CrowdStrike Holdings CRWD

Reference table

1. What the business is, and how the disclosure actually works

CrowdStrike describes itself in the 10-K as having "reinvented cybersecurity for the cloud era" with "the AI-native CrowdStrike Falcon platform – the first, true cloud-native unified platform built with artificial intelligence at the core, capable of harnessing vast amounts of security and enterprise data to deliver highly modular solutions through a single lightweight sensor." Founded 2011; IPO 2019-06-12; headquartered in Austin, Texas. Chief executive and founder George R. Kurtz; president Michael Sentonas. 10,698 full-time employees at 2026-01-31 — a figure the vendor profile reports as exactly 10698, matching the filing to the person.

The share structure is now simple, and the filing settles it. The 10-K states: "On December 11, 2024, all of our outstanding shares of Class B common stock automatically converted into an equal number of shares of Class A common stock." There is one class of common stock, one vote per share. The 8-K of 2026-06-22 confirms it from the other direction: at the annual meeting record date of 2026-04-24 there were 254,536,521 shares of Class A common stock outstanding and entitled to vote, with no Class B mentioned. There is no dual-class overhang here, and that is worth saying because dual-class structures are the norm among 2019-vintage software listings.

The revenue disclosure: two lines, and they reconcile exactly

This is one of the simplest and cleanest segment disclosures in the batch. The vendor's seg_prod supplies exactly two lines, and both sum precisely to reported revenue in every year.

The 10-K corroborates every line verbatim: "Total revenue increased by $858.4 million, or 22%, in fiscal 2026... Subscription revenue accounted for 95% of total revenue for each of fiscal 2026 and fiscal 2025. Professional services revenue accounted for 5%." And: "Subscription revenue increased by $803.2 million, or 21%"; "Professional services revenue increased by $55.2 million, or 29%." The vendor's arithmetic gives $858.381M, $803.203M and $55.178M respectively. This is an exact vendor-to-filing reconciliation on the revenue split, and it is stated as a clean finding.

The only labelling issue is cosmetic: "Subscription And Circulation" is a generic XBRL tag, not CrowdStrike's own vocabulary, which is simply "Subscription." We use the company's word.

Note what professional services actually is. The 10-K is explicit that it is not a profit centre: "We view our professional services business primarily as an opportunity to cross-sell subscriptions to our Falcon platform and cloud modules." It is incident response, forensics and consulting, sold on time and materials, and it exists as a lead generator. Its gross margin fell 1% in FY2026 per the filing. It is 5.1% of revenue and it grew faster than subscription in FY2026 (+28.7% vs +21.4%) — which, given that breaches drive demand for it, is a fact with two possible readings.

Geography (seg_geo, sums exactly to revenue in every year)

The United States is 66.8% of revenue and its share has fallen in each of the last four fiscal years, from 69.8% in FY2023. International is growing faster on every line. The 10-Q confirms the concentration profile: "No single country other than the United States represented 10% or more of the Company's total revenue during the three months ended April 30, 2026 and April 30, 2025." There is no China exposure question here of the kind that dominates the semiconductor names in this batch, and no single-customer concentration is disclosed anywhere in the file.

The two forward indicators that are in NO vendor field

This is the most important disclosure-structure point in the dive. The metrics that actually govern this business are not in the vendor payload at all. They are in the filings, and we extracted them by hand.

Read that table carefully because it contains both halves of the argument.

The bullish half: ending ARR compounded from $4.2B to $5.3B (+24%) and then to $5.5B, and net new ARR in the April 2026 quarter was $255.8M against $193.8M a year earlier — a 32% increase in the FLOW. That is acceleration in the leading indicator, not just the lagging one. The 8-K of 2026-04-21 gives management's own framing: ARR "from $3.44 billion in ending ARR in fiscal year 2024 to $5.25 billion in ending ARR in fiscal year 2026" — half-weighted as company voice, but consistent with the 10-K.

The bearish half: remaining performance obligations FELL sequentially, from $9.0 billion at 2026-01-31 to $8.8 billion at 2026-04-30, while revenue grew 6.1% sequentially in the same quarter. The company's seasonality disclosure explains part of it — "Net new ARR generation is typically greater in the second half of the year, particularly in the fourth quarter" — but a declining forward-contract balance in a business valued at 135.7x forward earnings deserves to be named, not smoothed. The mix shifted slightly toward the near term as well: 51% of RPO was expected within twelve months at 2026-01-31, and 52% at 2026-04-30.

Backlog is a third figure and it is separate: $4.2 billion at 2026-01-31, defined as contracted amounts CrowdStrike does not yet have the contractual right to invoice, "not recorded in deferred revenue or elsewhere in our consolidated financial statements." The 10-K adds, with unusual candour, "We do not utilize backlog as a key management metric internally."

Competition, per the 10-K, names no companies at all — only seven categories: legacy antivirus, alternative endpoint security, network security vendors extending into endpoint, cloud security vendors, identity security vendors, professional services providers, and legacy SIEM vendors. That is the competitive-disclosure posture of a company that believes it competes with a whole industry from a platform position. For a reader, it means the filings give no help at all in sizing rivalry, and the knowledge base is the only place a named competitor appears (Section 7).

2. The single most important thing to understand about these numbers: GAAP is negative and non-GAAP is the entire valuation

Every forward multiple in this dive, and every forward multiple you will see anywhere on this stock, is struck on a non-GAAP earnings number. The GAAP number is a loss. The gap between them is almost entirely stock-based compensation. This section quantifies it, because the honesty of the whole valuation depends on it.

Four things follow from that table, and together they are the honest valuation frame.

First, GAAP operating income has been negative in five of the last six fiscal years, and the one positive year (FY2024) was positive by $89.3 million of NET income on a −0.07% operating margin — that is, the profit came from interest on the cash pile, not from operations.

Second, the FY2026 GAAP loss got WORSE, not better, while revenue grew 21.7%. Operating margin went from −3.05% to −6.10%. The reasons are disclosed and are mostly one-time or investment: $44.8 million of Strategic Plan restructuring charges, continuing legal and professional expense from the July 19 Incident, R&D up 29% and sales and marketing up 20%, and stock compensation up $231.3 million.

Third — and this is the reconciliation the whole dive turns on — the GAAP-to-non-GAAP EPS gap for FY2026 is $1.09 per share, and stock compensation alone is $1.06 of it.

> GAAP diluted EPS FY2026: −$0.160

> Consensus/reported non-GAAP EPS FY2026: +$0.928 (the four earn_cal actuals of $0.1825, $0.2325, $0.2400 and $0.2750 sum to $0.930, and the est FY2026 epsAvg is $0.92842 — a 0.17% difference, which independently confirms that the est block is struck on the NON-GAAP basis and never says so)

> Gap: $1.088 per share

> Stock-based compensation FY2026: $1,096.679M ÷ 1,032.5M diluted shares = $1.062 per share

> Stock compensation explains approximately 97.6% of the entire gap. The residue is amortisation of acquired intangibles ($31.2M), the $44.8M Strategic Plan charge, acquisition-related costs and their tax effects.

Fourth, and least discussed anywhere: consensus non-GAAP EPS went DOWN in FY2026. $0.940 in FY2025 (31 analysts), $0.928 in FY2026 (36 analysts). A 21.7% revenue increase produced a 1.2% decline in the adjusted earnings the entire valuation is struck on. That is what the July 19 Incident aftermath, the Strategic Plan and the investment step-up cost. It is also why the FY2027 consensus of $1.231 — a 32.6% jump — is the single most load-bearing number in this file.

We state the basis explicitly and we do it once: every forward multiple in this dive is struck on consensus NON-GAAP earnings per share, because that is the only forward earnings basis the file contains. The trailing GAAP price/earnings ratio is negative (the vendor prints −4,446.7x) and is rejected as a lens. Where we quote a GAAP figure we label it GAAP.

And we say plainly what that means for the reader. At $211.22 you are paying 135.7x an FY2028 earnings figure that will only exist if you agree to treat roughly a fifth of revenue paid to employees in shares as not being a cost. Reasonable people disagree about that treatment. What is not reasonable is to quote the 135.7x without saying which side of the argument it sits on.

3. Growth and margin trajectory — the post-incident recovery, quarter by quarter

This is the strongest section of the case and it is entirely observable.

The trough is unmistakable and it is the quarter ended 2025-04-30, reported 2025-06-03: revenue growth 19.8%, gross margin 73.8%, operating margin −11.30%, and a double miss — non-GAAP EPS of $0.1825 against $0.2022 expected (−9.7%) and revenue of $1,103.4M against $1,104.6M (−0.1%). Since that quarter, four consecutive beats on both lines, revenue growth accelerating in every single quarter, and gross margin recovering 230 basis points from trough to peak.

The cost lines tell the same story. Research and development went from 30.3% of revenue in the trough quarter to 27.3% in the January 2026 quarter; selling, general and administrative from 54.8% to 47.8%. That is 950 basis points of combined operating expense leverage in three quarters, and it is what took operating margin from −11.30% to +1.06%.

Two honest qualifications on the most recent quarter.

First, Q1 FY2027 operating margin went back to −2.21% and gross margin dipped to 75.3%. The 10-K pre-announces exactly this pattern: "we also experience seasonality in our operating margin, typically with a lower margin in the first half of our fiscal year due to a step up in costs for payroll taxes and annual sales and marketing events." On a like-for-like basis the improvement is enormous — Q1 FY2027 gross margin of 75.3% against Q1 FY2026's 73.8%, and operating margin of −2.21% against −11.30%, a 909 basis point year-on-year improvement. We accept the seasonality explanation because it is disclosed in advance rather than offered afterwards.

Second, the GAAP profit in Q1 FY2027 is not an operating profit and we will not present it as one. Operating income was −$30.6 million. Pre-tax income was +$39.1 million, arriving via $40.5 million of interest income on the cash pile and, per the 10-Q, "an increase in net realized gains on our strategic investments of $36.4 million." Net income was then flattered further by a tax BENEFIT of $6.9 million, which the 10-Q attributes to "income tax benefits recognized in the current period resulting from the realization of deferred tax assets in connection with recent acquisitions." Strip the $36.4 million of investment gains and the quarter is roughly break-even before tax; strip the tax benefit as well and it is a loss. The company is not yet GAAP-profitable from operations and the dive says so.

Research and development was $1,384.8 million in FY2026, 28.8% of revenue, up from 27.2% and 25.2%. The 10-K attributes the 29% increase to headcount and stock compensation. A company spending 28.8% of revenue on R&D while gross margin holds at 74.7% is investing through the cycle, and it is also why the GAAP line is negative.

4. Balance sheet and the enterprise-value rebuild

The most recent filing-verified balance sheet is 2026-04-30, from the 10-Q filed 2026-06-04. The vendor's bal_a block stops at 2026-01-31 and is therefore one full quarter stale — and that quarter contained the two largest cash movements in the company's history.

The 10-Q settles the debt question completely. "On January 20, 2021, the Company issued $750.0 million in aggregate principal amount of 3.00%" senior notes, due in fiscal 2030, with interest payable semi-annually. "As of April 30, 2026, the Company was in compliance with all of its financial covenants under the Indenture." The notes are guaranteed by CrowdStrike, Inc. and CrowdStrike Financial Services, Inc. Their fair value was $715.4 million at 2026-04-30 against $750.0 million of principal — they trade below par, which is what a 3.00% coupon issued in January 2021 does in 2026, and it is worth noting that CrowdStrike's cost of debt is far below anything it could refinance at today.

There is no revolving credit facility. The 10-Q states the decrease in interest expense was "primarily due to no amortization of debt issuance costs in the current period related to our secured revolving credit facility, which expired in January 2026." The facility was not replaced. Given $4.6 billion of cash and $1.6 billion of annual operating cash flow this is not a solvency issue, but it is a reduction in contingent liquidity that no screen will show, and it is stated as a finding.

Correction 1 — bal_a[0] is stale and misses $1.06 billion of cash movement. Between 2026-01-31 and 2026-04-30 the company spent $881.4 million on acquisitions (Seraphic, closed 2026-02-03, $327.4M net of cash acquired; SGNL.AI, closed 2026-02-20, $627.9M net) and $175.6 million on share repurchases, against $590.9 million of operating cash flow. The vendor's bal_a shows net cash of $4,410.0M; the 10-Q's disclosed liquidity implies approximately $3,704M; we use the 10-Q figure.

Correction 2 — and this one is a verified CLEAN result worth stating. The recurring defect on this class of file is netDebt omitting short-term investments. Here shortTermInvestments is genuinely $0 and that is correct, not a hole. The 10-K states liquidity as "$5.2 billion in cash and cash equivalents, which mainly consists of cash on hand and highly liquid investments in money market funds, U.S. Treasury bills, and time deposits" — i.e., the entire portfolio is classified as cash equivalents, so there is nothing for the vendor to omit. The FY2024 balance sheet did carry $99.6M of short-term investments, and the FY2025 cash-flow statement records "proceeds from maturities of short-term investments of $97.3 million", which is that position running off. The zero is a real business fact, not a missing field. This defect class is CLEAN on CRWD.

The rebuilt enterprise value:

> Market cap $215.077B (1,018,259,280 shares × $211.22)

> + total debt ~$0.849B (senior notes $0.747B carrying + finance leases ~$0.102B)

> − cash and equivalents ~$4.553B (10-Q: "$4.6 billion")

> = rebuilt enterprise value ~$211.37B

The vendor reports $211.345B. The difference is $28 million, or 0.01% — the closest vendor-to-rebuild agreement in this batch. We do NOT add back the $76.8 million of long-term strategic investments (illiquid private-company stakes held through the consolidated Falcon Funds); they are real but not liquidity, and including them would move the enterprise value by 0.04%.

Correction 3 — every EBITDA-based ratio in the vendor file is arithmetically meaningless and is REJECTED in full. ratios_ttm.ebitdaMarginTTM of 6.57% on TTM revenue of $5.094B implies trailing EBITDA of $334.7M, producing evToEBITDATTM of 631.4x. But CrowdStrike's trailing operating income is NEGATIVE, so a positive EBITDA can only be manufactured by adding back interest income. We compute EBITDA the honest way — operating income plus depreciation and amortisation — and for FY2026 that is −$293.3M + $281.5M = −$11.8M. A negative EBITDA cannot support a multiple. No EV/EBITDA figure appears anywhere in this dive and none should appear on any screen for this name.

What the balance sheet actually says. Net cash of $3.70 billion, total debt of $849 million against $1.6 billion of annual operating cash flow, no revolver, no derivatives, no hedges, no off-balance-sheet guarantees (the 10-Q says so in terms), and $4.7 billion of deferred revenue — customer money already collected against services not yet delivered. That deferred revenue is a liability on the balance sheet and an asset in economic terms: it is 98% of a year's revenue, pre-paid, and it grew 27.5% in FY2026 against revenue growth of 21.7%, which is billings running ahead of recognition. This is a financially unbreakable company. The risk in CRWD is not its balance sheet.

Three items that ARE genuine risks and sit in no vendor field.

First, $4.3 billion of non-cancelable purchase commitments. The 10-K discloses "non-cancelable purchase commitments with various parties to purchase products and services entered in the normal course of business totaling $2.8 billion as of January 31, 2026, with remaining terms in excess of 12 months." The 10-Q updates it to $2.6 billion at 2026-04-30 — and then adds: "Subsequent to April 30, 2026, the Company entered into a non-cancellable purchase commitment in the amount of $1.7 billion, commencing from the second quarter of fiscal 2027 through fiscal 2037." The commitments are described as "data center capacity, advertising, technology, equipment, office renovations, corporate events, and consulting services." Roughly $4.3 billion of contracted spend, running eleven years out, against a company that produced $1.24 billion of free cash flow last year. It appears in no vendor field, no ratio and no screen.

Second, uncommenced leases. "As of April 30, 2026, the Company has entered into non-cancelable operating leases, with lease terms greater than 12 months that have not yet commenced, with undiscounted future minimum payments of $103.7 million." Plus disclosed real-estate obligations of $21.7 million due within twelve months and $64.2 million thereafter.

Third, unrecognised tax benefits of $50.7 million at 2026-04-30 (up from $47.3 million), classified long-term, and $34.8 million of unfunded loan commitments from the financing arrangements.

Equity of $4.47 billion against $11.09 billion of assets is a 40.3% equity ratio, and the accumulated deficit of $1.283 billion is the cumulative record of a company that has never made money on a GAAP basis in aggregate. The vendor's priceToBookRatioTTM of 46.3x, returnOnEquityTTM of −0.58% and interestCoverageRatioTTM of −7.27x are all arithmetically correct and all economically useless. They are rejected as lenses. The one return measure worth quoting is returnOnInvestedCapitalTTM of 5.96% — genuinely low, and honestly reported: this is a company whose returns live in the future, not the present.

5. Cash flow and capital returns

Correction 4 — the vendor's FY2026 free cash flow is overstated by $68.8 million because the capital-expenditure line is computed on a DIFFERENT basis from FY2025 and FY2024. For FY2025 the vendor's capitalExpenditure of $313.821M reconciles to the 10-K's "purchases of property and equipment of $254.9 million, capitalized internal-use software and website development costs of $59.0 million" — it includes both. For FY2024 the vendor's $237.112M likewise equals $176.5M + $60.6M. For FY2026 the vendor's capitalExpenditure of $302.108M is identical to investmentsInPropertyPlantAndEquipment and therefore EXCLUDES the $68.8 million of capitalised internal-use software the 10-K discloses.

The consequence is a fake acceleration in free cash flow. On the vendor's numbers FCF grew from $1,067.9M to $1,310.2M, +22.7%. On a consistent basis it grew from $1,067.9M to $1,241.4M, +16.2%. The vendor reports FY2026 free cash flow of $1,310.2M; the filing supports $1,241.4M; we use $1,241.4M and every free-cash-flow multiple in this dive is computed on it. FCF margin is 25.8%, not 27.2%.

Rule of 40, honestly computed: 21.7% revenue growth + 25.8% FCF margin = 47.5%. Good. Not extraordinary for a company at 42x sales.

Cash conversion is genuinely excellent and the reason is the business model. Operating cash flow of $1.612 billion arrives from a GAAP net loss of $161.2 million because of $1.097 billion of stock compensation, $449.4 million of amortised deferred contract acquisition costs, $281.4 million of depreciation and amortisation — and a $1.0 billion increase in deferred revenue. Customers pay up front. That is a structural funding advantage and it is why a company with negative accounting earnings has $4.6 billion of cash.

But note precisely what that means. Of the $1.612 billion of operating cash flow, $1.097 billion — 68% — is the add-back of compensation paid in shares. Free cash flow is real cash, but the shareholder pays for it in dilution rather than in cash outflow. Diluted shares (split-adjusted) went from 974.5M in FY2024 to 979.0M in FY2025 to roughly 1,010–1,033M in FY2026: approximately 3.9% growth over two years, or about 2% a year. That is the true cost line, and at $211.22 per share, 2% annual dilution is roughly $4.3 billion of value transferred per year — more than three times the free cash flow.

Capital returns: a buyback that has barely started, and no dividend, ever.

• Dividend: none, and the 10-K forecloses one. "We have never declared or paid any cash dividends on our capital stock. We currently intend to retain all available funds... and do not expect to pay any dividends on our capital stock in the foreseeable future." The vendor's lastDividend: 0 and dividendYieldTTM: 0 are correct.
• Buyback: authorised $1.0 billion on 2025-06-03, increased by $500 million on 2026-04-06 to $1.5 billion. The 10-K's Item 5(c) — "Issuer Purchases of Equity Securities: None" — confirms nothing was repurchased in the whole of FY2026. The first repurchases were in the quarter ended 2026-04-30: $175.6 million, with $1.3 billion still available.
• The vendor's cf_a shows commonStockRepurchased: 0 in all four years. That is CORRECT for FY2023-FY2026 and STALE thereafter — it does not capture the Q1 FY2027 repurchase. Flagged rather than silently corrected.

Total shareholder yield is therefore approximately 0.33% — $175.6 million annualised at one quarter's rate against a $215.1 billion market capitalisation, with no dividend. Against roughly 2% annual dilution from stock compensation, the net shareholder yield is meaningfully NEGATIVE. This is not an income story and it is not a return-of-capital story. Every dollar of the return has to come from growth and the multiple.

One observation on the buyback that cuts both ways. The board added $500 million of authorisation on 2026-04-06, when the stock was far below today's price, and the company has used only 12% of the enlarged authority. Whether management chooses to buy stock at $211.22 having declined to buy much of it at lower prices is a genuinely informative signal, and it is observable in the next 10-Q.

6. Valuation — priced in or room?

At $211.22 (market cap $215.077B, 1,018,259,280 shares, rebuilt enterprise value $211.37B):

Estimate coverage is excellent on the near years and collapses beyond them. FY2027 and FY2028 carry 38-39 analysts on revenue and 26-27 on EPS — among the deepest coverage in this batch, and the reason those two years carry all the weight in our fair value. FY2029 EPS rests on 6 analysts, FY2030 on TWO, and FY2031 on five. The FY2030 EPS row is discarded outright and no conclusion in this dive touches FY2029 or beyond.

Per the data contract, est.ebitdaAvg and est.ebitAvg are treated as unreliable and are not used. We verified them, and they fail decisively. In every one of the eight years in the block, netIncomeAvg EXCEEDS ebitdaAvg — FY2027 net income of $1,279.8M against EBITDA of $927.0M; FY2028 $1,639.5M against $1,129.1M; FY2029 $2,143.9M against $1,375.5M. Net income cannot exceed EBITDA for a company with depreciation, interest expense and tax, even allowing for $190 million of interest income. The only coherent reading is that netIncomeAvg and epsAvg are on a NON-GAAP basis while ebitdaAvg/ebitAvg are on a GAAP-like basis, and the two are stitched into the same row without a label. All forward valuation here runs on epsAvg and revenueAvg only.

We did verify epsAvg against actuals and it passes: FY2026 epsAvg of $0.92842 against the four earn_cal non-GAAP actuals summing to $0.930 — a 0.17% difference. We also checked for the split-corruption signature (the FY2024 row being unadjusted for the 4-for-1 split, as happened on another name in this batch): FY2024 epsAvg is $0.7389, consistent with the $0.928-$1.231 progression around it. The estimate block is correctly split-adjusted across all eight years. That sub-check is CLEAN.

Peer context, and a warning. The vendor peer set is unusable as supplied: Accenture, Analog Devices, KLA, Synopsys and Texas Instruments are not comparables to a security-software company, and CoreWeave is not comparable to anything. The three genuine peers in the set are Palo Alto Networks ($298.57B, $366.34), Fortinet ($123.48B, $168.29) and Cloudflare ($106.97B, $301.38). No peer multiples, revenue figures or estimates are supplied for any of them in this file, so no peer-multiple comparison is drawn. What can be said is that at $215.08 billion CrowdStrike is now the largest of the pure-play security names in the set by market capitalisation, having passed Palo Alto Networks.

6a. What today's price assumes (the inversion)

At $211.22 — 171.6x FY2027 consensus non-GAAP EPS, 135.7x FY2028, 41.5x enterprise value to trailing revenue — the price embeds the following falsifiable claims, each with a number and a date:

• Revenue reaches $5.944B in the twelve months to 2027-01-31 (+23.5%) and $7.241B by 2028-01-31 (+21.8%). (Consensus-derived; 38 and 39 analysts — the deepest coverage in this file.) That is $2.43 billion of added revenue in two fiscal years, roughly the entire revenue base of FY2023. The nearest check is the 2026-09-01 print: consensus revenue of $1,440.077M would be +23.2% year on year, which is a DECELERATION from the +25.6% just delivered. If the September quarter accelerates instead, the FY2027 line has upside; if it decelerates below 23%, the first leg of the price is already failing.
• Non-GAAP EPS more than doubles in two years, from $0.928 actual to $1.5567 by FY2028 — +67.7% cumulative, against +21.7% revenue growth last year and a 1.2% DECLINE in non-GAAP EPS in FY2026. (Consensus-derived, 27 analysts.) This is the most fragile assumption in the price, and it is fragile for a specific, disclosed reason: the 10-K states that a change in accounting estimate — extending sales-commission amortisation from four years to five, effective FY2027 — will "improve our fiscal year 2027 income (loss) from operations by $85.0 million to $95.0 million." Against consensus FY2027 non-GAAP net income of $1.280 billion that is roughly 7% of the number, and it is an accounting change, not operating leverage. It does not repeat as a growth contributor in FY2028.
• Stock-based compensation remains something the market is willing to exclude. (Our number, from the filings.) SBC was 22.8% of revenue in FY2026 and has been between 20.7% and 23.5% in every year in this file. The price requires the market to keep valuing this company on an earnings figure that treats $1.1 billion of annual compensation as a non-cost. That is a convention, not an arithmetic fact, and conventions re-price.
• The market keeps paying roughly 130x or better on the second-forward year at every date between now and then. (Our number.) At 100x FY2028E the stock is $160; at 170x it is $268. The multiple, not the earnings, is by far the widest source of variance in the outcome — and the multiple is the one thing we have no edge on.
• No adverse resolution of the Department of Justice and SEC information requests. (Filing-confirmed existence; our assumption on the outcome.) The 10-K states plainly: "The Company has received requests for information from the U.S. Department of Justice and the U.S. Securities and Exchange Commission relating to the Company's recognition of revenue and reporting of ARR for transactions with certain customers, the July 19 Incident and related matters." ARR is the metric this valuation is built on. The company has already revised fiscal 2025 and 2024 once, for a stock-compensation timing error found in the fourth quarter of FY2026. The price assumes both matters close quietly.

6b. The return bridge (why the multiple moves)

Expected return over the next twelve months decomposes as: EPS growth (+26.5%, from FY2027E $1.2309 to FY2028E $1.5567) + multiple drift (COMPRESSION, from 135.7x to roughly 105x on the respective forward year, −23%) + shareholder yield (+0.33%, less roughly 2% of dilution) ≈ −1% to −5% before any change in estimates.

Our base case assumes multiple COMPRESSION, and we should be explicit that this is not a rolldown artefact. Today the market pays 135.7x on the second-forward fiscal year. Our base assumes it pays roughly 105x on the then-second-forward year, which is a genuine de-rating of about 23%, not the mechanical effect of the calendar advancing.

We assume it for three reasons, stated so they can be argued with. First, growth decelerates on the consensus path itself: revenue +23.5%, +21.8%, +21.8%, then +15.8%; EPS +32.6%, +26.5%, +29.0%, then +14.2%. A multiple appropriate to 30% earnings growth is not appropriate to 14%, and the market normally begins discounting the deceleration a year ahead. Second, the FY2027 earnings step contains roughly 7% of accounting benefit that does not repeat. Third, and most importantly, at 135.7x the multiple is high enough that it can compress substantially without the company doing anything wrong at all — that is what a multiple this size means.

Almost none of the expected return over our horizon is delivered by the shareholder yield, and the dilution runs the other way. This is the opposite of a compounder setup. If the multiple simply HELD at 135.7x on FY2028E the price would be $215 (+1.8%) — i.e., today's price already IS today's multiple applied to earnings two years out. If it compressed to 100x, the price is $160 (−24%). If it expanded to 170x, $268 (+27%). The entire spread between a good outcome and a bad one sits in a multiple we do not control and have no edge on, and that is the honest reason this is a Watch.

The bull case at $269 requires no re-rating at all — it holds today's 135.7x and applies it to the FY2028 consensus high of $1.9533, then adds the net cash per share. That is a deliberately conservative construction: the bull case is an earnings beat, not a re-rating. The bear case at $105 is the mirror image: a de-rating to 70x applied to the FY2028 consensus low of $1.44762 — and it is independently corroborated by the street's own low price target of $103.25.

6c. Variant perception (where we differ, what would surprise)

• We differ from the price, and we do NOT differ from the street — which is itself the most unusual fact on this name. The vendor grades block shows 50 buy, 14 hold, 2 sell across 66 analysts with a consensus label of "Buy". The vendor pt block shows a consensus target of $185.33, a median of $187.50, and a high of only $237.00 — against a market price of $211.22. The sell side is rated Buy and priced Sell. Our base of $167 is 9.9% below their $185.33 and 20.9% below spot. When 66 analysts collectively think a stock is 12% too expensive and it makes an all-time high anyway on 0.64x average volume, the marginal buyer is momentum, not research. We are stating that plainly rather than manufacturing a differentiated view.
• We differ on the FY2027 margin story, and this is our sharpest genuine variant. The consensus path has non-GAAP EPS rising 32.6% in FY2027 after falling 1.2% in FY2026. The 10-K discloses that $85-95 million of that improvement is a change in the amortisation period for sales commissions — an accounting estimate change, effective FY2027, worth roughly 7% of consensus non-GAAP net income and non-repeating. We have seen no commentary anywhere in this file that separates it out. Watchable number: the FY2027 income-from-operations bridge in the FY2027 10-K, and any quarter where the commission amortisation line moves without a corresponding change in bookings.
• We differ on how to read the knowledge base's own best call, and this is where the KB is most useful. The lane's highest-conviction bullish claims were struck on 2026-03-03, the day after the FY2026 fourth-quarter print, and describe buying a "washout" at approximately $355 pre-split — which, adjusted for the 4-for-1 split effective 2026-07-02, is approximately $88.75, against a 52-week closing low in this file of $87.56. That claim was close to perfectly timed and the stock is up roughly 138% since. But the SAME channel, on the SAME date, carries the counterweight: "at 20x sales it must sprint to $10-20B revenue; if it stalls near $7B the stock gets murdered" and "at 20x sales it has no margin of safety." At $88.75 split-adjusted, the market capitalisation was roughly $90 billion on $4.8 billion of revenue — about 19x. Today it is $215 billion on the same revenue base: 44.7x. The knowledge base's own bull named the valuation ceiling and the price has since gone more than twice through it. That is the single most decision-relevant thing in the KB lane and it argues for Watch, not Buy.
• Positive surprise that would force a re-rate: September-quarter revenue materially above the $1,440.077M consensus with the year-on-year growth rate ACCELERATING past 25.6% rather than decelerating to 23.2%; net new ARR above $300 million in a seasonally weak quarter; remaining performance obligations recovering above $9.0 billion; or GAAP operating income turning positive without help from investment gains. Any two of those together would validate both legs of 6a at once.
• Negative surprise that would break the thesis: a September guide implying FY2027 revenue below roughly $5.85 billion; net new ARR below the $255.8 million just delivered; a second sequential decline in remaining performance obligations; an escalation of the DOJ or SEC matters from "requests for information" to anything formal; or an adverse development in the Delta litigation, which survived a motion to dismiss in part on 2025-05-16 and is in discovery. Any of these hits a stock at 135.7x forward earnings with negative GAAP income and no valuation floor.

Synthos fair values

All three anchors are multiples of the FY2028 consensus non-GAAP EPS distribution (mean $1.55669, low $1.44762, high $1.9533, 27 analysts — the best-covered EPS line in the file), plus the net cash of $3.66 per share, and cross-checked against FY2027E of $1.23093 and against EV/sales.

• Bear ~$105 — 70x the FY2028 consensus LOW of $1.44762, plus $3.66 of net cash. Cross-check: 85.3x FY2027E; enterprise value of $103.2B on FY2028E revenue of $7.241B is 14.3x sales. Independently corroborated by the street's own LOW price target of $103.25. The scenario: growth decelerates back through 20%, the AI-disruption argument gains traction, the accounting-driven part of the FY2027 margin step is seen through, and the multiple normalises toward the rest of large-cap software. −50.3%.
• Base ~$167 — 105x the FY2028 consensus MEAN of $1.55669, plus $3.66 of net cash. Cross-check: 135.7x FY2027E; enterprise value of $166.3B on FY2028E revenue of $7.241B is 23.0x sales, and on FY2027E revenue 28.0x. Sensitivity, stated openly: 90x gives $144; 120x gives $190; and 136x — today's multiple, held — gives $215, which is the price. The entire answer lives in a 90-136x band and we claim no greater precision than that. The scenario: consensus revenue lands roughly as modelled, the earnings step happens but is partly accounting, and the multiple de-rates as growth moves from the high 20s toward the mid teens. −20.9%.
• Bull ~$269 — today's 135.7x forward multiple, HELD, applied to the FY2028 consensus HIGH of $1.9533, plus net cash. Cross-check: 218.6x FY2027E; 36.6x FY2028E sales. This bull case requires no re-rating, only an earnings beat — the module attach rate keeps compounding, the four acquisitions convert into ARR faster than modelled, the enterprise AI-security hardening cycle the knowledge base describes actually arrives, and net new ARR breaks $1.3 billion for the year. +27.4%.

Base is 20.9% BELOW spot; asymmetry roughly 0.55:1 (50.3% down, 27.4% up). That ratio is unfavourable, and it is unfavourable for the plainest possible reason: the stock closed today at the highest price in the whole six-year series in this file. There is no such thing as a margin of safety at an all-time high on 135.7x forward earnings, and we are not going to invent one.

7. Knowledge base — 33 claims, genuinely about this company, and its own bull named the ceiling

Raw hits: 33. Entity matches: 32. Text-only matches: 1. Used: 19. Discarded: 10. Excluded from the conviction pool as company-management voice: 4.

The primary search covered CRWD and CrowdStrike across all 51,928 distilled claims, returning 32 entity matches and 1 text-only match. Supplementary sweeps were run on cybersecurity (1 entity, 107 text-only), endpoint (0 / 44), Falcon (4 / 7), SIEM (0 / 8), XDR (0 / 0), George Kurtz (0 / 0 — the founder-CEO's name appears nowhere in 51,928 claims), Palo Alto (8 / 4), SentinelOne (2 / 0), Zscaler (0 / 1), Fortinet (0 / 1), Okta (1 / 0), security software (0 / 4), cloud security (0 / 3), identity security (0 / 0) and a deliberate homograph sweep on crowd (29 / 281).

The homograph triage, because the brief warned about it and the warning was correct. The crowd sweep's 29 entity matches are all substring hits on "CrowdStrike" itself and are already counted; its 281 text-only matches are discarded in full as collisions on "crowded trade", "crowding out" and ordinary uses of the word. The Falcon sweep is worse and is instructive: all four entity matches are collisions — the Falcon open-source language model (twice), FalconX the crypto brokerage, and one more model reference — and six of the seven text-only matches are SpaceX's Falcon 9 and Falcon Heavy rockets. Only one Falcon hit is CrowdStrike's Falcon platform, and it was already captured by the CrowdStrike sweep. Not one Falcon-sourced claim added anything.

What was discarded, and why

• Four near-identical duplicates of one 2026-01-05 claim about enterprise platforms being accelerated by a chip vendor and adopting agentic interfaces — the same claim distilled five times. One retained.
• Three duplicates within the 2026-03-03 cluster — three unique claims appearing twice each. Three retained.
• One claim discarded as a product/partnership claim rather than an equity claim (2025-10-28: "CrowdStrike is a leading cybersecurity company and Nvidia enterprise-platform partner" — that is a fact about a partnership, not a view on the security).
• One claim discarded as a collateral entity tag (2026-03-08: a claim about AI-enabled state cyberattacks breaking RSA cryptography and Bitcoin becoming the essential protected sandbox. Its entities are Bitcoin and CrowdStrike, but its subject is Bitcoin. It is a crypto claim, not a CRWD claim).
• One claim discarded on its own quarantine flag (2023-07-03, marked QUARANTINED: NOT_VISSER — an attribution failure the system has already caught).
• 281 "crowd" text-only matches and the entirety of the cybersecurity/endpoint text-only sets discarded because none of them names CrowdStrike; a claim about the security category is not a claim about this security.

The attribution problem on this name, stated as a finding

Only 2 of the 32 entity matches carry a named speaker. The other 30 are attributed to a CHANNEL slug only — the weakest form of sourcing the 4-lane attribution policy admits, because a channel tells you where a claim was said, not who said it or what their record is.

And one channel slug is demonstrably wrong. Four claims dated 2022-07-13 carry the channel slug eli_lilly_ceo — a pharmaceutical-CEO channel — while their content is unambiguously CrowdStrike's own management describing CrowdStrike's own business in the first person: "still very early in our innings", "the Salesforce of security", gross margin "from ~35% to ~80% across 22 modules". These are company-management claims filed under an unrelated channel. Under the 4-lane policy they are excluded from the conviction pool entirely and quoted separately at half weight. They are listed below as management voice, and the mislabelling is logged as data-integrity finding 12.

The lane, by cohort

Cohort A — the 2023 structural thesis (6 claims, channel business_breakdowns, no named speaker, 2023-05-23). These are three years old and their numbers are stale, but their MECHANISMS are exactly what the FY2026 filings still describe:

> "Cloud-native endpoint leader growing 60% at $1.9B ARR, taking share from legacy AV; only Snowflake has matched this growth at scale." — bullish, conviction 85

> "Module expansion (10→22) is a durable growth engine; 20% of ARR is now non-endpoint and growing twice as fast, average customer uses ~5 modules." — bullish, conviction 80

> "Endpoint TAM is expanding via higher ASPs (legacy $1 vs CrowdStrike $16+), managed services, and agents on cloud servers—not just more devices." — bullish, conviction 70

> "Cross-customer threat-graph correlation across 18,000 customers is a data moat competitors like Microsoft cannot replicate within a single org." — bullish, conviction 75

> "Microsoft (bundled E5 ATP) and SentinelOne are CrowdStrike's most competitive rivals; Microsoft is 'good enough' free option, second-fastest growing." — neutral, conviction 65

> "Using an elite FBI-led incident-response team as proof-of-concept and marketing engine was CrowdStrike's key GTM differentiator versus rivals." — bullish, conviction 70

These are the only named-competitor statements available anywhere in this dive, because the 10-K names none. They are also the correct reference point for how much the growth rate has changed: the KB describes a 60% grower at $1.9B of ARR. The filings show a 24% grower at $5.3B. That is not a criticism — it is the arithmetic of scale — but a reader carrying a 2023 mental model of this company is carrying a model that no longer applies.

Cohort B — the 2026 selloff-and-recovery cluster (7 used claims, channel compound_and_friends, no named speaker except where noted). This is the most decision-relevant material in the lane and it is internally two-sided:

> 2026-01-06 · bullish · conviction 55: "Cybersecurity software is exempt from the enterprise-software selloff hitting workflow-automation names."

> 2026-02-03 · bullish · conviction 65: "Long-term CrowdStrike holder; it crushed earnings yet got dragged down $150 with the indiscriminate software/cyber selloff."

> 2026-02-24 · bullish · conviction 70: "Bought CrowdStrike on the AI-disruption selloff; market conflates [a frontier AI developer]'s bug-finder with incident response — CrowdStrike's business isn't actually threatened." [AI lab name redacted per house rule]

> 2026-03-03 · bullish · conviction 85: "Best-of-breed cyber name; $100B cap too small, clear path to a much bigger company as AI workloads need securing."

> 2026-03-03 · bullish · conviction 70: "Textbook Q beat; $5.25B ARR run-rate up 24%, fastest pure-play cyber to reach scale; bought the ~$355 washout."

> 2026-03-03 · NEUTRAL · conviction 60: "At 20x sales it must sprint to $10-20B revenue; if it stalls near $7B the stock gets murdered." / "Must sprint from $5B to $10-20B revenue or stock gets murdered; at 20x sales it has no margin of safety."

> 2026-04-10 · BEARISH · conviction 65: "Software is in a bloodbath on AI-disruption fear; market ignores record earnings/estimates and prices a much lower terminal value years out." (a basket claim listing CrowdStrike among eight software names)

> 2026-04-24 · bullish · conviction 60 · SPEAKER: Adam Parker: "In software buy the expensive names (security, Cadence, Synopsys, CrowdStrike, Palo Alto) not cheap ones — cheap means higher disruption risk; fast-growers lead the rebound."

The 2026-03-03 pair is the most important thing in this section and we treat it as one integrated call, because it was made on one day by one source. The bullish half bought a washout at approximately $355 pre-split — roughly $88.75 split-adjusted, against a 52-week closing low of $87.56, which is as close to the bottom as it is possible to get. The $5.25B ARR figure quoted matches the 8-K of 2026-04-21 exactly, so the claim is factually anchored. And the same source, the same day, said the ceiling is 20x sales and named $7 billion of revenue as the level at which the stock "gets murdered." Consensus has revenue reaching $7.241 billion in FY2028. The price is now 44.7x FY2026 revenue. We take both halves seriously and the second half is the one that binds today.

Cohort C — the AI-disruption debate (2 used claims, opposite signs).

> 2026-02-28 · BEARISH · conviction 70 · channel all_in · no named speaker: "Each [frontier AI developer] product release (legal, code security, COBOL modernization) tanked the incumbents it targets — IBM fell 13% on COBOL news." (entities include CrowdStrike and Okta) [AI lab name redacted per house rule]

> 2026-05-01 · bullish · conviction 52 · channel all_in · no named speaker: "AI cyber capability [frontier AI model names redacted] drives a big one-time hardening upgrade cycle — an incredible revenue opportunity for CrowdStrike and Palo Alto Networks, with a line out the door for the product."

These two claims are from the same channel, nine weeks apart, and they point in opposite directions. We report that rather than picking the one that suits the verdict. The bearish claim is a basket claim in which CrowdStrike is one of seven names and is not the subject; the bullish one names CrowdStrike as a direct beneficiary. Neither settles the question, and the fact that the same channel argued both sides within a quarter is itself the finding: the AI-disruption question is genuinely open, and a 135.7x multiple is not a price at which an open question is comfortable.

Cohort D — the one named, role-tagged independent voice.

> 2026-02-13 · bullish · conviction 75 · channel anthony_pompliano_show · SPEAKER: Dan Ives · role: independent: "The cybersecurity selloff (CrowdStrike, Palo Alto, Zscaler) is dislocated from fundamentals; he'll take the opposite side of the bet that [a frontier AI developer]'s tools unseat them." [AI lab name redacted per house rule]

Two weighting notes, both disclosed. The house rule that the show host's own voice is weighted low does not apply here — this is a GUEST claim, attribution: guest, and guests are first class. But the speaker is a persistently bullish technology voice, which is a promoter-class frame, and the claim is weighted accordingly. It was, for what it is worth, correct: it was made in the depth of the February selloff and the stock is far higher.

Cohort E — the basket claims we retained but discounted. One 2026-01-05 claim (of five near-duplicates) about enterprise platforms being accelerated by a chip vendor and adopting agentic interfaces, listing CrowdStrike alongside four other names. It is a claim about a hardware vendor's ecosystem, made on that vendor's own channel — promoter-class by construction — and it assigns no view to any constituent. Retained as sector colour only.

Cohort F — company-management voice, EXCLUDED from the conviction pool, quoted at half weight (4 claims, dated 2022-07-13, filed under a mislabelled channel).

> "Delivering security from the cloud via a single lightweight agent — 'the Salesforce of security' — beats legacy multi-agent on-prem architectures."

> "Once data is collected via one agent, each added module is near-pure profit; drove gross margin from ~35% to ~80% across 22 modules."

> "'No compression algorithm for experience' — running a profitable public cloud at scale requires hard-won tuning that's a real barrier to entry."

> "TAM is a percentage of total cloud spend — a massive market; CrowdStrike is 'still very early in our innings.'"

These describe the platform economics accurately — the FY2026 gross margin of 74.7% and the 95% subscription mix corroborate them — but they are the company describing itself and they carry no independent information about the security.

The one text-only match, from business_breakdowns 2023-05-23, is bearish on legacy antivirus rather than bullish on CrowdStrike, and mentions it only as the beneficiary: "Legacy AV (60% of market) is being pillaged; Broadcom's acquisition of Symantec abandoned most customers, ceding share to CrowdStrike." Retained as supporting colour.

Conclusion, stated without inflation. The Synthos knowledge base DOES have a view on CrowdStrike, and it is the richest single-name lane in this batch: 32 claims genuinely naming the company, spanning 2022 to 2026, covering platform economics, competition, the module attach mechanism, the AI-disruption debate on both sides, and a well-timed purchase into the 2026 washout. Net stance is clearly positive — 25 of the 32 are bullish, 2 bearish, 5 neutral. But three things stop it from lifting the verdict. First, 30 of 32 have no named speaker, so we cannot weight by record. Second, four are the company's own management on a mislabelled channel. Third, and decisively, the lane's own highest-conviction bull bought at roughly a quarter of today's price and named 20x sales as the point at which there is no margin of safety — and you now pay 44.7x. The conviction rating is Medium-high on direction, low on timing, breadth 6, and the KB's own valuation warning is a direct input to the Watch verdict.

8. Data integrity — what we rejected from the vendor file and why

Fourteen findings, plus five defect classes tested and found CLEAN. CRWD's vendor file is unusually good on the structural fields — segments, geography, share count and enterprise value all reconcile to the filings — and unusually bad on the derived earnings and cash-flow metrics, which is exactly what you would expect from a company whose GAAP earnings are negative.

1. The 4-for-1 stock split is real, and the vendor handled it correctly — verified to the dollar. A $211.22 price on a $215,076,725,122 market capitalisation implies 1,018,259,280 shares. The 10-Q cover page states "As of May 28, 2026, the number of shares of the registrant's common stock outstanding was 254,564,820" — pre-split. 254,564,820 × 4 = 1,018,259,280, and 1,018,259,280 × $211.22 = $215,076,725,122, which matches the vendor's reported market capitalisation EXACTLY, to the dollar. The 8-K of 2026-06-03 records the board's approval of "a four-for-one stock split of the Company's outstanding shares of Class A common stock, to be effected in the form of a stock dividend", record date 2026-06-25, distribution after the close on 2026-07-01, split-adjusted trading from 2026-07-02. Independent third check: the 8-K of 2026-06-22 records 254,536,521 shares outstanding at the annual-meeting record date of 2026-04-24; ×4 = 1,018,146,084, within 0.011% of the count above. Vendor and filing agree. Share-class check also clean: the 10-K states all Class B converted to Class A on 2024-12-11, so there is one class. This is stated because the price level and the recent split demanded the check, and the check passed.

2. bal_a[0] is one quarter STALE and misses $1.06 billion of cash movement — REJECTED in favour of the 10-Q. The vendor's most recent balance sheet is 2026-01-31. The 10-Q filed 2026-06-04 covers 2026-04-30, in which CrowdStrike spent $881.4 million on acquisitions (Seraphic closed 2026-02-03 at $327.4M net; SGNL.AI closed 2026-02-20 at $627.9M net — both disclosed as subsequent events in the 10-K) and $175.6 million on share repurchases. The vendor shows net cash of $4,410.0M at 2026-01-31; the 10-Q shows cash of "$4.6 billion" at 2026-04-30 against roughly $849M of debt, i.e. approximately $3,704M of net cash; we use the 10-Q. Note that km_ttm.enterpriseValueTTM DOES use the newer balance sheet (implied net cash $3,731M), so the enterprise value is current even though the balance-sheet block is not. That inconsistency between two blocks of the same file is itself worth flagging.

3. No financing event post-dates the statements — the 8-Ks were scanned and are CLEAN on this test. The three 8-Ks in the archive are 2026-04-21 (a performance-stock-unit award to the president), 2026-06-03 (first-quarter results and the split announcement) and 2026-06-22 (annual meeting results and a charter amendment limiting officer liability). None involves debt, equity issuance or any financing. The only debt in the company is the $750.0 million of 3.00% senior notes issued 2021-01-20 and due in fiscal 2030, unchanged. Independently, the secured revolving credit facility EXPIRED in January 2026 and has not been replaced — a disclosed reduction in contingent liquidity that appears in no vendor field.

4. Every EBITDA-based metric in the file is meaningless and is REJECTED in full. ratios_ttm.ebitdaMarginTTM of 6.57% implies trailing EBITDA of $334.7M and produces evToEBITDATTM and enterpriseValueMultipleTTM of 631.4x and netDebtToEBITDATTM of −11.15x. CrowdStrike's operating income is negative; a positive EBITDA can only be produced by adding back interest income, which is not an operating item. Computed properly, FY2026 EBITDA is operating income of −$293.3M plus D&A of $281.5M = −$11.8M. No EV/EBITDA figure appears anywhere in this dive.

5. est.ebitdaAvg and est.ebitAvg are internally impossible in ALL EIGHT years — NOT USED, per the data contract. In every year netIncomeAvg exceeds ebitdaAvg: FY2027 $1,279.8M vs $927.0M; FY2028 $1,639.5M vs $1,129.1M; FY2029 $2,143.9M vs $1,375.5M; FY2030 $2,368.8M vs $1,593.5M; FY2031 $2,694.9M vs $1,868.5M; and FY2026 $974.1M vs $749.1M. Net income cannot exceed EBITDA in the presence of depreciation, interest expense and tax, even with $190M of interest income. The only coherent reading is that the net-income and EPS rows are NON-GAAP while the EBIT/EBITDA rows are on a different basis, mixed into one row without a label. All forward valuation runs on epsAvg and revenueAvg.

6. est is on a NON-GAAP basis and never says so — the single most important labelling defect on this name. est FY2026 epsAvg is $0.92842 against actual FY2026 GAAP diluted EPS of −$0.160. The four earn_cal non-GAAP actuals for FY2026 sum to $0.930, confirming the basis to within 0.17%. The gap is $1.088 per share, of which stock-based compensation of $1,096.7M ÷ 1,032.5M diluted shares = $1.062 accounts for approximately 97.6%. Every forward multiple in this dive is labelled as struck on consensus non-GAAP EPS, and the trailing GAAP P/E (vendor: −4,446.7x) is rejected.

7. capitalExpenditure is computed on an INCONSISTENT basis between FY2026 and the two prior years — free cash flow REJECTED and rebuilt. For FY2025 and FY2024 the vendor's capex includes capitalised internal-use software ($313.821M = $254.9M + $59.0M; $237.112M = $176.5M + $60.6M). For FY2026 it does not — $302.108M is identical to investmentsInPropertyPlantAndEquipment and excludes the $68.8 million the 10-K discloses. The effect is fake acceleration: reported FCF growth of +22.7% versus +16.2% on a consistent basis. The vendor reports FY2026 free cash flow of $1,310.2M; the filing supports $1,241.4M; we use $1,241.4M and FCF margin of 25.8%, not 27.2%.

8. The FY2025 income statement and cash flow are the PRE-REVISION versions and are inconsistent with the post-revision FY2026 — the year-on-year SBC and net-loss comparisons are REJECTED. The 10-K discloses at Note 16: "during the fourth quarter of the fiscal year ended January 31, 2026, the Company identified an immaterial error related to the timing of recognition of stock-based compensation expense in prior periods associated with certain awards granted in the fiscal years ended January 31, 2022 and 2023", with a cumulative $29.3 million adjusted against accumulated deficit at 2023-02-01 and fiscal 2025 and 2024 revised. The 10-K's own MD&A states FY2025 operating cash flow "resulted from net loss of $12.6 million" with "$861.4 million in stock-based compensation expense." The vendor's cf_a FY2025 reports a net loss of $16.596M and SBC of $865.421M. The two differences — $4.0 million on each line — match exactly, which confirms the vendor carries the superseded figures. For FY2026 the vendor matches the filing precisely (net loss $161.165M vs "$161.2 million"; SBC $1,096.679M vs "$1.1 billion"; D&A $281.451M vs "$250.2 million... and $31.2 million of amortization of intangibles assets"). Filing wins on FY2025; we say so rather than smoothing the comparison.

9. inc_a FY2026 uses a DILUTED share count in a loss year, which GAAP forbids, and both annual weighted averages are copies of the fourth quarter's — flagged, not used. The FY2026 row shows basic 1,010,264,000 and diluted 1,032,532,000. In a loss year diluted must equal basic (dilutive securities are anti-dilutive), and both figures are byte-identical to the Q4 FY2026 quarterly row — a full-year weighted average cannot equal the fourth quarter's. The reported EPS of −$0.16 is computed on the basic figure and is correct. We use the 10-Q cover-page count × 4 for market capitalisation and the FY2026 diluted count only for the per-share SBC reconciliation, where the direction of the error is conservative. Separately, the FY2025 row reports a suspiciously round 979,000,000 for both basic and diluted, against a Q4 FY2025 basic count of 987,732,000.

10. inc_q Q4 FY2026 epsDiluted is arithmetically wrong — corrected. The vendor reports $0.0375. Net income of $59.377M ÷ 1,032.532M diluted shares = $0.0575. The correction is independently confirmed by the annual reconciliation: the four quarterly EPS figures as printed sum to −$0.185 against a reported annual −$0.160, while using $0.0575 gives −$0.165, essentially the annual figure. We use $0.0575 and flag the vendor value.

11. Trailing GAAP-derived ratios are meaningless with a near-zero denominator and are REJECTED as lenses. priceToEarningsRatioTTM −4,446.7x; incomeQualityTTM −433.4; effectiveTaxRateTTM 3.14 (i.e. 314%); interestCoverageRatioTTM −7.27x (the company is a net interest RECEIVER — $194.97M of interest income against $28.02M of interest expense in FY2026 — so a negative coverage ratio conveys the opposite of the truth); taxBurdenTTM −12.50; netIncomePerEBTTTM −12.50; priceToBookRatioTTM and priceToFairValueTTM both 46.26x; returnOnEquityTTM −0.58%. The vendor's own composite rating is C− with an overall score of 1 out of 5, driven by 1-out-of-5 scores on return on equity, return on assets, price/earnings and price/book. That is a rating artefact produced by negative GAAP earnings, not a judgement about the business, and it is rejected. returnOnInvestedCapitalTTM of 5.96% is the one honest return measure in the block and is what we quote.

12. The knowledge base contains four CrowdStrike-management claims filed under an unrelated pharmaceutical-CEO channel slug — an attribution failure, handled rather than ignored. Four claims dated 2022-07-13 carry channel eli_lilly_ceo and content that is unmistakably CrowdStrike management speaking in the first person about CrowdStrike. They are excluded from the conviction pool and quoted separately at half weight in Section 7. A fifth claim carries the system's own QUARANTINED: NOT_VISSER flag and is discarded. These are reported because attribution-by-channel is precisely the failure mode the 4-lane policy exists to catch, and this name is a live instance of it.

13. quote.yearHigh/yearLow disagree with tech.hi52/lo52 — we use tech and explain the divergence rather than calling it a defect. The quote block reports a 52-week high of $217.50 and low of $85.68; the computed tech block reports $211.22 and $87.56. The divergence is 3.0% and 2.2% and it has an obvious cause: tech is computed from the six-year CLOSE series that also produces the moving averages and drawdown, while quote.yearHigh/yearLow are INTRADAY extremes. An intraday high above the highest close is exactly what you expect. We use the tech figures throughout for internal consistency, and we state that the stock has traded as high as $217.50 intraday at some point in the last year — 3.0% above today's close.

14. tech.hi52 equals today's close of $211.22 AND max_dd_from_peak reads 0.0% — this is NOT a defect, it is the finding. The 252-day maximum close and the six-year peak coincide with the current price. Per the data contract, coincidence of hi52 and the peak when a multi-year high is recent is legitimate and must not be reported as an error. What it means is stated plainly instead: CrowdStrike closed 2026-08-04 at the highest price in the entire six-year price series held in this file. One caveat we state rather than hide: the series is six years and the company listed on 2019-06-12, so the file cannot see the first roughly fourteen months of trading history. "Highest close in six years" is what the data supports; "all-time high" is what it probably is, and we label the difference.

Defect classes tested and found CLEAN — each is a real finding and is stated as one:

• Currency mixing (class 5): CLEAN. inc_a[0].reportedCurrency, bal_a[0].reportedCurrency and profile.currency are all USD, and the quote is USD on NASDAQ. No multiple in this dive crosses a currency boundary.
• Net debt omitting short-term investments (class 2): CLEAN, and verified against the filing. shortTermInvestments is $0 because it genuinely is — the 10-K describes the entire $5.2 billion of liquidity as "cash and cash equivalents, which mainly consists of cash on hand and highly liquid investments in money market funds, U.S. Treasury bills, and time deposits." The FY2024 position of $99.6M ran off, as the FY2025 cash-flow statement confirms with "proceeds from maturities of short-term investments of $97.3 million." The rebuilt enterprise value of $211.37B differs from the vendor's $211.35B by 0.01%.
• seg_prod/seg_geo (class 8): CLEAN, and verified line by line. Both tables sum to $4,812.005M for FY2026 — exactly the reported revenue — and to reported revenue in every one of the seven years supplied. The 10-K confirms the split verbatim ("$858.4 million, or 22%" total; "$803.2 million, or 21%" subscription; "$55.2 million, or 29%" professional services) and the 95%/5% mix. The only issue is the generic XBRL label "Subscription And Circulation" where the company says "Subscription"; corrected in presentation.
• Split-corruption in the estimate block (class 9 sub-check): CLEAN. All eight years of epsAvg are consistently 4-for-1 split-adjusted ($0.7389 for FY2024 through $2.61 for FY2031), with no 4x discontinuity anywhere.
• Non-equity tripwire (class 13): CLEAN. CRWD is common stock, NASDAQ Global Select, CUSIP 22788C105, ISIN US22788C1053, single class since 2024-12-11. The price of $211.22 is not par-like; beta is 1.237, far above the sub-0.3 of a fixed-income-like instrument; there is no dividend at all and the 10-K forecloses one, so there is no fixed coupon; volume was 7.73 million shares (roughly $1.6 billion of turnover); and the 52-week band of $87.56 to $211.22 is a 141% range, the opposite of the narrow band a bond-like instrument shows. This is common equity.

Also correctly reported and independently confirmed: profile.fullTimeEmployees of 10,698 matches the 10-K exactly; stockBasedCompensationToRevenueTTM of 22.39% reconciles to the TTM SBC and revenue; researchAndDevelopementToRevenueTTM of 28.64% reconciles to FY2026's 28.8%; dividendYieldTTM of 0 and dividendPerShareTTM of 0 are correct and filing-confirmed; treasuryStock of $0 is correct because the 10-Q states repurchased shares are "subsequently retired"; inventory of $0 is correct for a software company; and cf_a.commonStockRepurchased of $0 is correct for FY2023-FY2026 per Item 5(c) of the 10-K.

Known extraction limits on this archive, stated rather than estimated around. Numeric tables are stripped from the extracted filing text throughout, which cost us four figures we would have used: the dollar-based net retention rate (the 10-K and 10-Q both discuss the metric at length; its VALUE sits in a stripped table), the July 19 Incident expense rollforward, the Strategic Plan charge-by-line-item table, and the revision tables showing the exact restated FY2025 and FY2024 figures. We report only prose-sourced numbers and we do not interpolate the missing ones. Separately, the 8-K of 2026-06-03 is present as the cover document only: it states that the first-quarter results press release "is furnished herewith as Exhibit 99.1", and Exhibit 99.1 was NOT extracted. The Q1 FY2027 non-GAAP reconciliation and any forward guidance are therefore absent from this archive. The 8-K of 2026-06-22 likewise has its voting-results tables stripped. By contrast the 8-K of 2026-04-21 is substantive narrative prose and is one of the most useful documents in the file — it is the source of the ARR history and the $20 billion ARR ambition.

9. Technicals

• Price $211.22. At the 52-week closing high of $211.22 and at the six-year peak — pct_from_hi is 0.0% and max_dd_from_peak is 0.0%. +141.2% above the 52-week low of $87.56. Position within the annual range: the 100th percentile.
• Moving averages, both supportive and both far below: +15.3% above a 50-day average of $183.18, which is rising; +58.2% above a 200-day average of $133.50, which is rising. There is no moving average anywhere near the price, which means there is no technical support anywhere near the price either.
• RSI 53.6 — and this is the most interesting technical fact in the file. A stock at an all-time high after an 80% three-month run would normally print an RSI in the 70s or 80s. 53.6 says the last two weeks have been sideways-to-slightly-up, not a blow-off. MACD +3.97, positive.
• Relative performance, and it is extraordinary: 3-month +80.1% vs SPY +7.6% and QQQ +7.7% — a 72-point outperformance in a quarter; 6-month +92.5% vs SPY +11.1% and QQQ +15.7%; 12-month +89.1% vs SPY +24.3% and QQQ +30.8%. Note that the 3-month and 12-month returns are nearly identical (80.1% and 89.1%), which means essentially the ENTIRE twelve-month return was earned in the last three months. The knowledge base independently locates the starting point: a washout bought at roughly $88.75 split-adjusted in early March 2026.
• Sentiment, and it is contradictory: 50 buy, 14 hold, 2 sell across 66 analysts, consensus label "Buy" — but a consensus target of $185.33, which is 12.3% BELOW spot, a median of $187.50 (−11.2%), a high of only $237.00 (+12.2%) and a low of $103.25 (−51.1%). The single highest price target on the street is 12% above the current price. There is no analyst in this file whose target implies meaningful upside from here.

Today's move and what it does to the entry

CRWD closed 2026-08-04 at $211.22, up 4.29% or $8.68 from a previous close of $202.54. It opened at $205.80 — a 1.61% gap — traded a $203.50 to $212.64 range, and closed within 0.67% of the day's high. Volume was 7,733,353 shares against a 12,070,000 average — 0.64x.

The honest read on the day, and it differs from the semiconductor names in this batch. 2026-08-04 was a violent sector-wide semiconductor melt-up — MRVL +12.8%, INTC +10.9%, LRCX +7.9%, MU +7.6%, KLAC +7.0%, AMD +7.0%, AMAT +5.5%, ASML +4.2%. CrowdStrike is not a semiconductor company and its +4.29% is not that trade. No company-specific news for CRWD appears anywhere in this file for 2026-08-04; the last company event was the 2026-06-03 earnings release, 62 days earlier. The file contains no same-day price changes for the software peers (Palo Alto Networks, Fortinet, Cloudflare), so we cannot say whether this was a security-software move or a CRWD move, and we will not guess. What we can say is that it happened on 0.64x average volume, which is not the participation profile of a decisive breakout.

Here is the thing that matters about the entry, and it is not the 4.29%. Elsewhere in this batch a gap-up means paying up from a drawdown. Here there is no drawdown to pay up from. The stock is at the top of its six-year range, it has risen 80.1% in three months, and there is no price in the last twelve months at which it traded higher. A buyer today is not paying a one-day premium; a buyer today is paying the highest price anyone has ever paid.

And the valuation does not soften that. At $211.22 the stock is 42.2x trailing sales, 171.6x FY2027 consensus non-GAAP EPS and 135.7x FY2028, with negative GAAP earnings, and our base fair value of $167 is 20.9% below the closing price and 9.9% below a street target that is itself 12.3% below the market.

What we are therefore telling the reader, plainly: we are not asking you to buy at all today, and we are not asking you to short it either. The business is genuinely accelerating and the balance sheet is unbreakable; those are not the reasons to decline. The reason to decline is that every dollar of forward return has to come from a multiple that is already 135.7x, on earnings that only exist after adding back 22.8% of revenue. The setups that would change this are the falsifiers:

• A price near $165 to $175 — roughly 105-113x FY2028E and 134-142x FY2027E — which would put our base at or above the price and make the risk/reward defensible. That is 17-22% below today's close, and this stock has traversed more than that in a fortnight at least twice in the last six months.
• Or the 2026-09-01 print landing above $1,440.077M of revenue with the year-on-year growth rate ACCELERATING past 25.6%, and net new ARR above $300 million in a seasonally weak quarter. That would validate both legs of 6a at once and would make paying a premium multiple defensible, because the uncertainty being paid for would have been resolved.

Buying today means paying an all-time-high price, 28 days before a print that consensus itself models as a deceleration, for a base case that is 21% below the price. That is not a trade with an edge in it.

10. Insiders — eight transactions, one person, one week, and it is selling

Every one of the eight insider transactions in this file is George Kurtz — founder, President and Chief Executive Officer — selling, on 2026-07-30 and 2026-07-31, all filed 2026-08-03 under a single Form 4 accession.

Total: 10,328 shares for approximately $1.883 million, at prices from $178.72 to $185.29.

The reading, stated carefully in both directions.

Against the seller: these are open-market sales (transaction type S-Sale), not tax withholding. Unlike the F-InKind withholding transactions that dominate most insider tables, these represent an actual decision to convert stock into cash. And the chief executive sold at $178.72 to $185.29 in the last week of July, against a closing price of $211.22 on 2026-08-04 — 14% to 18% higher. He did not capture the last leg.

For the seller, and this is the larger part of it: the eight sales total 10,328 shares against a residual holding of 8,093,216 — he sold 0.127% of his stake. The retained position is worth $1.709 billion at today's close. The laddered price structure (eight tranches across 24 hours at rising prices in $1 increments) is the signature of a pre-arranged plan execution rather than a discretionary decision, though the file does not state whether a Rule 10b5-1 plan applies and we will not assert one.

The limitation of this table matters more than its content, and we state it. The vendor supplies the last EIGHT insider transactions only. All eight were consumed by one person over two days. This file therefore contains NO information about whether any other officer or director bought or sold at any point, and no information about George Kurtz's own activity before 2026-07-30. We cannot say whether this is an isolated trim or the eighth week of a programme. On the evidence available, the signal is: the founder-CEO sold roughly one-eighth of one percent of his stake into strength, and retains $1.7 billion of it. That is closer to neutral than to negative, and we report it as neutral with the caveat attached rather than dressing a 0.127% trim as a warning.

11. Verdict, kill-criteria and flip conditions

Watch.

This is the best operating story in this batch and the worst entry price.

What is genuinely excellent, and almost none of it is in dispute: revenue growth that has ACCELERATED for four consecutive quarters, from 19.8% to 25.6%; ending ARR of $5.5 billion at 2026-04-30 with net new ARR up 32% year on year in a seasonally weak quarter; a 74.7% gross margin that recovered from a post-incident trough of 73.5% to 76.1%; GAAP operating margin from −11.30% to +1.06% in three quarters, with 950 basis points of operating-expense leverage behind it; 95% of revenue on subscription; $4.7 billion of pre-paid deferred revenue growing 27.5% against revenue at 21.7%; $8.8 billion of remaining performance obligations; net cash of $3.70 billion with the only debt being $750 million of 3.00% notes not due until fiscal 2030; $1.24 billion of corrected free cash flow at a 25.8% margin; and a knowledge-base lane that is the richest single-name coverage in this batch and clearly positive.

What we are declining to pay for: 42.2x trailing sales, 171.6x FY2027 consensus non-GAAP EPS, 135.7x FY2028, and 170x enterprise value to corrected free cash flow — on GAAP earnings that are NEGATIVE, with $1.097 billion of stock compensation (22.8% of revenue) accounting for 97.6% of the gap between the two bases. A base fair value of $167 that is 20.9% below spot and 9.9% below the street's $185.33 — which is itself 12.3% below the market price. A forward case that requires non-GAAP EPS to rise 67.7% in two years after FALLING 1.2% last year, of which roughly 7% of the FY2027 step is a disclosed change in commission-amortisation accounting rather than operating leverage. $4.3 billion of non-cancelable purchase commitments, in no vendor field, including a NEW $1.7 billion commitment entered after the last quarter end and running to fiscal 2037. Open Department of Justice and SEC information requests specifically about the recognition of revenue and the reporting of ARR — the metric this entire valuation is built on — from a company that has already revised two prior fiscal years for a stock-compensation timing error. And an entry at the highest close in the entire six-year series in this file.

The distinction that matters. We are not saying CrowdStrike is a bad business — it is one of the better ones we have looked at, and the recovery from the July 19 Incident is measurable and real. We are saying that the price now embeds all of it and more, that our fair value sits a fifth below the market, that the street's own target sits an eighth below the market, and that the knowledge base's own best-timed bull explicitly named 20x sales as the point at which there is no margin of safety — at which point you now pay 44.7x. In the Synthos frame, a name where our arithmetic AND the crowd's arithmetic both land below the price, on the day it makes an all-time high, is a Watch by construction.

Two things keep this at Watch rather than Avoid, and we name them so the boundary is legible. First, the operating acceleration is genuine and could continue — this is not a decaying business with a stretched multiple, it is a compounding business with a stretched multiple, and those resolve differently. Second, the knowledge-base lane is real, broad and positive, which is not something we could say about most names in this batch. If a reader owns CRWD, the honest advice is that there is no fundamental reason to sell and every valuation reason not to add. If a reader does not own it, wait.

Pre-registered KILL criteria — what would take this to Avoid:

• September-quarter revenue (2026-09-01) below roughly $1.40 billion, or a guide implying FY2027 revenue below $5.85 billion. That is the first leg of 6a failing, and 135.7x has nothing underneath it.
• Net new ARR below $255.8 million in the July quarter, or ending ARR growth falling below 22%. The ARR flow is the leading indicator and it has been accelerating.
• A second consecutive sequential decline in remaining performance obligations, taking the balance below roughly $8.5 billion. One decline is seasonality; two is a trend.
• Any escalation of the Department of Justice or SEC matters beyond "requests for information", or any restatement — particularly any touching ARR. The company has revised prior periods once already.
• An adverse development in the Delta litigation, which survived a motion to dismiss in part on 2025-05-16 and is in discovery, or a reversal on appeal at the Fifth Circuit of the airline class-action dismissal (the Fifth Circuit affirmed the dismissal on 2026-05-20, which is currently a positive).
• Stock-based compensation rising above 25% of revenue, which would mean the non-GAAP construct is being stretched rather than grown into.

Pre-registered FLIP conditions — what would take this to Buy — Tactical:

• Price below approximately $175 (roughly 112x FY2028E, 142x FY2027E, 24x FY2028E sales) with no deterioration in the ARR or growth picture. This is the single most likely path to a Buy and it is 17% below today's close. At $175 the base case is at parity and a further move toward $165 makes the risk/reward genuinely positive.
• The 2026-09-01 print beating $1,440.077M of revenue with the year-on-year growth rate accelerating past 25.6%, AND net new ARR above $300 million — both legs of 6a validated at once, at which point paying a premium multiple is defensible because the uncertainty has been resolved.
• GAAP operating income turning positive for two consecutive quarters without help from realised investment gains or discrete tax benefits. That would convert the non-GAAP debate from load-bearing to academic and is worth more to this thesis than any single revenue print.
• The company using the buyback aggressively. $1.3 billion of the $1.5 billion authorisation remains. Management declining to buy at $211 while having bought at lower prices would be informative in one direction; buying heavily here would be informative in the other.
• A resolution of the DOJ and SEC matters with no restatement, which would remove the one genuinely binary risk in the file.

Where CRWD fits in the Synthos Framework Portfolio. The cybersecurity / infrastructure-software sleeve, at 0% today with a 2-3% target on a fill near $175 or a confirmed September print. On the batch-overlap question: CRWD is the only pure-play security name in this batch and it does not overlap the wafer-fab-equipment complex (KLAC, LRCX, AMAT, ASML) or the memory names in any meaningful way — today's 4.29% was not the semiconductor trade. The relevant overlap is instead with any other high-multiple software exposure, because the 2026-04-10 knowledge-base claim in this lane is explicit that the whole category is being repriced on AI-disruption fear as a group: "Software is in a bloodbath on AI-disruption fear; market ignores record earnings/estimates and prices a much lower terminal value years out." A reader holding several 100x-forward-earnings software names is holding one macro bet on the durability of the software business model, expressed several ways. Within that bet, CrowdStrike is among the strongest expressions on fundamentals — accelerating, recurring, cash-generative, unlevered — and among the most expensive. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $211.22.

Single biggest risk: there is no valuation cushion under negative GAAP earnings. At 135.7x FY2028 consensus non-GAAP EPS, a 10% revenue disappointment does not produce a 10% price decline; it produces a de-rating on top of it, which is what the $105 bear case is — and that bear case is independently corroborated by the street's own low target of $103.25. The entire architecture — the 135.7x multiple, the 67.7% two-year EPS step, the $4.3 billion of purchase commitments already contracted, the four acquisitions bought for $1.26 billion of cash in fifteen months — rests on the growth rate staying above 20%. It has been above 20% for every quarter in this file. It has also decelerated from 66.0% in FY2022 to 21.7% in FY2026, and consensus itself has it at 15.8% by FY2030.

Most fragile assumption in the price: that non-GAAP EPS rises 67.7% in two fiscal years after falling 1.2% in the last one. Every other assumption is a revenue assumption, and revenue is observable quarter by quarter with a leading indicator (net new ARR) attached to it. This one is a margin assumption made against a company that has posted negative GAAP operating income in five of the last six fiscal years, that pays 22.8% of revenue in stock, and whose FY2027 operating-income improvement is disclosed as containing $85-95 million of accounting-estimate change. Operating leverage has genuinely arrived — 950 basis points of it in three quarters — so the assumption is not unreasonable. But if the FY2028 figure lands at $1.30 rather than $1.5567, then at 105x the price is $140, and at today's 135.7x it is $180. Both are below the price today.

Provenance & disclosures

• Traceability: 32 tagged knowledge-base claims name CrowdStrike by entity plus 1 text-only match — 33 raw, 19 used, 10 discarded, 4 excluded from the conviction pool as company-management voice (breadth 6, net conviction positive-high). Channels represented: business_breakdowns (6 claims, all 2023-05-23, the structural endpoint/module thesis, no named speaker); compound_and_friends (11 claims of which 8 unique, dated 2026-01-06 through 2026-04-24, including the 2026-03-03 washout purchase near ~$355 pre-split and the paired "20x sales, no margin of safety" warning; one claim carries speaker Adam Parker); all_in (2 claims, 2026-02-28 bearish and 2026-05-01 bullish, opposite sides of the AI-disruption question, no named speakers); jensen_huang (6 claims of which 5 are near-duplicates of one basket claim and 1 is a partnership statement — promoter-class by construction, discounted, 5 discarded); jordi_visser (1 claim, discarded as a Bitcoin claim with a collateral CrowdStrike entity tag); anthony_pompliano_show (1 claim, speaker Dan Ives, role independent, attribution guest — first-class as a guest claim, weighted for a persistently bullish frame); eli_lilly_ceo (4 claims dated 2022-07-13 that are demonstrably CrowdStrike management voice on a mislabelled channel — excluded from the conviction pool, quoted at half weight); and quarantine_misattributed (1 claim carrying the system's own NOT_VISSER flag — discarded). Only 2 of 32 entity matches carry a named speaker; the other 30 are channel-only attribution, the weakest sourcing the 4-lane policy admits, and this is stated as a finding rather than smoothed. Searches run: the primary sweep on CRWD and CrowdStrike (32 entity, 1 text-only); supplementary sweeps on cybersecurity (1/107), endpoint (0/44), Falcon (4/7), SIEM (0/8), XDR (0/0), George Kurtz (0/0), Palo Alto (8/4), SentinelOne (2/0), Zscaler (0/1), Fortinet (0/1), Okta (1/0), security software (0/4), cloud security (0/3), identity security (0/0); and a deliberate homograph sweep on crowd (29/281). The 281 "crowd" text-only matches are discarded in full as collisions on "crowded trade"/"crowding out"; ALL FOUR Falcon entity matches are collisions — the Falcon open-source language model, FalconX the crypto brokerage — and six of seven Falcon text-only matches are SpaceX's Falcon 9 and Falcon Heavy. Not one Falcon-sourced claim survived triage. All quotes are verbatim from the stored claim text except where an AI laboratory or model name has been redacted, which is marked in square brackets in every instance.
• Data as-of: fundamentals — income statement, balance sheet and cash flow for fiscal 2026 ended 2026-01-31 and the quarter ended 2026-04-30, cross-checked against the earnings-calendar entries; most recent filing-verified financials 2026-04-30, from the 10-Q filed 2026-06-04 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873601 = 2026-08-04T20:00:01Z ($211.22, +4.29%; 50-DMA $183.18; 200-DMA $133.50; RSI 53.6; MACD +3.97) · knowledge-base claims 2026-08-04. Fiscal/calendar note: CrowdStrike's fiscal year ends 31 JANUARY. "FY2026" means the twelve months to 2026-01-31 (revenue $4.812B, +21.7% on FY2025's $3.954B); "Q1 FY2027" means the quarter ended 2026-04-30; "FY2027E" means the twelve months to 2027-01-31. Every fiscal label follows the company's own convention. All figures come from the Synthos vendor data file for CRWD or from the SEC filings in the CRWD archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents and its gaps: 10-K filed 2026-03-05 (fiscal 2026, year ended 2026-01-31); 10-Q filed 2025-12-03 (October 2025 quarter); 10-Q filed 2026-06-04 (April 2026 quarter); 8-K filed 2026-04-21 (performance-stock-unit award to President Michael Sentonas — substantive narrative prose, and the source of the ARR history and the $20 billion ARR ambition); 8-K filed 2026-06-03 (first-quarter results and the four-for-one stock split); 8-K filed 2026-06-22 (annual meeting results and charter amendment). The 2026-06-03 8-K is present as the COVER DOCUMENT ONLY — it states the results press release "is furnished herewith as Exhibit 99.1", and Exhibit 99.1 was NOT extracted, so the Q1 FY2027 non-GAAP reconciliation and any forward guidance are absent from this archive. Numeric tables are stripped from all extracted filing text, which specifically cost us the dollar-based net retention rate value, the July 19 Incident expense rollforward, the Strategic Plan charge-by-line table and the exact restated FY2025/FY2024 revision figures. These are stated as gaps and are not estimated around.
• Where the filings contradicted or corrected the vendor (detailed in Section 8): the FY2026 free-cash-flow figure rebuilt from $1,310.2M to $1,241.4M because the vendor's FY2026 capital expenditure excludes the $68.8M of capitalised internal-use software that FY2025's and FY2024's include, turning +16.2% FCF growth into a reported +22.7%; the FY2025 income statement and cash flow identified as PRE-REVISION (vendor net loss $16.596M and SBC $865.421M against the 10-K's revised "net loss of $12.6 million" and "$861.4 million in stock-based compensation expense" — a $4.0M difference on each line that matches exactly), making the FY2025-to-FY2026 comparison a mixed-basis one; bal_a[0] at 2026-01-31 identified as one quarter stale, missing $881.4M of acquisition cash (Seraphic and SGNL) and $175.6M of buyback, with net cash rebuilt from $4,410.0M to approximately $3,704M using the 10-Q; every EBITDA-based ratio rejected in full (evToEBITDATTM 631.4x, netDebtToEBITDATTM −11.15x) because properly computed FY2026 EBITDA is negative $11.8M; all eight years of est.ebitdaAvg/ebitAvg rejected as arithmetically impossible against netIncomeAvg in the same row; the est block identified as NON-GAAP and unlabelled (FY2026 epsAvg $0.92842 against GAAP diluted EPS of −$0.160, confirmed against the four earn_cal actuals summing to $0.930); inc_a FY2026's diluted share count of 1,032,532,000 rejected for a loss year and both annual weighted averages identified as copies of the fourth quarter's; inc_q Q4 FY2026 epsDiluted corrected from $0.0375 to $0.0575; cf_a.commonStockRepurchased of $0 noted as correct through FY2026 but stale thereafter ($175.6M repurchased in Q1 FY2027); quote.yearHigh/yearLow ($217.50/$85.68) rejected in favour of tech.hi52/lo52 ($211.22/$87.56) with the intraday-versus-close cause explained. Where vendor and filing AGREED — worth recording: the share count (10-Q cover 254,564,820 at 2026-05-28 × 4 = 1,018,259,280, which × $211.22 gives $215,076,725,122, the vendor's market capitalisation to the dollar); the revenue split (subscription +$803.2M/+21%, professional services +$55.2M/+29%, total +$858.4M/+22%, 95%/5% mix — all verbatim in the 10-K); seg_prod and seg_geo both summing exactly to reported revenue in all seven years; employees at 10,698; the enterprise value (rebuilt $211.37B against the vendor's $211.35B, 0.01%); FY2026 net loss, SBC and D&A all matching the 10-K's MD&A; short-term investments of $0 confirmed as a real business fact rather than a missing field; and zero dividends, filing-confirmed and foreclosed.
• Basis note: analyst estimates are NON-GAAP and the vendor never labels them. Reported diluted EPS is GAAP and is NEGATIVE for FY2026 (−$0.160) against a non-GAAP figure of +$0.928 — a gap of $1.088 per share, of which stock-based compensation of $1.062 per share accounts for approximately 97.6%. All forward multiples in this dive use consensus non-GAAP EPS and are labelled as such at every appearance. The trailing GAAP price/earnings ratio is meaningless and is rejected; the trailing GAAP price/sales and enterprise-value/sales ratios are used because revenue is basis-independent. This is the single most consequential labelling issue on this name and no conclusion here is drawn without it stated.
• Estimate coverage: 38 analysts on FY2027 revenue and 26 on FY2027 EPS; 39 and 27 on FY2028 — the deepest coverage of any forward line in this file, and the reason both fair-value anchors sit on FY2028. The FY2029 EPS row rests on 6 analysts, FY2030 on TWO and FY2031 on five. The FY2030 EPS row is discarded outright and no conclusion in this dive touches FY2029 or beyond. All eight years of epsAvg are correctly four-for-one split-adjusted; that sub-check is clean.
• Peer note: the vendor peer set contains three genuine comparables — Palo Alto Networks ($298.57B), Fortinet ($123.48B) and Cloudflare ($106.97B) — alongside names that are not (Accenture, Adobe, Analog Devices, CoreWeave, KLA, Synopsys, Texas Instruments). No peer multiples, revenue figures or estimates are supplied for any peer in this file, so no peer-multiple comparison is drawn. The 10-K's own competition section names no individual company at all, only seven categories of competitor, so it offers no substitute. The only named competitors available anywhere in this dive come from a 2023 knowledge-base claim, and they are the large bundled-suite vendor and SentinelOne.
• Fair-value caveat: the $105 / $167 / $269 anchors are multiples of the FY2028 consensus non-GAAP EPS distribution — 70x the low of $1.44762, 105x the mean of $1.55669, and 135.7x (today's forward multiple, held) the high of $1.9533 — each plus net cash of $3.66 per share. Stated arithmetic, not a discounted cash flow. The base is deliberately sensitivity-disclosed: 90x gives $144; 120x gives $190; and 136x — today's multiple, held flat — gives $215, which is essentially the price today. The whole answer lives in a 90-136x band and no greater precision is claimed. The bear case at $105 assumes a de-rating and is independently corroborated by the street's own low price target of $103.25; the bull case at $269 assumes no re-rating at all, only an earnings beat.
• Timing: first-quarter fiscal 2027 results were released 2026-06-03, 62 days before this dive, so the April quarter is fully reflected. The next print is 2026-09-01, 28 days away — outside the five-day banner window, but close enough that the verdict is set with it as an explicit input, and consensus for that quarter ($1,440.077M of revenue, +23.2%) models the first deceleration after four quarters of acceleration. 2026-08-04 was a violent sector-wide SEMICONDUCTOR up-day (MRVL +12.8%, INTC +10.9%, LRCX +7.9%, MU +7.6%, KLAC +7.0%, AMD +7.0%, AMAT +5.5%, ASML +4.2%); CrowdStrike is not a semiconductor company and its +4.29% is not that trade. No company-specific news for CRWD appears in this file for 2026-08-04, and the file contains no same-day price changes for the software peers, so the driver of today's move is not determinable from these sources and is not guessed at.
• Legal and regulatory disclosure carried forward from the filings, because no vendor field contains it: the July 19, 2024 Incident (a faulty Falcon sensor content configuration update that crashed Windows systems) remains a live cost and a live risk two fiscal years later. The securities class action was dismissed with final judgment entered 2026-01-28 and no appeal filed; the consolidated airline class action was dismissed and the Fifth Circuit AFFIRMED the dismissal on 2026-05-20; six or more derivative suits are consolidated and stayed; Delta Air Lines' complaint survived a motion to dismiss in part on 2025-05-16 and is in discovery. Separately and most importantly, "The Company has received requests for information from the U.S. Department of Justice and the U.S. Securities and Exchange Commission relating to the Company's recognition of revenue and reporting of ARR for transactions with certain customers, the July 19 Incident and related matters." The 10-K states that a loss is "reasonably possible" but that "it is not possible to estimate the amount of any loss or range of possible loss." No accrual is recorded for these matters and no vendor field reflects them.
• Accessibility note: no information in this dive is conveyed by colour. All emphasis is carried by bold text, table structure and explicit labelling.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.