Palantir's US Commercial Revenue Grew 149% While Stock Pay Grew 66%

Palantir's US commercial revenue reached $764 million in the quarter to June 2026, up 149% on a year earlier. Stock-based compensation rose 66% to $265 million over the same period. On the annual comparison, revenue is outrunning stock pay by a wide margin. On the sequential comparison, the gap closed: stock pay grew 32% in the quarter against 28% for US commercial revenue. That second number is the one the Q3 2026 10-Q has to settle.

What the Q2 2026 10-Q actually reports

This analysis runs on the most recent quarterly filing Palantir has made. That is the 10-Q for the quarter ended 30 June 2026, filed on 4 August 2026. The company is due to report Q3 2026 after the market closes on 2 November 2026, and its 10-Q follows shortly after. When this post was written, no Q3 8-K had appeared on EDGAR.

So the honest framing is this. The numbers below are the baseline, and the final section turns them into a test you can run on the Q3 filing in about ten minutes. I would rather publish a clean baseline now than a rushed read of a release nobody has digested.

The Q2 filing has four numbers that matter for the question of whether AIP (Palantir's generative AI platform) is turning into durable enterprise revenue. US commercial revenue was $764 million. Stock-based compensation was $265.2 million. US commercial remaining deal value was $6.238 billion. Total customers on a trailing twelve-month basis were 1,049.

The 10-Q describes AIP as providing connectivity to third-party language models, a toolchain for building agents and automations, end user applications and an evaluations framework. It does not break out AIP revenue. That matters, because every claim about AIP's commercial traction is an inference from the US commercial line, not a disclosed figure. Palantir is not unusual here, as the ranking of which AI revenue disclosures are worth trusting shows.

US commercial revenue: four quarters of acceleration

Palantir's US commercial segment is the cleanest public proxy for enterprise AI demand at a single vendor. It is also small relative to the headlines it generates. At $764 million, it was 39% of total Q2 2026 revenue of $1.935 billion.

The quarterly sequence is what makes the case. Each figure below comes from the company's own earnings release for that quarter. The Q2 2025 figure is implied by the 149% growth rate rather than read off a release, so treat it as approximate.

Revenue and growth rates from Palantir earnings releases (8-K Exhibit 99.1) for each quarter. Stock-based compensation from each release's non-GAAP reconciliation table. The final column is our calculation. Stock pay funds the whole company, not the US commercial segment alone, so the last column is a scale ratio, not a cost allocation.

The full-year guide implies a steep second half

In August, Palantir raised its 2026 US commercial guidance to more than $3.424 billion, growth of at least 134%. The first half delivered $1.359 billion. That leaves at least $2.065 billion for Q3 and Q4 combined.

Spread evenly, that is an average of at least $1.03 billion a quarter, a 35% step up from Q2. The guide has been raised twice this year, from more than $3.144 billion in February to $3.224 billion in May and then to $3.424 billion. A company that keeps raising a guide this steep is either seeing deals close faster than planned or guiding conservatively on purpose. The Q3 number tells you which.

My view is that the guide is credible, because Q2 alone added $169 million of sequential US commercial revenue. Two more quarters of that pace roughly gets you there. The risk is not that the guide fails. It is that the market already prices in a beat.

Stock-based compensation grew 66%, and where it went

Stock-based compensation is the accounting cost of paying employees in equity rather than cash. It does not consume cash, which is why Palantir and most software companies exclude it from adjusted profit. It does transfer value from existing shareholders to employees, which is why critics add it back.

In Q2 2026 the 10-Q shows $265.2 million of stock-based compensation, against $160.0 million a year earlier. That is a 66% rise. Over six months the figure was $466.8 million, up 48%.

Sales and marketing took the largest share

The line-item split is more useful than the total. Sales and marketing stock pay rose from $56.0 million to $106.1 million, up 89%. Cost of revenue stock pay rose from $15.0 million to $30.9 million, up 106%. Research and development rose 81%. General and administrative rose only 23%.

That pattern fits a company paying for go-to-market capacity. Cost of revenue includes the engineers who deploy software inside customer sites. Sales and marketing includes the people who sell it. Equity flowing to those two lines is consistent with AIP demand requiring human deployment effort, rather than selling itself through a self-serve funnel.

I read that as a mild negative for the durability thesis. Software that needs more equity-paid deployers per dollar of revenue scales more like a services business. Palantir's gross margin is still exceptional, so this is a direction to watch, not a verdict.

Source: Palantir 10-Q for the quarter ended 30 June 2026, stock-based compensation note. Percentages are our calculation from the thousands-level figures.

Dilution is smaller than the dollars suggest

The dollar cost of an equity grant is set by the share price when it is granted. A high share price means fewer shares per dollar of compensation. That is why Palantir's share count barely moves despite large stock pay figures.

Weighted average diluted shares were 2.569 billion in Q2 2026 against 2.563 billion a year earlier, up 0.2%. Basic weighted shares rose 1.5%. Shares outstanding went from 2.391 billion at the end of 2025 to 2.403 billion on 27 July 2026, a rise of 0.5% in seven months. Repurchases over the first half of 2026 totalled $1.5 million, so buybacks are not what is holding the count down.

The forward pipeline is visible too. The 10-Q lists $1.41 billion of unrecognised stock compensation. Restricted stock units account for $916 million over three years, options for $361 million over five years and stock appreciation rights for $136 million over eight years. At the Q2 run rate, that is a little over five quarters of expense already granted.

The SBC test: stock pay per dollar of US commercial revenue

The test for whether growth is real or bought is simple. Does revenue grow faster than the equity it takes to produce it? On a year-over-year basis, Palantir passes clearly. On a quarter-over-quarter basis, Q2 2026 was the first quarter in the series where it did not.

A consensus has formed that stock pay is a solved problem at Palantir because adjusted margins are above 60%. I think that is too comfortable. The annual numbers look good partly because Q2 2025 was a heavy stock pay quarter. The sequential trend is where the next problem would show up first.

Against adjusted operating income

Adjusted operating income excludes stock pay and the employer payroll tax on equity awards. The difference between adjusted and GAAP operating income is exactly those two items. In Q2 2026 that gap was $282.5 million, against adjusted operating income of $1.194 billion.

So stock pay and its tax consumed 23.6% of adjusted operating income. A year earlier the share was 42.0%. That is a real improvement, and it is the strongest point on the bull side. It also flattened: Q1 2026 was 23.3%, so Q2 ticked up rather than down.

Against incremental revenue

The second cut looks at the additions. Between Q2 2025 and Q2 2026, quarterly US commercial revenue rose by roughly $457 million. Quarterly stock pay rose by $105.2 million. That is about 23 cents of added stock pay for each added dollar of US commercial revenue.

Against total revenue, which rose $931.8 million, the ratio falls to about 11 cents. Both are healthy for a company growing at this pace. For comparison, stock pay was 15.3% of total revenue in 2025 and 24.1% in 2024. In Q2 2026 it was 13.7%. Every one of these readings says the same thing: revenue is diluting the cost of stock pay faster than stock pay is growing.

The quarter-over-quarter cut is the exception. Stock pay rose 32% from Q1 to Q2. US commercial revenue rose 28% and total revenue 19%. One quarter does not reverse a trend. It does mean the improvement was not automatic.

Is AIP revenue durable? Customers, backlog and RDV

Growth is one question and durability is another. Durable revenue renews, expands and sits under contract. Palantir's filings offer three windows on it, and they point in slightly different directions.

Customer count grew a quarter as fast as revenue

The 10-Q defines a customer as an organisation that generated revenue in the trailing twelve months. On that basis Palantir had 1,049 customers at 30 June 2026, against 849 a year earlier. That is growth of 24%, against revenue growth of 93%.

The rest of the growth came from existing accounts getting bigger. The 10-Q puts average trailing revenue from the top 20 customers at $124 million, up 67% from $75 million. Twenty customers at that average is about $2.48 billion. Trailing twelve-month revenue, summed from the last four quarterly releases, is about $6.16 billion. So the top 20 account for roughly 40% of revenue.

Expansion inside large accounts is the most durable kind of growth software produces. Concentration is the price. A company with 40% of revenue in 20 relationships is exposed to any one of them pausing a programme. The release does not disclose a US commercial customer count for Q2 2026, so I cannot split this between government and commercial accounts.

RPO covers about a third of a year, and RDV coverage is drifting

Remaining performance obligations are the GAAP measure of contracted revenue not yet recognised. Palantir reported $4.9 billion of RPO, with about 43% due within 12 months. That is roughly $2.1 billion of revenue locked in for the next year, against trailing revenue of $6.16 billion. Contracted backlog covers about 34% of a trailing year.

The company's own remaining deal value metric is larger. US commercial RDV alone was $6.238 billion, above the entire company's GAAP RPO. The two measure different things, and RDV is defined by Palantir rather than by accounting standards. How different companies' backlog figures compare is covered in the comparison of cloud backlog and RPO disclosures, and the specific credit questions a large backlog raises are in the Oracle RPO analysis.

Divide US commercial RDV by quarterly US commercial revenue and you get a coverage ratio. It was 9.1 times in Q3 2025, 8.6 in Q4, 8.3 in Q1 2026 and 8.2 in Q2. Revenue is growing faster than the deal pipeline behind it. That is either fast conversion of deals into revenue, which is good, or a pipeline that is not refilling as fast as it drains, which is not.

Where this argument is weakest

The bear case on Palantir's stock pay is older than the current numbers. Critics, including a May 2026 piece in 24/7 Wall St, argue that GAAP profits are offset by equity issuance. That argument was stronger in 2024 than it is now, and I should say so plainly.

The SBC ratio is falling, not rising

On every annual measure, stock pay is shrinking relative to the business. It fell from 24.1% of revenue in 2024 to 15.3% in 2025 and 13.7% in Q2 2026. GAAP operating margin reached 47% in Q2 2026, which already counts stock pay as a cost. A company that is that profitable after stock pay is not hiding its costs behind an adjusted metric. My flag on the Q2 sequential rise is one data point against a strong year-long trend.

One quarter, one company, one non-GAAP metric

The sequential jump may be grant timing. Equity grants cluster around annual review cycles, and a one-quarter rise can reverse with no change in strategy. The RDV coverage ratio rests on a company-defined metric whose methodology I cannot audit from outside. And the whole analysis predates the Q3 2026 filing, which may make parts of it obsolete within days.

There is also a valuation question this post does not address. Stock pay at a company priced at a high revenue multiple costs fewer shares per dollar, so dilution stays low as long as the share price stays high. If the price fell sharply, the same dollar of stock pay would cost more shares. That is a share-price risk, not an operating one, and it is outside what a 10-Q can tell you.

What to check when the Q3 2026 10-Q lands

The Q3 release will carry most of these figures, and the 10-Q a few days later will carry the rest. Here is the test I will run, with the Q2 baseline next to each line. If you run a software business, the same five rows work on your own board deck.

Thresholds are our judgement, not company guidance. The Q3 2026 guide was revenue of $2.160 to $2.164 billion and adjusted operating income of $1.292 to $1.296 billion, per the Q2 release.

The Q3 revenue guide is $2.160 to $2.164 billion, about 12% above Q2. Palantir has beaten its own guide in each of the last four releases. A beat on revenue is therefore the expected outcome, not the news. The news is in rows two and three.

For broader context on how this quarter's guidance compares across the sector, see the SaaS earnings guidance scorecard. For how Palantir's margin profile sits against other AI-exposed companies, see the AI profitability and operating margin scorecard.

Frequently asked questions

Where to start

Open the Q3 2026 release when it posts and go straight to the reconciliation table at the bottom. Add stock-based compensation to the payroll tax line, divide by adjusted income from operations, and compare it with 23.6%. That one number tells you whether Q2 was a blip.

Then run the same calculation on your own company or your largest software vendor. If equity pay is growing faster than the revenue it is meant to produce, you have found the conversation for your next board meeting. The wider pattern of net revenue retention across Q2 filings is a useful benchmark for the expansion half of the test.

Related reading