Phase Space AI

Valuation

CrowdStrike [CRWD]

CrowdStrike Holdings, Inc. [CRWD] — Valuation

As of 2026-07-29 · spot $179.50 · Two outputs, both required and both given.

---

1. Verified inputs — and the 4.0x correction

Cross-checks: EPS 27.774 ÷ 257.881 (pre-split diluted) = $0.108 vs filed $0.11 ✓. TTM revenue matches the screen ✓. Net cash matches the screen ✓. Period 2026-04-30, filed 2026-06-04 — 55 days old ✓.

Note what the EPS cross-check does and does not catch. The pre-split share count and the pre-split EPS are internally consistent, so the mandated net income ÷ shares ≈ filed EPS test passes on the screen's own numbers. The error is only visible when the count is paired with a price. That is the lesson: the EPS check validates a share count against an income statement, not against a market. Both checks are needed, and the second one — does the implied market cap make sense against the last corporate action in the price series? — is the one that was missing.

---

2. The exit multiple — a SEPARATE anchor, and the EBIT basis is UNIDENTIFIED

2a. Why CRWD cannot share the cluster's anchor

Four names span 8.8%–14.2%. CRWD is 11.4pp above the top of that band, nearly double the next-fastest. Using the mature-software anchor (17.5x EV/EBIT, comparators growing 7.0–20.6%) on a 25.6% grower would repeat the exact defect criteria.md documents: an anchor set with no growth dispersion in the subject's range, extrapolated beyond its support. CRWD is anchored separately.

2b. The EBIT-basis exit multiple is UNIDENTIFIED — declared, not defaulted

The growth-matched comparator set for a 25.6% grower, every member's growth and EV computed from its own filings on this date:

Of the six comparators in the 20–34% growth band, three have negative GAAP operating income and the three positive ones span 49.9x to 177.8x — a 3.6x range with no usable central tendency. Per criteria.md: "An exit multiple may only be drawn from a comparator set whose growth brackets the subject's growth at the exit year. If no such comparator exists, the multiple is UNIDENTIFIED and must be declared so."

I declare the EBIT-basis exit multiple UNIDENTIFIED and run the primary test on a sales basis instead.

2c. The sales-basis anchor, which IS identified

The same 20–34% growth cohort on EV/sales:

HUBS 3.39x · ZS 7.52x · NOW 8.33x · MDB 8.99x · FTNT 15.52x · SNOW 19.29x

• Base exit multiple: 8.7x EV/sales (the median)
• Low: 3.4x (HUBS) · High: 19.3x (SNOW, at a −26.1% operating margin)

The set brackets the subject on growth (20.1% to 33.5% versus 25.6%). Growth-matched: SATISFIED. All six are $10bn+ enterprise software companies; none is a sub-$100m shell.

Cross-check on the year-5 anchor. By 2031 a 25.6% grower will have decelerated — plausibly to 15–18%. The comparators bracketing that range are VEEV at 7.97x (16.3% growth), MSFT at 8.81x (17.7%) and FTNT at 15.52x (20.1%), a median of ~8.8x. That is within 1% of the 8.7x base, so the anchor is robust to the deceleration assumption. Recorded because it is the check that would have caught the diagnostics-set failure.

Terminal operating margin is not required on a sales basis. It is nonetheless run as a sensitivity in §3, on an EBIT basis, to demonstrate that the FAIL does not depend on the choice of basis.

---

3. Implied-path test — the Valuation Criteria

assets/reverse_dcf.py. Terminal value is 100% of EV by construction, so the reverse DCF is the primary long-horizon output.

Solve 1 — required revenue CAGR, sales basis

Held fixed: WACC 9.5%, 5 years, revenue base $5,094.2m, shares 1,018.148m, net cash +$3,807m.

Solve 2 — required exit multiple at demonstrated growth

Held fixed: revenue CAGR 25.6%, WACC 9.5%.

• Year-5 revenue implied: $15,923m
• Required exit multiple: 17.7x EV/sales
• Today's multiple: 35.13x
• Implied compression: −49.6%
• Against the anchor set: 17.7x sits between FTNT's 15.52x and SNOW's 19.29x — i.e. at roughly the 90th percentile of the growth-matched cohort, and just below the single most expensive comparator, which carries a −26.1% operating margin.

So the price requires CRWD, having grown revenue 3.1x over five years to $15.9bn, to still trade at a multiple only the most expensive and least profitable name in its growth cohort commands today.

Sensitivity 1 — the exit multiple

The verdict only turns positive at the single highest observation in the cohort — SNOW's 19.29x, a name losing 26 cents of operating profit on every revenue dollar. At FTNT's 15.52x, the closest genuinely profitable comparator (31.1% GAAP operating margin), the margin is still −4.5pp.

Sensitivity 2 — the terminal margin, on an EBIT basis, to show the basis does not matter

Exit multiple held at the mature-cluster 17.5x EV/EBIT:

FAIL at every terminal margin from 10% to 35%, including above the guided non-GAAP figure. The sales-basis and EBIT-basis routes agree.

Verdict

FAIL. Margin −19.2pp on the identified sales-basis anchor. The EBIT-basis exit multiple is UNIDENTIFIED and is declared so rather than defaulted to a peer median.

A necessary distinction from PANW, which carries the same verdict. CRWD fails by 19.2pp; PANW by 46.2pp. CRWD's failure is a valuation failure on a genuinely good business: growth accelerating 19.8% → 25.6%, entirely organic, 115% dollar-based net retention disclosed exactly, gross margin the only one expanding in this cluster, and the best disclosure of the five. PANW's is a valuation failure on a decelerating business whose reported growth is two-thirds acquisition and which discloses no retention rate at all. Same verdict, very different names.

And unlike CRM at −0.9pp and WDAY at −0.0pp, this is not a case where the analysis fails to determine the answer. −19.2pp is outside any reasonable noise band, it holds across every terminal margin tested, and it turns positive only at the most expensive observation in the growth-matched cohort. The brief warned against manufacturing conviction where the margin is inside ±3pp; it equally requires stating a clear result where one exists, and here one does.

---

4. Twelve-month target

4a. Own-multiple history, with percentile stated

P/S at fiscal quarter ends, priced ~45 days after period end. Explicit split factor of 4.0 applied to all observations before 2026-07-02 so the series is internally consistent:

2020-04 75.8 · 2020-10 57.8 · 2021-01 52.2 · 2021-07 51.2 · 2022-01 32.1 · 2022-10 13.2 · 2023-01 13.9 · 2023-10 22.3 · 2024-01 25.6 · 2024-04 29.4 · 2024-07 19.1 · 2025-01 22.7 · 2025-04 28.8 · 2025-10 26.8 · 2026-01 22.6 · 2026-04 35.1

Read the two percentiles carefully. The full-history 72nd percentile is flattered by the 2020–21 tail, where CRWD traded at 52–76x sales during the pandemic software bubble. That regime is gone and using it as an anchor would be the definition of a regime-change error. The five-year figure (90th percentile) is the honest one, and the July-19 trough of 13.21x sets the floor of the relevant range.

CRWD is also the only name in this cluster to have re-rated upward recently: 22.6x (2026-01) → 35.1x (2026-04) → 35.88x today, a 59% multiple expansion in two quarters — which is what the +12.1% post-split session on 2026-07-14 and the raised net-new-ARR guidance produced.

4b. Regime-change test

The own-multiple history spans a 40.4pp growth-regime change — by far the largest of the five. Mean reversion within a history that includes a 66%-growth, 76x-sales company is not a valid anchor for a 25.6% grower. Reversion is declared UNIDENTIFIED per valuation.md.

4c. The stated target — with the same asymmetry caveat as PANW

• NTM revenue $6,281m. Derivation: FY27 guidance $5,914.7–5,958.7m (mid $5,937m), rolled forward one quarter at ~23%. Company guidance, disclosed; Alpha Vantage consensus is quota-blocked.
• Shares in 12 months ~1,033m. 1,018.1m rising ~1.5%: the buyback covers only 59% of SBC and the count rose 0.46% in Q1 FY27 alone. This is one of only two names in the cluster where the forward count is higher (PANW is the other).

As with PANW, "flat multiple" is not a neutral assumption for a name at the 90th percentile of its own history — it means holding a near-record multiple for another year. For CRM, ADBE and WDAY (5th–10th percentile) the flat case is a conservative floor; here it is an aggressive one.

Stated 12-month position: UNIDENTIFIED, with a defensible range of −14% to +22%, centred on approximately flat. The p75 reversion case lands at $179 against a $179.50 spot — essentially zero. That is the most defensible single reading: revenue growth of ~23% offset by a modest de-rating from the 90th percentile toward the 75th.

Cross-check against earnings, which is brutal and must be reported. At the current price the FY27 non-GAAP P/E is ~156x; on GAAP there are no meaningful earnings to divide by. Even at the −13.7% median-reversion case the non-GAAP P/E is ~135x. The P/S-based range and any earnings-based view do not merely disagree — they are not in the same universe. On ADBE the two cross-checks agreed; on WDAY they disagreed; here the earnings basis simply offers no support at any point in the range, and that is the honest report.

4d. Named events inside 12 months

Each is in the Catalyst Calendar with a date or a marked estimate. 1. Q2 FY27 results (~late Aug 2026) — guided to $1,436–1,442m revenue and $5,792.6–5,794.6m ARR. Tests whether net new ARR is tracking the raised 27.7% full-year growth guide, and whether GAAP operating income improves without further help from the commission-amortisation change. 2. Q3 FY27 results (~early Dec 2026) — third quarter of the new commission policy; the year-over-year comparison begins to normalise. 3. Q4 FY27 / FY2026 10-K (~early Mar 2027) — the annual dollar-based net retention disclosure. It has run 125.3% → 119% → 112% → 115%. This is the single most informative disclosure any of these five companies publishes, and it arrives once a year.

4e. External sanity band

No professional target is on file for CRWD in this book's reference set. The check cannot be run, and that is stated rather than substituted.

---

5. Downside case with a named cause

Type: MEASURED. Logged, scored, and it does not reject the name — the Valuation Criteria already does.

Scenario — a second platform incident, or Microsoft Defender bundling, compresses net retention below 110% while SBC at 22.8% of revenue cannot be cut. Probability 30%.

Named cause and mechanism, with precedent. The July 19, 2024 incident is not a hypothetical — it happened, and its effect is measurable in CRWD's own disclosure: dollar-based net retention fell from 119% to 112% and revenue growth troughed at 19.8%. The 10-Q still discusses it 55 times, two years later, and the FY2025 10-K states that following the incident the company "experienced delays in creating sales opportunities and longer sales cycles, including delays in customer purchasing decisions." A single-agent, kernel-level architecture on hundreds of millions of endpoints has this exposure structurally.

The second leg is competitive: Microsoft Defender is bundled into E5, and CRWD's expansion motion depends on module attach (51% / 35% / 25% at ≥6 / ≥7 / ≥8 modules). Bundling compresses expansion pricing without causing churn — so it would appear in net retention, not in gross retention or in the ARR headline. Net retention is disclosed only annually, so the signal would arrive up to twelve months late.

The third leg is why the impairment is severe: SBC is 22.8% of revenue and rising, and it cannot be cut quickly without losing engineers. A revenue-growth shock therefore hits an unprotected cost base — ex-SBC operating income of $804m is entirely a function of not paying cash for labour.

Quantified: growth falls from 25.6% to ~12% over two years and DBNRR settles near 105%. The multiple reverts toward the July-19 trough of 13.21x sales on NTM revenue of ~$6.6bn across ~1,050m shares → approximately $83 per share, a 54% permanent impairment. A milder version — growth to ~17%, multiple to the five-year median 25.48x — gives approximately $155, a 14% decline.

Not a going-concern case. Net cash of +$3,807m, FY26 operating cash flow of $1,612m, and only $745.8m of debt. CRWD's cash generation is genuinely strong — $591m of operating cash flow in Q1 FY27 alone, up 54%. Stated explicitly because criteria.md requires a going-concern case to be flagged where it exists, and here it does not.

One structural item flagged, not resolved: the non-controlling interest. $18.2m of Q1 FY27 pre-NCI income was attributable to minority holders, against −$0.8m a year earlier — a $19.0m swing on a $27.8m reported result, with $41.5m of NCI equity on the balance sheet. I have not established what entity this is and the 10-Q sections I read do not make it clear. It is disproportionately large relative to earnings and it is a genuine gap in my understanding of this income statement.

Interim control: inverse-volatility sizing. CRWD's 252-day volatility is 47.6%, the second-highest of the five (behind WDAY's 49.0%), so it sizes down automatically. Unlike PANW, the interim control works as designed here — CRWD's left tail is a genuine operational event risk with historical precedent, and it is a high-volatility name. The mechanism and the analysis point the same way.