UiPath (PATH) — Investment Tree v1
🟢 PRICE RE-ANCHOR (2026-05-30): scaffold anchored $14.00; live price $11.72 (−16%; 52-wk $9.2–19.84). The stock is now at the tree's own "~$12-14 melting-ice-cube floor" (FCF-yield + buyback-accretion support) — so the Watch-only verdict holds (the agentic-pivot thesis is still unquantified; at the floor the downside is more bounded but ownership still requires NRR stabilization or Maestro ARR proof). Primary cite: FY2026 10-K (accn 0001734722-26-000012, period 2026-01-31); PATH evidence is already Tier-A-heavy. (Down-move to the modeled floor = safe direction; light re-anchor.)
Date: 2026-05-25
0. Company Fundamentals — what UiPath is and how it earns
Figures FY2026 (ended Jan 31, 2026) unless noted.
What it is & how it earns. UiPath (PATH) is the leading robotic-process-automation (RPA) software vendor, now repositioning itself as the "agentic automation" layer — orchestrating AI agents, software robots and human workers under one governance stack (Maestro + AI Trust Layer). It sells on a subscription/ARR model: FY2026 revenue was $1,610.6M (+13%) against ARR of $1,852.6M (+11%), with dollar-based net retention of 107% (down from 110%) and gross margin a steady ~83%. The central question the tree resolves is whether agentic AI is a tailwind that re-accelerates ARR or a disruptor that erodes the RPA core — agentic ARR reached ~$200M (≈10.5% of ARR) by the Q1 FY2027 print.
Cash-flow anatomy. UiPath is capital-light, so adjusted FCF tracks operating cash flow closely and has been strongly positive and growing:
| FY2024 | FY2025 | FY2026 | |
|---|---|---|---|
| Operating cash flow | $299.1M | $320.6M | $371.2M |
| Capex | $7.3M | $14.9M | $19.0M |
| Free cash flow (adj.) | ~$292M | ~$306M | ~$352M |
| FCF margin (FCF/revenue) | ~22.3% | ~21.4% | ~21.9% |
Strong ~22% FCF margins, but stock-based comp ran $290.7M — ~18.1% of revenue (down from 25.1%) — the key earnings-quality caveat: a large share of "profit" is funded by dilution, not cash cost.
Balance sheet & capital allocation. UiPath holds ~$1.69B in cash, equivalents and marketable securities (of which ~$870M is cash/equivalents) against $0 debt — all net cash. It pays no dividend. Capital return runs entirely through an active buyback ($329M in FY2026, $391M in FY2025; new $500M authorization March 2026), which is partly an SBC offset rather than pure per-share accretion.
What drives it. The thesis turns on ARR/net-retention durability and whether the agentic-AI transition is incremental revenue versus an RPA rebrand, weighed against a modest ~3.4–3.8× EV/ARR multiple and ongoing SBC dilution. That single question — real new revenue layer or repackaged renewal — is what the tree exists to resolve.
Section I — The Central Question
Is UiPath's "agentic orchestration" platform a genuine architectural necessity for enterprise AI deployment — commanding a durable and growing share of enterprise automation budgets — or is it a rebranding exercise on top of a legacy RPA installed base that will gradually erode as AI-native tools absorb its use cases?
Stories lie, structure doesn't. The market has a story: "RPA is dead; AI agents replace workflows." The structure says something more nuanced. UiPath's $1.85B ARR is growing, not shrinking. Its 357 largest enterprise accounts are deepening their commitment (+13% YoY), not defecting — despite Microsoft actively bundling competing tools at near-zero cost. The financial architecture is clean: 83% gross margins, $352M FCF, no debt, $870M cash. None of this looks like a dying business.
The honest uncertainty: the agentic platform pivot (Maestro) has generated no quantifiable incremental ARR. Every dollar of that $1.85B ARR could be RPA renewal; none may be new-paradigm agentic orchestration. Without quantification, the bull thesis is a claim, not a fact. The verdict is: coherent, worth watching, not yet worth owning.
Section II — Company Snapshot
UiPath Inc. (PATH) is the global leader in robotic process automation (RPA) repositioning itself as the operating layer for the agentic enterprise. Founded in Bucharest, Romania (2005); IPO April 2021 (NYSE: PATH); co-founder Daniel Dines returned as CEO October 2023.
FY2026 (year ended Jan 31, 2026) — Primary source: 10-K filed 2026-03-25:
| Metric | Value |
|---|---|
| ARR | $1,852.6M (+11% YoY) |
| Revenue | $1,610.6M (+13% YoY) |
| Gross margin | 83% |
| FCF | ~$352M |
| Cash / debt | ~$870M / $0 |
| NRR | 107% (↓ from 110%) |
| ≥$1M ARR customers | 357 (+13% YoY) |
| SBC | $290.7M (18.1% of rev, ↓ from 25.1%) |
| S&M ratio | 42% (↓ from 52%) |
Revenue mix: Licenses 38% ($606M, +3%) / Subscription services 59% ($954M, +19%) / Professional services 3% ($50M, +23%)
Fiscal year: February 1 – January 31. FY2027 = Feb 2026 – Jan 2027.
Section III — The H-0 Thesis
PATH is priced as a declining RPA incumbent being disrupted by AI agents; reality is that UiPath is the only enterprise-grade orchestration layer capable of governing heterogeneous AI agents, software robots, and human workers in production — and this positioning will generate ARR re-acceleration to 13-17% from the current 11% trough as agentic enterprise adoption crosses early-majority in FY2027-2028, producing a 40-70% re-rating from current multiples.
Why non-obvious: The market's "AI victim" classification is not wrong about the threat — it is wrong about the response. PATH is not standing still while AI agents displace its bots; it is repositioning as the orchestration and governance layer for those AI agents. The analogy: Twilio didn't die when WhatsApp commoditized SMS — it built the communication-API layer on top. PATH is building the agentic-API-governance layer on top of RPA.
Five falsifiable conditions:
- F1: ARR re-accelerates ≥13% by Q2 FY2027 (Jan 2027)
- F2: NRR holds ≥107% through Q2 FY2027
- F3: Maestro ARR ≥$100M quantified by Q3 FY2027
- F4: Gross margin holds ≥80% through FY2027
- F5: ≥$1M ARR cohort grows ≥10% through Jan 2027
Pre-Stage-3 confidence: 45-55%. Post-Stage-3 confidence: 48%. F3 not firing is the primary downward pressure on H-0 confidence.
Section IV — Mispricing Mechanism
Type: AI-victim cognitive bias suppressing orchestration infrastructure re-rating
The market has correctly identified that AI agents can replace individual RPA tasks (the "victim" thesis) but has incorrectly extrapolated this to mean that all enterprise automation governance becomes worthless. The structural error: governance, audit trails, compliance workflows, and multi-vendor AI orchestration are NOT naturally provided by AI agents — they are the enterprise-grade infrastructure that wraps AI agents.
The closest historical analog: Nokia in 2010 was "a victim of the smartphone." But Nokia's network infrastructure business (post-Infinera) was not a smartphone victim — it was structural internet backbone. The market priced Nokia as a single entity, when the victim component (mobile handsets) was separable from the structural component (network infrastructure).
PATH's mispricing follows the same pattern: the market prices PATH as a single "RPA victim" when the correct structure separates (a) RPA task automation [genuine AI agent substitution risk] from (b) enterprise orchestration governance [genuine structural necessity for enterprise AI deployment].
What would falsify the mispricing narrative: If F3 fails (no Maestro ARR by Q3 FY2027), the "orchestration infrastructure" claim is brand, not substance — and the mispricing thesis collapses.
Section V — Investment Tree (ASCII)
PATH H-0 (48%)
│
├── L1A — ARR Growth Engine ⚠️
│ ├── L1A 1.1 — NRR ≥107% holds ⚠️A [F2 test]
│ ├── L1A 1.2 — ≥$1M cohort ≥12% growth ⚠️A [F5 test]
│ └── L1A 1.3 — New-logo ARR to ≥20% ✗C [new growth missing]
│
├── L1B — Unit Economics ✅
│ ├── L1B 1.1 — GM ≥80% sustained ✅A [2yr primary source]
│ ├── L1B 1.2 — ≥$400M buybacks FY2027 ⚠️A [below pace in FY2026]
│ └── L1B 1.3 — SBC ≤15% by FY2027 ⚠️A [trend right; target tight]
│
├── L1C — Agentic Platform Thesis ✗
│ ├── L1C 1.1 — Maestro ARR quantification ✗A [LOAD-BEARING MISS]
│ ├── L1C 1.2 — Subscription ≥18% FY2027 ⚠️A [FY2026 delivered 19%]
│ └── L1C 1.3 — 3 Maestro case studies ✗C [no primary source]
│
├── L1D — Competitive Moat ⚠️
│ ├── L1D 1.1 — Microsoft cohort check ✅A [fortress holding +13%]
│ ├── L1D 1.2 — Gartner MQ leadership ⚠️C [plausible not confirmed]
│ └── L1D 1.3 — OpenAI/Anthropic moat window ✅C [governance gap real]
│
└── L1E — Management Execution ⚠️
├── L1E 1.1 — R&D ≥22% FY2027 ⚠️A [declining ratio trend]
├── L1E 1.2 — CPTO product releases ✗C [too early to assess]
└── L1E 1.3 — No third restructuring ✅C [no signals]
Tally: 4 ✅ · 7 ⚠️ · 4 ✗ · 0 ⊗Section VI — Branch Analysis
L1A — ARR Growth Engine: ⚠️ Mixed — Fortress Holding, Growth Decelerating
The core business is not melting. ARR grew +11% to $1.85B; the 357 largest enterprise customers deepened at +13%. These are not the metrics of a company losing its installed base. But NRR compression (110%→107%) signals that expansion revenue from existing customers is decelerating — the primary mechanism by which AI substitution erodes the business.
The load-bearing question for L1A: can NRR stabilize? If Microsoft Power Automate wins mid-market and M365-centric accounts (low-complexity workflows), NRR could compress toward 103-104% without touching the ≥$1M enterprise cohort. The enterprise fortress (L1D 1.1) suggests UiPath's complex enterprise accounts are not defecting. But the mid-market compression is real and NRR doesn't distinguish by customer size.
L1A verdict for H-0: Provides a floor (not collapsing) but not a catalyst (not accelerating). The thesis requires F2 (NRR stable) to hold and F1 (ARR re-acceleration) to fire — neither is confirmed yet.
L1B — Unit Economics: ✅ Financial Architecture Is Sound
Gross margin at 83% for two consecutive years is the thesis pillar that requires no caveats. The financial architecture — FCF positive ($352M), no debt, $870M cash, buybacks running at $300-400M/yr — is the floor valuation case. Even if the agentic pivot fails, PATH generates sufficient FCF to buy back 5-7% of its market cap annually at current prices. This is not a zero-sum bet.
The SBC trajectory (25%→18% of revenue) is declining toward the 15% threshold. Not there yet, but the direction is unmistakably right. Buybacks at $329M in FY2026 fell below the $390M pace of FY2025 — the new $500M authorization signals intent to accelerate, but execution must be demonstrated in FY2027.
L1B verdict for H-0: If H-0 fails completely (agentic pivot is branding), the financial architecture still supports a ~$12-14 "melting ice cube" price (FCF yield + buyback per-share accretion on a stable base). L1B is the reason PATH isn't uninvestable.
L1C — Agentic Platform Thesis: ✗ Not Proven in Revenue Terms
This is the thesis branch that matters most for the H-0 re-rating — and it has zero primary-source confirmation. Maestro exists as a product. The AI Trust Layer exists as a product. Neither has been assigned a revenue number in UiPath's filings. The 10-K describes the agentic platform in terms of product capability, not customer revenue. Without quantification, Stage 3 cannot move L1C 1.1 above ✗.
The absence is the evidence. If Maestro were generating $100M+ incremental ARR, it would appear in Key Business Metrics (alongside total ARR). It doesn't. This could mean: (a) Maestro ARR is too small to separately disclose; (b) management is choosing not to break it out yet (analogous to MSFT delaying Copilot ARR disclosure for 2 years†); or (c) Maestro IS the RPA renewal — there is no "incremental" agentic ARR, just legacy RPA with a new name. The market is pricing (c) at 41% probability. We assign 25% probability to (c). The gap is the thesis.
L1C verdict for H-0: The most critical unfired condition. H-0 confidence cannot move above 55% until F3 fires.
L1D — Competitive Moat: ⚠️ Fortress Holds; Perimeter Under Pressure
The enterprise fortress (≥$1M ARR cohort) is holding at +13% while Microsoft is actively bundling. This is the most important single data point in Stage 3. It means: (1) the switching cost moat is real for complex enterprise deployments; (2) Microsoft's bundling has not materially eroded the highest-value accounts in FY2026.
What this doesn't tell us: what happens in FY2027-2028 as Microsoft's Copilot Studio matures, as AI agent task coverage expands to more complex workflows, and as enterprise IT budget cycles allow for tool consolidation. The moat is intact today — it is not expanding.
L1D verdict for H-0: L1D 1.1 ✅A is the strongest single positive finding. The enterprise fortress is real. But moat is not widening — it requires sustained L1C (Maestro creating new reasons to stay) to avoid slow erosion.
L1E — Management Execution: ⚠️ Improving Capital Discipline; Platform Execution Unproven
Daniel Dines returning as CEO + two completed restructurings + CPTO appointment from Microsoft = organizational intent to execute the platform pivot. The capital allocation improvement (S&M -10pp, SBC -7pp, GAAP profitable for first time) demonstrates that the previous over-investment in growth has been corrected. The floor valuation case works now.
But the platform pivot execution is entirely unproven in product terms. Malpani (CPTO, ex-Microsoft) was appointed March 2026 — 2 months ago from this assessment. His product impact will first be visible at UiPath Forward 2026 (H2 2026†). The CPTO appointment is a positive signal; the product delivery is a future state.
L1E verdict for H-0: Management is not a constraint on the bull thesis. The organizational correction is real. The execution gap (no Maestro product proof) remains.
Section VII — Scenario Analysis
Full detail in reports/PATH/scenarios.md
| Scenario | Probability | 12-mo Target | Key trigger |
|---|---|---|---|
| Bull — Agentic Re-rating | 30% | $22-25† | F1+F2+F3 all fire by Q2 FY2027 |
| Base — Stable Orbit | 45% | $14-17† | ARR 9-12%, NRR holds, no Maestro quantification |
| Bear — Melting Ice Cube | 25% | $8-10† | NRR < 105%, Microsoft wins enterprise accounts |
Probability-weighted E[r]: +16.3% Asymmetry: 2.3x favorable (bull contribution 20.4% vs bear contribution 8.9%)
Market-implied bull probability: ~14% (vs our 30%). The gap is the thesis — we see the enterprise fortress (L1D 1.1 ✅A) as stronger evidence than the market gives credit.
Section VIII — Valuation Context
All multiples Tier C†. Live quote required before any sizing decision.
Current pricing (at $14†/share):
- EV/ARR: ~3.4-3.8x on $1.85B ARR
- EV/FCF: ~18-20x on $352M FCF
- Market cap: ~$7.2B†
Peer benchmarks:
- Workiva (WK, stable SaaS, 3-4x EV/ARR†): PATH at "stable decay" price
- PTC Inc. (successful pivot, 7-9x EV/ARR†): PATH re-rating target if pivot proven
- ServiceNow (enterprise platform leader, 12-14x EV/ARR†): ceiling only if PATH achieves NRR parity
Re-rating math (bull case):
- ARR grows to ~$2.1B by Jan 2027 at 14% growth
- Multiple expands from 3.8x to 6.5x EV/ARR (PTC-comparable on proven pivot)
- EV = $2.1B × 6.5x = $13.65B; add $870M cash; divide by ~510M shares = ~$28†
- Conservative target = $22-25† (accounting for re-rating lag)
Floor valuation (bear case, FCF-floor):
- FCF ~$340M in FY2027 (modest FCF growth)
- SaaS software peers at FCF-contraction trade 12-15x FCF = ~$4B-5B EV
- Add cash $870M = $4.9-5.9B market cap / 510M shares = $9.6-11.6†
- Target: $8-10† (allowing for multiple compression in bear narrative)
Section IX — Risk Taxonomy (per K.4)
Primary risks for PATH:
R1 — Platform disruption risk (K.4 Technology displacement): AI agents replace RPA workflows natively. This is the bear scenario mechanism. Probability: 25% (our estimate). Most observable via NRR trajectory.
R2 — Competitive bundling risk (K.4 Competitive/market-share): Microsoft subsidizes Power Automate / Copilot Studio below cost to defend M365 ecosystem. Cannot compete on price against a deep-pocket bundler. Observable via ≥$1M ARR cohort retention.
R3 — Execution risk (K.4 Management/execution): Maestro pivot generates no incremental ARR — management over-promises and under-delivers (consistent with 2023-2024 guidance miss history). Observable via L1C 1.1 (Maestro ARR disclosure).
R4 — Governance concentration risk (K.4 Governance/ownership): Daniel Dines holds majority voting control via dual-class structure†. M&A optionality is Dines-dependent. This limits Option 2 (acquisition target) probability.
R5 — Evidence quality risk (K.4 Information asymmetry): Anchor price ($14†) and several valuation multiples are Tier C. If live price is materially different (e.g., $18-20 reflecting recent recovery), the asymmetry calculation changes. R2 verification required before any position decision.
Section X — Forward-Looking Catalyst Calendar
| Date | Event | Thesis relevance |
|---|---|---|
| ~Sep 2026 | Q1 FY2027 earnings | NRR (T1/RF1) — MOST IMPORTANT |
| H2 2026† | UiPath Forward 2026 | Maestro case studies (T2 first signal) |
| ~Dec 2026 | Q2 FY2027 earnings | ARR re-acceleration (T3) + Maestro ARR (T2) |
| Q4 2026† | Gartner MQ 2026 | Competitive positioning (L1D 1.2) |
| ~Mar 2027 | Q3 FY2027 earnings | F3 window closes (Maestro ≥$100M final deadline) |
Section XI — Long-Term Holdability Assessment
Durability: 17/25 (Medium). 0 fatal flags.
Per the Stage 4d durability test:
- Q1 Business model (3/5): Real 5-yr durability; 10-yr depends on Maestro
- Q2 Moat trajectory (3/5): Holding at top end; narrowing at edges
- Q3 Capital allocation (3/5): ROIC > WACC confirmed; improving
- Q4 Disruption survival (2/5): Material AI disruption risk; not fatal yet
- Q5 Reinvestment runway (3/5): Conditional on agentic TAM
- Q6 Optionality (3/5): M&A option, Maestro-as-standalone, AI governance mandate
Conclusion: PATH is NOT a primary compounder for a 5-10 year hold at current evidence quality. It is a conditional hold — investable IF F2+F3 fire by Q3 FY2027. Without those catalysts firing, the medium durability (17/25) and high disruption risk (Q4 = 2/5) make it unsuitable as a 5-10 year compounder.
Section XII — Investment Scorecard
Format B (analytical-tree scorecard). 15 questions. Per K.3.5, K.3.6.
The 15 Questions
| # | Question | Verdict | Tier | Notes |
|---|---|---|---|---|
| Q1 | Does the business model make sense in 5-10 years? | ⚠️ | A | Conditional on Maestro pivot |
| Q2 | Why is this a good time to look at it? | ✅ | C | Post-restructuring efficiency + valuation trough |
| Q3 | What is the strongest bull case? | ✅ | C | Maestro = agentic orchestration infrastructure |
| Q4 | What is the most credible bear case? | ✅ | A | NRR compression + Microsoft bundling (factual) |
| Q5 | Is the valuation creating a margin of safety? | ⚠️ | C | 3.8x EV/ARR modest cushion; not screaming cheap |
| Q6 | Will revenue grow reliably for 5 years? | ⚠️ | A | ARR growing 11%; re-acceleration unconfirmed |
| Q7 | Are profits and margins durable? | ✅ | A | 83% GM confirmed stable; FCF $352M growing |
| Q8 | Is the balance sheet clean? | ✅ | A | $870M cash, $0 debt, FCF positive |
| Q9 | Is the debt level manageable? | ✅ | A | No debt; cash balance exceeds short-term needs |
| Q10 | Who are the competitors and do they threaten the moat? | ⚠️ | A | Microsoft is existential threat; currently not winning enterprise cohort |
| Q11 | What would trigger an exit? | ✅ | A | RF1/RF5/RF2 defined; clear exit thresholds |
| Q12 | Can the thesis be falsified? | ✅ | A | F1-F5 are falsifiable; NRR is measurable quarterly |
| Q13 | Is this relevant in a 10-year context? | ⚠️ | C | Uncertain; depends on whether agentic pivot succeeds |
| Q14 | Does the moat trajectory point up, flat, or down? | ⚠️ | A | Flat-to-narrowing; fortress holds but perimeter compresses |
| Q15 | Has management deployed capital at ROIC > WACC? | ✅ | A | FCF-ROIC ~35%†; buybacks sensible; no capital destruction |
K.3.5 Weighted-Score Derivation
| Q | Description | Tier (K.3.5) | Weight | Verdict | Weighted |
|---|---|---|---|---|---|
| Q12 | Falsifiability (fatal flags 0, thesis testable) | Critical | 5 | ✅ (1.0) | 5.0 |
| Q8 | Balance sheet clean | Critical | 5 | ✅ (1.0) | 5.0 |
| Q14 | Moat trajectory | Critical | 5 | ⚠️ (0.5) | 2.5 |
| Q4 | Bear case credibility | Load-bearing | 3 | ✅ (1.0) | 3.0 |
| Q5 | Valuation / entry asymmetry | Load-bearing | 3 | ⚠️ (0.5) | 1.5 |
| Q10 | Competitive position | Load-bearing | 3 | ⚠️ (0.5) | 1.5 |
| Q15 | ROIC > WACC (capital allocation) | Load-bearing | 3 | ✅ (1.0) | 3.0 |
| Q3 | Bull case strength | Important | 2 | ✅ (1.0) | 2.0 |
| Q6 | Revenue growth reliability | Important | 2 | ⚠️ (0.5) | 1.0 |
| Q7 | Margins and profits | Important | 2 | ✅ (1.0) | 2.0 |
| Q13 | 10-year relevance | Important | 2 | ⚠️ (0.5) | 1.0 |
| Q1 | Business model persistence | Confirming | 1 | ⚠️ (0.5) | 0.5 |
| Q2 | Timing rationale | Confirming | 1 | ✅ (1.0) | 1.0 |
| Q9 | Debt level | Confirming | 1 | ✅ (1.0) | 1.0 |
| Q11 | Exit trigger clarity | Confirming | 1 | ✅ (1.0) | 1.0 |
| Total | 39 max | 21.0 / 39 = 53.8% |
K.3.5 xii_score: 54% — "Wait or skip (weak)" band (≥45% to <65%)
Bands per K.3.5: ≥85% high-conviction (strong) / ≥65% moderate buy with sizing (partial) / ≥45% wait or skip (weak) / <45% avoid (weak).
PATH at 54% sits in the "wait" band — not avoid (the financial architecture is real), not "buy" (the thesis is unproven).
Scorecard Summary
Bull/Base/Bear implied probability: 30 / 45 / 25
Prob-weighted expected return at $14†: +16.3%
Asymmetry ratio: 2.3x favorable (bull contribution 20.4% / bear contribution 8.9%)
Durability: 17/25 Medium. 0 fatal flags.
H-0 confidence: 48% (below 50% threshold for conviction buying)
Final Verdict: Watch-only / Avoid compounder framing
Position sizing: 0% now → 1% on T1 confirmation (NRR ≥107% at Q1 FY2027) → 1.5% on T1+T2 (Maestro ARR ≥$50M) → 2% max on T1+T2+T3 (ARR ≥13%).
Never above 2%: durability 17/25 Medium caps K.3.2 at 1-3%, and the incumbent-under-threat archetype with Q4 disruption risk 2/5 warrants the lower end of that range.
Do NOT size as a compounder. PATH is a catalyst play, not a decade hold.
2-Minute Pitch
PATH is priced as a victim. The enterprise data says it is a survivor, possibly a re-rater. The most complex enterprise accounts are deepening their UiPath commitment at +13% while Microsoft bundles for free — this is the market's biggest analytical blind spot.
The re-rating requires one proof: management quantifying Maestro/agentic ARR. Everything else (financial floor, enterprise fortress, no debt, FCF positive) is already in the price. The single unresolved question — is Maestro a real new revenue layer or a rebrand? — is what separates $14 from $22.
Watch the September 2026 NRR print and the December 2026 earnings call for Maestro quantification. If both fire, this is a 1.5-2% selective position. If either misses, it is a clean exit.
Risk Types (K.4)
- Primary: Technology displacement (AI agent substitution of RPA workflows)
- Primary: Competitive/market-share (Microsoft bundling below cost)
- Secondary: Management/execution (Maestro pivot failure)
- Secondary: Information asymmetry (anchor price Tier C, Maestro ARR absent)
- Tertiary: Governance/ownership (Dines dual-class; M&A optionality limited)
When NOT to Buy (K.5 Anti-patterns)
❌ Do NOT buy on "cheap" alone — 3.8x EV/ARR is cheap relative to history but not relative to risk. A melting ice cube can stay cheap for 3+ years.
❌ Do NOT buy expecting a Microsoft acquisition — Dines dual-class voting control makes unwanted M&A nearly impossible.
❌ Do NOT buy as a "AI winners" play — PATH is ai-capex-low exposure; it benefits from enterprise AI adoption, not from the hyperscaler GPU capex cycle. It is not NVDA/TSM/AVGO.
❌ Do NOT buy before verifying the anchor price live — $14† is Tier C. If live price is $18+, the E[r] and asymmetry calculation changes materially; this scorecard was built at $14†.
❌ Do NOT size to 3% without F1+F2+F3 all firing — durability 17/25 + Q4 disruption 2/5 cap the position at 2% maximum under K.3.1/K.3.2 rules.