Palantir Q2 2026 Results: Revenue +93% to $1.94bn and a ~$500m Guidance Raise — Why the Raise Mattered More Than the Beat

Palantir reported second-quarter revenue of $1.935bn on Monday 3 August 2026, up 93% year-on-year, against its own guidance of $1.797bn to $1.801bn and consensus near $1.812bn. Adjusted EPS was $0.41 against roughly $0.35 expected. But the number that carried the information was not the beat: full-year revenue guidance rose to $8.150bn–$8.158bn from $7.650bn–$7.662bn, an increase of about $498m at the midpoint — roughly three and a half times the size of the quarter's own beat.

That ratio is the whole story, and it is the test this note set out in advance. A beat compares a finished quarter to a published estimate. A raise changes what is expected of quarters that have not happened. When the second is much larger than the first, management has lifted the shape of the year rather than re-banking revenue that already arrived.

What actually happened

Every figure below is from the company's own results release, filed with the SEC on Form 8-K on 3 August 2026 (primary source: the Q2 2026 earnings release as filed; the company also publishes it through Palantir Investor Relations).

Alongside the revenue lines, the company reported closed total contract value of $3.373bn, up 49% year-on-year, including a record $2.132bn of US commercial contract value, up 153%. It closed 220 deals of at least $1m, 98 of at least $5m and 73 of at least $10m, and finished the quarter with $9.2bn of cash, cash equivalents and short-term US Treasury securities. Adjusted free cash flow was $1.220bn, a 63% margin.

Chief executive Alex Karp, quoted in the release, described the quarter as "otherworldly" and said that "demand for AI sovereignty has now been unleashed". Read as disclosure rather than as language, the substance behind it is the US commercial line and the contract value behind that line.

The test, and the subtraction that settles it

This note argued before the release that the beat was close to pre-determined and therefore uninformative — consensus sat about half a percent above the ceiling of the company's own guided range — and that the live variable was whether any raise would exceed the beat that funded it. That is now a matter of arithmetic rather than interpretation.

The beat, measured properly against the guided midpoint of $1.799bn, was about $136m. Measured against consensus of $1.812bn it was about $123m, or 6.8%. The full-year raise, midpoint to midpoint, was about $498m. The raise is therefore roughly 3.6 times the quarter's beat.

The number nobody guides to, recomputed

The most informative figure is still not printed in the release, and it still has to be derived. Before Monday the derivation produced an implied second half of roughly $4.22bn, about 63% above the second half of 2025 — a decelerating growth rate embedded in the company's own annual figure. Here is the same subtraction on the new numbers.

The new full-year midpoint is about $8.154bn. The first half is now reported in full: $1.633bn in the first quarter and $1.935bn in the second, or $3.568bn. That leaves an implied second half of roughly $4.586bn. The second half of 2025 produced about $2.588bn — what remains of the $4.475bn full year once the $883.9m of the first quarter, the $1.004bn of the second and the $1.407bn of the fourth are removed. Implied growth is therefore around 77%, against the 63% the previous guide contained.

The sequence now reads: 85% in the first quarter, 93% delivered in the second, about 83% guided for the third — the $2.162bn midpoint against $1.181bn a year earlier — and roughly 72% implied for the fourth, since $4.586bn less $2.162bn leaves about $2.424bn against the $1.407bn of the fourth quarter of 2025.

That is still a decelerating rate, and it is worth being precise about why: the denominator is growing at nearly twice the pace it was a year ago, so an identical amount of new revenue produces a smaller percentage. Deceleration of that kind is arithmetic, not deterioration. What changed on Monday is the level from which the deceleration starts. Treat the derived figures as approximate — the 2025 base is rounded and a guidance range is a range rather than a point, so the implied rates carry perhaps half a percentage point either way.

The steepest bar was raised, not reiterated

Of the three branches this note set out in advance, the third was the one that would have mattered most while showing up least in the headline: total guidance rising while US commercial guidance was merely reiterated, which would have signalled the mix drifting back toward government work. The release settled that directly. Full-year US commercial guidance moved to in excess of $3.424bn, at least 134% growth, from at least $3.224bn and at least 120%. The company raised the line with the least slack in it.

The quarter behind that guide: US commercial revenue of $764m, up 149% year-on-year and 28% quarter-on-quarter, against $306m in the second quarter of 2025. US government revenue grew 90% to $809m, so this was not a case of one side of the business carrying the other — but commercial is now within $45m of government inside the United States, which is the composition shift the whole strategic claim rests on.

There is a live argument running against the entire category, and the release speaks to it more usefully than any commentary does. The argument is that as general-purpose model providers absorb capability into their own products, some functions that once required a dedicated application-layer vendor can be assembled with less bought software — a genuine mechanism, and one plausible reading of why enterprise software has been repriced this year while its revenue kept growing. The observable that bears on it is not sentiment but retention and expansion: whether existing customers keep enlarging their commitments. US commercial remaining deal value of $6.238bn, up 124% year-on-year and 27% quarter-on-quarter, and closed US commercial contract value of $2.132bn, up 153%, are contracted future revenue rather than opinion. For this vendor, in this quarter, customers enlarged their commitments. That is one data point against a structural argument that will take several more quarters to resolve either way, and the same tension is visible from the hardware side in the memory-chip selloff we covered this week.

How this reaches an index you trade — and how much of it doesn't

Do the arithmetic openly, because it is the antidote to the "Palantir moved the market" headline.

Palantir's market capitalisation is roughly $300bn. Against an S&P 500 aggregate measured in the tens of trillions of dollars, that is a weight under half a percent, so even a 10% single-day move contributes only a few hundredths of a percentage point to that index directly — well inside a normal session's range. In the Nasdaq-100 the same move lands considerably harder, because that index is a fraction of the S&P 500's aggregate size and excludes financials entirely, concentrating weight in exactly this kind of company. Both figures are approximations: index weights are float-adjusted and aggregate capitalisations are moving targets.

The larger channel is correlation. A demand read from one enterprise software vendor is immediately applied to every company thought to sell into the same budget, which is why a single report can produce a sector-wide session while the reporting company's own index contribution stays small. The reverse case ran through the market days earlier, when one company's capital-spending guidance repriced its entire supplier chain — the mechanics of which we set out in the note on that record session. The same week's Apple and Amazon reports demonstrated the divergence in both directions at once, and AMD reports on 4 August, into a tape that now has this read on enterprise AI budgets in front of it.

On sequencing, the practical point stands: the release landed after the US close, the reaction forms in after-hours trading on Monday evening, and it is Tuesday's cash open that transmits any of it into a Nasdaq-100 or S&P 500 level. The shares had closed July at $123.06 and traded around $126 during Monday's session, still roughly 30% lower across 2026 against a 52-week high of $207.52 — which is the context in which a report this far above its own guided range is being received.

What this is not is a currency event. One company's quarterly revenue is not an input to interest-rate paths, growth, inflation, terms of trade or positioning. The only equity-to-currency channel is risk sentiment, and it needs breadth to operate. The genuine connection is the discount rate: cash flows expected years after the spending that produces them are more sensitive to the level of rates than near-term earnings are, and the policy rate that sets that discount is scored through the interest-rate factor on the USD currency page, with our coverage of the July Federal Reserve decision taking that side of the ledger.

What would change the picture

Three things, none of them a forecast.

A third-quarter print that merely meets the new guide. The $2.160–2.164bn range is now the published bar, and consensus will assemble at or just above its ceiling exactly as it did for the second quarter. The informative question in November will again be the relationship between the beat and the raise, not the beat alone.

US commercial growth below the pace the annual guide requires. At least 134% for the full year against 149% delivered in the quarter leaves less headroom than the previous bar did. A raised guide is a tighter guide, which is the trade-off management accepted on Monday.

Remaining deal value flattening while revenue stays strong. Contracted future revenue leads reported revenue. If expansion slowed while the reported line still looked strong, that would be the earliest observable of the structural argument in the section above beginning to bite — and it would appear in the RDV line long before it appeared in a revenue headline.

The takeaway

The bar was published in advance and the quarter cleared it: $1.935bn against $1.797–1.801bn guided, 93% growth where 79% was guided, $0.41 of adjusted EPS against roughly $0.35 expected. None of that was the news, because a beat is a comparison with history.

The news was the subtraction. A $498m raise on a $136m beat lifted the implied second half from roughly $4.22bn to roughly $4.586bn, and raised the steepest published bar in the release rather than reiterating it. Whether the price agrees is a separate question with a separate reference point — the same distinction that let a 6% beat and a 5.7% share-price fall coexist last quarter without either being wrong (contemporaneous coverage of that first-quarter report: CNBC).

To learn how Pip Theory builds its fundamental scores, see the methodology overview.

Educational macro context only — not investment advice.