Palantir: our view and what our model says about 2030
Our view
Palantir is one of the fastest-growing and most profitable large software companies in the world, and we found no restatement or open control weakness in its filings through Q3 2026. Our simulation puts the middle value at end-2030 at $410 a share, against a hurdle of $222 (today's price grown at 9% a year). 85% of simulations beat that hurdle and 5% end below today's price. The result rests mostly on our choice to value Palantir on its own past premium, with the centre capped at 70 times operating profit after stock pay.
- Price, Nov 5 close
- $154.99
- Hurdle, end-2030
- $222
- Median value, end-2030
- $410
- P10–P90 range, end-2030
- $195–$820
- Simulations beating hurdle
- 85%
- End below today's price
- 5%
Prices: official close, Yahoo and CNBC, Nov 5, 2026 · Company data: 10-Q, Sep 30, 2026 · What these numbers mean
Market data as of the Nov 5, 2026 close ($154.99). The hurdle and the probabilities depend on that price. Values per share at Dec 31, 2030, undiscounted, from 50,000 simulations.
Snapshot at the Nov 5, 2026 close
- Market value
- $373.4B
- Enterprise value
- $363.1B
- Price/sales
- 51.2×
- EV/sales
- 49.8×
- P/E
- 101.8×
- Price/FCF
- 92.1×
- Revenue growth
- +87%
- FCF margin
- 56%
- Net cash
- $10.4B
Our calculation from the company's filings (latest: 10-Q for the quarter to Sep 30, 2026, filed Nov 3, 2026); not investment advice. Revenue, earnings and cash flow: 12 months to Sep 30, 2026. Price $154.99 (official close, Yahoo and CNBC); 2409.5 million shares (common shares on the cover page of the 10-Q filed Nov 3, 2026). n/m = not meaningful (e.g. negative earnings).
Earlier versions
| Version | Price used | Horizon | Median value | P10–P90 | Beat chance |
|---|---|---|---|---|---|
| Nov 5, 2026THIS VERSION | $154.99 (Nov 5, 2026) | end-2030 | $410 | $195–$820 | 85% |
No earlier versions yet. Each row is frozen as published: when we update this view, the previous numbers stay here.
Version 1.20 · Nov 5, 2026
Editors hold this stock. Updated for the Q3 2026 results. Market data as of the Nov 5, 2026 close ($154.99); market value $373.4B on 2,409.5M common shares (Q3 10-Q cover). Including options, stock units and share-appreciation rights on the treasury method (about 2,587M diluted shares): $401B. Latest filings: the 10-K for 2025 and the Q3 2026 10-Q.
Labels used below: Fact (filings or market data), Company (management claims), Analyst (named or consensus views), Our view (our own reasoning).
Simulation result (value per share at Dec 31, 2030, undiscounted; 50,000 simulations):
- P10 $195 · P50 $410 · P90 $820 (middle half $275–$600; mean $465).
- Hurdle $222 (the end-2030 price that earns 9% a year from today): 85% of simulations beat it (about +26% a year at the median); 5% end below today’s price.
What would move the result (median value per share at end-2030 in our simulation)
- Toward the bull case:
- US commercial revenue grows fastest (top fifth of simulations): the median would be about $602.
- The market keeps a high premium (top fifth of Palantir’s own-history multiple): about $735.
- No recession and steady AI spending (neutral economy): about $425, with 88% of simulations beating the hurdle.
- Golden Dome missile-defence software becomes a funded role (45%): about +$1B of 2030 revenue.
- Main downside risks:
- US commercial grows slowest (bottom fifth): the median would be about $279.
- The premium is at the low end of its history (bottom fifth): about $220. At a Microsoft-like 24×, about $155 (our arithmetic on the median simulation), about today’s price.
- AI labs and cloud giants take the platform layer (15% of simulations): about $307 when it happens.
- A recession or an AI-spending pause (the most recession quarters): about $365.
Our view in brief
- The business is exceptional. Revenue grew 96% in Q3 at a 47% operating margin, driven by US commercial customers and the US government. Only about $2.2B of next year’s revenue is locked in by non-cancellable contracts.
- The simulation. It replays 2026–2030 50,000 times. Half of the simulations end above $410 a share at end-2030 (the median). A quarter end below $275, above the hurdle, and a quarter above $600.
- The odds. 85% of simulations beat the $222 hurdle and 5% end below today’s $154.99. That is about 26% a year at the median.
- The multiple decides it. Palantir keeps its own historical premium over the stock market (about 4 times the S&P 500’s earnings multiple), adjusted for growth against expectations, with the centre capped at 70 times next year’s operating profit after stock pay. The stock is at about 66 times now and averaged about 78 over the last four quarters, so the centre now sits between the two. The middle half of simulations runs from 49× to 97×.
- The main caution. Mature software companies such as Microsoft and Meta trade at 18–25×. Our simulation allows deep falls in the premium, but they remain rare; in reality the stock fell 48% between Nov 2025 and Jun 2026 with no change in the business.
- Profit quality has caveats. GAAP profit is flattered by a near-zero tax rate and by gains on equity stakes, so we value it at a normal 22% tax rate. Stock pay is falling as a share of revenue but rising in dollars.
- The next tests: whether contract signings recover after slowing in Q3 (total contract value +6%, after +49% in Q2), the Dec 11 funding deadline, the UK’s NHS decision around year-end and the first 2027 guide, expected in February.
1. The business in one paragraph
Palantir sells software that ties an organisation’s data into one working model and lets people and AI agents act on it; in US defence software it is the incumbent platform. For our model the mix matters most: US commercial and the US government drive the growth paths, while international is small and slow.
The full facts are in Palantir’s Long View → /stocks/pltr/long-view/
The facts behind this
Palantir sells software that pulls an organisation’s data into one model of how the organisation works (the “Ontology”) and then lets people and AI agents act on it. It has four platforms: Gotham for defence and intelligence, Foundry for commercial data operations, AIP (2023) for connecting large language models to that data, and Apollo for deploying and updating the software in any environment, including classified and disconnected ones. Customers pay subscriptions, recognised ratably over the term, and contracts usually run 1–5 years. Palantir’s own engineers work on customer sites, and pilots and “bootcamps” are often run at its own expense. In US defence software it is the incumbent platform (Maven, the Army enterprise agreement). In commercial AI it competes with hyperscalers, data platforms and now the AI labs themselves.
Fact:
- Revenue was $4.48B in FY2025 (+56%), and guidance for FY2026 is $8.53B (+91%). Q3 2026 revenue was $2.31B (+96%).
- Mix (Q3): the US is 84% of revenue. Government is about half, and the US government alone is 41%. US commercial is 43% and grew 148%; by the company’s account it passed US government for the first time.
- It had 1,090 customers on a trailing 12-month basis.
Confidence: high.
2. Where growth comes from (3–5 years)
Growth is coming mostly from existing customers spending much more, in US commercial and US defence. The contracted backlog covers only a small part of the revenue the consensus expects, so the forecast rests on Palantir continuing to expand accounts as fast as it has in the last two years. Our model stays within about 2% of consensus to 2028 and then slows growth; the record so far (guidance beaten every year since 2023) supports that path, but visibility is thin, and contract signings slowed in Q3.
The facts behind this
Drivers (Fact/Company):
- US commercial AIP adoption. Revenue grew 148% in Q3, and US commercial remaining deal value (RDV) was $6.52B (+79%), only 4% above June after a 27% rise the quarter before. Company (Q3 call): there were 686 US commercial customers (+29%), and net dollar retention was 165% (up from 157% in Q2), meaning customers from a year earlier generated 65% more revenue over the trailing 12 months.
- Contract signings slowed. Total contract value (TCV) was $2.92B (+6%, after +49% in Q2), and US commercial TCV $1.46B (+11%, after +153%). The CFO said multiple eight-figure deals were shorter than usual as customers adopted early Sovereign AI capabilities, and he expects their length and size to grow. Palantir does not say how much of the slowdown that explains, and signings also swing widely from quarter to quarter ($4.26B in Q4 2025, $2.41B in Q1 2026).
- US defence. The 10-year Army enterprise agreement (July 2025) has a $10B ceiling, which is a maximum rather than committed spending. Maven is being made an official program of record. The FY2027 request seeks more than $1.5B to expand Maven access (DefenseScoop). A Deputy Secretary memorandum in August directs funding of up to $243.9M for Palantir services through March 2027; it is a no-bid funding directive, not a contract order (DefenseScoop/The Register). Palantir also has an undisclosed role in the Golden Dome missile-defence software layer, with Anduril.
- International. This is the weak spot. International commercial grew only 21% in Q3 (1% from Q2) and international government 31%. European “sovereignty” politics cut both ways (section 4).
Size of the opportunity. Company: management offers no TAM figure, only that “every institution” faces the problem it solves. Independent: Gartner (2026 forecast, via SaaStr) puts AI software spending at $453B in 2026 and $638B in 2027. Palantir’s 2026 revenue is under 2% of that. The market is growing fast, so Palantir’s growth does not need it to take much share. The real question is whether it captures the platform layer, or whether model vendors and hyperscalers do.
Consensus path, with a fallback.
| FY2024 | FY2025 | FY2026E | FY2027E | FY2028E | FY2029–30 | |
|---|---|---|---|---|---|---|
| Revenue ($B) | 2.87 | 4.48 | 8.55 (guide 8.53) | 13.00 | ~17.85 | not public |
| Growth | 29% | 56% | ~91% | ~52% | ~37% | our path: see below |
| Adj. operating income ($B) | 1.13 | 2.25 | 5.25 (guide) | n/a | n/a | |
| Adj. EPS (consensus) | 0.41 | 0.75 | 1.70 | 2.48 | n/a |
FY2026–27: S&P Global via StockAnalysis, Nov 5. FY2028: Simply Wall St, 17 analysts, from before the Q3 results (thin, low confidence). For FY2029–30, our own path eases growth to about +28% and +23% before events and the economic cycle. That is not a consensus figure.
Backlog quality (Fact/Our view). The headline backlog figures overstate the contracted base:
| Measure (Sep 30, 2026) | Value | What it includes |
|---|---|---|
| Total remaining deal value (Company, Q3 call) | $13.4B (+55%; +2% from June) | All options assumed exercised; most contracts cancellable for convenience |
| US commercial RDV | $6.52B (+79%; +4% from June) | Same basis |
| Remaining performance obligations (RPO) | $4.9B (+88%; flat from June) | Non-cancellable only; 44% (~$2.2B) due within 12 months |
| Contract liabilities | $1.2B | Cash billed in advance |
Our view: the non-cancellable RPO due in the next 12 months is about $2.2B, roughly a sixth of next year’s consensus revenue. RPO structurally understates the recurring base for a vendor whose government contracts are typically cancellable for convenience and funded yearly. It also leaves out contracts of 12 months or less and the $513M of cancellable customer deposits.
Most of FY2027’s $13.0B has to come from renewals, options and expansions not yet signed. Conversion has been excellent so far: in Q3, $891M of the $1.13B revenue increase came from customers already on the books at the end of 2025. The weakness is visibility, not the record.
Track record (Fact). Financial guidance has been conservative since 2023 and beaten every time:
- FY2024’s initial guide of $2.66B ended at $2.87B (+8%).
- FY2025’s initial guide of $3.75B ended at $4.48B (+19%).
- FY2026’s initial guide of $7.19B has been raised three times, to $8.53B. Quarterly revenue has come in about 6–8% above guidance in 2026.
- The older record is weaker. In Feb 2021 (Q4 2020 results), management set a target of 30%+ annual growth through 2025, but growth was 24% in 2022 and 17% in 2023 before re-accelerating.
On product timelines there are no published milestones to grade. Management states its goals in qualitative terms.
Confidence: medium. The history is strong, but less than a quarter of the next year’s revenue is contracted, and long-range consensus is thin.
3. The edge (moat)
The edge is real and probably widening in defence. In commercial it rests on the Ontology and on deployment know-how. AI labs are now attacking that layer directly, which is why the platform-layer event is one of the biggest single risks in our simulation (section 7).
The facts behind this
Why customers choose it (Company/Our view):
- Palantir ties data, logic and actions into one operating model. AIP then lets any model act on it, so customers can switch LLMs without rebuilding. Karp’s pitch is “AI sovereignty”: the customer keeps its data and model weights. A newer Sovereign AI offering, launched this year, has customers fine-tune open-weight models on their own data inside Palantir’s software (Company).
- It has two decades of accreditation for classified, disconnected and allied networks. Few vendors can deploy there.
- Forward-deployed engineers learn each customer’s operations. Gartner told service providers to “embrace the Palantir FDE model”.
- Switching costs rise as workflows are built on the Ontology. L3Harris, for example, says it unified more than 30 ERP systems on it.
Signs it is widening:
- Net dollar retention rose to 165%, and the top-20 customers averaged $146M each over the trailing 12 months (+76%).
- Margins are expanding while growth accelerates (section 5).
- Maven is becoming the Pentagon’s standard, with more than 100,000 users (Company, Q3 call), and NATO declared it fully operational in June 2026.
Signs it is narrowing or capped:
- Contract signings slowed in Q3 (section 2). One quarter does not show whether that is timing or the start of slower growth; the first 2027 guide is the next test.
- Model vendors are moving up the stack. Anthropic’s Claude Managed Agents launched in April 2026, and OpenAI and the hyperscalers offer agent platforms. Investor Michael Burry argued that Anthropic is “eating Palantir’s lunch”. Our view: the threat is not head-on replacement. It is CIOs sending AI budgets to model vendors and hyperscalers first, with Palantir left for the hardest deployments. Karp’s answer on the Q3 call: Palantir is in a “symbiotic relationship” with Anthropic commercially, the labs are growing the market, and Palantir “cannot satiate demand”.
- In the UK NHS deal, at least 16 trusts and health boards say they do not use the £330M Federated Data Platform, some calling it “duplicative” (Guardian, FOI responses). That is evidence that the Ontology does not win where local systems are adequate. An unpublished NHS England analysis seen by City AM found that trusts using the platform’s discharge tool cut delays, though it does not prove the tool caused the fall, and Palantir says it handles patient data only on the NHS’s instructions.
- In Europe, Germany’s army has ruled out Maven, and France renewed its DGSI contract late in 2025 as a stopgap while it looked at a sovereign tool. In June France said ChapsVision will replace Palantir at the DGSI over several years, and a September letter from a UK minister suggests the NHS break clause is unlikely to be used in time (both reported). “US vendor” is a handicap in Europe.
Competitors (Our view; market share is not disclosed or meaningful): Microsoft (Fabric, Copilot Studio), Amazon’s AWS and Alphabet’s Google agent platforms; Databricks and Snowflake in data platforms; ServiceNow, Salesforce and SAP in workflow agents; Anthropic and OpenAI enterprise offerings; C3.ai. In defence: Anduril (also a partner), Scale AI, primes and integrators such as Lockheed, Booz Allen and Accenture.
Confidence: medium.
4. The worst realistic risk
In our view survival is not in question. The worst realistic case is that the stock re-rates while the business keeps growing: growth slows from about 90% toward 30% faster than the price assumes, possibly helped by a US political or budget turn or by AI-lab disintermediation, and the stock halves or worse even with revenue still rising. Our simulation carries this through the wide spread of the exit multiple and through the political-turn and platform-layer events.
Customer concentration and the key contracts are in Palantir’s Long View → /stocks/pltr/long-view/
The facts behind this
| Risk | How likely (3–5 yrs) | How bad | Notes |
|---|---|---|---|
| Multiple compression as growth normalises | Medium–high | Severe for the stock | The multiple went from about 125× trailing sales (Nov 2025 high) to about 47× (June 2026 low) in seven months with no business problem (Our estimate) |
| AI labs and hyperscalers take the platform layer | Medium | High | Section 3. Hardest to see early, because Palantir’s customer counts can keep rising |
| US government politics and budgets | Medium | High | 41% of revenue is US government. Democrats won Congress in November’s midterms, as Reuters reported, and a new administration arrives in Jan 2029; our model keeps its 25% chance of a political turn, mostly in 2029–30. Funding runs on a continuing resolution to Dec 11, 2026. Growth relies on the commercial-item procurement law (FASA), which Palantir says has “significantly” increased federal business |
| European backlash and sovereignty | Medium | Medium | NHS break clause; Germany and France reluctance; activist pressure |
| Execution (engineer capacity, expense ramp) | Medium | Medium | Adjusted expenses rose 20% from Q2 and 53% from a year earlier in Q3, which the CFO ties to AI platform investment and technical hiring; he says the investment continues. Deployment capacity depends on forward-deployed engineers |
| Key person and governance | Low–medium | High | Karp is central to strategy, sales and culture. The founders control voting (section 8) |
Legal and regulatory docket. The Cupat case is the one the Q3 2026 10-Q describes (contingencies note). Palantir accrues a loss when it is probable and can be estimated, and the filing gives no accrual amount for it. The filing says that at the end of September Palantir knew of no pending matter expected to have a material adverse impact on its financial statements. The other rows are policy and regulatory matters from the sources listed below:
| Case or issue | Status | Amount | Next date |
|---|---|---|---|
| Cupat securities class action (2022 claims about SPAC deals and growth) | Dismissed with prejudice Apr 2025; 10th Circuit appeal argued Mar 16, 2026 | Unspecified damages; no accrual amount given | Ruling any time; the Q3 10-Q reports none, and we found none in a web search this month |
| NHS Federated Data Platform (UK) | A parliamentary committee recommended using the early-2027 break clause; whole-life cost revised up to £1.1B (Computing); more than 44,000 people filed legal objections (Guardian, via TNW) | £330M contract | The government must decide on an extension by December 2026 (House of Commons Library) |
| Pentagon’s “supply-chain risk” designation of Anthropic | Palantir not a party. Maven had to replace Claude. A California court ruled the designation unlawful in August; in September the DC federal appeals court upheld a separate exclusion of Claude under the supply-chain security law | Rebuild cost not disclosed | Further appeals possible |
| FY2027 appropriations | Stopgap to Dec 11, 2026; request includes $1.5B+ for Maven access | — | Dec 11, 2026 |
| EU AI Act and GDPR | Phasing in; fines up to 7% or 4% of global revenue | — | Ongoing |
| Shareholder human-rights proposals | Defeated June 2026 (about 12% of votes cast for the impact-assessment proposal) | — | 2027 AGM |
Dependence map (Fact):
- US government: 41% of Q3 revenue. One unnamed customer (“Customer I”, probably a US agency) is 27% of receivables, up from 25% at the end of 2025.
- Concentration: the top 20 customers averaged $146M each over the trailing 12 months, which is about 40% of trailing revenue (Our calculation). The top three were 16% of FY2025 revenue.
- Cloud: a $5.6B minimum commitment over ten years to one unnamed cloud provider (amended March 2026), at $268–979M a year.
- Model vendors: AIP runs third-party LLMs. The Anthropic episode showed that model switches carry real rebuild costs in classified systems.
- Related parties and customer financing: the CEO’s aircraft ($11M in the first nine months of 2026). The 2021 SPAC “strategic commercial contracts” ($326M total value) have dwindled to about $400,000 of revenue in Q3. No new SPAC investment agreements were disclosed.
Product-programme risk (Fact/Our view): there is no hardware and there are no published product deadlines. The live programme risks are the Claude replacement inside Maven (timing undisclosed) and the move of Maven to a program of record, which changes how it is budgeted.
Survival check (Fact): it passes easily. Palantir has $10.4B of cash and Treasuries, no debt, an undrawn $500M revolver (matures Mar 2027), operating cash flow of $3.3B in the first nine months, and about $1.2B of customer prepayments. Its cash is far larger than its prepayments. It could get through several bad years without raising money.
Confidence: medium–high.
5. Financial quality (read the notes)
Quality is high, with three caveats: GAAP profit is temporarily flattered by near-zero tax and, earlier in the year, by gains on equity stakes; stock-based pay is falling as a share of revenue but rising in dollars; and the backlog metrics count options and cancellable contracts. For our model we use a normal 22% tax rate and treat stock pay as a real cost, which is why the multiple is applied to operating profit after stock pay.
Dilution has slowed (basic shares +2.2% in 2025 and +0.8% in the first nine months of 2026, almost all from employee awards), and the model uses about 2.6B fully diluted shares. The share bridge, cash flow and funding facts are in Palantir’s Long View → /stocks/pltr/long-view/
The facts behind this
Latest quarter (Q3 2026), the first nine months of 2026, and full years FY2024–FY2025 (10-K for FY2025).
| FY2024 | FY2025 | Nine months 2026 | Q3 2026 | |
|---|---|---|---|---|
| Revenue growth | 29% | 56% | 92% | 96% |
| Gross margin (GAAP) | 80% | 82% | 85% | 83% |
| GAAP operating margin | 11% | 32% | 47% | 47% |
| Adj. operating margin | 39% | 50% | 61% | 62% |
| Stock-based pay (% revenue) | 24% | 15% | 13% | 14% |
| Adj. free cash flow ($B) | 1.25 | 2.27 | 3.36 | 1.22 |
| Receivables, days of quarterly revenue | ~64 | ~68 | — | ~75 |
Sources: GAAP lines from the 10-K (FY2024–25) and the 10-Q (nine months and Q3 2026). Adjusted operating margin and adjusted free cash flow are the company’s non-GAAP measures (company reporting). Receivable days are our calculation using quarter-end receivables from the filings and quarterly revenue from the filings and earnings releases.
Standard checks (Fact from the 10-Q and 10-K unless marked):
- Gross profit is growing about as fast as revenue (Q3: +98% vs +96%; 10-Q), though the GAAP gross margin slipped to 83% from 85% in Q2. Cloud hosting costs (+$135M in Q3) are the main cost driver, which fits usage growth.
- Receivables ran ahead of revenue in Q3. They rose 81% in nine months, and days outstanding are about 75 against about 70 at the end of June and about 68 at year-end. Operating cash flow was 52% of revenue, against 63% in Q2. One customer is 27% of receivables. There is no inventory.
- Stock pay against free cash flow (10-Q and company adjusted FCF; ratios our calculation): nine-month stock pay of $786M equals 23% of adjusted FCF. That is a big improvement on 2024, when stock pay ($692M) was 55% of adjusted FCF. But Q3 stock pay rose 85% year on year, and $1.4B of RSU expense remains unrecognised, plus about $0.6B on options and share-appreciation rights. “Adjusted FCF” also adds back employer payroll taxes on stock pay, a real cash cost ($72M paid in cash in nine months; $68M expensed).
- Adjusted operating income excludes stock pay and related payroll taxes; the original measure also excluded non-recurring direct-listing charges. Adjusted EPS applies a notional 23% tax rate, while GAAP tax was 1.3% in the first nine months. GAAP EPS ($0.44) and adjusted EPS ($0.44) happen to match in Q3 because the stock-pay add-back roughly cancels the missing tax.
- Tax. A full $3.5B valuation allowance (10-K, as at Dec 31, 2025) sits against US and UK tax assets, including $9.0B of federal NOLs created largely by stock-pay deductions. The 10-Q says a release is a “reasonable possibility”.
- Our view: a release would produce a large one-time GAAP gain, and taxed GAAP earnings would then fall by about a fifth. Valuation should use a normal tax rate (we use 22%).
- Cash taxes are likely to stay low for years even after a release. The $9.0B federal NOL ($8.3B with no expiry), $578M of federal R&D credits and continuing excess deductions from stock pay all shelter taxable income (10-K). The 10-K says US results were a cumulative tax loss over three years once those deductions are counted.
- A 22% tax rate is therefore conservative for cash flow, though right for normalised GAAP earnings.
- Non-operating help (10-Q; SpaceX per management on the Q2 call). In the first nine months, interest income ($233M) plus other income ($134M, mostly gains on equity stakes earlier in the year; management cited SpaceX in Q2) made up 12% of pre-tax income. Other income turned negative in Q3 (−$26M), including a net unrealised loss of $25M on publicly traded stakes, which we put at about $185M; most of them face short-term sale restrictions.
- Estimate catch-ups: none disclosed. Revenue is recognised ratably, with no percentage-of-completion accounting. Capitalised costs are negligible ($65M of property and equipment; no capitalised software). Goodwill is immaterial against $11.2B of equity.
- There are no material weaknesses (10-K). The Q3 10-Q says disclosure controls were effective at a reasonable-assurance level at the end of September and reports no material change in its controls over financial reporting, the auditor is EY, the H1 2026 10-Q updates the stock-compensation accounting-policy disclosure, including SAR valuation, and no factoring or receivable sales are disclosed.
- Backlog definitions (company definitions; RPO from the 10-Q). TCV and RDV assume every option is exercised and nothing is cancelled. Only RPO is non-cancellable (section 2). The legacy SPAC contracts show that Palantir removes doubtful contracts from RDV, which is the right practice.
- The $5.6B cloud commitment is off balance sheet and fixed. It is cheap at current revenue but becomes a fixed cost if growth stalls.
Pending events that could move reported figures (no merger or spin-off): a release of the $3.5B tax valuation allowance would be a one-time GAAP gain and would start normal tax charges after it, and Maven is becoming a Pentagon “program of record”.
Share-count bridge (Fact, cover page and quarter-end awards from the Q3 10-Q; treasury-stock method at $154.99 is Our calculation):
| Item | Shares (M) |
|---|---|
| Class A (1 vote) | 2,306.7 |
| Class B (10 votes) | 101.8 |
| Class F (founder voting trust) | 1.0 |
| Basic total | 2,409.5 |
| Options (146.8M at average $10.16, mostly Karp’s 2020 grant at $11.38), net | +137 |
| RSUs and performance RSUs | +36 |
| SARs (18.5M at average $134.77, including capped awards), net estimate | +2 |
| Fully diluted | ~2,587 |
There are no convertibles, warrants, preferred shares or earn-outs (10-K and 10-Q). The $1B buyback was terminated in January 2026 after only $75M was used in 2025. Basic shares grew 2.2% in 2025 and 0.8% in the first nine months of 2026, mostly from RSU vesting (13.2M shares) and option exercises (4.8M), both already counted above.
Confidence: high on the facts, medium on interpretation.
6. Valuation today
On any metric Palantir trades at one of the highest multiples among large stocks, though measured against its growth it is not the highest among software names. On Palantir’s own premium the price needs little; on a mature software multiple it needs a great deal, so the choice of multiple decides the result.
At $154.99 (about $390B diluted EV, about 46× 2026 sales), our simulation gives $275 / $410 / $600 (P25 / median / P75) against a $222 hurdle: 85% of simulations beat it and 5% end below today’s price. That rests on our own-history multiple, with the centre capped at 70×.
The facts behind this
About 46× this year’s sales and about 91× this year’s adjusted EPS.
Fact/Our calculation: the market value is $373B on 2,409.5M shares, or about $401B fully diluted. Enterprise value (EV) is about $390B fully diluted after $10.6B of cash and investments. Data sites use basic shares, which gives an EV of about $363B.
| Company (Nov 5 close) | EV/sales (TTM) | Forward P/S | Forward P/E | EV/FCF | 3-yr rev. growth (cons.) |
|---|---|---|---|---|---|
| Palantir | 50× | 32× | 69× | 91× | ~59% (2025–28) |
| CrowdStrike | 41× | 34× | 153× | 137× | ~23% |
| Cloudflare | 34× | 27× | 180× | 253× | ~30% |
| Snowflake | 17× | 13× | 107× | 78× | ~31% |
| ServiceNow | 7× | 6× | 23× | 23× | ~20% |
Source: StockAnalysis (basic shares, adjusted P/E basis varies), Nov 5, 2026; peers’ 3-year growth from our October consensus check; medium confidence. Our view: forward sales divided by growth puts Palantir (about 0.5) between ServiceNow and Snowflake (0.3–0.4) and Cloudflare and CrowdStrike (0.9–1.5). On forward earnings Palantir sits below CrowdStrike, Cloudflare and Snowflake, because its margins are far higher, and well above ServiceNow.
On a fully diluted basis it trades at about 74× the FY2026 adjusted operating income guide (company definition, which excludes stock pay) and about 97× our estimate of FY2026 GAAP operating income (about $4.0B). Against its own history (Our estimate): about 125× trailing sales at the November 2025 high ($207.18 close), about 47× at the June 2026 low ($107.27 close), and about 50× now. The stock set a new record close of $209.05 in October. At $154.99 it is 26% below that record and 13.1% below its close before the Q3 results (official closes, our calculation). It fell 9.7% on Nov 5. The S&P 500 fell 4.4% and the Nasdaq Composite 5.5% that day (Yahoo Finance closes). Yahoo Finance tied the fall in AI-linked software to the selling in chipmakers after Democrats won Congress, following campaigns in which candidates in both parties criticised AI data centers; it named no Palantir-specific news, and we found none in the filings.
On our exit basis (EV to next-four-quarter operating profit after stock pay), it is about 66× at the Nov 5 close. By quarter, on daily-average prices, it was about 112×, 80×, 63× and 68× over the last four quarters, an average of about 78× after adjusting for the market’s own multiple (Our calculation).
What the price implies (reverse arithmetic, our arithmetic). We use our simulation’s exit basis: 70× next year’s operating profit after stock pay (2031), about $52B of net cash and 2.60B shares.
- To be worth today’s $154.99 at end-2030, Palantir needs about $5.0B of 2031 operating profit, about $10B of revenue at a 48% margin. That is less than analysts expect for 2027.
- To beat the hurdle it needs about $7.5B, about $15.6B of revenue, between what analysts expect for 2027 and 2028.
- At a Microsoft-like 24×, the hurdle would need about $22B of operating profit, about $46B of revenue, well above the top tenth of our 2030 revenue range (about $34B).
- The simulation’s median is about $15B of 2031 operating profit. Middle value $410 against the $222 hurdle: above it.
Analyst views. Consensus is Buy (S&P Global, Nov 2026, 33 analysts), with an average 12-month target of about $218, about 41% above today’s price.
Ratings, targets and the main dissenters
S&P Global, Nov 5, 2026, 33 analysts: 22 Buy or Strong Buy, 9 Hold, 2 Sell or Strong Sell. The average 12-month target is about $218 and the median about $232, with a range of $80–$273.
Targets rose after the Q3 results: Barclays’ Anthony Valentini to $273 from $265, the high in this data; Freedom’s Almas Almaganbetov to $250 from $230; Deutsche Bank’s Brad Zelnick to $240 from $225; Goldman Sachs’ Gabriela Borges to $235 from $230; and Northland’s Michael Latimore to $210 from $200. Morgan Stanley’s Sanjit Singh stayed at Equal Weight with a $233 target and raised his 2027 revenue-growth forecast to 65%, above the consensus of about 52% (TipRanks). The dissenters are Jefferies’ Brent Thill (Sell, $80), who has said valuation leaves no room for slippage, and RBC’s Rishi Jaluria (Underperform, about $90 before the results; targets reported inconsistently), who has pointed to slowing international growth.
Short interest is low at about 2.3% of shares.
Confidence: medium.
7. Value at the end of 2030 (simulation)
In our simulation the middle outcome is well above the hurdle, but about one in seven outcomes falls below it, and the result depends on the multiple. Our model values Palantir at its own past premium, with the centre capped at 70×.
Method: Valued directly at end-2030.
Implied end-2030 multiple: our end-2030 value implies 68.8× (P25–P75 48.6–96.9×) 2031 operating profit after stock pay (EV/EBIT, the basis of our exit multiple); last four quarters’ market-adjusted average, on daily-average prices: 78.0×; at the Nov 5 close on the next four quarters’ figures, not trailing: 66.0×. Also: 33.0× (P25–P75 23.1–46.9×) 2031 sales (at the Nov 5 close, on the next four quarters: 32.3×) and 85.5× (P25–P75 61.5–118.6×) 2031 earnings (at the Nov 5 close, on the next four quarters: 70.4×).
How to read the simulation, and how the multiple is set
How to read it: the simulation replays the business quarter by quarter 50,000 times, drawing each uncertain input from a sourced range. P25, P50 and P75 are the values below which a quarter, half and three-quarters of the simulations end. The hurdle is today’s price grown at 9% a year to Dec 31, 2030: $222. Values are at end-2030 and not discounted.
The multiple method, in one sentence: the 2030 value applies Palantir’s own average premium over the stock market since 2022 (about 4 times the S&P 500’s earnings multiple, measured on all economic shares and including the latest year, roughly 101 times next year’s operating profit after stock pay). Because the stock’s four-quarter market-adjusted average on daily-average prices (about 78×) is below that, the anchor becomes a geometric blend of 75% that average and 25% the longer-run figure, about 83×; its weights are our judgment. It adjusts that premium for growth against what the market expects and for market mood, applies no fade toward peers, and caps the centre at 70 times, which is still what sets the centre, so the blend leaves the median essentially unchanged. Each simulation varies around it (middle half 49–97×).
The result. The full distribution is in the Our view box at the top. The median ($410) is above the hurdle ($222), an implied return of about 26% a year (P25 +15%, P75 +39%). The lower quartile ($275) sits above the hurdle. The result holds only while Palantir keeps a multiple near its own history.
Top 5 assumptions. Our model uses 111 inputs in total; the main ones are shown here.
- The exit multiple is own history with the centre capped at 70×, plus the spread from Palantir’s own history and market mood. Microsoft and Meta trade at 18–25×.
- The revenue path stays near consensus to 2028, then slows on our own path (about +28% and +23% before events and the economic cycle; about +26% and +19% at the simulated median). The simulated 2030 revenue median is $26B, with the middle half of runs between $23B and $30B.
- Costs are tied to revenue. The 2030 operating margin after stock pay is about 48% at the median.
- AI labs and cloud giants take the platform layer in 15% of simulations.
- US government revenue follows a shared defence-budget cycle, plus a 25% chance of a political turn against Palantir from 2029.
The events that matter most (chance that each happens by end-2030, and what it does in the model):
- AI labs and cloud giants take the platform layer (assumption 4): US commercial growth 10–25 points lower for the remaining years; the largest single event effect.
- A political turn against Palantir (25%, mostly 2029–30): US government growth 8–20 points lower for two years.
- A recession or an AI-spending pause: part of the shared economic cycle (effect in the box at the top).
- Golden Dome becomes a funded role (45%): about +$1B of 2030 revenue.
- European setbacks: the NHS break clause (about 15–20%, in 2027; a September 2026 UK minister’s letter makes its use less likely) and a wider sovereignty backlash (about 40%, mostly 2027–28; France’s DGSI is already moving to ChapsVision) cut international revenue and growth by a few points.
Sensitivities. The multiple is the biggest swing factor (see the box at the top). Three other settings:
| Value per share, Dec 31, 2030 | P10 | P25 | P50 (median) | P75 | P90 | Mean | Chance to beat the hurdle | Chance of a loss |
|---|---|---|---|---|---|---|---|---|
| Multiple centred at 56×, not 70× (0.72 of the 78× four-quarter average; our test) | $160 | $225 | $330 | $490 | $690 | $385 | 76% | 9% |
| Costs not tied to revenue | $170 | $255 | $395 | $600 | $845 | $460 | 81% | 8% |
| Neutral economy (no cycles) | $210 | $293 | $425 | $613 | $824 | $475 | 88% | 3% |
Why we centre the multiple at 70×: our 4-year back-test
Our view: we set the 70× cap when Palantir traded near 100×. In our 4-year back-test, the highest-multiple third of 44 tech and growth companies kept about 0.72 of their multiple at the median, and 70× out of about 100× matched that. The stock has since fallen: 70× is now about 0.9 of its four-quarter average of about 78× and above the 66× at the Nov 5 close, so the centre no longer assumes the fade the back-test found. Keeping 0.72 of the average would put the centre near 56×; our test of that (in the table above) gives a median of about $330, with 76% of simulations beating the hurdle and 9% ending below today’s price. That test is not our main result. The range around the multiple comes from Palantir’s own volatility: a tenth of simulations end below 36× and the chance of ending below today’s price is 5%. That is closer to what high-multiple stocks have shown, though still milder than Palantir’s own falls of 48% in 2025–26 and 84% in 2021–22.
Confidence: medium for 2026–27 revenue. Low for the 2030 multiple, which dominates the result.
8. Management and governance
The founders control the company, and their equity pay comes mainly from a 2020 stock grant, so it tracks the share price rather than per-share profit. They are steady sellers rather than buyers, although most of the founders’ own sales are automatic tax sales at each quarterly vesting. Guidance credibility has been excellent since 2023 after a poor 2021–23 stretch, which is why we anchor the near-term path on guidance and consensus.
The facts behind this
- Team (Fact). Alex Karp (co-founder, CEO), Stephen Cohen (co-founder, President), Shyam Sankar (CTO, at the company since 2006), David Glazer (at Palantir since 2013, now CFO), Ryan Taylor (CRO and CLO). It is a long-tenured team with little turnover.
- Credibility events (Fact/Our view). On the negative side: the 2020 talk of 30%+ annual growth to 2025 was missed in 2022–23, and the 2021 SPAC investments plus linked “strategic contracts” faded (revenue of $87M in 2023, $15M in 2025) and led to the 2022 lawsuit. On the positive side: guidance has been beaten by wide margins since 2023. Press releases are unusually promotional (“Crushing Consensus Expectations”), so read the 10-Q, not the headline.
- What pay rewards (Fact, proxy). Karp’s 2025 pay was $8.6M: $1.1M salary and $7.5M of other items ($2.5M security, $2.0M aircraft, a $2.39M HSR filing fee, $0.36M tax services and other items). He has had no new equity since 2020.
- His 2020 grant of 141M options at $11.38 plus 39M RSUs vests on service alone, quarterly until about 2031, with partial acceleration on a change of control. There are no operating metrics in it.
- The CFO and CRO received capped share-appreciation rights at a $150 strike in 2025.
- Our view: pay rewards staying and share-price gains, not per-share profit or returns on capital. Say-on-pay passed with about 85% of votes cast, a total that includes the founders’ votes.
- Pledging (Fact). Karp has pledged 3.0M Class A shares and director Alexander Moore 978,630 shares as collateral for personal loans (proxy, as of Apr 6, 2026). The insider-trading policy bans pledging but allows limited exceptions for directors and the CEO.
- Voting control (Fact). Through the Class F trust, the founders (Karp, Cohen, Thiel) control up to 49.999999% of votes as long as they keep owning at least 100M shares. Officers and directors together hold 50.8% of votes. Thiel chairs a seven-person board. Outside holders cannot change control or strategy.
Insider activity, Oct 2025–Oct 2026: about $0.88B sold and no purchases; more than half were automatic tax sales
Fact, Form 4s, our tally. Insiders sold about $0.88B in total, and there were no open-market purchases. About $478M (54%) were automatic sales to cover tax withholding when RSUs vested; the Form 4s footnote these as required, not discretionary. About $402M were 10b5-1 plan sales.
- Thiel: $290M (2.0M shares, Mar 2026, plan).
- Karp: $272M, of which $244M was quarterly tax sales on the 20th of Feb, May, Aug and Nov and only about $28M discretionary plan sales.
- Cohen: $144M, all tax sales.
- Sankar: $102M ($73M tax sales, $30M plan).
- Others: about $73M.
Since that tally, through Nov 5: three officers reported stock awards in late October (about 987,000 shares in total), and one director’s plan sold 1,358 shares.
Trading plans: several replaced in 2026, not a red flag on their own; Karp now has a further plan running to June 2027
Fact/Our view (10-Q Item 5, Form 4s).
- Karp, Sankar, Glazer, Taylor and the chief accounting officer adopted new 10b5-1 plans in March 2026. Karp and Sankar each ended an existing plan and replaced it the same day or the next.
- That was about six weeks after Q4 results and seven weeks before the Q1 report, which raised guidance by 10 points. The plans are subject to cooling-off periods, so this is not a red flag on its own.
- Replacing a plan restarts the cooling-off period. The replacement pattern is what regulators look at; so far it has been followed by small discretionary sales (see insider activity above).
- More changes came in August and September 2026 (Q3 10-Q). Sankar ended his March plan and adopted a new one for up to 1.45M shares, running to May 2027. Glazer ended his and adopted a new one for up to 82,800 shares, to June 2027. Taylor adopted one for up to 78,000 shares, to June 2027, and director Lauren Stat one for up to 39,000 shares, to the end of 2027.
- Karp’s March plan allows up to 7.08M shares (about $1.1B at today’s price) until Dec 12, 2026. In August he adopted a further plan, also for up to 7.08M shares, running to June 2027. The 10-Q lists no termination of the March plan and does not say how the two relate. Form 4s through Nov 5 show about 0.49M shares sold since the March plan could start, of which only about 90,000 were discretionary plan sales; the rest were tax sales. Reported sales under the March plan remain small against its maximum. Watch whether sales accelerate.
- PLTR Holdings II LLC, whose shares may be deemed owned by chairman Peter Thiel, adopted a plan in August for potential purchases of up to 4M shares, running to December 2027. We saw no Form 4 showing a purchase under it through Nov 5.
Confidence: high.
9. Signposts
Most near-term tests fall in the next four months (dates in the table below). Over the longer run, US commercial and international commercial growth each quarter show whether Palantir is holding the platform layer, the main question behind our range.
The eight signposts in full, with dates
| # | Signpost | On track | Broken | When |
|---|---|---|---|---|
| 1 | Contract signings | TCV and remaining deal value growth pick up; deals lengthen as management expects | A second quarter of weak TCV; remaining deal value flat or falling; net dollar retention falls | Each quarter, next in Feb 2027 |
| 2 | US politics and budget | The new Congress leaves procurement rules alone; FY2027 defence bill funds Maven expansion | Long stopgap or shutdown; Maven money cut | Dec 11, 2026, then 2027 |
| 3 | Karp’s trading plans | Sales stay modest. So far only about 90,000 discretionary shares have been sold under the March plan; the other ~0.4M were tax sales | Large sales near the Dec 12 expiry or under the further plan | Dec 12, 2026, then to June 2027 |
| 4 | Q4 results, FY2026 10-K and FY2027 guide | Q4 above the $2.65B guide; FY2027 guide near or above $12.5B (about +47%; consensus $13.0B); tax treatment clarified | Guide well below consensus; growth below 40% | Early Feb 2027 |
| 5 | NHS break clause | Contract kept | Exercised, especially if other European bodies follow | Decision by Dec 2026; break clause early 2027 |
| 6 | Platform-layer competition | US commercial growth above 80% and customer count above 30% through 2027 (Q3: +148% and +29%) | US commercial growth halves within a year; new pilots stall | Each quarter |
| 7 | International commercial | Growth back above 30% (Q3: +21%) | Flat or falling | Each quarter |
| 8 | Maven program of record, Golden Dome, Army EA | Multi-year funded lines; Golden Dome value disclosed | Re-competition or budget cuts | 2027 |
10. Open questions
The biggest gaps in what is disclosed bear directly on our range: how much of the $13.4B remaining deal value is committed, who “Customer I” is, and when the tax valuation allowance will be released.
All nine open questions
- Who “Customer I” is (27% of receivables), and how much of US government revenue sits under the Army enterprise agreement versus Maven.
- How far the Claude replacement in Maven has progressed, and what it cost.
- When the valuation allowance will be released, and the effective tax rate after that.
- The split of $13.4B RDV between committed amounts and options, and the historical renewal rate of options.
- How much of the Q3 slowdown in contract signings comes from shorter contracts.
- The size of the SpaceX and other equity stakes, and their lock-up dates.
- The unit economics of forward-deployed engineers: revenue per deployed engineer and how fast it scales, since this is the capacity constraint critics cite.
- Which cloud provider holds the $5.6B commitment, and how much hosting cost rises with AIP usage.
- A real FY2028–30 consensus. We saw only one thin aggregator figure for FY2028.
Sources
Sources (accessed Oct 25 and Nov 5, 2026): 26 filings, releases and data sources
- Palantir 10-K FY2025 (Feb 17, 2026)
- 10-Qs Q1–Q3 2026 (May 5, Aug 4, Nov 3, 2026)
- 8-K earnings releases Q4 2023–Q3 2026
- DEF 14A (Apr 24, 2026)
- 8-K annual meeting results (Jun 9, 2026)
- Forms 4 Oct 2025–Nov 2026 (SEC EDGAR)
- Q2 2026 call transcript (Motley Fool, Aug 2026) and Q3 2026 call transcript (Stock Analysis, Nov 2026)
- CNBC (Aug 3–4, 2026)
- Yahoo Finance and CNBC (PLTR official closes to Nov 5, 2026)
- StockAnalysis (PLTR, CRWD, NET, SNOW, NOW statistics and forecasts, Nov 5 close; analyst actions, Nov 3–5, 2026)
- DC Circuit opinion, Anthropic v. Department of War (Sep 25, 2026)
- City AM (NHS analysis, Aug 11, 2026) and TNW via the Guardian (NHS objections, Sep 30, 2026)
- TipRanks (Morgan Stanley, Nov 3, 2026)
- Reuters via MarketScreener (midterm result, Nov 4, 2026) and Yahoo Finance (market closes and the software selling, Nov 5, 2026)
- House of Commons Library and Science, Innovation and Technology Committee report (NHS platform)
- Simply Wall St (FY2028)
- Gartner AI spending forecast (May 19, 2026) and SaaStr (Gartner AI software figures)
- DefenseScoop (Aug 13, 2026) and The Register (Feinberg memorandum)
- Reuters via The Star/Euronews (NATO Maven, Oct 22, 2026)
- Guardian via secondary reports (NHS FOI, Oct 17, 2026)
- Computing.co.uk (NHS costs)
- Business Wire (DGSI renewal, Dec 15, 2025)
- Law360/Bloomberg Law (10th Circuit argument)
- Axios, Fortune, BBC via secondary reports (Pentagon–Anthropic)
- Yahoo Finance/247wallst (Burry, Apr 2026)
- Benzinga, Investing.com (analyst actions, record high)
About Palantir Technologies: Website (external site)Investor relations (external site)SEC filings (external site)Wikipedia (external site)
+ 9 more
Accessed Nov 5, 2026: Palantir 10-Q for Q2 2026 · House of Commons Library: NHS Federated Data Platform · Science, Innovation and Technology Committee report · Reuters via MarketScreener: Wall Street after the midterms · DC Circuit: Anthropic v. Department of War, Sep 25, 2026 · City AM: NHS data counters claims on Palantir deal · TNW: 44,000 object to Palantir's NHS data platform · Yahoo Finance: Palantir and CrowdStrike tumble, Nov 5 · TipRanks: Morgan Stanley on Palantir
Quoting this? You may quote short excerpts under our reuse policy. Please credit WhyMyStockMoved.com and link to this page.