Axon cut its full-year revenue guidance to USD 3.61bn–3.63bn, about 30% growth, down from the 32–34% it had previously indicated, and attributed the reduction to three delayed federal Dedrone deployments representing roughly USD 130m of revenue it no longer expects to recognise in 2026. Third-quarter revenue of USD 909m was up 28% year on year but short of a consensus near USD 940m, and the shares fell about 10% on the results. For anyone weighing Axon as a long-term investment, the only question that matters is whether USD 130m of slipped federal procurement is a timing event or a structural one.
The rest of the quarter argues it is timing. Software and services revenue accelerated to 43% growth and reached USD 437m, annual recurring revenue stands at roughly USD 1.8bn, and AI annual recurring revenue passed USD 100m for the first time. Non-GAAP diluted earnings of USD 1.72 a share were up from USD 1.24 a year earlier, though below consensus. Axon also disclosed a USD 1.2bn investment taking its stake in the drone manufacturer Skydio to 37%. The arithmetic is worth doing: USD 130m is about 3.6% of the guided full-year figure, which very nearly accounts for the entire three-point reduction in the growth rate on its own. What the long-term case has to price is not whether the revenue exists, but how often this company's largest contracts sit on a federal calendar it does not control.
What actually happened
Results for the three months to 30 September 2026 were released after the US close on 5 November 2026. Revenue of USD 909m grew 28% and set a quarterly record, but landed below the roughly USD 940m analysts had modelled. Non-GAAP diluted EPS of USD 1.72 was a substantial year-on-year increase from USD 1.24, and also a miss; consensus compilations differ on the exact expectation, so the size of the shortfall varies by source, but the direction does not.
The segment detail is where the quarter diverges from its headline. Software and services revenue grew 43% to USD 437m, an acceleration driven by newer offerings including AI products, real-time operations and Axon Records. That is the highest-margin, most recurring part of the business, and it grew considerably faster than the group. AI annual recurring revenue crossing USD 100m is the first time Axon has put a hard number on that line, and total annual recurring revenue of around USD 1.8bn gives the recurring base real scale against a roughly USD 3.6bn annual revenue run rate.
The guidance cut is specific and narrow. Full-year revenue moves to USD 3.61bn–3.63bn, roughly 30% growth, from 32–34%. Management named the cause: three large federal Dedrone counter-drone deployments worth approximately USD 130m that are no longer expected to land in 2026. Fourth-quarter revenue is guided to USD 990m–1.01bn with adjusted EBITDA of USD 245m–250m.
Separately, Axon announced a USD 1.2bn investment increasing its holding in Skydio to 37%. For a company of Axon's size that is a significant capital commitment, and it signals that counter-drone and autonomous airborne systems are being treated as a core platform rather than an adjacent experiment.
Why the market reacted the way it did
A roughly 10% fall on a quarter that grew 28% and accelerated its highest-quality segment to 43% needs explaining. Three factors did the work.
First, this is a stock that had already been de-rated hard. Axon shares were down about 33% year to date going into the print, against a gain of roughly 12% for the S&P 500 over the same period. A company that has spent the year losing the benefit of the doubt gets less of it on a miss. The reaction was not the market discovering a problem; it was the market extending a trend it had already started.
Second, the miss was on both lines. Revenue short of consensus and EPS short of consensus, in the same quarter as a guidance reduction, removes the usual defence that one of the two was a quality beat. Investors could not point at the income statement and say the shortfall was investment-driven.
Third — and this is the part that goes to the long-term question — it is the second time federal timing has interfered with this story. The Department of Homeland Security counter-drone award arrived in October only as a place on a shared USD 1.5bn framework: terms for future orders, not booked revenue. A single slipped contract is noise. A pattern of federal awards arriving later than planned starts to look like a characteristic of the revenue mix rather than an accident, and the market prices characteristics differently from accidents. That is the re-rating that happened on the night, and it is a Reward/Risk judgement rather than a Growth one.
It is worth being precise about what did not happen. No contract was cancelled. No customer was lost. Management did not say the Dedrone revenue has gone away — it said it is not expected in 2026. The software business accelerated. On the evidence in the release, this is a shift in the calendar, not in the demand.
The Openbook read
Our five-factor framework is built to separate a timing event from a deterioration, which is precisely the distinction this print demands.
Growth is reduced but remains the strongest factor here, and the quality of it improved even as the quantity fell. Thirty per cent guided full-year growth is a derating from 32–34%, and a derating that the company's own guidance now fixes. But 43% growth in software and services — the part of the business that recurs, carries the better margin and does not depend on a procurement cycle to be recognised — is a better growth input than the group number. A business whose recurring software line is accelerating while its hardware-linked federal deployments slip is a business whose growth mix is improving even as its growth rate falls. Scoring Growth off the headline alone misses that.
Profitability holds and is the quietly encouraging part of the release. Non-GAAP EPS rising from USD 1.24 to USD 1.72 is roughly 39% earnings growth on 28% revenue growth, which is operating leverage arriving. Guided fourth-quarter adjusted EBITDA of USD 245m–250m on revenue of USD 990m–1.01bn implies a margin in the mid-twenties, consistent with the trajectory rather than a break from it. The standing caveat applies with force at this company: USD 1.72 is a non-GAAP figure and the gap to reported earnings is wide. Anyone scoring Profitability should fix the basis before comparing it to a peer, because conflating the two here produces a very different answer.
Momentum is weak and gets weaker. Down about 33% year to date before the print and down about 10% after it, with guidance revised downwards, means both legs of the factor — price trend and estimate revisions — are negative. This is the factor doing most of the damage to the composite, and it will not repair on fundamentals alone; it needs a quarter where the federal revenue actually lands.
Solvency is where the Skydio transaction belongs. A USD 1.2bn investment is a material deployment of capital against a business of this size, and it converts balance-sheet flexibility into a strategic position in counter-drone hardware. That is defensible on strategy — Dedrone and Skydio sit in the same flow — but it reduces the cushion at the same moment the revenue calendar has proved less predictable than assumed. The factor is not impaired, but it is less slack than it was, and a second large investment before the federal revenue converts would change that assessment.
Reward/Risk is the factor this print genuinely moves, and it moves in a way that is not simply bad. The USD 130m is almost certainly not lost — counter-drone capability is not a discretionary purchase for the agencies involved, and the deployments were delayed rather than cancelled. So the reward side improves: the same revenue, a lower price, and a software line growing at 43%. The risk side is where the change bites. Concentration in a small number of very large federal awards means revenue recognition depends on departmental timetables, appropriations and leadership continuity — none of which Axon influences, and all of which have now interfered twice. A long-term holder is being asked to accept lumpier revenue than the smooth compounding narrative implies. Our read: Reward/Risk is better-priced than it was a week ago and more honestly described than it was a quarter ago, and those are both improvements, but the variance in the forward revenue path is structurally higher than the market had been assuming. That is the thing to underwrite, not the quarter.
The read-across
The clearest implication is for anyone holding the defence and public-safety technology complex, Motorola Solutions included, on the assumption that government revenue is dependable revenue. It is dependable in aggregate and unreliable in timing, and the difference between those two statements is exactly what cost Axon three points of growth. Companies whose revenue is concentrated in a handful of large federal programmes carry a quarter-to-quarter variance that has nothing to do with demand, and investors should expect to see the same shape appear elsewhere — particularly in counter-drone, where procurement is new, budgets are being established rather than renewed, and award processes have no long track record to pace against.
The second read-across is positive and more specific. AI annual recurring revenue passing USD 100m, inside a software line growing 43%, is one of the few hard disclosures available anywhere about AI revenue actually being recognised in a vertical software business rather than promised. Peers selling AI into regulated, evidence-handling or records-management workflows now have a benchmark. Comparable names can be screened on our screener, and the full data on this one sits on the Axon equity page.
What to watch next
- Fourth-quarter results, due late February 2027. Guidance is USD 990m–1.01bn of revenue and USD 245m–250m of adjusted EBITDA. The number to look for is not the revenue but any update on the three Dedrone deployments: whether they are dated, undated, or re-scoped.
- The 2027 guidance issued alongside it. If the USD 130m is simply pushed into 2027, the full-year growth guide should be flattered by roughly three points. If 2027 is guided without that benefit, the revenue has been lost rather than delayed, and the Growth factor requires a larger adjustment than this quarter's.
- Software and services growth. Forty-three per cent is the factor doing the heavy lifting. A deceleration here matters far more to the long-term case than another quarter of federal slippage, because it is the part of the business that is supposed to be insulated from procurement timing.
- AI ARR disclosure. Having now put USD 100m on the record, Axon has created an expectation of an updated figure. The rate of increase will say more about pricing power in its software stack than any other line it publishes.
For a reader whose question was what Axon's numbers now say about the long-term case: the recurring software base is accelerating at 43%, AI revenue has crossed USD 100m, earnings are growing faster than revenue, and the guidance cut is fully explained by USD 130m of federal work that has slipped rather than vanished. What would have to go right is those deployments converting on a dated schedule. What would have to go wrong is a third instance of the same thing.
Discussion
Log in to join the discussion