
Palantir
Enterprise + government SaaS/usage software; large multi-year contracts, land-and-expand via AIP, net dollar retention above 150% (Q1-Q2 FY26); in Q3 FY26 US government was ~41% of revenue, US commercial ~43% and international ~16% (computed), with US commercial the growth engine and, for the first time, larger than US government
The thesis on this name
State of AI Compute
AI-native operating-system software (Gotham/Foundry/Apollo + AIP) that lets governments and enterprises run mission-critical decisions on an ontology over their data; the breakout US-commercial AIP story.
State of Enterprise AI SaaS
Best fundamental momentum in the cohort by a wide margin — Q1'26 rev +85% to $1.633B, US commercial +133%, op margin 60%, Rule of 40 = 145% (Palantir Q1'26, May 2026). AIP is the clearest 'agents-replace-software' winner: it sells outcomes/ontology, not seats. The catch is purely valuation, NOT the business — hence 'high potential' but LOW conviction: at ~140x forward / 200x+ trailing P/E, the multiple already prices a flawless decade. Classified here as high-potential because the growth is real and durable; the conviction tax is the entry price.
State of Enterprise AI SaaS
Best enterprise-AI product-market fit (AIP + ontology + FDEs) at a price that leaves little room for error. The June avoid was retired when its own test fired on the Q3 FY26 print (2 Nov 2026), so the board holds the 3% capped long its book carries and adds only on a 30%+ drawdown.
State of Enterprise AI SaaS
Best fundamentals, worst valuation (~140x fwd); tiny so a multiple reset is survivable. Add only on 30%+ drawdowns.
State of Frontier AI
Best listed application-layer + verticals operator (AIP, ShipOS, GE Aerospace 26% production lift; FY26 guide raised to ~$7.65B / ~71% growth; US revenue +104% YoY) — the fundamentals are genuinely excellent. But the stock prices perfection: fwd P/S >40 and fwd P/E ~76 at ~$107-112 / ~$260B+ cap. The 'short_avoid' tag with LOW conviction means: do NOT short the operator (the business is real and could grow into it), but the entry is a valuation trap at these multiples — avoid chasing, accumulate only on a deep reset.
State of Frontier AI
Excellent application-layer operator (AIP, ShipOS, FY26 guide ~$7.65B / ~71% growth, US rev +104%) but priced for perfection at fwd P/S >40 — own a starter, scale ONLY on a deep multiple reset, never chase.
State of Frontier AI
Best listed application-layer + verticals operator (AIP, ShipOS, GE Aerospace 26% lift, FY26 guide ~$7.65B / ~71% growth). But fwd P/S >40 and fwd P/E ~76 price perfection. Starter sized at a deep haircut; the size is in the staged-entry plan, not here.
State of Physical AI
Listed proxy for the ANDURIL/SHIELDAI defense-autonomy thesis — AIP coordinates autonomous drone swarms; the autonomy-software moat rhymes with Lattice/Hivemind.
State of Physical AI
The way to ride the ANDURIL/SHIELDAI defense-autonomy thesis on the public tape — AIP coordinates autonomous drone swarms; the autonomy-software moat rhymes with Lattice/Hivemind. US rev doubling YoY. Richly valued, so a 2% theme-proxy, not a core.
Earnings, margins, COGS & capex
Hyper-growth re-accelerating: FY25 +56% to $4.5B, then Q3 FY26 +96% Y/Y to $2.313B (Q1 +85%, Q2 +93%) with US revenue +119% and US commercial +148%, which surpassed US government revenue for the first time. GAAP-profitable ($1.128B net income attributable to common stockholders, 49% margin in Q3), 53% adjusted FCF margin, Rule of 40 at 158%, $10.4B cash and Treasuries, no debt. The debate is entirely valuation, not fundamentals — though new contract value cooled in Q3 (US commercial TCV +11% Y/Y against US commercial revenue +148%).
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~17¢ is cost of goods and ~51¢ operating expense, leaving ~32¢ of operating profit (~49¢ net).
Revenue trend
Margins
expanding
up sharply
up Y/Y (46% in Q3 FY25); down from 63% in Q2 FY26
stable/high
best-in-class
COGS structure
Primarily cloud/hosting, deployment & support engineers; gross margin ~83% in Q3 FY26 (computed). Low marginal cost per incremental seat/usage once ontology is deployed.
Capex
Immaterial — asset-light. Growth funded entirely from operating cash; capital intensity near zero vs revenue.
Latest earnings
Beat against guidance: revenue $2.313B vs $2.160-2.164B guided (about $150M above the top of the range, computed); adjusted income from operations $1.427B vs $1.292-1.296B guided; GAAP and adjusted diluted EPS $0.44; FY26 revenue guide raised to $8.530-8.534B from $8.150-8.158B and the US commercial guide to above $3.559B (fact). Consensus figures were not checked.
Q4: revenue $2.649-2.653B, adjusted income from operations $1.640-1.644B. FY26: revenue $8.530-8.534B (+91%); US commercial >$3.559B (at least +143%); adjusted income from operations $5.245-5.249B; adjusted FCF $4.8-5.0B; GAAP operating and net income expected in every quarter of the year
- Revenue Q3 FY26
- $2.313B (+96% Y/Y, +20% Q/Q); first quarter above $2B
- US revenue growth
- +119% Y/Y to $1.937B
- US commercial growth
- +148% Y/Y and +29% Q/Q to $984M, above US government revenue for the first time
- US government growth
- +96% Y/Y and +18% Q/Q to $953M
- GAAP net income
- $1.128B attributable to common stockholders (49% margin)
- Adj FCF
- $1.216B (53% margin)
- Net dollar retention
- 150% in Q1 FY26 and 157% in Q2 FY26 (press reports of the call; not in a filing); not given in the Q3 FY26 release
- Customers
- 1,007 at Q1 FY26 (+31% Y/Y); no company-wide count in the Q3 FY26 release
- Contract value
- Q3 FY26 total contract value closed $2.918B (+6% Y/Y); US commercial TCV $1.463B (+11% Y/Y); US commercial remaining deal value $6.517B (+79% Y/Y, +4% Q/Q); 212 deals of at least $1M, 107 of at least $5M and 64 of at least $10M (fact)
- Cash / debt
- $10.4B cash and US Treasuries at Sep 30 2026 / no borrowings on the balance sheet (fact)
- Rule of 40
- 158%
Growth drivers
- US commercial AIP land-and-expand (US comm +148% Y/Y in Q3 FY26, FY26 guide raised to >$3.559B, at least +143%)
- Net dollar retention 150% in Q1 FY26 and 157% in Q2 FY26 (press reports of the call; not given in the Q3 FY26 release) — existing accounts expand fast
- Customer count 1,007 at Q1 FY26 (+31% Y/Y — no company-wide count in the Q3 FY26 release), 212 deals of at least $1M closed in Q3 FY26, bootcamps converting pilots to production
- US government re-acceleration (+96% Y/Y in Q3 FY26) and defense/AI mandates
- Forward-deployed engineering + ontology lock-in deepening per account
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-02-17. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’25
Bull & bear
The only large-cap software name compounding 85-96% (Q1-Q3 FY26) with GAAP profits, 53-63% adjusted FCF margins and a deepening ontology moat — if AIP becomes the enterprise AI operating system, today's 'expensive' multiple is paid down fast.
- 96% Y/Y growth in Q3 at a ~$9.3B annualized run-rate (computed) with the FY26 revenue guide raised to ~91% growth — defies the law of large numbers
- Rule of 40 at 158% and 150%+ net retention (Q1-Q2 FY26) prove durable, expanding economics
- US commercial +148% (to $984M, larger than US government revenue) shows the commercial flywheel compounding beyond government
- $10.4B cash and Treasuries, no debt, 53% adjusted FCF margin — funds growth and buybacks without dilution pressure from financing
- Structural AI tailwind: enterprises need a decision/agent layer over their data, and Palantir's ontology is years ahead
A fundamentally excellent business at an indefensible price — ~103x FY26 consensus non-GAAP EPS and ~50x the FY26 revenue guide (estimate; computed) leave zero margin of safety after a ~55% rebound from late June; any deceleration, heavy insider selling and rising SBC make the risk/reward asymmetric to the downside (the stock is -1.3% YTD at $175.40).
- Forward P/S ~50 on the FY26 revenue guide (~33 on FY27 consensus revenue) and forward P/E ~103 on FY26 consensus EPS (~71 on FY27) vs software median ~18 (estimates; computed) — prices in a decade of flawless execution
- Stock -1.3% YTD and ~18% below the $212.94 52-week high after a ~55% rebound from late June; the multiple has re-expanded as fundamentals beat
- Insiders selling ~9:1 with zero open-market buys (Thiel offloaded 2M+ shares; Karp selling on RSU vesting) — bear/Citron fair-value calls near $40-$103
- SBC of $318.9M in Q3 FY26 (+85% Y/Y, 13.8% of revenue) is a real cost behind the adjusted margins, although weighted diluted shares were flat Y/Y at 2.574B
- Government concentration + contract lumpiness means a single soft quarter could re-rate the stock violently
What it is worth
Reverse-DCF / multiple sanity check on ~$422.6B market cap at $175.40 (Nov 3 2026 close). Each case is FY2027 consensus non-GAAP EPS ($2.48; estimate, aggregator) times a named multiple: bull is the multiple the average analyst target implies, base is today's multiple (up to the named-peer median), bear is the cheapest named software peer. Peer multiples are Nov 3 2026 closes over consensus EPS for the fiscal year nearest calendar 2027, estimates from an aggregator.
~$58 (-66.7% from the close)
Microsoft's 23.6x, the cheapest named peer that sells enterprise software and AI platforms (Snowflake is dearer), on FY27 consensus EPS of $2.48 (estimates, aggregator); consensus is held and the multiple carries the downturn. Booz Allen (~12.9x, ~$32) is cheaper but a government services contractor, so it is not used. Driver: growth decelerates faster and/or AI multiples compress further and the premium unwinds even with strong fundamentals (the price, not the business, is the risk). The June bear of ~$40-80 (15-25x sales, 40-60% downside) equals 16-32x today's FY27 estimate (computed); this bear sits inside it. It is 45% below the 52-week low of $106.37 (Jun 25 2026) and below the lowest analyst target ($80; estimate, aggregator): it prices Palantir losing its whole growth premium and trading like a mature software platform, and published bear fair values run $40-$103.
~$175 (0.0% from the close)
today's multiple, 70.7x FY27 consensus EPS of $2.48 (estimate, aggregator); the named software peers (Microsoft ~23.6x, Snowflake ~97.4x) have a median of ~60.5x, below today's multiple, so there is no step up. Driver: growth normalizes from 96% toward ~40-50% then ~30%, FCF margins ~50%, returns track FCF compounding net of dilution. The June base let the multiple compress to ~25-35x sales; here the multiple is held, because today's ~32.5x FY27 consensus revenue (computed) is already inside that range. The June base price of ~$110-150 equals 44-60x today's estimate, below today's multiple, and did not hold with the stock at $175.40.
~$218 (+24.3% from the $175.40 close): the multiple the average analyst target implies, 87.9x FY27 consensus EPS of $2.48 (target $217.97 across 33 analysts; estimates, aggregator, Nov 4 2026). Driver: AIP becomes the enterprise AI operating system; ~45-50% revenue CAGR sustains for years, FCF margin >50%, premium multiple holds. The June bull of ~$180-210+ ('upside re-rate toward prior highs') equals 72.6x and up on today's estimate, below the 87.9x the average target implies; the 52-week high of $212.94 is 85.9x (computed).
At $175.40 and a ~$422.6B market cap (computed), the stock trades at ~103x FY26 and ~71x FY27 consensus non-GAAP EPS ($1.70 and $2.48, estimates, aggregator) and ~50x the FY26 revenue guide of $8.53B (~33x FY27 consensus revenue of $13.0B, estimate), with an adjusted FCF yield of ~1.2% on the FY26 guide midpoint (computed). Named software peers on consensus EPS at the Nov 3 close: Microsoft ~23.6x (calendar 2027, average of the fiscal years to Jun 2027 and Jun 2028) and Snowflake ~97.4x (year to Jan 2028); Booz Allen trades at ~12.9x but is a government services contractor, so it is not used as a software comparison (computed; estimates, aggregator). The price implies Palantir sustains ~38% revenue growth a year for five years while holding adjusted FCF margins above 45% and a premium multiple: at an assumed ~10x sales multiple in the early 2030s, ~$42B of revenue (~5x the FY26 guide) is needed to support today's market cap (analysis; the 10x exit multiple is an assumption). Fundamentals (96% growth in Q3, 53% adjusted FCF margin, Rule of 40 at 158%) can support an above-market multiple, but ~50x sales bakes in near-flawless execution, and new-contract growth lagged revenue in Q3 (US commercial TCV +11% against US commercial revenue +148%); even a deceleration to 'merely great' (30-40%) de-rates the stock. The June cases (bull ~$180-210+, base ~$110-150, bear ~$40-$80) did not hold: they were framed on ~$107 (the base and bear on 25-35x and 15-25x sales), and the stock closed Nov 3 at $175.40 (fact). Consensus (aggregator) averages a $217.97 target across 33 analysts, range $80-$273. Bear/Citron fair-value calls cluster $40-$103; bull cases extend on AIP-as-AI-OS optionality.
SWOT
Strengths
- Re-accelerating 96% growth at a ~$9.3B run-rate (Q3 revenue x 4, computed) — extremely rare at scale
- GAAP-profitable with 53% adjusted FCF margin and 158% Rule of 40 (Q3 FY26)
- $10.4B cash and Treasuries, no debt, self-funding growth
- Deep ontology + forward-deployed-engineering moat → 150%+ net retention (Q1-Q2 FY26)
- Entrenched, hard-to-displace US government/defense position
Weaknesses
- Extreme valuation (~103x forward P/E on FY26 consensus, ~50x forward P/S on the FY26 guide; estimates) prices perfection
- Heavy stock-based compensation ($318.9M in Q3 FY26, +85% Y/Y — 13.8% of revenue), with no share repurchases in the nine-month cash-flow statement
- Still government-exposed (US government ~41% of Q3 FY26 revenue, computed) with lumpy contract timing
- Persistent, large insider selling (Thiel, Karp) — no open-market buys
- Concentration of largest commercial deals; long, bespoke sales cycles historically
Opportunities
- US commercial AIP TAM expansion as enterprises operationalize LLMs on their own data
- Allied-government and defense AI budgets rising globally
- Warp Speed / supply-chain and healthcare verticals scaling
- Becoming the default 'AI operating system' / agent-orchestration layer for the enterprise
Threats
- Hyperscalers (Microsoft Fabric, AWS, Google) and Databricks/Snowflake bundling AI cheaper
- Multiple compression if growth merely decelerates to 'great' from 'spectacular'
- US government budget / political volatility and contract delays
- AI-platform commoditization eroding the differentiation premium
Moats, dependencies & bottlenecks
Moats
Once a customer's operations run on the Foundry/Gotham ontology, switching means re-platforming the business — drives 150%+ net retention (Q1-Q2 FY26).
FDE model embeds Palantir in the customer's hardest workflows; hard for cloud-bundled tools to replicate.
Security clearances, IL-class accreditations and a 20-year track record are a multi-year barrier for new entrants.
Real lead in operationalizing LLMs on enterprise data, but hyperscalers are racing to bundle equivalents.
Bootcamps compress sales cycles and seed expansion; a process advantage more than a structural one.
Dependencies
Customer concentration ~41% of Q3 FY26 revenue (computed); subject to appropriations, political shifts and contract timing.
Foundry/AIP run on public cloud; hosting is the main COGS and also a competitive frenemy relationship.
Anthropic, Google, open models) AIP orchestrates third-party LLMs; model availability/pricing/quality flows through, though Palantir is model-agnostic by design.
Demand environment Commercial demand rides the enterprise-AI spending wave; a pullback would hit the highest-growth segment.
Engineering talent retained largely via SBC; dilution is a structural drag on per-share value.
Advantages
- Re-accelerating growth at scale that no other large-cap software matches
- GAAP profitability + 53% adjusted FCF margin while growing 96% (Q3 FY26)
- $10.4B cash and Treasuries, no debt — fortress balance sheet
- Switching costs from ontology lock-in (150%+ NDR, Q1-Q2 FY26)
- Entrenched, accredited government/defense franchise
Weaknesses
- Valuation leaves no margin of safety (~103x fwd P/E on FY26 consensus, ~50x fwd P/S on the FY26 guide)
- Heavy SBC and ongoing dilution
- Persistent large insider selling signals
- Government revenue concentration and contract lumpiness
Bottlenecks
- Forward-deployed-engineer capacity gates how fast large deployments scale
- Long, bespoke enterprise sales cycles for the biggest accounts
- Security-clearance and accreditation lead times on government work
- Customer change-management — operationalizing ontology requires deep org buy-in
Top signals & trends
Top signals
FY26 revenue guide went from $7.65-7.66B (May) to $8.15B (Aug) to $8.53B (Nov 2) and US commercial to >$3.559B (at least +143%); Q3 revenue of $2.313B beat the $2.160-2.164B guide by about $150M (fact).
Best-in-class expansion + efficiency — fundamentals not the problem.
The multiple has re-expanded: ~103x FY26 consensus non-GAAP EPS of $1.70 and ~50x the FY26 revenue guide at $175.40, 17.6% below the $212.94 52-week high (computed; consensus is an aggregator estimate).
Thiel/Karp continued selling; bearish sentiment signal even if Rule 10b5-1 scheduled.
A quality-of-earnings flag in dollars, though the burden eased as a share of revenue (14.6% a year earlier) and weighted diluted shares were flat Y/Y at 2.574B (computed from the Q3 release).
US commercial total contract value closed was $1.463B, up 11% YoY and down from a record $2.132B in Q2, and US commercial remaining deal value was $6.517B, up 79% YoY but only 4% on the quarter, even as US commercial revenue grew 148% (fact).
Trends
Core tailwind — companies need a decision/agent layer over proprietary data; Palantir's headline growth driver.
Government segment re-accelerating (+96% Y/Y in Q3 FY26) on AI/defense mandates.
Microsoft Fabric, Databricks, Snowflake Cortex commoditizing parts of the stack and competing on price/ease.
Sector-wide de-rating of richly valued AI names hit the most expensive names hardest in June-July 2026; Palantir has since recovered ~55% from its Jun 26 close (computed), so a repeat would again fall hardest on the most expensive names.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Cloud infrastructure host for Foundry/AIP deployments; primary COGS input.
Cloud host (esp. government/IL-class) — supplier and competitor simultaneously.
Cloud infrastructure + Gemini models available through AIP's model-agnostic layer.
GPU compute underpinning the model inference Palantir orchestrates (indirect, via clouds).
Foundation-model providers whose LLMs AIP orchestrates; technology inputs, not exclusive.
Anchor government customer; Gotham + AI mission software; +96% US gov growth in Q3 FY26.
Long-standing classified deployments — the original franchise.
manufacturing, healthcare, financial) 1,007 total customers at Q1 FY26; US commercial +148% Y/Y in Q3 FY26 is the growth engine.
International government deployments expanding with sovereign-AI demand.
Lakehouse + AI platform; strongest technical alternative for data/AI workloads; pre-IPO.
Cloud data platform; Cortex AI surpassed $100M AI run-rate, 9,100+ accounts using AI features; competes on data + ease.
Bundles data + AI + Copilot at enterprise scale; the most dangerous distribution-led competitor.
Enterprise AI apps (predictive maintenance, supply chain, defense analytics); smaller, direct AIP rival.
Government IT primes and integrators competing for the same defense/intel budgets.
Defense-tech software/hardware; competes for next-gen DoD mission software mandates.