Atlassian Corporation TEAM

Key data

The business

Atlassian sells the tools software and IT teams use to run their work: Jira to track every task and bug, Confluence to write down what the team knows, Jira Service Management to run the help desk. More than 350,000 customers, none more than 5% of revenue. The company lands cheap through self-serve sign-ups and grows as teams add seats and products around the one they came for. The moat is switching cost: Jira becomes the system of record for engineering, wired into daily workflow, and ripping it out means re-teaching everyone. The money is subscription, almost all recurring, and the company is now retiring its legacy Data Center product, the version customers ran on their own servers, to push everyone to the cloud. The thing a developer actually touches all day is the Jira ticket.

The numbers

Revenue compounded about 24% a year, from $2.8B in FY2022 to $6.6B in FY2026, and FY2026 is where the story turns.

Operating income went positive for the first time since 2022, and the June quarter did the heavy lifting.

On the derived June figures, the quarter earned about $139M net and roughly $210M of operating income on revenue up 28%, right after a March quarter that lost $98M. One honest caveat sits inside the same numbers: revenue actually slipped from $1.79B to $1.77B quarter to quarter, so the profit came as much from cost discipline as from demand accelerating.

Free cash flow is steadier than the profit line but it did not grow this year, $1.3B on a 20% margin, and stock compensation runs 24% of revenue, so the owner keeps less than the cash statement shows. The $10,000-plus Cloud customer count grew 10%: expansion from existing accounts is doing the work now, not a flood of new logos. Wall Street models about $10B of revenue by FY2029 and non-GAAP earnings near $8.50 a share. The variant is plain: if June's operating profit is the run rate and not a year-end, DX-acquisition artifact, then 7.6 times sales is too cheap, and the print that settles it is operating income in the September quarter.

Management

Michael Cannon-Brookes runs it as solo CEO now; co-founder Scott Farquhar stepped back to Former Co-CEO. Insiders bought nothing over the last year and sold about $6.6M across 97 transactions, tiny against a $50B company, with the largest a $747K sale the day the 10-K filed; plan status not disclosed. The capital allocation is louder than the selling: FY2026 buybacks jumped to $1.8B from $0.8B, and $1.0B of it landed in the March quarter with the stock near $100 to $130, well before today's $190, which reads as conviction that paid off. Against that, buybacks mostly offset the 24%-of-revenue stock compensation rather than shrinking the count much. The pay plan is literally scored on "Rule of 40," growth plus margin, which at least points management at the thing that matters.

How it fails or surprises you

The profit was one quarter. June's +$210M of operating income sits against a −$56M March quarter and a full-year total of only +$10M. If the September quarter reverts to a loss, the turn was year-end seasonality plus DX synergy timing, not a new baseline, and 7.6 times sales stops looking like a gift. First print: Q1 FY27 operating income, late October.

Data Center end-of-life. The company itself flagged that FY2027 gross margin declines slightly on "mechanical drag" as it stops selling the on-server product, with no new term-license sales since March 2026. If cloud migration does not fully recapture that revenue, growth slips and the compounding case thins. Watch total revenue growth and gross margin the next two quarters.

The re-rate (right tail). At 7.6 times sales, the cheapest in twelve years against an 11.8x-to-21.8x historic band, a business now earning an operating profit on 26% growth is priced as if the turn cannot hold. If operating income sustains above $150M for two straight quarters, even a partial re-rate on a larger revenue base is a large move nobody is paying for today.

Closing thoughts

This is a name where one specific print settles the argument, and the print is Q1 FY27 operating income, reported in late October. If it holds above $100M with revenue growth still above 20%, the profit turn is real and the cheapest-ever sales multiple has clear room to lift. An ambiguous print, say $40M of operating income on low-twenties growth, leaves it unresolved and you wait another quarter with the Data Center drag pressing on FY27. If it swings back to a loss, June was an artifact. I lean toward the turn being real rather than not: years of investment finally showing operating leverage, and management buying back stock hard into the low. But the sequential revenue dip and the self-flagged FY27 headwind are real, so the right tail is fatter than the left without being a sure thing.

The bet is still that software teams keep running on Jira and Confluence and paying more as they move to the cloud, and that Atlassian now keeps a profit while doing it. What breaks it is the September quarter reverting to an operating loss, or revenue growth decelerating materially below the mid-twenties rate. The one pair of numbers that tells you first: Q1 FY27 operating income against the $100M line, and year-over-year revenue growth.