Combined average of Moat (AI Resilience), Growth, and Valuation scores.
Moat ScoreWidening
Ontological data integration platform with deep government and enterprise switching costs.
Palantir's moat is built on Operational Data Depth and AI Integration Complexity:
- Ontology Platform: Palantir's Ontology links raw data to real-world business operations. Once deployed, it becomes the operational backbone of an organization — replacing it requires years of re-integration.
- AIP (Artificial Intelligence Platform): AIP Boot Camps turn prospects into customers within days by demonstrating real ROI on their own data. This compressed sales cycle is creating a commercial moat with extraordinary velocity. The June 2026 NVIDIA alliance extended AIP downward into the model layer and that layer is now in production: Nemotron open models run inside customer security boundaries, and post-training — supervised fine-tuning and reinforcement learning — accumulates intelligence in weights the customer owns. The pitch has shifted from analytics to AI sovereignty, and Q2 2026 was the first quarter where the market paid for it.
- Government Lock-In: 15+ year relationships with US DoD, CIA, and allied intelligence agencies embed Palantir at the mission-critical layer. Switching costs are measured in years, not months. In July 2026 the US Army standardised its Next Generation Command and Control (NGC2) cloud data layer on Foundry as that programme moved from prototype to wide deployment. Q2 2026 went a step further: the first government program of record chose to run inside Maven itself, inheriting its open data standards, ontology and developer tooling, and Maven now carries over 25,000 builders across uniformed services, civilians and contractors. Department of War trailing-12-month revenue is still under 25 basis points of the Pentagon budget.
Moat Verdict
Palantir's moat is highly AI-resilient because AI is the product, not a threat to it. The Ontology platform deepens in value as more AI models are layered on top, making Palantir the connective tissue of enterprise AI operations. Learned interfaces were marked down to intact — training cost is not Excel-level UI intensity; the fortress is the ontology, not the console.
70.7 resilient · 65.0 vulnerable · 80/20 = 69.6 · = 70 · blended with 75 projected = 72
Open a moat to read its note.
Foundry and AIP require significant analyst and operator training — real personal switching costs for power users, which is intact. strong on this pillar is category-defining UI intensity; Microsoft rates the Office surface intact on the same bar. Marked down from strong on a software-sector consistency review. The ontology lock-in is scored on businessLogic (strong, routed resilient), not on the console.
Palantir encodes customer-specific workflows, decision trees, and ontologies directly into the platform — making it nearly impossible to replicate in a competitor tool. AIP strengthens this moat: every additional AI agent on Foundry deepens the ontology coupling, since the LLM needs the customer's bespoke business logic to reason. Routed to resilient via aiExposure override — AI consumption of the ontology compounds the moat rather than eroding it. The substitution thesis that drove the 2026 enterprise-software de-rating — that frontier models subsume the application layer — got its first clean test in Q2 2026 and failed it: a major Silicon Valley technology company ran a bake-off between a frontier lab with its own deployment team and AIP with Palantir's forward deployed engineers, same customer, same models, same timeline; the lab delivered nothing against the problem and the work converted to a $10M ACV contract. Net dollar retention of 157%, up 700bps sequentially, says existing customers are deepening the coupling rather than substituting away from it. Downgrade trigger unchanged and still not met — two or more named ontology displacements.
Palantir does not derive its moat from public data; its edge comes from the customer's own operational data, integrated through the ontology. A pillar that never applied is not applicable rather than weakened. Previously weakened.
Forward deployed engineers (FDEs) who embed with clients are a scarce talent model that competitors have not successfully replicated at scale. Q2 2026 produced the first clean head-to-head test: a frontier lab brought its own deployment team against Palantir's FDEs at the same customer, on the same models and timeline, failed to deliver, and Palantir booked the resulting $10M ACV contract. Held at intact rather than upgraded, because that is one company-disclosed anecdote and the scarcity claim needs a pattern. Upgrade trigger: a second and third disclosed displacement of a frontier-lab deployment team, or FDE-led accounts sustaining net dollar retention above 150% through FY27.
Foundry + AIP + Gotham + Apollo creates an integrated stack that is increasingly difficult to unbundle once adopted enterprise-wide. Upgraded from intact in August 2026 because both halves of the two-sided July rationale resolved the same way. The deepening half shipped: the NVIDIA alliance's model layer is in production, with Nemotron running inside customer security boundaries and post-training — supervised fine-tuning and reinforcement learning — accumulating intelligence in weights the customer owns. That is a bundle component that did not exist two quarters ago. The eroding half did not materialise: if Microsoft Fabric were winning commercial deals on bundled Azure pricing, US commercial TCV would not have set a record at $2.132B (+153% YoY) and net dollar retention would not have jumped 700bps to 157%. The clearest single data point is a customer that started at one operating company in Q4 2025 and converted to a three-year ~$370M deal spanning its full portfolio. Held at intact (M9, October 2026): net dollar retention is businessLogic's proof, the record TCV is a growth fact, the ~$370M deal is one customer expanding across its own companies rather than a second product attaching, and no named unbundler has failed; ORCL and SNOW sit at intact on the same kind of evidence. Upgrade trigger: a disclosed attach rate for AIP on Foundry, or a named Fabric bake-off loss. Downgrade trigger: net dollar retention falling back below 130%, or a disclosed Fortune 500 loss to a bundled hyperscaler AI suite.
The AIP loop captures operational telemetry inside each customer's security boundary and post-trains weights — but the note's own point is that those weights and ontologies are customer-owned, so the asset is the customer's, not a Palantir dataset rivals cannot replicate. Re-rated from strong to intact in the proof-point pass.
Classified-environment accreditation (IL5/IL6 and air-gapped deployments) is a barrier few defense software vendors clear, and procurement consolidates around it: the Army's enterprise agreement folds up to 75 contracts into one vehicle worth up to $10B over 10 years. In July 2026 the DIA withdrew a military-intelligence solicitation after a Palantir protest.
Limited direct network effects — value is primarily per-customer depth rather than cross-customer scale. The first real counter-evidence arrived in Q2 2026: Maven now carries over 25,000 builders — uniformed service members, civilians, contractors and outside companies developing agents and applications on a shared platform — and a government program of record chose to run inside Maven specifically to inherit its open data standards, ontology, developer tooling and peering. That is other parties building on the platform rather than another per-customer deployment. Held at weakened on a single quarter of disclosure confined to the defense channel; upgrade trigger is a second and third program of record hosting on Maven, or a disclosed third-party developer ecosystem outside government.
The Ontology sits in the operational path — logistics decisions, targeting workflows and supply-chain actions run through it daily. Maven now carries over 25,000 builders across the services, and NGC2's move to wide deployment in July 2026 puts Foundry in the Army's live command-and-control path, where removal is an operational risk, not a migration project. Held at intact (M3, October 2026): the ontology lock-in is already scored on businessLogic, Maven's builder count is the networkEffects fact and NGC2 is the systemOfRecord fact, and the file has no transaction or retention figure of its own, which DDOG, NOW and ORCL each carry at strong.
The Army standardised the NGC2 cloud data layer on Foundry and the first program of record moved onto Maven, so Palantir's schema is becoming the one defence programmes are written against — becoming, not yet the record across the enterprise. Held at intact until more programmes of record run on it. Re-rated from strong to intact in the proof-point pass.
Widening over 5–10 years. Watch: More defence programmes of record move onto Maven after the first one, and the Army's NGC2 data layer on Foundry spreads under the enterprise agreement worth up to $10B over 10 years; the trend stops if Maven stays at a single programme of record and NGC2 does not extend past the Army.
Software at near-zero marginal cost: scale shows up as network effects, data or bundling, which are rated there. No separate unit-cost lead to credit.
Buyers are enterprises choosing on switching cost, integration and performance, which the other pillars rate. The name carries reputation, not a price premium it could hold on brand alone.
Moat in 5–10 years
The defence system of record is the one pillar heading up: the Army standardised NGC2's data layer on Foundry and the first programme of record moved onto Maven, and each further programme written against Palantir's schema makes it the record rather than one vendor among several. The ontology lock-in and the classified accreditation hold at strong, the Foundry-AIP bundle holds at intact, and Microsoft Fabric has not shown up as a commercial loss while net dollar retention rose to 157%. Network effects stay where they are until Maven's builder base shows up outside government.
70 today → 75 if the trend completes · scored 72 (70/30)
Combined average of Moat (AI Resilience), Growth, and Valuation scores.
Moat ScoreWidening
Ontological data integration platform with deep government and enterprise switching costs.
Growth Score
Q3 2026 (reported November 2) beat again and accelerated: revenue $2.31B, +96% YoY (from +93% in Q2) against ~$2.18B consensus and the company's own $2.160–2.164B guide. US commercial revenue was $984M (+148% YoY, +29% QoQ) and passed US government ($953M, +96% YoY, up from +90%) for the first time. Adjusted and GAAP EPS were both $0.44 against ~$0.41 expected. FY26 revenue guidance went to $8.53–8.534B (+91% YoY) from $8.150–8.158B, US commercial growth to +143% from >+134%, and Q4 is guided at $2.64–2.65B (~+88%) against ~$2.46B consensus. The soft line is bookings: total contract value was $2.9B (+6%) and US commercial contract value $1.5B (+11%), far below Q2's record $2.132B US commercial TCV, so the forward book grew much more slowly than revenue this quarter. Q2 2026 (reported August 3) beat on every line and was the strongest quarter in the company's history: revenue $1.935B (+93% YoY, +19% QoQ) against ~$1.81B consensus and the company's own $1.797–$1.801B guide; US revenue $1.573B (+115% YoY) now 81% of the total; US commercial $764M (+149% YoY, +28% QoQ); US government $809M (+90% YoY, up from +84% in Q1). GAAP net income $1.062B (55% margin, $0.41 diluted), adj FCF $1.220B (63% margin), adj operating margin 62%, Rule of 40 at 155%. Net dollar retention 157%, up 700bps sequentially. US commercial TCV bookings set a record at $2.132B (+153% YoY); US commercial RDV $6.238B (+124% YoY); total RDV $13.1B (+83%) and RPO $4.9B (+103%). FY26 guidance was raised to $8.150–$8.158B (+82% YoY) from $7.650–$7.662B — an 11-point increase and the largest raise the company has ever made — with US commercial to >$3.424B (+134%) and adj FCF to $4.5–$4.7B. The quality of the raise matters more than its size: only ~$136M of the ~$498M increase came from the Q2 beat itself, so ~$362M is a genuine mark-up of the second half. Q3 guide $2.160–$2.164B (~+83% YoY) against ~$2.0B consensus, implying ~$2.424B in Q4 (~+72%). The soft spot is international commercial at $182M, +26% YoY and only +2% sequentially.
Valuation Score
Refreshed November 3, 2026 after the Q3 print, marked at the $184.11 close on October 30, 2026 (the last dated close available; Reuters had the stock down about 3.1% on November 2 before the report, roughly $178, which scores 67): 65 on the held $100 / $185 / $300 ladder. The ladder is not re-struck on this print. The FY26 raise is about 5% on revenue, the bookings line slowed, and at ~$184 the stock sits on the base case already. Re-marked at the $186.03 close on October 16, 2026 (the last mark before the 2026-10-19 rebalance): 65 on the unchanged $100 / $185 / $300 ladder, from 64. Only the price moved; the ladder and the thesis are not re-underwritten here. Re-marked at the $189.67 close on September 25, 2026 (the last IM25 mark before the rebalance): 64 on the unchanged $100 / $185 / $300 ladder, from 74 at the $156 reference the text below was written at. Only the price moved; the ladder and the thesis are not re-underwritten here. At ~$156 (August 4, 2026, intraday — up ~24% on the Q2 print) the stock sits ~66% of the way from bear ($100) to base ($185). The score falls from 78 to 74 even though fair value rose: the price moved further than the estimates did. The de-rating the last two reviews were tracking has fully reversed — ~46x FY26 guided revenue and ~37x NTM sales, against ~38x and ~32x a week ago — and the stock is now ~12% below where it started the year after having been down ~30%. The scenario ladder is rebuilt on a raised FY27 base of ~$12.2B (roughly +50% on the guided $8.154B; the $11.2B street figure predates the print and Citi already models +53%): bear $100 is ~20x FY27 revenue, base $185 ~36x, bull $300 ~59x — the same bull multiple as the prior ladder, rolled onto higher revenue rather than expanded. Base $185 sits deliberately just under the $187 street average and below the $200 median. What the quarter bought is contracted visibility rather than cheapness: $13.1B of remaining deal value (+83%) and $4.9B of RPO (+103%) underwrite the near-term guide in a way that was not true in July, but at ~82x forward earnings and ~81x guided FY26 adj FCF the entry is materially worse than the $123 of the July 30 review. Next catalyst: Q3 2026 earnings in early November against the $2.160–$2.164B guide.
The Ontology Moat (AIP)
Palantir's moat is built on Operational Data Depth and AI Integration Complexity:
- Ontology Platform: Palantir's Ontology links raw data to real-world business operations. Once deployed, it becomes the operational backbone of an organization — replacing it requires years of re-integration.
- AIP (Artificial Intelligence Platform): AIP Boot Camps turn prospects into customers within days by demonstrating real ROI on their own data. This compressed sales cycle is creating a commercial moat with extraordinary velocity. The June 2026 NVIDIA alliance extended AIP downward into the model layer and that layer is now in production: Nemotron open models run inside customer security boundaries, and post-training — supervised fine-tuning and reinforcement learning — accumulates intelligence in weights the customer owns. The pitch has shifted from analytics to AI sovereignty, and Q2 2026 was the first quarter where the market paid for it.
- Government Lock-In: 15+ year relationships with US DoD, CIA, and allied intelligence agencies embed Palantir at the mission-critical layer. Switching costs are measured in years, not months. In July 2026 the US Army standardised its Next Generation Command and Control (NGC2) cloud data layer on Foundry as that programme moved from prototype to wide deployment. Q2 2026 went a step further: the first government program of record chose to run inside Maven itself, inheriting its open data standards, ontology and developer tooling, and Maven now carries over 25,000 builders across uniformed services, civilians and contractors. Department of War trailing-12-month revenue is still under 25 basis points of the Pentagon budget.
Moat Verdict
Palantir's moat is highly AI-resilient because AI is the product, not a threat to it. The Ontology platform deepens in value as more AI models are layered on top, making Palantir the connective tissue of enterprise AI operations. Learned interfaces were marked down to intact — training cost is not Excel-level UI intensity; the fortress is the ontology, not the console.
70.7 resilient · 65.0 vulnerable · 80/20 = 69.6 · = 70 · blended with 75 projected = 72
Open a moat to read its note.
Foundry and AIP require significant analyst and operator training — real personal switching costs for power users, which is intact. strong on this pillar is category-defining UI intensity; Microsoft rates the Office surface intact on the same bar. Marked down from strong on a software-sector consistency review. The ontology lock-in is scored on businessLogic (strong, routed resilient), not on the console.
Palantir encodes customer-specific workflows, decision trees, and ontologies directly into the platform — making it nearly impossible to replicate in a competitor tool. AIP strengthens this moat: every additional AI agent on Foundry deepens the ontology coupling, since the LLM needs the customer's bespoke business logic to reason. Routed to resilient via aiExposure override — AI consumption of the ontology compounds the moat rather than eroding it. The substitution thesis that drove the 2026 enterprise-software de-rating — that frontier models subsume the application layer — got its first clean test in Q2 2026 and failed it: a major Silicon Valley technology company ran a bake-off between a frontier lab with its own deployment team and AIP with Palantir's forward deployed engineers, same customer, same models, same timeline; the lab delivered nothing against the problem and the work converted to a $10M ACV contract. Net dollar retention of 157%, up 700bps sequentially, says existing customers are deepening the coupling rather than substituting away from it. Downgrade trigger unchanged and still not met — two or more named ontology displacements.
Palantir does not derive its moat from public data; its edge comes from the customer's own operational data, integrated through the ontology. A pillar that never applied is not applicable rather than weakened. Previously weakened.
Forward deployed engineers (FDEs) who embed with clients are a scarce talent model that competitors have not successfully replicated at scale. Q2 2026 produced the first clean head-to-head test: a frontier lab brought its own deployment team against Palantir's FDEs at the same customer, on the same models and timeline, failed to deliver, and Palantir booked the resulting $10M ACV contract. Held at intact rather than upgraded, because that is one company-disclosed anecdote and the scarcity claim needs a pattern. Upgrade trigger: a second and third disclosed displacement of a frontier-lab deployment team, or FDE-led accounts sustaining net dollar retention above 150% through FY27.
Foundry + AIP + Gotham + Apollo creates an integrated stack that is increasingly difficult to unbundle once adopted enterprise-wide. Upgraded from intact in August 2026 because both halves of the two-sided July rationale resolved the same way. The deepening half shipped: the NVIDIA alliance's model layer is in production, with Nemotron running inside customer security boundaries and post-training — supervised fine-tuning and reinforcement learning — accumulating intelligence in weights the customer owns. That is a bundle component that did not exist two quarters ago. The eroding half did not materialise: if Microsoft Fabric were winning commercial deals on bundled Azure pricing, US commercial TCV would not have set a record at $2.132B (+153% YoY) and net dollar retention would not have jumped 700bps to 157%. The clearest single data point is a customer that started at one operating company in Q4 2025 and converted to a three-year ~$370M deal spanning its full portfolio. Held at intact (M9, October 2026): net dollar retention is businessLogic's proof, the record TCV is a growth fact, the ~$370M deal is one customer expanding across its own companies rather than a second product attaching, and no named unbundler has failed; ORCL and SNOW sit at intact on the same kind of evidence. Upgrade trigger: a disclosed attach rate for AIP on Foundry, or a named Fabric bake-off loss. Downgrade trigger: net dollar retention falling back below 130%, or a disclosed Fortune 500 loss to a bundled hyperscaler AI suite.
The AIP loop captures operational telemetry inside each customer's security boundary and post-trains weights — but the note's own point is that those weights and ontologies are customer-owned, so the asset is the customer's, not a Palantir dataset rivals cannot replicate. Re-rated from strong to intact in the proof-point pass.
Classified-environment accreditation (IL5/IL6 and air-gapped deployments) is a barrier few defense software vendors clear, and procurement consolidates around it: the Army's enterprise agreement folds up to 75 contracts into one vehicle worth up to $10B over 10 years. In July 2026 the DIA withdrew a military-intelligence solicitation after a Palantir protest.
Limited direct network effects — value is primarily per-customer depth rather than cross-customer scale. The first real counter-evidence arrived in Q2 2026: Maven now carries over 25,000 builders — uniformed service members, civilians, contractors and outside companies developing agents and applications on a shared platform — and a government program of record chose to run inside Maven specifically to inherit its open data standards, ontology, developer tooling and peering. That is other parties building on the platform rather than another per-customer deployment. Held at weakened on a single quarter of disclosure confined to the defense channel; upgrade trigger is a second and third program of record hosting on Maven, or a disclosed third-party developer ecosystem outside government.
The Ontology sits in the operational path — logistics decisions, targeting workflows and supply-chain actions run through it daily. Maven now carries over 25,000 builders across the services, and NGC2's move to wide deployment in July 2026 puts Foundry in the Army's live command-and-control path, where removal is an operational risk, not a migration project. Held at intact (M3, October 2026): the ontology lock-in is already scored on businessLogic, Maven's builder count is the networkEffects fact and NGC2 is the systemOfRecord fact, and the file has no transaction or retention figure of its own, which DDOG, NOW and ORCL each carry at strong.
The Army standardised the NGC2 cloud data layer on Foundry and the first program of record moved onto Maven, so Palantir's schema is becoming the one defence programmes are written against — becoming, not yet the record across the enterprise. Held at intact until more programmes of record run on it. Re-rated from strong to intact in the proof-point pass.
Widening over 5–10 years. Watch: More defence programmes of record move onto Maven after the first one, and the Army's NGC2 data layer on Foundry spreads under the enterprise agreement worth up to $10B over 10 years; the trend stops if Maven stays at a single programme of record and NGC2 does not extend past the Army.
Software at near-zero marginal cost: scale shows up as network effects, data or bundling, which are rated there. No separate unit-cost lead to credit.
Buyers are enterprises choosing on switching cost, integration and performance, which the other pillars rate. The name carries reputation, not a price premium it could hold on brand alone.
Moat in 5–10 years
The defence system of record is the one pillar heading up: the Army standardised NGC2's data layer on Foundry and the first programme of record moved onto Maven, and each further programme written against Palantir's schema makes it the record rather than one vendor among several. The ontology lock-in and the classified accreditation hold at strong, the Foundry-AIP bundle holds at intact, and Microsoft Fabric has not shown up as a commercial loss while net dollar retention rose to 157%. Network effects stay where they are until Maven's builder base shows up outside government.
70 today → 75 if the trend completes · scored 72 (70/30)
Growth Analysis
Growth Drivers
Key Risk
Both channels of the prior risk receded in Q2 2026 and neither triggered: US commercial accelerated to +149% rather than guiding below 70%, and the substitution test ran head-to-head — a frontier lab brought its own deployment team against AIP at the same customer on the same models and lost the work to a $10M ACV contract — so no ontology displacement has been disclosed. The live risk is now the comparison base, not the funnel. Q2 2027 laps +93% growth and a record $2.132B US commercial TCV quarter, and the company's own FY26 exit rate (~+72% implied in Q4) already assumes deceleration. Falsifiable test by the Q2 2027 print: if US commercial RDV net adds fall below the ~$1.3B quarterly pace set in Q2 2026 for two consecutive quarters, the record bookings quarter was a pull-forward, the 25-35% band loses its contracted backing, and the ~46x FY26 guided revenue multiple has no cover. A second, narrower channel is concentration: international commercial grew 26% YoY and 2% sequentially, so 81% of revenue and effectively all of the growth is now a single-geography bet that a federal budget disruption or another enterprise-software rotation would hit undiversified.
Score Derivation
90.0 base + 2.7 trajectory + 4 margin − 5 risk = 92
November 3 refresh: US commercial moves to stable (+149% → +148% YoY) and total revenue to accelerating (+93% → +96%), so the trajectory term is unchanged at two of three drivers accelerating and the computed score stays 92. The 25–35% band is held: the FY26 raise lifts the base year about 5%, but the decay path from FY27 on is unchanged and the Q3 bookings slowdown (TCV +6%) argues against pulling it forward. Base 90 (30% CAGR midpoint of the unchanged 25-35% band) + 4 trajectory (all three drivers accelerating, government having joined at +90% YoY) + 4 expanding margins (62% adj operating margin, Rule of 40 at 155%) - 5 moderate key-risk severity = 93. The band was held deliberately through a +11-point guidance raise. Q2 lifted the near-term path (FY26 +71% to +82%, FY27 ~+46% to ~+50%), but rolling that series forward on the same decay moves the five-year midpoint only ~29% to ~31% — inside the existing band and short of the 3pp threshold that would justify re-basing it. Charging the beat to cagrEstimate as well as to the trajectory flip would ratchet the pillar twice on one quarter, and the base is ~78% of the score's variance. The move from 92 to 93 is therefore attributable to one thing: government stopped being stable. primaryType is descriptive and no longer scores.
Price Scenarios (12–24 Months)
Price vs Targets
Dashed lines mark the 12–24 month bear, base and bull targets. Daily closes.
Valuation Multiples
| Trailing P/E (GAAP) | ~132× |
| Forward P/E (NTM) | ~82× |
| PEG Ratio | ~1.8× |
| Price / Sales (NTM) | ~37× |
| Price / FCF | ~81× |
A forward P/E of ~82× is close to 4× the S&P 500 (~21×) and more than double the growth-tech band (28–35×), and the Q2 print made the stock more expensive rather than less: forward earnings estimates rose roughly 7% while the price rose ~24%. The PEG has slipped from ~1.5× to ~1.8× against a ~45% EPS CAGR, still inside the "premium, requires execution" bracket but in the upper half of it. The gap from ~132× trailing to ~82× forward remains an earnings ramp rather than a deceleration signal — GAAP net income margin was 55% in Q2 and the company now guides $4.889–$4.897B of adjusted operating income on $8.154B of revenue — but Price/FCF moved the wrong way, from ~69× to ~81×, because guided FY26 adj FCF rose only ~7% against the re-rating. At ~37× NTM sales the multiple prices sustained 40%+ growth well beyond FY27, which is still the single assumption the whole valuation rests on.
Approximate figures as of August 2026.
Where We Are vs Targets
Loading live price…
Growth decelerates hard against FY27 comparisons that lap +93% and a record bookings quarter, and the multiple compresses toward ~20x forward revenue as the AI-software cohort de-rates again.
- US commercial RDV net adds fall below the ~$1.3B pace set in Q2 2026, revealing the record $2.132B TCV quarter as a pull-forward rather than a run-rate
- A federal budget disruption or extended continuing resolution stalls US government renewals, which now carry 42% of total revenue at +90% YoY
- Multiple compresses from ~46x FY26 guided revenue toward ~20x FY27 as Microsoft Fabric bundles AI orchestration into signed Azure agreements and Databricks Unity Catalog/Mosaic AI take the cost-sensitive tier of commercial deals
FY26 lands inside the raised $8.150–$8.158B guide and FY27 grows ~50% to ~$12.2B; the multiple settles at ~35-38x forward revenue, just under the street average target.
- FY26 revenue lands within the $8.150–$8.158B guide (+82% YoY) with US commercial clearing >$3.424B (+134%)
- FY27 sustains ~50% growth to ~$12.2B with net dollar retention holding above 140% and Rule of 40 above 120
- Adj FCF reaches the raised $4.5–$4.7B FY26 outlook and the $13.1B of remaining deal value converts on schedule
Sovereign AI becomes a distinct enterprise category and Palantir owns it; the US business doubles again in 2027 as management has signalled, and the multiple holds near 60x forward revenue.
- US revenue doubles again in 2027, taking the US business from ~$6.6B toward $13B and total revenue past $14B
- The Nemotron sovereign stack makes AIP the default control plane for air-gapped and regulated AI, extending the bundle into the model layer at scale
- Maven's 25,000-builder platform converts into further programs of record across the Joint Force, taking Department of War revenue past 50bps of the Pentagon budget from under 25bps today
In our long-term outlook
PLTR: common questions
Is Palantir Technologies stock a buy?
We rate Palantir Technologies Accumulate, with a composite score of 77/100: moat 72, growth 92 and valuation 65. The valuation score is struck against our bear, base and bull targets and moves with the price.
What is Palantir Technologies' moat?
Palantir Technologies scores 72/100 on moat. Ontological data integration platform with deep government and enterprise switching costs.
How fast is Palantir Technologies expected to grow?
We estimate 25-35% a year over the next three to five years, for a growth score of 92/100.
What is Palantir Technologies' price target?
Our base case is $185, with $100 in the bear case and $300 in the bull case.