the deferred-revenue tailwind did not carry the quarter. The one item to keep an eye on is receivables, which grew from $1,042M at year-end to $1,485M at June 30, faster than the revenue base did over the same span.

Segment Performance

Palantir also reports the book on a commercial-versus-government cut, which produces a different but equally informative picture: total commercial revenue of $945M grew 110% year over year and 22% sequentially, while total government revenue of $990M grew 79% and 15%. Government remains the larger of the two by $45M, but commercial is closing the gap at roughly 30 points of growth-rate differential per quarter and will cross over in the second half on any reasonable extrapolation. That crossover matters for the multiple: a business that is majority-commercial trades differently from one that is majority-government, and Palantir has spent five years arguing it should be valued as the former.

U.S. Commercial: Acceleration Off the Harder Comp

The U.S. commercial trajectory through the AIP era now reads +93% (Q2 2025), +121% (Q3 2025), +137% (Q4 2025), +133% (Q1 2026), +149% (Q2 2026). Q1's modest step-down was widely read as the beginning of the long-awaited deceleration; Q2 says otherwise, and it says so against a materially harder comparison base ($307M in Q2 2025 versus $255M in Q1 2025). At $764M of quarterly revenue, this segment alone is now larger than Palantir's entire company was as recently as two years ago, and it is compounding at a rate normally seen only in the first two years after product-market fit.

The composition of the growth is what makes it credible rather than merely impressive. U.S. commercial customer count reached 653, up 35% year over year but only 6% sequentially. Revenue in the same segment grew 28% sequentially. The arithmetic is unavoidable: roughly three-quarters of the sequential growth came from existing customers spending more, not from new logos. That is exactly what a 157% net dollar retention rate describes, and it is a fundamentally healthier growth mix than a land-heavy quarter would be, because expansion revenue carries lower acquisition cost and higher retention. Revenue per U.S. commercial customer rose from roughly $967K in Q1 to roughly $1.17M in Q2, a 21% sequential increase in account size.

The forward book is stronger still. U.S. commercial total contract value of $2,132M grew 153% year over year and 81% sequentially, landing $788M above the previous record bookings quarter. Trailing-twelve-month U.S. commercial TCV reached $5,964M, up 117%. Remaining deal value in U.S. commercial hit $6,238M, up 124% year over year and 27% sequentially, which holds the coverage ratio against in-quarter revenue at roughly 8.2x even as the denominator grew 28%. Holding an eight-times coverage ratio while the base compounds at this rate is the single most important durability signal in the release.

"We closed a record-setting $2.132 billion of U.S. commercial TCV bookings, representing growth of 153% year-over-year."
— David Glazer, Chief Financial Officer

"We closed 220 deals worth $1 million or more, of which 98 deals were worth $5 million or more and 73 deals were worth $10 million or more, record highs across the board."
— Ryan Taylor, Chief Revenue Officer and Chief Legal Officer

Assessment: We modeled FY26 U.S. commercial at +130–140% coming out of Q1 against a guide of at least +120%. First-half actuals of $1,359M represent +142% and the guide has moved to at least +134%, so we are raising our estimate to +140–150%, implying $3.5–3.7B for the year. The expansion-led composition of the growth, the 8.2x RDV coverage, and management's explicit statement that net dollar retention should move higher from here together push the deceleration question out of FY27 and into FY28. That is a full year further than we were willing to say in May.

U.S. Government: Ninety Percent, and the Platform Argument Lands

U.S. government revenue of $809M grew 90% year over year and 18% sequentially, extending an acceleration sequence that has now run for four consecutive quarters: +52% (Q3 2025), +66% (Q4 2025), +84% (Q1 2026), +90% (Q2 2026). We wrote in May that we were raising our FY26 government estimate to +55–65% from +35–45% and flagged that we had been badly too conservative. We were too conservative again. First-half government revenue of $1,496M is running at +87%, and the guide implies no meaningful step-down.

The qualitative development this quarter is more consequential than the growth rate. Palantir disclosed that a government program of record chose Maven as the platform on which it will operate, the first time an external program has been built inside Maven rather than alongside it. That is a category change: it converts Maven from a widely-deployed application into infrastructure that other programs are built on, which is the same transition that made Palantir's commercial ontology defensible. The builder count of over 25,000 across uniformed service members, civilians and contractors is the leading indicator for how many more such programs are plausible.

"We had our first program in the Maven platform launched this past quarter, where a government program of record chose Maven as the platform that they will operate their program in, taking advantage of our open data standards, ontology, developer tooling, peering, security and other platform primitives to go faster and deliver seamless experiences and capabilities to the department's chosen command and control platform."
— Shyam Sankar, Chief Technology Officer

"And for all that growth, our Department of War trailing 12-month revenue is still less than 25 basis points of the Pentagon's budget."
— Shyam Sankar, Chief Technology Officer

Assessment: The 25-basis-points framing is a rhetorical device, but it is a defensible one: penetration this low means the constraint on government revenue is procurement velocity and delivery capacity rather than addressable budget. We are raising our FY26 U.S. government estimate to +75–85% from +55–65%. The risk we continue to carry is that this segment's growth is now three quarters of a lap ahead of where any procurement-cycle model would put it, which means we do not have a reliable framework for calling the top. Being repeatedly too conservative on a line is not the same as understanding it.

International: Reaccelerating From a Very Low Base

The combined international book of $363M grew 33% year over year, split between international commercial at $182M (+26%) and international government at $181M (+42%). Sequentially the two lines added just $12M combined, so international remains a rounding error in the growth story, and its share of revenue fell to 18.7% from 21.5% in Q1. But the year-over-year rates are the strongest international prints of the AIP era, and international government in particular at +42% is doing real work as European and NATO defense modernization budgets convert.

Management did not address international at all beyond reciting the numbers, which is now a multi-quarter pattern. There is no re-engagement timetable, no named strategy, and no indication that forward-deployed engineering capacity is being reallocated toward it while U.S. demand exceeds supply.

Assessment: Our post-Q1 assumption of +12–15% international growth for FY26 was too low and we are raising it to +25–35%. The strategic point stands unchanged, though: Palantir is deliberately conceding the international commercial market during the exact window when enterprise AI platform decisions are being made, and those decisions have five-to-ten-year half-lives. This is a 2029 problem being created in 2026. It is the right trade given that U.S. demand exceeds delivery capacity, but it is a trade, not a free option.

Key KPIs

Two rows in that table carry most of the analytical weight. The first is the deal-size distribution: deals over $10M grew 55% sequentially while deals over $1M grew 7%. Palantir is not signing more contracts, it is signing much larger ones, which is the signature of a platform moving from departmental pilots to enterprise-wide commitments. The second is the gap between total TCV growth of 49% and dollar-weighted-duration TCV growth of 129%. That spread says the contracts being signed are substantially longer than the ones they replaced, which is both a positive (revenue visibility, switching costs) and a caution (a longer-duration book converts to revenue more slowly, so bookings growth of 153% will not translate into revenue growth of 153%).

One more observation that runs against the usual concentration worry. Trailing-twelve-month revenue from the top 20 customers grew 67% year over year, against total revenue growth of 93%. The largest accounts are growing fast in absolute terms and yet are becoming a smaller share of the whole. For a business whose bear case has historically leaned on customer concentration, that is a quiet but meaningful data point.

Key Topics & Management Commentary

Overall Management Tone: This was the most assured call of our coverage period, and the structure of it showed how assured: prepared remarks ran long and increasingly ideological, the analyst session ran to three questions, and not one of them tested the quarter's central claims. Where the Q1 call was confident with evidence, this one was confident with conviction, and the difference matters because the falsifiable content of the call went down as the numbers went up. The single most consequential statement made on the call was an unquantified 18-month growth commitment from the chief executive that nobody was in a position to interrogate.

1. The Largest Guidance Raise in Company History

FY26 revenue guidance moved to $8,150–8,158M from $7,650–7,662M, a $498M increase at midpoint and an 11-point step-up in the implied growth rate to +82%. In February the framework was $7,182–7,198M at +61%. Two quarterly raises later, the FY26 number is $964M higher than the original and the implied growth rate is 21 points higher. This is the second consecutive quarter in which the FY26 guide has been raised by roughly 6.5% at midpoint, and it repeats the cadence the company ran against the FY25 guide through 2025, which converts what we called a hypothesis in May into an observed operating pattern.

The raise is broader than the headline. U.S. commercial guidance moved to at least $3,424M from at least $3,224M, a $200M increase and a 14-point step-up to at least +134%. Adjusted operating income guidance moved to $4,889–4,897M from $4,440–4,452M, which is a $447M raise and, more importantly, lifts the implied full-year adjusted operating margin from 58.1% to 60.0%. Adjusted free cash flow guidance moved to $4,500–4,700M from $4,200–4,400M. The company continues to commit to GAAP operating income and GAAP net income in every quarter.

"We're also raising our full year 2026 revenue guidance midpoint to $8.154 billion, representing 82% growth year-over-year and an 11-point increase over our full year 2026 revenue guidance from last quarter and our largest ever full year revenue guidance raise."
— David Glazer, Chief Financial Officer

Assessment: Our post-Q1 note said we expected the Q2 print to raise FY26 to $8.0B or better. It came in at $8.154B, above the top of the $7.85–8.10B range we were modeling. The margin element of the raise is the underappreciated part: a 190-basis-point lift in the full-year adjusted operating margin guide on a $8.15B base is worth roughly $155M of incremental operating income, which is a larger contribution to the FY26 profit framework than the revenue raise itself.

2. Sovereign AI Becomes the Organizing Frame

Palantir reframed its entire pitch this quarter around what it calls AI sovereignty: the argument that enterprises using frontier-lab APIs are paying to transfer their operational know-how into a third party's model weights, and that the durable form of enterprise AI is one where the customer owns the data, the logic, the actions and ultimately the weights themselves. Every speaker on the call built on this frame, and it is the connective tissue between the product roadmap (post-training, supervised fine-tuning, reinforcement learning inside the customer's security boundary) and the commercial results.

"In contrast, enterprises that are not using Palantir are seeing their token meters spinning endlessly just to get slop without any correlation to value. This token model may be working for the labs, but it is not working for anyone else."
— Ryan Taylor, Chief Revenue Officer and Chief Legal Officer

"Last quarter, I said tokens are the new coal and AIP is the train. Now our customers can build their own locomotives."
— Shyam Sankar, Chief Technology Officer

The strategic significance is that this is a repositioning against the AI labs rather than against traditional software vendors. For three years the competitive question put to Palantir was whether hyperscaler platforms and data-cloud vendors would commoditize the ontology. That question was not raised once on this call. The question management has chosen to answer instead is whether the frontier labs will disintermediate the application layer, and the sovereignty argument is the answer.

Assessment: This is a stronger frame than the "no-slop zone" positioning of Q1 because it identifies a concrete economic harm to the customer (the transfer of proprietary advantage into someone else's weights) rather than a quality attribute of Palantir's product. It is also self-serving in a way that is worth naming: sovereignty is an argument for buying an orchestration layer, and Palantir sells the orchestration layer. The evidence that customers are buying the argument is the bookings, not the rhetoric, and the bookings are unambiguous.

3. The Chief Executive's 18-Month Growth Commitment

Near the end of his prepared remarks, the chief executive made a forward statement that is materially more aggressive than anything in the formal guidance, and it was not clarified, quantified or questioned.

"I am driving the business to grow at a rate equal or above to what we have in U.S. commercial for the next 18 months, which is a very high goal, but it is one we can actually get to because we are fully aligned with what's right and what's good and what actually works well in an enterprise."
— Alexander Karp, Chief Executive Officer

The sentence admits two readings. The narrow one is that U.S. commercial itself will sustain its current pace for another six quarters. The broad one, which the syntax more naturally supports, is that the total business will be driven to grow at or above the U.S. commercial rate. U.S. commercial grew 149% this quarter; the total business grew 93%; the company's own FY26 guide implies +82% for the year and roughly +72% in Q4. Under the broad reading, the chief executive publicly committed to a growth rate that exceeds his own company's guidance by more than 60 points.

Assessment: We are not going to model to a sentence, and we do not think anyone should. But the statement is not costless. Management has now set a public bar that formal guidance does not support, on a stock that just re-rated 29.5% in a session, and has left the ambiguity unresolved. If the intended reading was the narrow one, a single clarifying sentence in the next investor communication removes a real overhang. If it was the broad one, then FY27 consensus is about to be built on a promise rather than a framework, and the first quarter that comes in at merely excellent will be scored as a failure. This is the largest new risk created on this call, and it was created by the company, not by the business.

4. Net Dollar Retention at 157%, With More Guided

Net dollar retention reached 157%, up 700 basis points sequentially and the third consecutive quarterly increase on our record, from 134% in Q3 2025 to 139%, 150% and now 157%. At this level, existing customers are expanding at a 57% net rate before a single new logo is counted. What made the disclosure more interesting than the number was the unprompted forward statement attached to it: management said the metric should move higher, and gave a mechanism.

"By the way, one of the reasons the NDR number is so strong, ridiculously strong. I mean people always write these things about customer adoption and maybe -- but the NDR number is anomalously strong that is also going to shift as hard as to believe, become even more positive because some of our older partners we haven't really interacted with, they also showed up."
— Alexander Karp, Chief Executive Officer

"So customers that were only using Foundry, now want Ontology, now want to be part of the sovereign AI stack."
— Alexander Karp, Chief Executive Officer

Assessment: The migration of dormant Foundry accounts onto the AIP and sovereignty stack is the most specific and most underpriced growth mechanism disclosed on the call. It is a cohort that has already been sold, already been implemented, and carries near-zero acquisition cost. We are raising our long-term net dollar retention assumption to 135–150% from 130–145%, and we note that a business sustaining 150% net retention needs almost no new logos to hit its numbers.

5. The Gross Margin Step-Down Nobody Asked About

Adjusted gross margin fell to 86% from roughly 88% in Q1. The cause was disclosed clearly and then never revisited.

"Adjusted gross margin, which excludes stock-based compensation expense, was 86% for the quarter and reflects an increase in costs associated with taking on cloud hosting for one of our government customers. While this change led to higher cost of revenue in Q2, going forward, we believe it will power faster time to value, drive greater efficiency, provide greater cost certainty to the customer and enable us to expand their future workflows."
— David Glazer, Chief Financial Officer

What is not disclosed: the dollar magnitude, whether the arrangement is one customer or the template for a class of government contracts, whether it is priced to recover the hosting cost, and whether it recurs at this level in Q3. Cost of revenue grew 54% year over year against 93% revenue growth, so the year-over-year picture still shows leverage; the sequential picture does not. On a $1.9B quarterly base, 160 basis points of adjusted gross margin is roughly $31M.

Assessment: Small in isolation and entirely defensible as a commercial trade. The reason to watch it is structural rather than arithmetic: if hosting government workloads becomes standard practice, Palantir's gross margin ceiling resets two to three points lower and the operating margin story becomes wholly dependent on opex leverage. Adjusted operating margin still expanded this quarter, so nothing is broken. But we would like the Q3 disclosure to tell us whether 86% is the new floor or a one-quarter artifact.

6. The Bake-Off: Palantir's Competitive Evidence Against the Labs

The most concrete competitive datapoint on the call was a head-to-head evaluation against a frontier AI lab at a large technology customer, described in enough operational detail to be checkable in principle.

"A major Silicon Valley tech company recently ran a bake-off, a Frontier lab and its deployment team against AIP and our forward deployed engineers. Remember, only Palantir has FDEs. Everyone else has sparkling sales engineers. The lab picked a ticketing automation problem and failed to deliver anything of value against it. We built agent swarms for each of our customers' customers, proactively recommending marketing, packaging and pricing changes to drive revenue and utilization. This work converted into a $10 million ACV contract. The lab was shown the door. Same customer, same time line, same models."
— Shyam Sankar, Chief Technology Officer

The "same models" clause is the analytically important part. Palantir's argument is that its advantage sits in deployment tradecraft and the surrounding orchestration layer, not in model access, which means model-layer commoditization is neutral-to-helpful rather than threatening. If the models are a commodity input that everyone can buy, the differentiator is who can convert them into a production workflow, and Palantir's forward-deployed engineering model is a fifteen-year-old organizational asset that cannot be replicated by hiring.

Assessment: This is the cleanest articulation we have heard of why Palantir is not disintermediated by better models, and it is consistent with the bookings. The obvious caveat is that it is a single self-reported anecdote from the winning side, with the customer and the lab both unnamed. Treat it as illustrative of the argument rather than as evidence for it. The evidence is the $2.1B of U.S. commercial bookings.

7. From Benchmaxing to Benchmaking, and the Open-Weights Pivot

The product narrative moved decisively toward post-training and customer-owned weights. Management argued that public benchmarks have been optimized against to the point of uselessness and that the relevant benchmark is the customer's own operational definition of success.

"Within 24 hours of bringing Nemotron Ultra into our stacks, we found 5 production tasks where a standard Nemotron Ultra model without post-training beat frontier models. This underscores that a handful of common benchmarks can be gamed. That era, benchmaxing has ended. The new era is benchmaking."
— Shyam Sankar, Chief Technology Officer

The commercial expression of this is that Palantir now sells a model factory operating inside the customer's security boundary, accumulating intelligence in weights the customer controls, with the platform providing the control plane to trade cost, performance and latency workflow by workflow. The chief executive confirmed the company is entering the fine-tuning market and is already doing so in the classified environment, and referenced a partnership with NVIDIA supporting the stack.

Assessment: Strategically coherent and commercially significant, because a customer whose proprietary weights live inside Palantir's platform is a customer with a switching cost measured in years rather than quarters. It also creates a new cost vector: post-training and inference infrastructure is capital-intensive in a way that Palantir's historical business has not been, and the gross margin step-down discussed above may be an early signal of that. Capital expenditure remains negligible at $22M for the half, so nothing has changed yet.

8. Operating Leverage and the Q3 Expense Ramp

Adjusted expense of $741M grew 14.2% sequentially and 37% year over year, against 18.6% and 93% revenue growth respectively. That gap is the entire operating margin story. On a GAAP basis, total operating expenses grew 34.2% year over year while revenue grew 92.8%, taking opex from 53.9% of revenue to 37.5%. General and administrative expense grew just 19.7%, sales and marketing 39.3%, and research and development 42.6%. Stock-based compensation of $265M grew 66%, which is fast in dollars but represents 13.7% of revenue against 15.9% a year ago.

Management guided explicitly to a steeper expense quarter ahead. The Q3 adjusted operating income guide of $1,292–1,296M against revenue of $2,160–2,164M implies adjusted expense of roughly $868M, up 17% sequentially and above the 14% pace of Q2.

"As in prior years, we expect a significant ramp in expense in the third quarter due to the seasonality of new hire starts and other product and marketing initiatives."
— David Glazer, Chief Financial Officer

Assessment: The guided Q3 adjusted operating margin of 59.9% is 180 basis points below the Q2 print, and the implied Q4 margin of roughly 58.6% is lower again. Read plainly, the company is telling you that 61.7% is a seasonal high rather than a run rate. That is consistent with prior years and it is not a negative surprise, but it does mean anyone extrapolating a 62% margin into FY27 is extrapolating a peak. We model 60–62% for FY26 and hold 60–64% as the long-run range.

9. Non-Operating Income Is Becoming Material

Interest income of $77.5M and other income of $91.8M together contributed $169.3M of pre-tax income, against $912.0M from operations. Other income grew from $6.6M a year ago, and management identified the driver.

"Unrealized gains from our holdings in SpaceX resulted in a $0.03 tailwind to GAAP EPS and a $0.02 tailwind to adjusted EPS in the quarter."
— David Glazer, Chief Financial Officer

Palantir now carries $7,379M of marketable securities against $2,030M of cash, and is generating meaningful reported earnings from positions in private companies whose size, cost basis and valuation methodology are not disclosed in the release. Combined with a 1.4% effective tax rate, a material share of reported net income is coming from outside the operating business. To management's credit the non-GAAP construction is conservative here: adjusted net income of $1,047M is below GAAP net income to common of $1,062M precisely because the reconciliation applies a $297M normalized tax charge.

Assessment: Not a concern at current magnitude, and disclosed clearly rather than buried. It is worth tracking for two reasons. Marks can reverse, so an unrealized gain contributing three cents this quarter can contribute negative three cents in another. And a growing pool of private-company holdings inside an operating company invites a valuation question that has nothing to do with the software business. We would rather see the position sized in the next filing.

10. Three Questions

The analyst session consisted of three questions. For a company that had just reported 93% revenue growth, raised full-year guidance by nearly half a billion dollars, disclosed a step-down in gross margin, made an unquantified 18-month growth commitment, and reported a material contribution to earnings from private-company marks, that is a remarkably light examination. None of the three questions touched guidance, margins, the expense ramp, capital allocation, competition from data-platform vendors, or the durability of the bookings.

Assessment: This is not a criticism of the business and it is not a governance failing, but it is a fact about the information available to investors, and it cuts against the reader. The prepared remarks were the call. When a company is executing this well, scrutiny declines exactly when the numbers being extrapolated forward are largest, and the gap between what management asserted and what management was made to defend was wider on this call than on any other in our coverage period.

Guidance & Outlook

The FY26 framework has now been raised twice since it was issued in February, from $7,182–7,198M to $7,650–7,662M to $8,150–8,158M. In growth-rate terms that is +61% to +71% to +82%. Each raise has been roughly 6–7% of the prior midpoint, and each has followed a quarter in which the in-period revenue beat the company's own guide by 6–8%. The mechanism is consistent enough to be modeled: Palantir guides to a number it is confident of, beats it by six to eight points, and rolls most of the beat plus a further increment into the full year.

Implied quarter-over-quarter ramp: the Q3 guide midpoint of $2,162M is +11.7% sequentially off Q2's $1,935M, and the residual Q4 at roughly $2,424M is +12.1% off Q3. Both are materially below the +18.6% Palantir just printed and below the +16% it printed in Q1, so the guide embeds a visible sequential deceleration. Against the prior-year base, the guide steps down from +93% actual to +83% guided in Q3 to +72% implied in Q4. If the historical 6–8% beat pattern holds in Q3, actual revenue lands near $2.30–2.33B, which would be +95–97% year over year rather than +83%, and the FY26 number gets raised again in November.

Street position: pre-print consensus for FY26 sat close to the prior $7.66B guide; the new $8.154B midpoint resets that base by more than 6%, and the sell-side revisions published within a day of the print moved twelve-month targets into a $200–245 cluster for the bullish majority against a low target of $80 held by the one desk still rating the stock underperform. The average target moved from roughly $181 pre-print to roughly $189 post-print, which is only a 4% revision against a 6.5% guide raise and a 29.5% stock move. In other words the Street raised numbers by less than the stock moved, so the multiple did the work, not the estimates.

Guidance style: unchanged and conservative. Palantir has beaten its own quarterly revenue guide by roughly 6–8% in each of the last three quarters and has raised the full year on each print. Nothing in the Q3 guide suggests a change in posture. We treat the guide as a floor and model to a 3–5% beat.

Analyst Q&A Highlights

The session ran to three questions in total. Because two of them produced answers spanning genuinely distinct topics, we treat those as separate entries below.

What the Sovereignty Boot Camp Revealed About Latent Demand

The opening question asked what management learned from customers at the company's sovereignty event that it had not expected. The answer described an inbound-demand dynamic in which attendance vastly exceeded invitations and, more consequentially for the model, disclosed that dormant legacy accounts have begun migrating onto the newer stack. Management tied that migration directly to the retention metric and stated it should move higher from here.

Q: "Can you walk through -- at the sovereignty boot camp, what did you learn with talking to customers that maybe you weren't expecting? I mean, obviously, just an overwhelming sort of group of executives. Can you just talk about that, Alex?"
— Dan Ives

A: "By the way, one of the reasons the NDR number is so strong, ridiculously strong. I mean people always write these things about customer adoption and maybe -- but the NDR number is anomalously strong that is also going to shift as hard as to believe, become even more positive because some of our older partners we haven't really interacted with, they also showed up. They're like, 'Oh, okay, now we get why we would need you. And not just Foundry. They're migrating across our stack. So customers that were only using Foundry, now want Ontology, now want to be part of the sovereign AI stack."
— Alexander Karp, Chief Executive Officer

Assessment: The most model-relevant disclosure of the entire Q&A, and it came unprompted inside an answer to a soft question. A re-activating base of previously-implemented accounts is the cheapest growth a software company can buy, and it explains why the retention metric can keep climbing even as the revenue base compounds. We have raised our long-term retention assumption on the strength of it.

The Capacity Constraint Behind the Growth Push

Later in the same answer, management explained the reasoning behind the aggressive forward growth ambition in operational rather than commercial terms: setting an internal growth target above what the business can currently deliver is a mechanism for forcing the partner and delivery capacity to scale. This is the clearest statement to date that the binding constraint on Palantir's growth is delivery capacity rather than demand, and that the company intends to relieve it through third-party partners rather than headcount alone.

Q: "Can you walk through -- at the sovereignty boot camp, what did you learn with talking to customers that maybe you weren't expecting?"
— Dan Ives

A: "Behind the scenes now, we're trying to find partners. Now partners, we need technically exceedingly competent partners. Partners doesn't mean a vassal. We don't have to agree on every issue. We don't even have to agree on every client. They can occasionally compete against us... Why am I pushing the company to grow not to just the end of the year, but next year? It's because that also forces us to find ways to scale to meet the demand that's out there."
— Alexander Karp, Chief Executive Officer

Assessment: This is the operational context for the 18-month commitment and it makes the statement more legible: the growth target is partly an internal forcing function for capacity building rather than a forecast. It also flags a genuine strategic shift. A partner-delivered implementation model dilutes the forward-deployed-engineering advantage that the same call spent several minutes arguing is the company's core moat. Reconciling those two positions is the thing to watch over the next several quarters.

Why the Sovereignty Thesis Took Three Years to Land

A question probed why the market took so long to recognize what Palantir had positioned for, given that the importance of proprietary data was obvious from the start of the AI cycle. The response drew a distinction between the initial phase, in which the priority was demonstrating that AI could produce economic value at all, and the current phase, in which customers have realized that the asset at risk is not just their data but the reasoning traces and operational metadata generated by using the systems.

Q: "But then we're like 3 years into that revolution and now everyone started to realize, owning my data and whatever or learn from my data is important. What happened there? Why you think that you position yourself back then in a different approach to AI that enterprises weren't able to see?"
— Mariana Perez Mora, Bank of America

A: "I also am going to need to control the weights, that the alpha that is being generated isn't simply the data that's resident in my enterprise. It's also the metadata, the reasoning traces, the exhaust, the usage of this, which I don't yet have mechanisms to control. And now I understand that's actually probably more valuable than just the data in my enterprise. And that's been a clarion call, I think, for the market over the last quarter or 2 quarters."
— Shyam Sankar, Chief Technology Officer

Assessment: The "last quarter or 2 quarters" timing claim is the useful part, because it dates the demand inflection to roughly the start of 2026 and is consistent with the bookings curve, where U.S. commercial contract value went from $1.18B in Q1 to $2.13B in Q2. If the market shift is genuinely two quarters old, the bookings response has barely begun to convert to revenue, and the remaining deal value of $6.2B in U.S. commercial is the queue.

Model Lock-In and the Ability to Switch

The final question reframed the sovereignty argument around continuity risk rather than competitive risk: a customer that buys orchestration from a frontier lab is exposed if that lab's model is withdrawn or degraded, and the questioner noted this had already happened more than once during the year. Management confirmed model-switching is already a shipped capability across U.S. government deployments and extended the answer into the argument that customer-specific benchmarks make switching a continuous optimization rather than an emergency procedure.

Q: "The topic of sovereignty, you focused a lot on how dangerous it is to give the keys to the labs because they could choose to compete with you. Is there another aspect of this as well, though, that if you choose a lab and you buy the orchestration and the consulting and the harnesses from them, you're beholden to their models. And if something happens, if it's not the best model anymore, if the model gets pulled, then you as a customer are stuck and you may have a mission-critical system fail."
— Gil Luria, D.A. Davidson

A: "No. I mean I'll let Ryan and Shyam comment here, but we're already doing that across the USG, U.S. government. We have a product that allows you to switch out models. Look, at the end of the day, if you are locked into a product, you're going to -- I mean, that's the nature of monopoly capitalism. That's why people want to lock in because then they can raise the prices and reduce the quality."
— Alexander Karp, Chief Executive Officer

Assessment: The strongest of the three exchanges, because it produced a verifiable product claim rather than a positioning argument. Model portability inside government deployments is a concrete artifact, and it converts the sovereignty pitch from a philosophical position into a procurement checklist item that competitors have to answer. The follow-on point from the technology side, that the customer's own benchmark turns model selection into a continuous rather than episodic decision, is the more durable version of the same argument.

What They're NOT Saying

1. Which base the 18-month growth commitment refers to. The chief executive committed to growing "at a rate equal or above" the U.S. commercial pace for six quarters, without specifying whether that means U.S. commercial sustaining 149% or the total company reaching it. Formal guidance implies +82% for FY26 and roughly +72% in Q4. The gap between the statement and the guide is more than 60 points and no one resolved it.
2. The dollar size and duration of the government cloud-hosting arrangement. Adjusted gross margin fell roughly 160 basis points sequentially on this item, which is about $31M at the current revenue base. Not disclosed: whether the arrangement is priced to recover cost, whether it recurs at this level, and whether it is the template for other government contracts. This is the difference between a one-quarter artifact and a permanent two-to-three-point reset of the gross margin ceiling.
3. Any FY27 framework, again this quarter. With FY26 now guided to $8.15B and growing 82%, the Street is being asked to build FY27 with no company framework and, this quarter, with a chief-executive statement that points far above any reasonable extrapolation. The absence of a framework was tactically useful when the guide was conservative. It is less useful now that an unquantified verbal commitment is filling the vacuum.
4. The size, cost basis and valuation methodology of the SpaceX position. The earnings contribution was disclosed to the cent. The position was not. Marketable securities stand at $7,379M and other income has gone from $6.6M to $91.8M year over year, so this is no longer immaterial.
5. Anything about capital allocation. No buyback update for the third consecutive quarter, despite $9.2B of cash and short-term treasuries and a guide to $4.5–4.7B of adjusted free cash flow this year. Cash is compounding faster than any disclosed plan for it. Financing activities for the half were a net inflow of $8M, so no material repurchase occurred.
6. Any competitive discussion of data-platform vendors. Sixth consecutive quarter with no mention of hyperscaler platforms or data-cloud competitors. The competitive frame has been redirected entirely toward the frontier labs, which is a different fight. As long as U.S. commercial compounds at 149% this is academic; it becomes the central question the moment growth normalizes.
7. International strategy. Both international lines reaccelerated this quarter, to +26% and +42%, and neither was discussed beyond the recitation of the numbers. There is still no re-engagement plan, no capacity-allocation commentary, and no acknowledgement that the segment moved.
8. How the partner-led delivery model reconciles with the forward-deployed-engineering moat. The call spent several minutes arguing that Palantir's engineers are the differentiator that frontier labs cannot replicate, then described a plan to scale delivery through third-party partners. Both cannot be maximally true.
9. Remaining performance obligations versus remaining deal value. Management continues to lead with the $13.1B remaining deal value figure while noting that the $4.9B RPO number excludes most government work. RDV is the more flattering measure and the less contractually binding one. The company has never quantified how much of RDV sits behind termination-for-convenience clauses.

Market Reaction

• Pre-print setup: PLTR closed at $125.65 on August 3, down 29.3% year to date against an S&P 500 up 11.0%, down 18.6% over the trailing twelve months, and 39% below its 52-week closing high of $207.18. The trailing 30-day move was -2.8%. The 52-week closing range was $107.27 to $207.18, so the stock entered the print in the lower third of its own range. The options market priced a 12.12% post-print move against a 7.39% four-quarter average realized move.
• After-hours August 3: the stock traded up roughly 15% to about $144.45 on the release, an unusually orderly reaction given the magnitude of the beat and the guidance raise.
• Reaction session, August 4: opened at $145.15 (+15.5%), traded a range of $143.28 to $164.52, and closed at $162.66, up 29.5% or $37.01. The stock gained a further 12.1% from the opening print through the close and finished 1.1% off the session high. This was not a gap and fade.
• Volume: 172.4M shares against a 30-day average of 40.3M, or 4.3x normal.
• Realized versus implied: the 29.5% move was roughly 2.4x the 12.12% the options market had priced, and roughly 4x the trailing four-quarter average reaction.
• Where it leaves the stock: $162.66 is still down 8.5% year to date and 21.5% below the 52-week closing high, but 51.6% above the 52-week closing low. Market capitalization on 2,568.7M diluted shares is approximately $417.8B. The S&P 500 rose 1.8% on the same session.

The size of this move is a story about positioning at least as much as it is a story about the quarter. Palantir delivered an operationally similar print in May, beating its guide by 6.5% and raising the full year by $470M, and the stock fell 6.9% the next session, from $146.03 to $135.91. The difference is entirely the entry point. In May the stock came into the print at $146.03 having already rallied off the February lows; in August it came in at $125.65 having spent the year as a de-rating AI-momentum casualty, 39% below its high, with the options market pricing a 12% move and the average sell-side target sitting 44% above the spot price. A crowded short-duration base plus a genuinely superior print plus a guidance raise that cleared the widely-discussed threshold produced a violent repricing rather than an orderly one.

The mechanical trigger was the guide. Pre-print commentary had converged on roughly $7.83B as the FY26 revenue number required for a positive reaction; the company delivered $8.154B. Everything else in the release supported it: 93% growth on the top line, 149% in the segment the entire bull case rests on, first billion-dollar GAAP net income quarter, first billion-dollar free cash flow quarter, and a Rule of 40 of 155 that has no comparable in large-capitalization software. There was no line in the print for a seller to point at other than the two-point gross margin step-down, and that was disclosed with a rationale attached.

Street Perspective

Debate: Does the Repeated Full-Year Raise Justify a Higher Multiple, or Just a Higher Base?

Bull view: two consecutive 6.5% raises to the FY26 guide, on top of the same cadence run against the FY25 guide last year, is no longer a coincidence, it is a demonstrated operating pattern, and a company that reliably raises its own forecast by that increment each quarter should be valued off a forward number well above the printed guide. Applying the pattern to the remaining two quarters exits FY26 near $8.6–8.8B and puts FY27 comfortably above $13B. On those numbers the multiple looks materially less demanding than the headline suggests.

Bear view: the raise cadence is arithmetic, not alchemy. Each raise has been funded by the in-quarter beat rolling forward, which means the company is capitalizing the same dollar twice in the market's mind. Meanwhile the guide itself embeds visible deceleration, from 93% actual to 83% guided to 72% implied, and the comparison base doubles over the next four quarters. Paying a higher multiple for a pattern of conservative guidance is paying for the guidance philosophy rather than the business.

Our take: the bull side has the better of the near-term argument and the bear side has the better of the framing. The pattern is real and we model to it, which is why our FY26 estimate sits above the guide. But a raise cadence justifies a higher base, not automatically a higher multiple, and the market applied it as a multiple this week: the sell-side average target rose about 4% while the stock rose 29.5%. That gap is the whole debate in one number.

Debate: Is the 18-Month Growth Commitment an Asset or a Liability?

Bull view: a founder-chief-executive publicly committing to sustaining this growth rate for six more quarters, with the delivery-capacity plan to back it, is exactly the kind of conviction signal that should be taken at face value from a management team that has beaten its own guidance for four straight quarters. Insiders can see the pipeline; the statement is information.

Bear view: the statement is unquantified, unclarified as to base, and sits more than 60 points above the company's own formal guidance. It converts every future quarter into a referendum on a sentence rather than on the framework, and it hands the bear case a scoring rubric that the company itself wrote. A merely excellent Q3 will now read as a shortfall.

Our take: liability, at the margin, and an avoidable one. We do not model to it and we would advise no one else to. The commitment adds no information that the bookings do not already provide, and it creates an expectations anchor that is unhelpful precisely because the underlying business does not need it. A clarifying sentence at the next investor event would remove the ambiguity at zero cost.

Debate: Has the Gross Margin Step-Down Changed the Long-Term Margin Structure?

Bull view: it is $32M on a $1.9B quarter, disclosed proactively, taken deliberately to win a wider workflow footprint at a government customer, and adjusted operating margin still expanded 150 basis points sequentially and 1,540 basis points year over year. Cost of revenue grew 54% against 93% revenue growth. There is no margin problem here.

Bear view: it is the first sequential gross margin decline in our coverage period and it arrives in the same quarter the company describes a product roadmap built on hosting, post-training and inference inside customer boundaries. Those are structurally capital-intensive and cost-of-revenue-heavy activities. If hosting government workloads becomes standard, the gross margin ceiling resets and the entire operating margin story rests on opex leverage that must eventually flatten.

Our take: the bear framing identifies the right question and overstates the current evidence. One quarter of 160 basis points on a disclosed, rationalized customer-specific arrangement is not a structural break, and capital expenditure of $22M for the half says the inference-infrastructure build has not started. But it is the correct thing to watch, and the Q3 disclosure will tell us which it is. We have not changed our long-run margin range on one data point.

Debate: Is the Stock Investable at 50 Times Current-Year Revenue?

Bull view: multiples have to be paid for something, and the something here is a Rule of 40 of 155, a 60% guided full-year operating margin, a 63% free cash flow margin, 157% net dollar retention, and a forward book growing faster than revenue. No large-capitalization software business has ever occupied this quadrant. At 30x our FY27 revenue estimate the multiple normalizes quickly on any continuation of the growth, and the 39% drawdown into the print showed the market will pay for it once conviction returns.

Bear view: the enterprise value went from 41x to 50x the current-year revenue guide in a single session while estimates moved 6.5%. The buyer at $162.66 is paying a 22% higher multiple than the buyer at $125.65 was on Monday for a set of facts that were, in expectation, largely knowable. Trailing price-to-sales above 60 and a trailing price-to-earnings ratio near 140 leave no cushion, and the one desk still rating the stock underperform carries an $80 target, implying the dispersion in this name is 3-to-1 between the high and low targets.

Our take: both are right, which is why we are maintaining Outperform rather than upgrading conviction. The operating case genuinely improved this quarter and our estimates went up 6–10%. But the stock went up 29.5%, so the return that was available on Monday is not available today. At $162.66 we see roughly 23% to our target over twelve months, against 37% from the $146.03 close at the time of our May note. That is still an above-market expected return and it justifies holding a full position. It does not justify adding one.

Model Update Needed

Valuation: at the reaction-session close of $162.66 and 2,568.7M diluted shares, market capitalization is approximately $417.8B. Net of $9.2B in cash and short-term treasuries, enterprise value is approximately $408.6B. Against the FY26 revenue guide midpoint of $8.154B that is 50.1x enterprise value to current-year revenue; against our own FY26 estimate of $8.45B it is 48.4x; against our FY27 estimate midpoint of $13.4B it is 30.5x. On cash flow, the FY26 adjusted free cash flow guide midpoint of $4.6B puts the stock at 88.8x, falling to roughly 53x on our FY27 estimate. For context on the move, the same enterprise value calculation at Monday's $125.65 close against the then-current $7.656B guide produced 41.0x, so the multiple expanded 22% in one session while the revenue denominator rose 6.5%.

Price target: $200, unchanged. Range $150 to $260. The target is anchored on approximately 38x enterprise value to our FY27 revenue estimate of $13.4B, plus net cash, across roughly 2.60B diluted shares. The upside case of $260 assumes FY27 revenue of $14.5B at 46x, reflecting the 18-month growth ambition proving directionally right. The downside case of $150 assumes FY27 revenue of $12.0B at roughly 32x, reflecting the guided deceleration continuing on schedule and the multiple compressing toward the pre-print level. Holding the target unchanged after raising estimates 6–10% is deliberate: the higher revenue base is offset by a target multiple that must compress as the stock re-rates, and the honest summary is that the upside we identified in May has now largely been paid out. The $200 target implies approximately +23% from $162.66, against +37% from the $146.03 close when we last published.

Thesis Scorecard Post-Earnings

Overall: the operating thesis strengthened materially. Four of five bull pillars were confirmed outright, two of them well beyond what we modeled in May, and one bear point was actively challenged for the first time in five quarters. Against that, two bear points deteriorated: valuation moved from a tightening constraint to a binding one after a 29.5% session, and management's own communications introduced an expectations risk that the business did not require. The net is a thesis that is stronger on fundamentals and weaker on price, which is the ordinary consequence of a good quarter arriving into a de-rated stock and getting paid for immediately.

Action: hold a full position; do not add at $162.66. The rating stays Outperform on a twelve-month view because a 23% implied return with a Rule of 40 of 155 behind it remains above-market, and because the bookings and retention data give unusually high visibility into the next four quarters. But the asymmetry that justified the upgrade in February and the maintenance in May has now been consumed by the price. Our conditions for adding are a re-test toward $135–145 without thesis impairment, or a Q3 print that resolves the gross margin question and clarifies the growth commitment. Our conditions for reconsidering the rating are a second consecutive quarter of gross margin compression without explanation, or a Q3 revenue print that fails to beat the guide by at least the historical 6%.