PLTR: 93% Growth at 50x Sales

The core results:

Line the last four quarters up in a row: $1.18 billion in Q3 2025, $1.41 billion in Q4, $1.63 billion in Q1 2026, $1.935 billion in Q2. The corresponding year-over-year growth rates are 63%, 70%, 85% and 93% (quarterly 8-Ks). A company already past a $7 billion revenue run rate has now accelerated for four straight quarters instead of decelerating, and there are very few precedents for that in enterprise software history. Scale usually does the decelerating on its own. Here it hasn't.

So how does the company see the back half? Full-year guidance is $8.150 billion to $8.158 billion, a $8.154 billion midpoint. The first half is already banked at $3.564 billion ($1.629 billion in Q1 plus $1.935 billion in Q2). Third-quarter guidance is $2.160 billion to $2.164 billion, which backs into $2.428 billion for the fourth quarter — up 12.3% sequentially. The quarter just reported grew 19% sequentially. Translated into year-over-year terms, the Q3 guide implies 83% (against $1.18 billion in Q3 2025) and the implied Q4 works out to 72% (against $1.41 billion in Q4 2025). The company's own line describes a deceleration from 93% all the way down to 72%.

Don't read that as bearish just yet. Actual second-quarter revenue came in $136 million, or 7.6%, above the midpoint the company gave back in May (TIKR calculation, 2026-08). Apply the same beat to the third-quarter guide and you get $2.33 billion — which would be 97% year over year, an acceleration. There is a persistent, stable wedge between this company's guidance and its results, and the market priced that wedge in long ago: consensus from eight analysts for the third quarter sits at $2.17 billion, above the top end of the company's own range (consensus snapshot, 2026-08-05). So the real contest was never over whether guidance is high enough. It's over whether the wedge narrows.

There is exactly one number that matters in the next report: US commercial revenue. The full-year line requires it to exceed $3.424 billion at growth of at least 134% (8-K, 2026-08-03). The first half delivered only $1.359 billion ($595 million plus $764 million), leaving $2.065 billion to make up in the back half. Do $950 million in Q3 (up 24% sequentially) and the fourth quarter only needs $1.115 billion (up 17%). Deliver just $880 million in Q3 (up 15%) and Q4 has to produce $1.185 billion — a 35% sequential jump this company has never posted. That arithmetic turns directly into an entry condition in Section 9.

The demand-side evidence chain itself is intact. Total bookings for the quarter were $3.373 billion, up 49%; US commercial bookings within that were $2.132 billion, up 153% and described by the company as a record. US commercial remaining deal value — signed contract backlog that will be recognized as revenue in the future — stands at $6.238 billion, up 124% year over year and 27% sequentially. The quarter closed 220 deals worth more than $1 million, 98 worth more than $5 million and 73 worth more than $10 million (all from the 8-K, 2026-08-03). The largest deal disclosed on the call was a three-year contract with a multinational technology company worth nearly $370 million. Management's own words: "We are raising our revenue guidance to between $8.150 $8.158 billion." (8-K, 2026-08-03)

GAAP net income was $1.062 billion, a 55% net margin. When a software company posts a 55% GAAP net margin, the instinct to lift the hood is the right one. Here's what's underneath:

A 47% operating margin, plus interest thrown off by a $9.2 billion cash pile, plus a quarter with almost no tax — that's what stacks up to 55%. Only the first layer is earned by the business. The second follows interest rates. The third ends sooner or later — and the 10-Q says so plainly: given recent and expected profitability, the company may reasonably release all or part of the valuation allowance previously recorded against its deferred tax assets. In the quarter that happens, GAAP net income jumps once. After that, the effective tax rate climbs toward normal and every subsequent quarter's GAAP net margin gets pressed back down by it. Modeling 55% as a steady state for the next several years will systematically overstate earnings.

The stock-compensation side is actually cleaner than the market's reputation for it. Second-quarter SBC was $265 million, or 13.7% of revenue, up about 1.4 percentage points sequentially; the first half totaled $466.8 million. Equity-related payroll taxes were $17 million. As of June 30, unrecognized RSU expense still stood at $916 million (10-Q, 2026-08). Diluted share count runs about 7% above basic.

Now translate dilution into what it actually costs existing holders: $466.8 million of first-half SBC, converted at $176.63 per share, is roughly 2.64 million shares — 0.11% of the 2.403 billion shares outstanding. Eleven basis points of dilution in six months is close to negligible for a company still growing 93%. But there's a trap buried in that conversion: the higher the share price, the fewer shares the same dollar expense converts into, and the lighter the dilution looks. Cut the share price in half and the same expense becomes twice the shares. That flattering dilution ratio is itself paid for by a high stock price.

Cash conversion leaves nothing to complain about: operating cash flow of $1.216 billion and adjusted free cash flow of $1.220 billion, both at a 63% margin and up 115% year over year (8-K, 2026-08-03). Adjusted free cash flow ($1.220 billion) exceeding adjusted operating income ($1.194 billion) means cash is arriving ahead of recognized revenue and deferred revenue is expanding — an independent second read on the order book. Full-year adjusted free cash flow guidance is $4.5 billion to $4.7 billion, and adjusted operating income guidance is $4.889 billion to $4.897 billion (8-K, 2026-08-03). Cash and equivalents on the balance sheet, including short-term Treasuries, total $9.2 billion, with effectively no interest-bearing debt.

The most notable upstream change shows up in the adjusted gross margin: 86% this quarter, which CFO Glazer attributed directly to "taking on increased cloud hosting costs for a government customer," saying the move would "accelerate time to value, improve efficiency, give customers stronger cost certainty, and open up space for expanding workflows in the future" (call, 2026-08-03). A pure software company voluntarily stepping one layer down the stack and putting the cloud bill on its own income statement buys stickiness and renewal certainty. What it costs is sustained downward pressure on gross margin. This doesn't read like one-off accounting noise — it reads like a direction. If the same move gets replicated with other large customers, that 86% has further to fall.

The second upstream input is foundation models. The company positions itself as a model-agnostic orchestration layer, and Karp used the call to make "sovereign AI" the core pitch, on the logic that customers don't want their data and processes locked into a frontier lab that may someday compete with them. That positioning is genuinely collecting a premium right now, and the risk is equally clear: whether the orchestration layer is a defensible long-term position depends on whether the model vendors decide to build that layer themselves. There is no quantifiable disclosure on this anywhere — it is a purely qualitative judgment.

The third upstream input, and the real bottleneck today, is people. Delivery depends on forward deployed engineers — FDEs, the engineers who embed on site and turn a customer's processes into software. On the call, management named FDEs and the ontology (the object model that organizes enterprise data so AI can act on it) as the decisive advantage competitors cannot copy. The cost shows up in expenses: adjusted operating expenses rose 37% year over year, and management said outright that third-quarter expenses "will step up meaningfully due to the seasonality of new employee onboarding as well as product and go-to-market investment" (call, 2026-08-03). What constrains this company has never been demand. It's how many people it can put in the field.

Downstream splits in two. The government side did $809 million this quarter, up 90%, with more than 25,000 developers now on the Maven platform including active-duty personnel and contractors. Sankar noted that defense-related revenue is still under 25 basis points of the Pentagon's budget (call, 2026-08-03) — part TAM rhetoric, but it does establish that the government ceiling is nowhere near being touched. The commercial side did $764 million, with 653 US commercial customers, up 35% year over year and 6% sequentially; total company customers reached 1,049, up 24% (8-K / call, 2026-08-03).

Concentration is the number in this section that deserves the most caution: the top 20 customers contributed an average of $124 million each over the trailing twelve months, up 67% year over year (call, 2026-08-03). Twenty times $124 million is $2.48 billion against roughly $6.15 billion of trailing-twelve-month revenue — the top 20 customers carry about 40% of revenue. And that single near-$370 million three-year contract accounted for 11% of the quarter's total bookings on its own.

It's worth being precise about what the backlog figure does and doesn't mean. The $6.238 billion of US commercial remaining deal value is signed contract balance, which is a firmer measure than "total contract value ceiling" style disclosures — but it still isn't a fixed collection schedule. How fast that balance converts to revenue depends on delivery pace, and delivery pace depends on the staffing bottleneck above. The company did not disclose average contract duration on this call, and gave no guidance on backlog conversion cadence. That data isn't available.

The most fragile link sits where delivery staffing meets the government budget cycle: demand has already stacked up two years of orders, and whether they convert to revenue on schedule depends on whether the engineers onboarding in the third quarter can actually carry the load.

Choosing comparables is a real problem here. Most infrastructure software companies that guided in the same window are small caps worth a few hundred million to a few billion dollars, and comparing them to a $424.5 billion market cap company is meaningless. Take instead a group genuinely comparable in scale and business model, using each company's own disclosed basis:

Set side by side, Palantir is growing more than two and a half times as fast as the second-place name on this list, and its revenue base is already in CrowdStrike's league. No other public software company at scale matches a Rule of 40 score of 155% — strong performers normally land between 50% and 70%, and companies clearing 100% are typically newly public with a tiny base.

Now the valuation multiples for the same group (EV/Sales):

Divide each multiple by its own growth rate and the ranking turns counterintuitive: Palantir comes out at 0.62, Cloudflare at 0.96, Snowflake at 0.42. Adjusted for growth, the stock everyone agrees is the most expensive is a third cheaper than Cloudflare. The genuinely expensive part isn't this year's multiple. It's that the 82% growth rate is one year's number, and the price being paid requires it to persist.

Conventional P/E basically breaks down here — the 152x trailing multiple has a denominator mixing a near-zero-tax quarter with $169 million of interest income, so the output neither reflects the business nor supports a comparison. Two rulers are worth using instead: enterprise value against full-year revenue guidance, to line up against peers; and enterprise value against free cash flow, to answer the question of how many years the market has prepaid.

The method and the results:

What does 90x free cash flow actually mean? Translate it into a falsifiable statement. For a top-quality software company whose growth has settled into the twenties or thirties, the mature multiple the market is typically willing to pay is around 30x EV to free cash flow. For Palantir to fall from here to a "expensive but normal" 30x, its free cash flow would have to reach $13.8 billion. Back that out at the company's own guided 56% free cash flow margin and it implies revenue of roughly $24.6 billion — three times this year's $8.15 billion. Compounding at 30% from this year takes four and a half years; at 25%, a little over five.

So what today's $176.63 quote has already prepaid is this: that this company gets revenue above $24 billion around 2030, holds a 56% free cash flow margin the entire way there, and still commands 30x in that year. All three conditions have to hold simultaneously for the market merely to break even. Miss any one of them and today's buyer is providing somebody else's exit liquidity.

Sell-side disagreement on this is unusually wide. Twenty-seven analysts carry a 12-month average target of $189.90, with a high of $255 and a low of $80 (Investing.com, August 2026); on another basis, twenty analysts average $200.05 (Public.com, 2026-08-12). Citi's Tyler Radke is at $260. Mizuho raised its target from $94 to $116 while maintaining an Underperform rating, and was candid about why — the raise reflected only "strong recent execution, materially higher estimates" and "an expansion in comparable company multiples." The high and the low are more than three times apart, and that spread is the most honest valuation conclusion available: nobody knows which ruler to use.

Software has no commodity cycle, but enterprise AI deployment does — from budgeted pilot, to single-point production, to full workflow rebuild, to the budget getting taken by the next technology cycle. That curve is as real as the four-year semiconductor cycle. Three sets of readings locate us on it:

Deal size direction: up, and the slope is steep. The quarter closed 73 deals above $10 million, 98 above $5 million and 220 above $1 million, all three of which the company called records (8-K, 2026-08-03). Net dollar retention of 157%, up 700 basis points sequentially — existing customers not only aren't leaving, they're accelerating what they spend. Among the specific deals disclosed on the call: a global asset manager at $35 million over three years, a nonprofit health system at $37 million over three years, and a global software services company at $15 million over five months. A $15 million deal closing in five months says delivery cycles are compressing and customers no longer need a long proof phase.

Capacity tightness: tight, and it's the supply side that's tight. Adjusted operating expenses rose 37% year over year, and the company said explicitly that third-quarter expenses step up meaningfully on the seasonality of technical hiring. Adjusted gross margin fell to 86% because the company absorbed cloud hosting costs for a government customer. There is no sign of crowding on the demand side — the crowding is in delivery.

Payback and "inventory": two years of orders in hand. US commercial remaining deal value of $6.238 billion against annualized US commercial revenue of $3.056 billion ($764 million x 4) is more than two years of stock. It's like a restaurant that has already collected two years of reservation deposits — the only remaining question is whether the kitchen can get the plates out on time.

The positioning call: mid-to-early in the up leg, not the tail end. Three reasons. Net dollar retention rose from 150% to 157% — still accelerating, not rolling over. US commercial customer count is up 35% year over year but only 6% sequentially, far below the 149% revenue growth, which means this leg is running almost entirely on existing customers expanding and the spread to new customers hasn't started. International commercial revenue is up only 26% year over year and 2% sequentially, at under a tenth of the company, so the second foundation is essentially still bare ground. Mid-cycle usually looks exactly like this: value per customer rises first, customer count second, geographic expansion last.

Two trackable leading indicators of a top. First, net dollar retention rolling over. If 157% falls back over the next two quarters, it means the expansion room within existing customers has bottomed out while the spread to new customers hasn't yet picked up the slack — the precursor to a growth air pocket. Second, sequential growth in US commercial remaining deal value dropping below sequential revenue growth. This quarter it was bookings up 27% against revenue up 28% — already close. If orders persistently lag revenue, the company is eating into the inventory on hand and new signings aren't replacing what's being consumed.

This section is a week-scale signal about entry and exit timing over the coming weeks. Readers holding for a year or more can skip straight to Section 8.

Start with the stock's own lane. Enterprise software as a group is strong — it has sat in the top bucket of the cross-subsector strength ranking for 8 consecutive trading sessions, and volume is firm, with a volume-direction reading in the 93rd percentile market-wide (as of 2026-08-13). But the technology parent sector carrying it has fallen into the weakest bucket of the cross-industry ranking (2026-08-13); the top bucket that same day held materials, health care and consumer discretionary. Strong lane, weak parent sector — the classic island structure. An isolated move with no systematic money underneath it.

Now the money itself. Enterprise software took in $4.0 billion of net inflow over the last 5 sessions and $5.9 billion cumulatively over 20 sessions (as of 2026-08-13). Those two numbers only mean something together: the last 5 days account for 68% of the 20-day total, which means the prior 15 sessions were essentially empty and the money only came back this past week. The money-flow intensity reading is just 18 (which the scoring system classifies as weak inflow), weak inflow has now run 7 days straight, and the rotation quadrant reads deteriorating. The subsector is up 4.6% over the same 5 days.

Against that, the stock's own move: from the $139.92 after-hours quote following the August 3 report (Investing.com) to $176.63 on August 13, it gained 26.2% in 8 sessions and was still up 3.3% on the day. The lane's money moved one notch; the stock moved six. The price is running more than twenty percentage points ahead of the money. That isn't a reason to be bearish, but it tells you something specific: at this level, the marginal buyer is momentum chasing, not sustained institutional accumulation.

Finally, the direction of the broader environment. The top three inflows are gold (score 57), intermediate Treasuries (43) and silver (38); the top three outflows are retail REITs (-70), short-term volatility (-67) and Mexico (-63) (all as of 2026-08-13). Within macro assets, 21% of the inflow is concentrated in gold alone. The three rotation families now forming are materials (confirmed), health care (confirmed) and industrials (forming) — this stock's sector is in none of them. The three strongest money-migration paths are utilities into consumer discretionary, utilities into materials, and consumer staples into materials. Not one leads to technology.

The overall market remains in a bullish state (score 43, 2026-08-13), and a VIX z-score of -1.64 shows very low fear. But macro liquidity reads "cautious": net liquidity of $5.84 trillion, up 0.3% week over week — essentially flat — with the 10-year Treasury yield at 4.7%. For a stock at 50.9x sales, a 4.7% long-end rate is a persistent headwind. Every notch up in the discount rate hits the most expensive assets hardest.

Put it in a trader's terms: money is running to gold, Treasuries and hard assets, technology as a whole is being trimmed, enterprise software is an island in that weakness, and this stock has outrun even the island by twenty percentage points.

Long-term (quarters to years): worth following, but the price of admission has to be acknowledged first.

The reasoning chain, first: 93% revenue growth alongside a 62% adjusted operating margin, a 63% free cash flow margin, 157% net dollar retention, and US commercial backlog up 124% year over year. Those four are mutually independent, they point the same direction, and every one of them comes from the primary 8-K and call disclosures rather than from narrative. A company can construct a story; it cannot construct four independent dimensions that corroborate each other. The demand-side evidence here is clean.

Three costs to accept, stated plainly. First, 50.9x EV/Sales means the price already prepays revenue reaching roughly $24 billion around 2030 while holding a 56% free cash flow margin, and any single year of stalling will hit the share price far harder than the fundamental miss itself. Second, a meaningful slice of the 55% GAAP net margin comes from interest income and a near-zero tax rate; once the deferred tax asset valuation allowance releases, the effective tax rate normalizes upward and reported earnings will grow more slowly than the business. Third, the top 20 customers carry roughly 40% of revenue, so losing any single large customer leaves a visible hole in a quarter's numbers.

How to hold it, driven by a verification point: test the position against US commercial revenue in the November 9 report, not against the share price. The full-year line requires $2.065 billion in the back half; $950 million in the third quarter keeps it on track. Two consecutive quarters below that arithmetic line means the long-term thesis itself needs to be re-evaluated, regardless of where the share price happens to be at that moment.

Near-term (days to weeks): don't chase, wait for a pullback.

The reasoning is entirely from the previous section: the stock ran 26.2% in 8 sessions while the lane's money-flow intensity reads only 18, weak inflow has persisted 7 days, and the rotation quadrant reads deteriorating; the technology parent sector has fallen to the weakest bucket in the cross-industry ranking; market money is migrating into gold, Treasuries, materials and health care, and this stock's sector isn't in any of the forming rotation families; a 4.7% long-end rate is a persistent headwind for the most expensive assets. Price-volume divergence doesn't predict direction, but it describes the current odds precisely — chase in at this level and you are paying full price for good news that has already been delivered.

What to wait for: either the money flow comes up on its own and realigns price with capital, or a pullback to a level with a rationale behind it. Two conditions, three price levels, and what to watch daily follow in order below.

Entry condition one (money flow, trackable daily): the enterprise software subsector's money-flow intensity reading recovers from the current 18 to above 40 and holds for 3 consecutive sessions, while the technology parent sector climbs out of the weakest bucket of the cross-industry ranking. Current state: 18, weak inflow for 7 straight days, rotation quadrant deteriorating, parent sector in the weakest bucket (2026-08-13). Satisfying this condition means the marginal buyer has shifted from momentum chasing to capital allocation, and only then is the twenty-point divergence from Section 7 actually filled in.

Entry condition two (earnings number, tied to November 9): third-quarter US commercial revenue ≥ $950 million. That number comes straight out of the company's own full-year guidance: full-year US commercial has to exceed $3.424 billion, the first half delivered only $1.359 billion, so the back half owes $2.065 billion. After $950 million in Q3 (up 24% sequentially), the fourth quarter only needs $1.115 billion (up 17%) — a normal curve. If Q3 lands between $880 million and $920 million, hitting the full-year target requires a sequential gain above 30% in Q4, a magnitude this company has never achieved, and condition two is judged not met. A supporting line to check alongside it: if total revenue comes in below $2.25 billion, it says the wedge between guidance and results is narrowing — an earlier deceleration signal than the headline growth rate.

Three reference levels

The failure level sits 24% below the current price — too far away — so pair it with an earlier warning: a close below $152.19 that fails to recover for three consecutive sessions. That trigger says the act of filling the gap has turned into a trend reversal, and there's no need to wait for the 50-day to break before deciding.

Two numbers to check daily. First, the enterprise software subsector's money-flow intensity reading — back above 40 and holding for 3 days satisfies entry condition one; conversely, if it drops into negative territory and the lane's strength ranking falls out of the top bucket, the island structure has broken and the near-term case shifts from waiting to avoiding. Second, the closing price relative to $152.19; the trigger rule is the early warning from the previous section.

Two numbers to check weekly. First, whether this stock's sector has entered any forming rotation family — currently materials confirmed, health care confirmed, industrials forming, technology in none. If technology shows up on the forming list, that adds a layer of confirmation to condition one. Second, the direction of the 10-year Treasury yield relative to 4.7%: down toward 4.4% and the discount pressure on a 50x sales multiple eases; above 5% and this class of asset gets sold ahead of any fundamental change.

Three numbers to check quarterly, all in the November 9 report. First, US commercial revenue, threshold $950 million, decision rule per Section 9. Second, net dollar retention, at 157% this quarter and up 700 bps sequentially; if it rolls over below 150%, the "mid-to-early cycle" placement from Section 6 has to be pushed back and the long-term thesis needs re-evaluation. Third, adjusted gross margin, at 86% this quarter after absorbing a government customer's cloud hosting costs; a further slide below 82% would say that stepping one layer down the stack has gone from a single case to a pattern, and the 62% operating margin moat gets ground thinner over time.

Two other variables could appear at any time. Once the deferred tax asset valuation allowance releases, that quarter's GAAP net income jumps — but it's a one-time event, and the effective tax rate normalizes upward afterward. When you see that headline, judge the underlying trend on adjusted operating income rather than GAAP net income. And any renewal or loss announcement involving one of the top 20 customers matters: that 40% revenue concentration means a single piece of news can change the full-year model.

If any of the above triggers, we'll publish a follow-up.

Catalyst calendar

We track every item on that checklist daily — whether the subsector's money-flow intensity gets back above 40, whether a close below the 200-day moving average fails to recover for three straight sessions, or whether US commercial revenue in the November 9 report clears the $950 million arithmetic threshold. If any of them triggers, we publish a follow-up that goes out with that day's market review, so readers don't have to sit refreshing the data pages themselves. Every number in this piece has a source: the company-side revenue, margins, bookings, backlog and guidance all come from the 8-K, the 10-Q and the call transcript, with dates in parentheses. The money-flow and sector-strength readings can all be verified line by line on the corresponding data pages on the site under the date 2026-08-13. Judgments can differ. The numbers shouldn't.

This article was produced on the NextPick research workbench; every money-flow, tier and valuation figure it cites can be verified on the corresponding data surfaces of the site, and the moment any tracking condition listed in the report triggers, an addendum will go out together with that day's daily review.