Report contents
Industry landscape
Where the company competes: endpoint security sold as a widening platform
CrowdStrike sells cloud-delivered security software. Its Falcon platform places one lightweight agent on laptops, servers, containers and cloud workloads; that agent streams telemetry to CrowdStrike's cloud, where detection models and human analysts grade it. Customers buy the platform as subscriptions, module by module: endpoint detection and response (EDR), next-generation antivirus, threat intelligence, exposure management, identity protection, cloud workload protection, and a log-management product the company calls Next-Gen SIEM. Subscriptions supplied about 95% of the $3.95 billion in revenue CrowdStrike reported for the fiscal year ended January 31, 2025, according to its 10-K. There is no appliance to ship and no hardware to install.
The upstream chain is short: hyperscale cloud capacity, principally AWS, plus operating-system compatibility work for Windows, macOS and Linux agents, and a large research and engineering payroll. Downstream sits the buyer — a chief information security officer and a security operations center that staffing levels rarely keep pace with. Between CrowdStrike and that buyer sit resellers, distributors and managed security providers, which the company relies on for a meaningful share of bookings, and, increasingly, the hyperscaler marketplaces where customers draw down committed cloud spend.
The core market is corporate endpoint security, and the boundary is moving. Precisely, CrowdStrike competes in software that prevents, detects and responds to attacks on employee devices and server workloads inside large organizations. That market was rebuilt over the past decade as signature-based antivirus gave way to behavioral detection, and CrowdStrike's share gains came largely at the expense of the incumbent suites. The company now sells into the broader platform budget where endpoint telemetry feeds extended detection and response, SIEM and cloud security. That widening is a business decision with a competitive consequence: it puts CrowdStrike into more purchase decisions, and into more purchase decisions where Microsoft, Palo Alto Networks and Splunk already sit.
Geographically, demand is global and revenue is not. About two-thirds of fiscal 2025 revenue came from the United States, per the 10-K, with the remainder spread across Europe, Asia-Pacific and the rest of the Americas. Product development is amortized across a worldwide installed base, which is a real scale advantage, but buying is organized nationally. FedRAMP authorization in the US, data-residency expectations in the EU, and sovereign-preference rules in Japan, Australia and India each require local legal entities, local data handling and local certifications. Those requirements raise the cost of entering a country; they do not create domestic champions of comparable scale.
Three features of the structure shape everything else. First, the buyer is concentrated and technically sophisticated. A single Fortune 500 account may run 100,000 endpoints, and procurement runs through proof-of-concept bake-offs against two or three shortlisted vendors. Second, displacement is expensive once an agent is deployed: replacing it means re-imaging machines and retraining analysts, so incumbency sticks, though large buyers routinely run two endpoint vendors side by side to keep pricing honest. Third, and most important, the largest competitor gives its core product away inside an operating-system license.
| Vendor | Route into the account | Where it pressures CrowdStrike |
|---|---|---|
| Microsoft | Defender bundled with Windows and M365 E5 licensing | Sets a near-zero marginal price for basic endpoint protection |
| Palo Alto Networks | Firewall installed base, Cortex XSIAM | Captures platform-consolidation budgets at existing customers |
| SentinelOne | Standalone agent, aggressive midmarket pricing | Price-sensitive deals and smaller accounts |
| Trellix, Broadcom (Symantec) | Legacy antivirus renewals | Incumbent displacement, though the renewal pool is shrinking |
The practical test is how many vendors can be counted on one hand. Globally, roughly five to seven suppliers hold the bulk of enterprise endpoint spending, and Microsoft is the largest of them on most third-party counts, though it does not disclose Defender revenue separately. Regional players and venture-funded startups fill the tail. That is a moderately concentrated market with high technical barriers — building a detection engine requires years of telemetry and validation in independent tests — and high trust barriers, since a failed detection is a public event. Capital intensity is low, so new entrants keep appearing; what they lack is data and reference accounts.
The structure therefore points in two directions at once. CrowdStrike's technology and brand position look defensible against like-for-like competitors. Against a bundled Microsoft, the defense has to be breadth: each additional module raises the cost of ripping out the agent and makes the platform a harder line item to delete. Whether module attachment has actually raised retention, or merely raised revenue per customer, is the question the next section has to answer.
Competitive advantage
Does the company have a moat? Moderate, and it rests on switching costs
Moat: Moderate. CrowdStrike clears the concentration test in enterprise endpoint security without much argument. It does not clear the sustained-profitability test on the accounts it publishes. GAAP operating income was negative in each of the last five fiscal years, and the company has not put together a profitable fiscal year since it listed in June 2019. That combination caps the rating. What keeps it off Weak is the closest thing to a controlled experiment this industry has offered: on July 19, 2024, a faulty sensor update disabled Windows machines worldwide, and the customer base largely stayed. Net revenue retention was 115% at January 31, 2026, down from roughly 119% before the incident, while ARR growth had re-accelerated to 25% by July 31, 2026, the company reported. An installed base that keeps spending more after an outage of that visibility is evidence of real friction in leaving.
Market position: Fairly Strong. CrowdStrike does not publish market share, and the third-party counts of endpoint security differ in what they include, so the trend has to be read from revenue. Revenue rose from $1.45 billion in fiscal 2022 to $4.81 billion in fiscal 2026, about 3.3 times, or roughly 35% a year, against competitors that were either bundling (Microsoft), defending an adjacent install base (Palo Alto Networks) or managing decline (the Symantec and Trellix renewal pools). If the market grew slower than 35% a year, which most industry estimates imply, CrowdStrike took share in every one of those years. That is an inference from revenue, not a measured share series, and it matters because the framework's rough threshold is a five-year share move of more than 5 percentage points, which it treats as evidence that barriers are probably absent. CrowdStrike's revenue path points the other way, but revenue cannot separate share gains from a growing market.
Stability: Fairly Strong. Growth decelerated steadily, from 54% in fiscal 2023 to about 22% in fiscal 2026, which is what a maturing base looks like whether or not share is still being won. The more useful stability measure is inside the account: 51% of subscription customers ran six or more modules at July 31, 2026, per the company, up from a smaller base in earlier years, and the average contract runs one to three years with billing in advance. Deepening beats new-logo growth for durability, because a six-module customer is a harder account to displace than a one-module customer.
| Fiscal year | Revenue (USD bn) | Growth | Gross margin | GAAP operating margin | Free cash flow (USD bn) |
|---|---|---|---|---|---|
| FY2022 | 1.45 | 73.6% | (9.7)% | 0.44 | |
| FY2023 | 2.24 | 54.5% | 73.2% | (8.5)% | 0.68 |
| FY2024 | 3.06 | 36.6% | 75.3% | (0.1)% | 0.93 |
| FY2025 | 3.95 | 29.1% | 74.9% | (3.0)% | 1.07 |
| FY2026 | 4.81 | 21.7% | 74.7% | (6.0)% | 1.31 |
Source: company annual filings. Growth for fiscal 2023 is calculated from fiscal 2022 revenue.
Profitability: Moderate. The gross margin has sat between 73% and 75% for five years, which is what subscription software with pricing power looks like, and free cash flow rose from $440 million to $1.31 billion, a 27% margin on fiscal 2026 revenue. The GAAP line runs the other way: a $293.3 million operating loss and a $162.5 million net loss in fiscal 2026, with the operating margin widening as revenue grew. The bridge is compensation paid in stock. Non-GAAP operating income of $1.05 billion against that GAAP loss implies about $1.3 billion of adjustments, mostly equity pay, or 28% of revenue, and free cash flow adds it back. The cash is real; part of it is funded by shareholders through dilution, and the 15%-25% after-tax ROIC that the framework treats as clear evidence of advantage is absent. ROIC is in fact not calculable on a net basis: $5.23 billion of cash against $760 million of debt leaves invested capital near zero, so any ratio computed from the filings is an artifact. The public record is also seven fiscal years, not the decade the benchmark assumes. The two quarters reported since, about $30 million and $10 million of net income, are too small to change that.
Two tests, two answers: concentration passes, sustained excess returns do not, on GAAP. The rating holds at Moderate because the durability evidence is behavioral, in retention and module attachment, while the financial evidence is still mostly non-GAAP. A GAAP operating margin sustained in double digits with stock pay below 15% of revenue would move profitability to Fairly Strong; a net retention reading back above 119% would strengthen the market-position case.
Where the advantage comes from and how long it can last: a sensor that is expensive to remove
In endpoint security the purchase decision is made once and renewed for years, detection quality is table stakes for making the shortlist, and the buyer's real risk is operational. Advantage therefore accrues to whoever raises the cost of removal and spreads fixed development across the most machines. CrowdStrike has one strong source and one weaker one.
Customer captivity: Fairly Strong, tested by the July 2024 outage
The mechanism starts with architecture. Falcon runs as a single lightweight agent on each protected machine, reporting to CrowdStrike's cloud, and the company sells 33 modules through that one agent, per its fiscal 2026 10-K. Adding a module is a license switched on centrally; it needs no new deployment, no new vendor relationship and no new procurement cycle. That is why the count of customers at six or more modules is the number that matters. Each module raises the installed cost of ripping the agent out without adding a minute of customer labor in the meantime, and it raises the share of the security budget that would have to be re-sourced at once.
Removal is expensive for reasons that have little to do with contract terms. Replacing the agent means re-imaging machines, re-tuning detection rules, retraining the security operations staff and re-running a proof of value in front of a chief information security officer who carries the reputational risk if the cutover fails. Search costs work the same direction: buyers typically shortlist two or three vendors after bake-offs, so the incumbent starts each renewal with an information advantage.
The July 2024 outage was the test case, because it handed every rival a reason to call. Revenue still grew about 22% in the fiscal year containing the event, and net retention, though down from roughly 119% to 115% by January 2026, stayed comfortably above 100%. That is the best available evidence that captivity is real and not just brand familiarity.
The limit is pricing. Captivity protects the base; it does not give CrowdStrike a free hand on price, because every renewal is priced against a Microsoft bundle that costs the customer nothing extra. The post-outage discounting the company absorbed shows where the ceiling sits.
Scale in detection and development: Moderate, and contested
The second source is data scale. Telemetry from more than 88,000 organizations feeds the models that grade detections, so each new customer adds attack variants that improve the product for everyone else. More coverage produces faster detection of new techniques, which shows up in independent test results, which gets the vendor on shortlists, which adds more endpoints. Fixed development costs spread across $4.81 billion of revenue work the same way: a new module is built once against one codebase and one cloud, then sold to the whole base, which is cheaper per dollar of revenue than a rival building separate products for separate consoles.
The problem is that this advantage is relative, and the largest competitor holds a larger dataset. Microsoft sees more endpoints through Windows and Defender than CrowdStrike will ever install, and Palo Alto Networks brings network telemetry that CrowdStrike does not have. A data advantage that a rival with a bigger base can match is a tempo advantage, not a structural one.
The accounts support that reading. A genuine scale advantage in cost should show up as rising gross margin as revenue triples, and the margin instead held flat at 73%-75% from fiscal 2022 through fiscal 2026. Flat margin is consistent with scale economies being competed away in price, which is what a market with a free alternative at the low end would produce. Scale here improves the product and the go-to-market reach; the evidence does not show it compounding into widening unit economics.
Durability: the price ceiling is a bundled product
The substitute that constrains this moat is not a new detection technology. It is a license line. Microsoft includes basic endpoint protection in Windows and in the M365 E5 bundle, so a large share of the addressable market already owns a good-enough product at zero marginal cost, and every CrowdStrike renewal has to justify the incremental spend against that. That caps price, and it limits how far CrowdStrike can push into accounts that are not driven by regulation, breach history or a sophisticated security team.
Consolidation cuts both ways. Buyers want fewer security vendors, which helps CrowdStrike when it is the consolidator absorbing point products into its 33 modules, and helps Microsoft when the buyer decides the operating-system vendor is consolidation enough. Which force wins varies by account size, and the company does not disclose the split.
AI is the open question. It raises the volume and speed of attacks, which supports demand and rewards whoever has the most telemetry. It also lowers the cost of building a credible detection model, which weakens the data lead over time. Both effects are plausible; the evidence so far only shows that ARR growth recovered to 25% while the model-building cost fell, which is suggestive but not decisive.
The duration case is that the switching cost lives in deployment, workflow and retraining, none of which a competitor can erase with a better algorithm or a lower price. That kind of friction decays slowly, over product cycles of five to ten years. The pricing case is weaker: the moat protects the installed base and the renewal, and it may not protect the growth rate against a bundle priced at zero.
Two readings would settle the direction. Net retention back above 119% would show the expansion engine works without post-outage discounting, and a GAAP operating margin that stays positive while stock pay falls below 15% of revenue would put the profitability test within reach. A net retention reading below 112%, or a gross margin under 72%, would say the bundle is taking price and that the advantage is narrower than the retention record suggests.
Generated by DeepVal AI. Data and models may contain errors or be outdated. For research only; not personalized investment advice.
Continue reading the full report
- Comparison and summaryHow Crowdstrike Holdings Inc compares with its closest rivals, and our moat rating
Free to sign up · 20 research points · unlocking this report uses 10
20 points cover 2 full reports. No card needed.
Research reflects its generation date. Preview the first two modules.
More on Crowdstrike Holdings Inc
Crowdstrike Holdings Inc
Business, moat and key questions
Open company page