$332.8m of Net New ARR, +51% (26 August 2026): Why CrowdStrike's Headline Number Grew Twice as Fast as the ARR Underneath It

CrowdStrike reported fiscal second-quarter results after the New York close on Wednesday 26 August, and the number the company led with was a record $332.8 million of net new annual recurring revenue, up 51% year on year. The ARR that figure was added to grew 25%, and revenue grew 26%. Those three rates describe the same business over the same thirteen weeks — the gap between them is arithmetic, not disagreement, and understanding why it opens is the most portable thing in the release.

What actually happened

CrowdStrike released results for the three months to 31 July 2026 after the close, with the call at 5:00 p.m. Eastern. Every company figure below comes from the filed earnings release; the consensus figures and the share reaction come from CNBC's report.

Chief executive George Kurtz called it "the best quarter in CrowdStrike's history" and attributed demand to what he has been calling the Mythos moment — a reference, per CNBC, to the release of Anthropic's Mythos model, which is capable of exploiting previously unknown software vulnerabilities and has driven enterprise demand for new security tooling. The shares rose more than 11% in extended trading. Per-share and share-count figures are split-adjusted throughout: a four-for-one split took effect on 2 July 2026, inside the reported quarter.

Why 51% and 25% are the same number

ARR is a level: the annualised value of the subscription book on a given date. Net new ARR is the first difference of that level — how much the book grew in thirteen weeks. Because a difference is measured against a base roughly twenty times smaller than the level, its growth rate amplifies whatever the level's growth rate is doing.

The mechanics are worth doing once, because they apply to every subscription business that reports this pair. CrowdStrike added $111.7 million more ARR this quarter than in the same quarter last year. Against last year's $221.1 million of net new ARR, that increment is 50.5%, which the company rounds to 51%. Against the $4.66 billion ARR base it was added to, the same increment is 2.4%. Neither number is spin. They are the same dollars divided by different denominators, and the choice of denominator is what makes one of them headline-shaped.

Run that on the last two years and the picture is clearer than either headline rate.

Opening balances are derived by chaining the reported net new figures off the reported ARR levels, which the company rounds to three significant figures — so these percentages carry a little rounding noise, but not enough to change the shape.

Two things fall out. First, the 51% is not an artefact: on the base-free measure the quarter added 6.0% of the opening book against 5.0% a year earlier, a genuine step of about one percentage point. Second, the first quarter's headline was much softer than it looked. Net new ARR grew 32% in the April quarter, but on the base-free measure that was 4.9% against 4.6% — an acceleration of three-tenths of a point. Judged on headline growth the second quarter looks 1.6 times better than the first; judged base-free it is roughly three times better. The second quarter is where the change in trend actually sits.

The guidance raise is smaller than the quarter

The release raised the full-year net new ARR growth outlook by 630 basis points to 34% at the midpoint, from the 27.7% set in June. That raise can be sized in dollars, and doing so is the most informative arithmetic in the release.

Fiscal 2026 ended with ARR of about $5,254 million against about $4,245 million a year earlier, so fiscal 2026 added roughly $1,009 million of net new ARR. The new full-year ARR guidance of $6,603.0 to $6,611.9 million puts the midpoint at $6,607 million, implying about $1,353 million of net new ARR for the year — 34.1% growth, which reconciles to the company's stated midpoint. The prior 27.7% guide implied roughly $1,289 million. The raise is therefore worth about $64 million of additional full-year ARR.

Set that against the quarter just delivered, which came in $111.7 million ahead of the year-ago quarter on its own. The company did not extrapolate the second-quarter step-change through the second half; it added roughly 57 cents of full-year guidance for every dollar by which the quarter beat its own comparison.

The implied second half decelerates to roughly 29% growth from the 42% the first half delivered. That is not a forecast of anything; it is a description of what the company's own range commits it to. Two readings are consistent with it, and the release does not distinguish between them. Either the guide is conservative in the way this company's guides have tended to be, in which case the second half beats it. Or the second-quarter impulse contained something that does not repeat at the same size — the Falcon Flex book is the obvious candidate, since ending ARR from accounts that have adopted Flex passed $2.29 billion and grew 101% year on year, and a consumption-style commitment model can change when ARR lands as well as whether it does. What would settle it: whether the third-quarter print comes in near the implied $344 million or well above it, and whether the base-free rate holds near 6.0% or slips back toward 5%.

The $399.0m between the two profit measures

CrowdStrike reported an operating loss of $33.2 million and adjusted operating income of $371.6 million for the same quarter. The bridge between them is $404.8 million, and it is almost entirely one line.

Two movements matter here and they run in opposite directions. The non-stock portion of the bridge collapsed from about $83.8 million to about $5.8 million, which is the mechanical reason reported profitability improved so sharply: the amortisation, incident-related costs and other excluded items that used to sit between the measures have largely run off, so the two are converging on everything except stock. Meanwhile stock-based compensation grew 44% against revenue growth of 26%, taking it from 23.7% to 27.1% of revenue.

The charge is non-cash, which is why it does not touch the cash line — operating cash flow was $530.3 million and free cash flow $377.4 million, a 25.7% free-cash-flow margin against 24.3% a year ago, with $845.9 million of free cash flow across the first half against $563.0 million. But dilution is the cash-free form of the same cost, and it shows up in the share count: basic shares rose about 2% year on year to 1,019.4 million, and additional paid-in capital rose $641.3 million in six months. Neither presentation is the true one. Reported profit charges the cost when it is granted; adjusted profit sets it aside and lets the share count carry it instead.

Does the company's own metric check out?

ARR is a company-defined measure. It is not audited, its definition can change, and it is not a GAAP line. The standard discipline when a business leads with one is to check it against something that is audited — and the natural cross-check for a subscription book is deferred revenue, which records amounts billed in advance of delivery.

Three measures built on different rules — a company definition, an audited balance-sheet liability and an audited income-statement line — land within a percentage point of each other. That is the result you want, and it is not guaranteed: a book growing on paper faster than it is being billed for would show up here as a widening gap. It does not. Note the comparison has to be year-on-year rather than against the January balance sheet, because deferred revenue peaks at the fiscal year end when the largest billing quarter closes, and drifts down through the first half by construction.

What reads across, and what does not

The demand explanation on offer is that frontier AI capability has created a new category of enterprise security spending. If that is the real driver, the read-across is not to this company alone but to the budget line — and CNBC notes the cybersecurity group broadly has been making new highs on demand for AI security tools. A shared-driver read-across is a hypothesis, though, and it holds only to the extent peers sell into the same budget and the same buying committee. Platform consolidation cuts the other way: a customer folding several tools into one vendor is a positive for that vendor precisely because it is a negative for the ones displaced. Module adoption rates of 51%, 35% and 26% for six, seven and eight or more modules describe exactly that dynamic. The company also agreed during the quarter to acquire the technology assets of XM Cyber from Schwarz Digits, alongside an expanded European partnership — consolidation of the same kind, one level up.

To the indices, the channel is weight and nothing more exotic: an 11% move in one large constituent contributes to US500 and NAS100 in proportion to its share of each, and the more durable transmission is whether the sector's multiple re-rates rather than what one print did overnight. Compare the two other large software and AI-infrastructure prints of the same week — Nvidia's quarter, where the binding constraint is physical supply, and Salesforce's, where it is seat-based software meeting consumption pricing. The metric-definition question raised here — a company-defined headline measure checked against an audited one — is the same discipline applied to the reported-versus-adjusted gap at Analog Devices.

To currencies, essentially nothing. A single enterprise software quarter is not a macro input. Software and security capital spending eventually reaches the growth factor that the meter scores across the eight majors, but the lag is long, the signal is small, and on any given week the dollar's path is set by the rates calendar rather than by a subscription book in Austin, Texas. The discipline that keeps both readings useful is to file them separately: read this print for what the enterprise security cycle is doing, and read the rates calendar for what the dollar is doing.

Educational macro context only — not investment advice.