ServiceNow, Inc. (NOW) — Investment Tree v1

Stage 7 final essay. Bilingual companion: tree_v1_zh.md. Date: 2026-06-08 · Anchor price: $112.45 (2026-06-07 close, split-adjusted) · Market cap: ~$116B (~1.0B shares) · Forward P/E: ~27-28× FY2026E · EV/FCF: ~26.5× · Rule of 40: ~52-56 · Dividend: none Archetype: Best-in-class enterprise-workflow compounder de-rated ~40-50% on an agentic-AI fear its own numbers (50% non-seat ACV, NRR 120-125%, Now Assist ACV doubling) are actively rebutting

SOURCE QUALITY: FY2025 financials Tier-A (10-K accn 0001373715-26-000007; Q4/FY2025 press release accn 0001373715-26-000005). Q1 2026 Tier-A (8-K accn 0001373715-26-000054 + 10-Q). Price/multiples/consensus Tier-B (2026-06-07). AI ACV / Investor Day targets Tier-B (company IR + Investor Day coverage). All per-share figures are SPLIT-ADJUSTED for the 5-for-1 forward split effective 2025-12-18. Forward scenario targets are analytical (†). See evidence_2026-06-08.jsonl.

R2-VERIFIED (2026-06-08): The load-bearing numbers were re-checked against primary / IR sources (the original build's WebFetch was 403-blocked, forcing Tier-B/C triangulation). CONFIRMED: FY2025 subscription ~$12.8B (+~20%), cRPO +21%, FCF margin ~32%; Now Assist ACV >$600M in 2025 (Q4 net-new ACV >2× YoY, 35 deals >$1M; $1B 2026 target); Investor-Day $30B+ (up to $32B) 2030 subscription target with AI ~30% (~$9B). NEW since build: Q4 FY2025 board authorized an additional $5B buyback. No material deviation — the Hold-with-sizing / initiate-3-5% thesis is intact. See evidence_2026-06-08.jsonl (NOW-R2-2026-06-08-014).

0. Company Fundamentals — what ServiceNow is and how it earns

Figures FY2025 (calendar 2025) unless noted.

What it is & how it earns. ServiceNow is the dominant enterprise-workflow SaaS company: its single Now Platform runs IT (ITSM/ITOM), employee/HR, customer-service and increasingly CRM and security workflows on one data model for ~85% of large enterprises. It earns almost entirely through multi-year subscriptions — FY2025 total revenue ~$13.3B of which subscription was ~$12.88B (~97%) — at a ~98-99% renewal rate and 120-125% net revenue retention. The newer monetization angle is agentic AI: customers buy "Now Assist" AI agents on consumption pricing (Now Assist ACV >$600M and doubling), so ~50% of net-new ACV is now non-seat-based.

Cash-flow anatomy. ServiceNow is a high-margin, low-capital subscription engine — free cash flow runs near a third of revenue and is expanding:

A ~32% FCF margin on ~21% growth reflects the recurring, prepaid subscription model; FY2026 is guided to a 35% FCF margin. (*FY2025 OCF/capex are estimated from the reported FCF margin pending the 10-K cash-flow statement; the existing tree body's "FY2025 OCF $4.267B" is in fact the FY2024 figure.)

Balance sheet & capital allocation. ServiceNow held ~$10.1B of cash and marketable securities at year-end 2025 against only ~$1.5B of legacy notes — a strong net-cash position (~$8.6B). It pays no dividend and returns capital via buybacks (an additional $5B authorized Oct 2025) that mainly offset stock-based comp. R&D was ~$2.96B (~22% of revenue). To fund the ~$7.75B all-cash Armis acquisition (closing H2 2026), NOW issued its first material debt — $4.0B of senior notes in 2026, comfortably covered by FCF but ending the pristine net-cash era.

What drives it. The bottom line turns on whether seat-plus-AI-agent expansion (cRPO ~+21% cc) keeps compounding revenue ~20% against a premium ~27-28× forward multiple. The tree resolves whether agentic AI is a tailwind (consumption offsetting seat compression) or a threat.

I. One-sentence verdict

ServiceNow at $112.45 is a best-in-class enterprise-workflow compounder — ~21% subscription growth, free-cash-flow margin expanding to 35%, Rule of ~56, 98-99% renewal, 120-125% net revenue retention — that has been de-rated ~40-50% to a near-decade-low multiple (~27-28× forward, ~26.5× EV/FCF) on a category-level "agentic-AI kills per-seat SaaS" fear, while ServiceNow's own firm-specific numbers (~50% of net-new ACV is already non-seat-based, Now Assist AI ACV >$600M and doubling, a credible $30B/2030 plan with ~$9B from AI) indicate it is far more likely an AI beneficiary than a victim; with the Street at a Strong Buy / $141.86 target (+26% above spot), the analyst scenario set produces +29% expected value and a 2.7× FAVORABLE asymmetry on a 0-fatal-flag, 21/25-durability business — making this a Hold-with-sizing → initiate/build a 3-5% core long-term position, scaling toward the top of the band as the agentic-AI evidence keeps resolving toward "beneficiary" (non-seat ACV ≥50%, NRR ≥120%), with the residual risks being the genuine-but-unresolved 10-year disruption tail and the unproven $10.6B Moveworks+Armis M&A step-change (first material debt).

II. Company snapshot

ServiceNow is the dominant enterprise-workflow SaaS company (Santa Clara, CA; NYSE: NOW; CIK 0001373715). Its single product, the Now Platform, runs on one data model and spans IT service management (ITSM — the founding category), IT operations (ITOM), employee/HR workflows, customer service (CSM), and increasingly CRM (Cport) and security operations — making it the cross-departmental "system of action" for the enterprise. ~97% of revenue is recurring subscription. ~85% of large enterprises already run on it; renewal is ~98-99% and net revenue retention ~120-125%.

FY2025 (Tier-A): total revenue $13.3B (+21%); subscription revenue $12.883B (+21%); GAAP net income ~$1.75B (+23%); GAAP diluted EPS $1.69 (split-adjusted; FY2024 $1.39); non-GAAP operating margin 30.5%; operating cash flow $4.267B (+39%); FCF margin ~32% (≈$4.25B FCF). RPO $28.2B (+26.5%); cRPO $12.85B (+25% / +21% cc) — the backlog is growing faster than revenue. Q1 2026: subscription $3.671B (+19% cc); cRPO $12.64B (+21% cc); non-GAAP op margin 32% (beat); FCF margin 44%; EPS $0.97 (met). FY2026 guide (raised): subscription $15.735-15.775B (+20.5-21%); FCF margin 35%.

The stock trades at $112.45 (2026-06-07, split-adjusted; ~$116B market cap) — down ~30% over six months, ~40% YTD, ~47% off its 52-week high of $211.48. The dominant market view is that agentic AI will invert ServiceNow's per-seat economics and that ~20% growth justifies only a compressed multiple. The opposing view — supported by this analysis — is that ServiceNow is the orchestration layer agents run on, is already 50% non-seat-based, is inflecting a new AI revenue engine (Now Assist), and has been over-sold on a sector-wide fear it is uniquely positioned to absorb.

The H-0 thesis names what the market is mispricing: a best-in-class compounder has been priced as a mature, decelerating, soon-to-be-disrupted seat business — but its non-seat ACV transition, NRR, and Now Assist ramp say it is the AI winner, not the AI casualty.

III. The Tree

H-0: NOW de-rated ~40-50% on an agentic-AI-disruption fear its own numbers rebut;
     ~21% growth + FCF margin to 35% + Rule of 56 + 98-99% renewal at a decade-low multiple;
     EV +29%, asymmetry 2.7× FAVORABLE; 5-10yr hold suitable as a 3-5% core compounder position
│
├── L1A — Growth Durability  ✅A supported; 20%+ growth + backlog ahead of revenue
│   ├── 1.1 FY2026 subscription guide raised +20.5-21%                    ✅A strongly supported
│   ├── 1.2 cRPO +21% cc (Q1'26); +19.5% cc Q2 guide                      ✅A strongly supported
│   └── 1.3 $30B/2030 plan (~17-19% implied CAGR) internally consistent   ⚠️B plausible, multi-year
│
├── L1B — AI Monetization  ✅B supported; Now Assist inflecting, non-seat ACV the rebuttal
│   ├── 1.1 Now Assist ACV >$600M doubling; $1.5B 2026 target            ✅B supported
│   ├── 1.2 ~50% of net-new ACV non-seat-based                           ✅B supported (key rebuttal)
│   └── 1.3 $9B / 30%-of-sub AI by 2030                                  ⚠️B load-bearing assumption
│
├── L1C — Disruption Survival  ⚠️A near-term ✅; 10-year tail genuinely open
│   ├── 1.1 No seat-collapse in data (NRR 120-125%, renewal 98-99%)      ✅A strongly supported
│   ├── 1.2 "Control tower" orchestration vs hyperscaler + peer agents   ⚠️B contested
│   └── 1.3 No marquee model-native replacement of NOW observed          ⚠️B absence-of-evidence
│
├── L1D — Moat + Capital Allocation  ⚠️A moat ✅A; capital ⚠️ on M&A step-change
│   ├── 1.1 Switching-cost moat intact (98-99% renewal, 2-4yr replace)   ✅A strongly supported
│   ├── 1.2 M&A/debt discipline — Armis $7.75B cash + $4B notes + DOJ    ⚠️A unproven (WATCH)
│   └── 1.3 Armis deal outcome (not yet closed)                          ⊗ not testable yet
│
└── L1E — Valuation + Financial Quality  ✅B valuation supportive; ⚠️ SBC
    ├── 1.1 Forward ~27-28× / EV/FCF ~26.5× = decade-low; Street PT +26% ✅B supportive
    ├── 1.2 FCF margin expanding 32%→35%                                 ✅A strongly supported
    └── 1.3 SBC keeps GAAP net margin ~13% vs non-GAAP op 30.5%          ⚠️A quality caveat

Total: 8 ✅ / 5 ⚠️ / 0 ✗ / 1 ⊗ across 14 leaves
H-0 verdict: SUPPORTED, ~64% confidence — beneficiary evidence dominates; residual risk is the
L1C 10-year tail + L1D M&A step-change (both ⚠️, not ✗); see §V/§XI

IV. Key findings

Finding 1 — Every operating metric beat or held while the stock fell ~40-50%

This is the cleanest mispricing signal in the whole dossier. Through the ~40-50% drawdown, ServiceNow raised FY2026 subscription guidance (+20.5-21%), grew cRPO +21% cc, beat on operating margin (32% vs 31.5% guide), and guided FCF margin UP to 35%. The decline is pure multiple compression — a sentiment de-rating, not a fundamentals break. The Q1 2026 print (beat every metric, stock still fell ~17%) is the textbook case: the market is pricing a narrative, not the numbers.

Finding 2 — The single fact that breaks the bear case: ~50% of net-new ACV is non-seat-based

The entire "SaaS-pocalypse" thesis rests on per-seat economics: fewer human seats → less recurring revenue. But ServiceNow is already ~50% non-seat-based on net-new ACV (consumption/usage/hybrid pricing), and the agentic-AI shift drives more transaction volume (agent-actions), not less. Combined with NRR of 120-125% and renewal of 98-99%, there is no seat-collapse in the data. The bear case is a forecast; the beneficiary case is a measurement.

Finding 3 — Now Assist is a real, inflecting AI revenue engine

Now Assist ACV has surpassed $600M and is more than doubling YoY; customers with >$1M Now Assist ACV are up >130%; the 2026 AI ACV target is ~$1.5B; and the Investor Day $30B/2030 plan assumes ~$9B (~30%) from AI. ServiceNow sells AI into a captive installed base that already trusts it with mission-critical workflows — structurally the cheapest AI distribution in enterprise software. This is the asymmetric upside: the market prices AI as a threat, so any confirmation it is a tailwind re-rates both the numerator (growth) and the multiple.

Finding 4 — The moat is the orchestration layer, and it is widening

Switching costs are extreme: 12-18 month implementations, 2-4 year rip-and-replace costing tens of millions, ~85% large-enterprise penetration. The strategic bull point is that agents need a governed system of action to operate within — permissions, audit, a single data model, integrations to every other system. ServiceNow owns that layer. On a 5-year view the moat is durable-to-widening (NRR 120-125%). The honest caveat (L1C 1.2/1.3, ⚠️) is the 10-year question: could the orchestration layer itself be abstracted away by hyperscaler-native agent stacks or foundation-model platforms? Unresolved — and the reason durability is 21/25, not 24/25.

Finding 5 — The one genuine new risk: a $10.6B M&A step-change with first debt

ServiceNow's capital-allocation record is a decade of high-ROIC organic compounding plus disciplined tuck-ins. That posture changed in 2025-2026: Moveworks $2.85B (now under DOJ probe) + Armis $7.75B cash (closing H2 2026), funded by the company's first material debt — $4.0B of senior notes plus a term loan. This is not a fatal flaw (FCF ~$4.25B covers the debt easily; investment-grade), but it is a real, unproven deviation from the pristine-compounder narrative, and it is the standing WATCH (L1D 1.2 ⚠️A; durability Q3 4/5).

Finding 6 — Long-term durability 21/25 = "Medium (top edge)," 0 fatal flags

The long-term-durability-test produces 21/25 with 0 fatal flags — the top edge of "Medium," one point below "High." The single point separating it from a high-conviction compounder is Q4 (agentic-AI disruption tail, 3/5) — precisely the fear the market has already priced. If the AI evidence keeps resolving toward beneficiary (Now Assist toward $1.5B, non-seat ACV ≥50%, NRR ≥120%), Q4 upgrades to 4/5 → 22/25 = High. This is a materially higher-conviction long-term hold than Ford (17/25, 1 fatal flag): NOW is a 3-5% core position, not a 1-3% rented position.

V. Valuation regimes

(Anchored at $112.45, 2026-06-07. Forward FY2026E non-GAAP EPS ~$4.0 → ~27-28× fwd. Must match scenarios.md. Targets are analytical †.)

Adopted analyst probability vector: 30 / 50 / 20 (Bull / Base / Bear). Implied market vector (back-solved from $112.45): ~10-15 / 35-40 / 45-55 — the market is pricing ~45-55% probability on the agentic-AI bear case. Expected value (analyst): $145.0 = +29% from $112.45. Asymmetry: upside +65% vs downside −24% = 2.7× FAVORABLE.

The analyst assigns the bear case 20%; the market implies ~45-55%. That ~25-35-point gap is the mispricing — and it exists because the bear case is a 10-year forecast that current data (50% non-seat ACV, NRR 120-125%, Now Assist doubling) directly contradicts. The Street agrees (Strong Buy, PT $141.86 = +26%).

Contrast with F (the discipline check): Ford's R2 re-anchor found EV −17%, asymmetry 0.35× UNFAVORABLE, Street PT 21% below spot — a realized thesis, "avoid." ServiceNow is the mirror image: EV +29%, asymmetry 2.7× favorable, Street PT +26% above spot — a live thesis, "own." Same discipline, opposite verdict.

Key Tension Point: the bear leg hinges entirely on seat-collapse. The empirical falsifier is the non-seat ACV mix + NRR. As long as non-seat ACV ≥50% and NRR ≥120%, the bear case lacks confirming data. Q2 2026 (~late July) cRPO + Now Assist ACV is the first test. Do NOT chase above ~$160, where the asymmetry compresses toward the F-style "thesis-realized" trap.

VI. Long-term durability test

Full execution at reports/NOW/durability_test.md. Summary:

No long-term-unsuitability warning (21/25 ≥ 17; 0 fatal flags). The single binding variable is the agentic-AI tail (Q4) — already priced by the market. Practical implication for Ming: ServiceNow is suitable as a 3-5% core long-term-compounder position (higher conviction than Ford's 1-3% rented hold). Size toward the top of the band as the AI evidence resolves toward beneficiary; re-test quarterly; the swing metrics are non-seat ACV %, Now Assist ACV, and NRR.

VII. Triggers and Red Flags

🟢 Triggers (would CONFIRM bull/base regime)

• T1 — Now Assist ACV tracking to $1.5B (2026) / non-seat ACV ≥50% (Q2/Q3 2026): falsifies seat-collapse; confirms L1B
• T2 — cc subscription + cRPO growth hold ≥19% (each quarter): confirms L1A durability
• T3 — Multiple re-rate toward ~34×+ on growth re-acceleration (12-24mo): bull re-rating toward $185
• T4 — Armis closes + integrates cleanly; DOJ Moveworks cleared (H2 2026): removes the L1D watch

🔴 Red Flags (would CONFIRM bear regime)

• RF1 — NRR < 110% OR renewal < 95% for 2 quarters: seat-collapse / churn confirmed (highest severity)
• RF2 — cc subscription growth < 15% for 2 quarters: deceleration-to-decline confirmed
• RF3 — Non-seat ACV mix < 40% AND Now Assist ACV growth < 30%: AI not offsetting seat pressure
• RF4 — Armis/Moveworks write-down >25% OR FCF margin < 28% for 2 years: M&A step-change destroys value
• RF5 — Marquee enterprise publicly replaces NOW with a model-native build: commoditization confirmed

应对 (Response) playbook

• If T1 fires: the core bear thesis is empirically falsified; raise bull weight to 35-40%; scale toward the 5% cap.
• If T2 fires (quarterly): maintain the 3-5% core; the compounder is intact.
• If T3 fires: the re-rating is happening; hold but stop adding above ~$160 (asymmetry compresses — the F lesson).
• If T4 fires: upgrade durability Q3 → 5/5 → aggregate 22/25 = High; comfortable at full 5%.
• If RF1 fires: exit or cut to ≤1% — the moat is breaking; this is the thesis-killer.
• If RF2 or RF3 fires: cut ~50%; re-test L1A/L1B; the AI-beneficiary thesis is failing.
• If RF4 fires: re-test capital allocation; trim; the M&A bet failed.
• If RF5 fires: re-test the entire L1C branch; reduce materially.
• Auto-exit if any 2 of RF1/RF2/RF3/RF5 fire.

Continuous tracking metrics: see dashboard.md.

VIII. Investment logic conclusion

The H-0 thesis reframes ServiceNow from "mature seat-based SaaS about to be disrupted by agentic AI" to "best-in-class workflow compounder and likely AI beneficiary, de-rated to a decade-low multiple on a category-level fear." The reframe holds against ~64% of the evidence: growth durability (L1A ✅A), AI monetization (L1B ✅B), and the switching-cost moat (L1D 1.1 ✅A) are strongly supported; the residual ⚠️ sit on the genuinely-unresolved 10-year disruption tail (L1C) and the unproven M&A step-change (L1D 1.2).

For the 12-24 month horizon: at $112.45 the expected value is +29% with a 2.7× FAVORABLE asymmetry; the Street is a Strong Buy at $141.86 (+26%). Every operating metric beat or held through the drawdown, so the entry is created by sentiment, not by deteriorating fundamentals. This is a positive-EV setup at spot — the inverse of Ford's R2 finding.

For the 5-10 year horizon: durability 21/25 with 0 fatal flags places ServiceNow at the top edge of long-term-hold suitability — a genuine compounder. The binding uncertainty is the agentic-AI tail (Q4 3/5), which is precisely what the market has over-priced. Forward confirmation (non-seat ACV ≥50%, NRR ≥120%, Now Assist toward $1.5B) upgrades it to High durability.

Recommended position sizing for Ming:

• New capital: INITIATE / BUILD a 3-5% core long-term position at $112.45. Favorable asymmetry, 0 fatal flags, top-edge durability, Street +26%.
• Scale toward the 5% cap as the AI-beneficiary evidence confirms (T1/T2); start at ~3% and add on confirmation rather than all at once.
• Do NOT chase above ~$160, where the asymmetry narrows toward the F "thesis-realized" trap.
• Re-test quarterly on earnings (Q2 ~late July). Auto-exit if any 2 of RF1/RF2/RF3/RF5 fire.

IX. System warnings — what could falsify H-0

The H-0 is supported, not certain (64%). It would be falsified by:

1. cc subscription growth < 15% for 2 consecutive quarters (FF1) — confirms the deceleration-to-decline bear read; re-test L1A to ✗.
2. NRR < 110% OR renewal < 95% (FF2) — confirms seat-collapse/churn; falsifies L1C 1.1 ✅A and L1D 1.1 ✅A; the thesis-killer.
3. Non-seat ACV mix < 40% AND Now Assist ACV growth < 30% (FF3) — AI is not offsetting seat pressure; falsifies L1B.
4. A marquee enterprise publicly replaces NOW with a foundation-model-native workflow build (FF4) — commoditization confirmed; collapses L1C.
5. Armis/Moveworks written down >25% OR FCF margin < 28% for 2 years (FF5) — the M&A/capital-allocation step-change destroys value; re-test L1D.
6. Forward multiple re-rates above ~40× without growth re-acceleration (FF6) — the cheapness disappears and the thesis is realized (the F outcome — at that point trim, don't add).

If 2+ of FF1-FF5 fire, the H-0 collapses and ServiceNow should be cut materially or exited.

X. Reading guide / cross-reference

Source materials in reports/NOW/:

For general stock concepts, read MANUAL_en.md / MANUAL_zh.md at the repo root.

XI. Position conclusion (verdict for Ming)

Verdict: HOLD-WITH-SIZING → INITIATE / BUILD a 3-5% core long-term position at $112.45. ServiceNow is the rarer-than-Ford case: a genuinely high-quality compounder that has been handed to you cheap by a fear it is well-positioned to absorb.

Why own it here:

• ~21% subscription growth + FCF margin expanding to 35% + Rule of ~56 = best growth-quality in large-cap SaaS, INTO a ~40-50% drawdown.
• The bear case (agentic AI kills per-seat SaaS) is contradicted by the company's own numbers: ~50% of net-new ACV is non-seat-based, NRR is 120-125%, renewal is 98-99%, Now Assist ACV is >$600M and doubling.
• Favorable structure: EV +29%, asymmetry 2.7× favorable, Street Strong Buy at $141.86 (+26%), multiple at a decade low.
• Durability 21/25 with 0 fatal flags — a true compounder, not a rented yield.

Why not maximum conviction (the honest ⚠️):

• The 10-year agentic-AI tail is real and unresolved (durability Q4 3/5): could model-native workflows or hyperscaler bundling eventually abstract away the orchestration layer?
• The $10.6B Moveworks+Armis M&A + first debt is a genuine, unproven step-change (DOJ probe on Moveworks; Armis the largest bet in company history).
• Even at a decade low the multiple is premium (~27-28× fwd vs Salesforce ~9× NTM FCF) — it requires ~20% growth to keep compounding.

Position management:

• Initiate at ~3%, build toward 5% on confirmation (T1/T2). Do not deploy all at once.
• Do NOT chase above ~$160 (asymmetry compresses — the F lesson).
• Re-test quarterly; auto-exit if any 2 of RF1/RF2/RF3/RF5 fire.

Stories lie, structure doesn't. The ServiceNow story right now is "SaaS is dead, agents will eat it." The structure is a 5-branch tree with 8 ✅, 5 ⚠️, 0 ✗ — a compounder whose growth, margin, backlog, retention, and AI ACV all beat through the drawdown, priced as if the disruption is the base case. The 5-year hold is a bet that the orchestration layer wins the agentic era; the 10-year hold adds the open question of whether that layer can itself be abstracted away — monitor it, but own the asymmetry.

Don't read news; update your tree.

XII. Investment Scorecard (15-question long-term-hold checklist)

Per MANUAL_en.md Part K.6. Bottom-line decision artifact.

Verdict tally (from narrative answers): 13 ✅ · 2 ⚠️ · 0 ✗ — Q1✅A, Q2✅B, Q3✅C, Q4⚠️C, Q5✅B, Q6✅A, Q7✅A, Q8✅A, Q9✅A, Q10⚠️B, Q11✅B, Q12✅C, Q13✅C, Q14✅B, Q15✅B.

K.3.5 Weighted-score derivation

Applying the 4-tier weighting from MANUAL §K.3.5 (verdict values: ✅ = 1.0, ⚠️ = 0.5, ✗ = 0.0; theoretical max 39):

95% — high-conviction structural score (≥85% band). Only Q4 (the agentic-AI disruption tail) and Q10 (the Microsoft/Salesforce competitive set) carry ⚠️; both are load-bearing/confirming uncertainties, not structural defects, and fatal_flags: 0. Two reads, pointing the same way:

• K.3.5 (95%, structural quality): the business — moat, balance sheet, growth, FCF, ROIC — is genuinely best-in-class.
• H-0 (64%, thesis confidence): lower than the structural score because the thesis (that the de-rating is an overcorrection) rests on the unresolved 10-year disruption tail. The 31-point gap between 95% structural and 64% thesis names the headroom that agentic-AI confirmation (non-seat ACV, NRR, Now Assist) would close — which is also the upside the bull case captures.

Scorecard summary

Final verdict: HOLD-WITH-SIZING → INITIATE / BUILD 3-5% core position at $112.45

For Ming specifically:

• ✅ Initiate a 3-5% core long-term position at $112.45 — favorable asymmetry (EV +29%, 2.7×), 0 fatal flags, top-edge durability, Street Strong Buy +26%.
• ✅ Build toward 5% on confirmation (T1/T2: non-seat ACV ≥50%, NRR ≥120%, Now Assist toward $1.5B) — start ~3%, add rather than lump-sum.
• ❌ Do NOT chase above ~$160 — asymmetry compresses toward the F "thesis-realized" trap.
• 📅 Re-test quarterly (Q2 ~late July); auto-exit if any 2 of RF1/RF2/RF3/RF5 fire.

The 2-minute pitch (per MANUAL Part K.1):

"ServiceNow is the best compounder in enterprise software — 21% growth, free-cash-flow margin heading to 35%, 98-99% renewal, 120-125% net retention — and it's down ~40-50% to its cheapest multiple in a decade because the market decided agentic AI will kill per-seat SaaS. But ServiceNow's own numbers say it's the AI winner, not the casualty: half of net-new contract value is already non-seat-based, Now Assist AI revenue is past $600M and doubling, and management has a credible $30B-by-2030 plan with $9B from AI. The Street is a Strong Buy at $142, ~26% above the price; my scenarios say +29% expected value with a 2.7× favorable asymmetry on a business with zero fatal flags and 21-out-of-25 durability. So I'd build a 3-5% core position here and add as the AI-beneficiary evidence confirms — the opposite of Ford, where the thesis had already played out. The one thing I'm watching is the 10-year tail (could agents eventually bypass the workflow layer?) and the new $10.6B of debt-funded M&A."

Risk types most relevant (per MANUAL Part K.4):

• Disruption/technology risk (agentic AI commoditizing the workflow layer; the binding 10-year tail)
• Competition risk (Microsoft bundling; Salesforce Agentforce)
• Execution risk (Armis $7.75B integration; DOJ Moveworks probe; first debt)
• Valuation risk (premium multiple requires sustained ~20% growth)

"When NOT to buy" anti-patterns (per MANUAL Part K.5) — DO ANY APPLY?

• ❌ Online hype: the opposite — NOW is out of favor (SaaS-pocalypse sell-off)
• ❌ Stock went up fast: the opposite — down ~40-50%, ~47% off high
• ⚠️ Has "AI" attached: YES — but as the de-rating cause, not a hype tailwind; the AI angle here is "is it a victim," which is the bear case being analyzed, not a promotional claim
• ❌ Looks "cheap" because price is low: the $112 absolute is a split artifact; the relevant metric (decade-low forward multiple) is genuinely cheap for the franchise
• ❌ Brand familiarity: thesis is grounded in metrics, not brand
• ❌ "Next [Nvidia]": no such framing

Net: zero buy-discipline anti-patterns fire. This is a fundamentals-grounded, favorable-asymmetry entry into a high-quality compounder during a sentiment-driven drawdown — the textbook setup the anti-pattern list is designed to protect, not flag.

Tree v1 written 2026-06-08. Next refresh: Q2 2026 earnings (~late July 2026). Refresh cadence: quarterly per the long-term-investability research workflow in CLAUDE.md.