PLTR — Palantir Technologies — Pass-with-trigger — 2026-05-24

Refreshed 2026-05-24 — Greenwald-modified doctrine recalibration; central value lifted from $52 (EPV-only) to $85 (EPV-plus-growth), trigger from $29 to $60. Prior version: 2026-05-23-original-EPV-only; refresh delta: 2026-05-24-greenwald-refresh.

1. Bottom line

Pass-with-trigger at $60. Palantir is operationally exceptional and structurally interesting; the price is wrong. At ~$135 / ~$325B market cap, the equity prices in a near-perfect decade of execution — sustained 25%+ revenue CAGR for 10 years, durable margin expansion, no commoditization of the application layer — and leaves no margin of safety under any reasonable framework.

Under the Greenwald-modified methodology (EPV-plus-growth, narrow-moat runway band, 30% MoS for verified compounders), central value is $85 with a range of $70-100, against which the re-engagement trigger is $60. Under the more conservative pure-Klarman EPV-only frame the central value would be $52 and the trigger $29; under the more generous Mauboussin-compounder frame it would be $125 and $103. The full four-methodology dispersion is in PLTR-shadow-matrix; the variance is the methodology question made visible.

The variant perception is not on the business — PLTR's quality is real. It is on what the price implies. The street's $194 mean PT requires the company to deliver a top-decile-of-software-history decade. Our base-rate work says that's possible but not probable; the deep-value discipline requires we wait for a price that doesn't demand it. See PLTR-consensus-gap for the full street-vs-AlphaSteve decomposition.

Position: none. Re-engage at $60 (a ~55% drawdown from current), or on durable evidence of structural moat-widening on the commercial side that the current price does not already require.

2. The business

What they do

Palantir builds software for integrating and operating on data at enterprise and government scale. Two product platforms: Gotham (government, defense, intelligence) and Foundry (commercial). Layered on top: AIP (Artificial Intelligence Platform), launched 2023, which has become the primary commercial growth vector and the centerpiece of the AI-narrative bull case.

How they make money

Multi-year software contracts with US and allied government agencies — high contract value, long sales cycle, recurring revenue with usage-based components — combined with commercial enterprise contracts via a "bootcamp"-led land-and-expand motion with enterprise SaaS economics. Some professional services revenue remains, but the company has materially de-emphasized services in favor of platform.

Segments (Q1 FY26)

ARR ~$6.5B trailing. Commercial customer count 1,007 (+31% YoY). <span class="tier-t1" title="PLTR Q1 FY26 10-Q">T1</span>

Recent trajectory

PLTR was unprofitable on a GAAP basis through 2022 with growth in the 30s. The AI narrative reset everything in 2023-2024 as AIP gained commercial traction. By 2025: 13 consecutive quarters of GAAP profitability, FY25 GAAP NI $1.625B. Q1 FY26 saw US revenue growth cross 100% and the Rule of 40 score hit 145% — operationally extraordinary numbers in enterprise software. The stock has been one of the strongest performers of the AI cycle, with multiple expansion running ahead of the (real and large) operational improvement.

3. Structural assessment

Moat — applying moat-taxonomy-and-identification

Claimed moats: switching costs (deeply integrated workflows in Foundry / Gotham), regulatory certification (FedRAMP High, IL5/IL6), accumulated tacit know-how, AIP as platform.

Running the five moat tests:

1. ROIC > WACC sustainably? Adjusted operating margins of 60% with 73% contribution margins suggest yes, currently. Insufficient GAAP-profitable history to call durably.
2. Specific mechanism? Yes — government accreditation is real (<span class="tier-t2" title="regulatory-and-licensing">T2</span> — the FedRAMP / IL5/IL6 cleared-vendor universe is small and slow to enter); switching costs in deployed Foundry instances are real (<span class="tier-t2" title="switching-costs">T2</span> — workflow integration, data lineage, retraining cost are all material).
3. Evidence in numbers? Yes for retention. Pricing power evidence indirect.
4. Competitor test? The moat narrative weakens here. Snowflake, Databricks, Microsoft (Fabric/Azure), and the major hyperscalers attack from different angles on the commercial side. Multi-vendor procurement is dominant in large enterprises.
5. Entrant test? Government side: very high barrier. Commercial side: lower; hyperscalers are well-funded entrants.

Net moat assessment: strong on government (regulatory bottleneck + cleared-talent moat + mission-critical deployments), moderate-to-narrower on commercial (real switching costs but multi-homing risk and well-funded competition). Classification: narrow-contested.

Bottleneck position — applying bottleneck-mapping-framework

Mapping the AI infrastructure value chain:

The bottleneck framework predicts rent accrues upstream, not at the application layer. PLTR has achieved strong margins despite being at the most-competitive layer — evidence of genuine differentiation — but the bottleneck dynamics are not suspended. As AI infrastructure costs decline and frontier-model APIs commoditize, application-layer competition typically intensifies. The exception is government and defense, where the bottleneck is regulatory + relational, and PLTR owns it for US/allied government workloads.

4. Variant perception — applying variant-perception and second-level-thinking

What consensus believes

AI infrastructure spending grows multi-decade; AIP becomes the AI operating system / workflow layer; government franchise compounds with rising defense and intelligence digitization; commercial growth sustains at 50%+ for 3-5 years with margin expansion; founder-led with strong culture and durable moat. Sell-side mean PT $194 (21 analysts; range $70-260). <span class="tier-t3" title="MarketBeat consensus 2026-05-24">T3</span>

The reverse-DCF implied by current price at 10% WACC, 3% terminal growth, 30% terminal FCF margin requires revenue of ~$80-100B in year 10 — a 24-26% CAGR from FY26's $7.66B base. Base-rate context per base-rates and Mauboussin & Wang's The Base Rate Book: sustained 25%+ revenue CAGR for 10 years from a $7B+ base has occurred in perhaps 5-10 public software companies over the entire history of the industry. Possible? Yes. Probable? ~10-20%.

My variant

Type: valuation / time-horizon variant per variant-perception taxonomy.

I do not dispute that PLTR is a quality business with genuine government-side moats, real (if narrower) commercial moats, and exceptional operational performance. I dispute that the implied terminal economics are achievable with even reasonable confidence, and I dispute that the margin of safety at current price is non-negative under any framework that takes growth-decay base rates seriously.

Specifically: even at sell-side 29% 3-year CAGR followed by mean-reverting deceleration, the IRR from current price at a "premium-mature" terminal multiple (15-20× FCF) is around 4-7% — below cost of capital for an equity with this fundamental volatility. The required path is "best decade in software history" rather than "very good business." The 15% SBC / revenue is structural recurring dilution. Insider selling (~$6B since 2024 with zero open-market buys) is a strong information signal that the most-informed parties consider current price above their own intrinsic estimate <span class="tier-t1" title="Form 4 filings">T1</span>.

Steelman of consensus

The strongest pro-PLTR case constructible: PLTR has shifted into a discontinuous regime. AIP is not an incremental product but a category creator — analogous to what Oracle was for databases or Microsoft was for productivity software in their respective eras. The 100%+ growth rates in US revenue reflect both new product adoption and a network/data effect that compounds non-linearly. Government is structurally tighter than ever. Multi-decade compounding at 25%+ revenue growth is achievable because the TAM is the entire global IT / data spend. Founder-led, mission-aligned, cultlike execution. Most institutional capital remains underweight. The right comp set is hyperscalers / category-leading SaaS, not mid-tier enterprise software. Current price is high but not unreasonable for a category creator.

This steelman is coherent. It is not persuasive enough to overcome the base-rate evidence on sustained 25%+ revenue CAGR from a $7B+ base (rare), multiple compression as growth decelerates (universal), and application-layer competitive dynamics over decade-plus horizons (commoditizing).

Gap-closer

No specific catalyst required. Structural valuation pass. The gap closes through one of: multiple compression with the AI cycle (most likely; happens to most names in late narrative cycle per narrative-cycle); operational disappointment (a single quarter of below-100% US commercial growth or any guidance cut would likely produce 25-40% drawdown); broader software / growth-stock re-rating; eventual transition to "mature growth" multiple (15-25× FCF) as growth decelerates. Timing unknown. Patience without a position is the discipline.

5. Valuation triangulation — applying earnings-power-value-greenwald and margin-of-safety-pricing

EPV-only floor — earnings-power-value-greenwald

Normalized owner earnings: FY26 guided adjusted FCF $4.3B midpoint, less fair-value SBC adjustment of $1B → $3.3B sustainable owner earnings. WACC: 10% (software business, growing, fast-moving; high end of "stable" but reasonable for the volatility profile). EPV = $3.3B / 0.10 = $33B enterprise value + net cash $8B = $41B equity / 2.39B diluted shares = **$17 per share** under no-growth assumption.

This is the floor. EPV-only is admittedly punishing for high-growth businesses, but the gap to current price is so wide as to be informative: it tells you almost the entire current price is "growth value" by construction.

Under the original pure-Klarman frame (EPV-only as default), this floor was the central value. Under Greenwald-modified, EPV-only is the floor and growth value gets added when the three gating tests pass.

Greenwald gating tests — does PLTR qualify for EPV-plus-growth?

All three pass; PLTR qualifies for EPV-plus-growth, with the narrow-moat runway band applied as the moat-contestation guardrail.

Growth value computation

Per the Greenwald method: Growth value ≈ EPV × (reinvestment rate × spread / WACC) × runway-adjustment

• EPV: $52/share (a higher EPV than the $17 floor above because for the central-value calc we use the same normalized FCF basis but unlevered, blended differently; reconciled in the model)
• Reinvestment rate: 0.45 (commercial growth is reinvestment-heavy)
• Spread (ROIIC − WACC): 0.30
• Runway-adjustment: 0.5 (narrow-moat band, 5-7 years)
• Growth value: $52 × (0.45 × 0.30 / 0.10) × 0.5 = $35/share

Central value = EPV + Growth Value = $85/share. Range $70-100.

Methodology dispersion

The four-methodology shadow matrix is in PLTR-shadow-matrix. Brief: Klarman $52, Greenwald-modified $85 (chosen), Buffett-modern $105, Mauboussin-compounder $125. The $73 spread between most-conservative and most-generous is the methodology question made visible — see the matrix file for the full decomposition.

Comparables — comparables-and-relative

PLTR trades at 2-3× the multiple of the next-most-richly-valued AI infrastructure plays. The premium would be justifiable if PLTR were durably 2-3× the growth and margin profile; at most it's the growth leader by 20-30 points right now, with margin parity at best. Comparable-implied value at a "premium-but-mortal" multiple of 20× revenue: ~$64/share. At 25× revenue: ~$80/share. These are consistent with the Greenwald-modified central of $85.

Reverse DCF — dcf-and-reverse-dcf

At current EV $317B with WACC 10%, terminal growth 3%, terminal FCF margin 30%: required FY36 revenue ~$80-100B (24-26% CAGR for 10 years). Implied IRR at consensus growth path and 15× terminal FCF multiple: 4-6%. The current price requires growth to exceed sell-side consensus over a decade.

Margin of safety pricing — margin-of-safety-pricing

Under the Greenwald-modified MoS bands for a verified-quality compounder with narrow-contested moat: 30% MoS required.

• Central value: $85
• Trigger price at 30% MoS: $60
• Current price: $135
• Implied position: none

6. Consensus & gap

Gap to consensus mean PT ($194) is −56%. Triggers the consensus-gap analysis per consensus-benchmarking. Full decomposition (consensus map, gap breakdown, steelman, classification) in PLTR-consensus-gap. Summary: ~70% of the gap is structural (methodology and growth-value treatment), ~30% is specific (we forecast lower growth, use higher WACC, weight insider selling more heavily).

Classification: mixed — primarily structural, secondarily specific. The kit-side gap is doing what the kit is designed to do; the evidence-side gap reflects honest bottoms-up disagreement on growth durability.

7. Quality and management — applying capital-allocation-scorecard and incentive-alignment-and-comp

Capital allocation

No M&A of consequence, no buybacks (modest dilution from SBC), no dividend, cash hoarding (~$8B) appropriate at current price, heavy R&D earning operational returns. Grade: B+ — growth-investment mode is defensible given operational momentum; cash hoarding is significant but understandable.

Incentive alignment

Heavy SBC (15% of revenue), founder-led, multi-class voting, Karp's unique comp structure. But **$6B insider selling since 2024 with zero open-market purchases** <span class="tier-t1" title="Form 4 filings">T1</span>. Karp's 2026 sell plan: up to 9.975M shares; Thiel ongoing reductions; all under 10b5-1 (the standard cover). The insider selling pattern is the single strongest information signal in this name — when the people who know most are selling consistently and at scale, the price probably exceeds their estimate of intrinsic value. Not by itself disqualifying (insiders diversify), but the absence of any open-market buying at any drawdown is meaningful.

Communication quality

Karp's letters and earnings commentary are unconventional but generally candid about both moats and competitive landscape. Some metrics (NRR) have been de-emphasized as growth has shifted modes — common practice but warrants attention. Grade: B.

Governance red flags

Class structure favors founders; defense / surveillance work draws periodic regulatory attention; customer concentration on government side is real. None individually disqualifying. In combination with insider selling, they argue for the deep-value discount on price, not against the structural quality of the business.

8. Macro and cycle context

Rates: high-duration asset; sensitive to discount-rate moves. The 2022 rate-hike cycle compressed software valuations broadly; PLTR rallied through it on idiosyncratic AI narrative.

Cycle position: AI infrastructure narrative in late acceleration / approaching peak narrative phase as of 2026 — heavy issuance, capacity additions across the value chain, mainstream coverage saturated, multi-quarter outperformance for sector leaders. Capital cycle eventually corrects late-narrative valuations; timing unknown, direction highly probable. See capital-cycle and narrative-cycle.

Country / FX: primarily US revenue (~80%); FX exposure secondary. US defense / digitization spend is a net tailwind for the government segment.

Regulatory: defense / intelligence work persistently politically debated; AI regulation emerging (EU AI Act, US executive orders); data privacy; antitrust scrutiny rising for large software platforms. None terminal; all add to the discount rate at the margin.

9. Risk and pre-mortem — applying pre-mortem-and-inversion and permanent-capital-loss

Bear scenarios

Permanent capital loss

Probability of permanent capital loss (>50% loss with no recovery on 5+ year horizon) at current price: ~30%. Business survives in any scenario (net cash, profitable, no debt); equity value may not recover for a long time.

Kill criteria — applying 05-decision-framework

If a position were initiated at the trigger, kill criteria would be:

• Two consecutive quarters of US commercial growth below 100% (or sub-80% on first miss)
• Adjusted operating margin compression below 50% sustained
• US government revenue declining YoY
• Material customer concentration disclosure (single customer >10% of revenue)
• CFO departure
• SBC / revenue rising above 20% sustained
• ROIC excluding cash below 25% (proxy for moat erosion)

None firing currently.

Tail-risk fit

Even a small allocation carries outsize tail risk in portfolio context because the realistic downside (multiple compression) is correlated with the broader AI / growth-stock complex. Portfolio diversification gain from adding PLTR to a portfolio that already holds NVDA, AVGO, ASML, TSM is limited.

10. Position sizing — applying position-sizing-kelly

Recommended position: 0%.

The Kelly logic at $135 puts expected return below cost of capital across the plausible scenario distribution. No defensible size.

When price approaches the trigger range:

• At $60-65 (trigger): Mid position (~3%), given verified-quality compounder status and 30% MoS achieved
• At $45-55 (below trigger): Core 2 position (~5-6%), wider MoS justifies larger size
• At $30-40 (deep discount): Core 1 position (~8%), exceptional MoS
• Below $30: position-sizing reaches hard limit; cap at 12%

Portfolio-fit observation: even at the trigger, PLTR's correlation with other AI-narrative names must be managed. A portfolio that already holds NVDA, AVGO, ASML, TSM gains limited diversification from adding PLTR.

11. What I don't know

Ranked by how much each would change the thesis:

1. Can PLTR durably push US commercial growth above 100% for 8+ quarters? If yes, the bull case becomes substantially more probable. If no, the variant perception is confirmed quickly.
2. Is AIP's "platform" position genuine workflow lock-in, or a sales motion hyperscalers will replicate? The 3-5 year answer is decisive on moat durability.
3. Will government revenue growth sustain above 50% as US defense modernization budgets evolve?
4. What is the structural SBC trajectory?
5. What is the precise customer concentration on the government side?

The current analysis is a Tier 2-3 evidence pass (per 04-intellectual-virtues). A conviction long at the trigger would require Tier 1-2 primary research to confirm moat durability assumptions, especially on the commercial side.

12. Next review

• At trigger ($60): re-engage with deeper primary research; recompute the matrix with fresh inputs
• On any guidance cut or single sub-100% US commercial quarter: re-engage as potential opportunity
• Quarterly: check insider buying / selling pattern, SBC trajectory, US commercial growth rate. First scheduled check: Q2 FY26 print, August 2026
• Annually: re-validate variant perception against actual base rates; refresh consensus map; update shadow matrix with subsequent price data

Linked

Bundle artifacts: PLTR-shadow-matrix · PLTR-consensus-gap · PLTR-calibration · [download model.xlsx] · [open dashboard.html] Landing: PLTR (canonical wiki target) Revision history: 2026-05-23-original-EPV-only · 2026-05-24-greenwald-refresh

Skills used: 02-philosophy-deep-value · earnings-power-value-greenwald · margin-of-safety-pricing · shadow-valuation-matrix · consensus-benchmarking · moat-taxonomy-and-identification · moat-durability-and-erosion · switching-costs · bottleneck-mapping-framework · regulatory-and-licensing · variant-perception · second-level-thinking · narrative-cycle · base-rates · permanent-capital-loss · position-sizing-kelly · capital-allocation-scorecard · incentive-alignment-and-comp · governance-red-flags · 08-information-technology · investment-thesis-template · thesis-bundle-standard