Atlassian closed Friday up 30.8%, the second-best move in the Russell 1000 on a day when the S&P 500 set a record. The quarter deserved a good day. Fiscal fourth-quarter revenue landed at $1.77 billion, 6.3% ahead of consensus and up 28% from a year earlier. Adjusted earnings of $1.87 a share beat by roughly 47%. Cloud revenue grew 31% to about $1.2 billion. Subscription annual recurring revenue rose 23% to $6.6 billion, and remaining performance obligations — contracted revenue not yet recognized — jumped 44% to roughly $4.8 billion.
Then the company told investors what next year looks like: total revenue growth of about 13%, subscription ARR growth of about 18%, cloud growth of about 25.5%.
Read those two paragraphs back to back. Atlassian just delivered 28% growth and guided to less than half of it. On the call, management attributed the gap to prudence, the lapping of its DX acquisition, and caution about the back half of the year. Analysts pushed on the ARR line specifically — why 18% next year when the quarter just printed 23%. The stock went up a third anyway.
What actually got repriced
Not the earnings. The story.
Atlassian has spent the past year carrying an AI-casualty discount. The bear case was mechanical and easy to say out loud: Jira and Confluence are sold by the seat, AI agents reduce the number of seats, therefore the revenue line eventually bends. Every quarter of decelerating cloud growth fed it. Investors had already marked the company down for a future they could describe but not yet see in the numbers.
This quarter interrupted that. Cloud growth accelerated to 31%. The company said its AI developer tools crossed a million monthly users, that usage of its Rovo AI product rose about 50% quarter over quarter, and that it signed its largest deal ever. None of that proves the seat model survives agents. It proves the erosion is not visible yet — and when a stock is priced for erosion, the absence of erosion is worth 30%.
Our take: This is not a beat-and-raise rally, it’s a discount-removal rally. The numbers improved modestly; the narrative flipped completely, and the narrative was carrying the multiple. That cuts both ways. Because almost none of Friday’s move came from the guidance, almost all of the downside sits in it. If 13% turns out to be the real number rather than a sandbagged one, there is no earnings surprise left to grow into — the multiple has to do all the work. Conservative guides that get beaten are a management style. Conservative guides that turn out accurate are a business condition. You find out which one this is in about ninety days.
The pattern is not just Atlassian
Friday was full of this. Doximity rose 33.3% on 7% revenue growth and an earnings miss. Twilio gained 30.6%. FIGS added 27.7%, Natera 18.4%, Cloudflare 7.9%. Software names that had been sold on the assumption AI would eat them all moved violently in the same direction on the same day, and the macro helped: July payrolls fell 23,000, odds of a September Fed hike collapsed to 42% from 58%, and duration-heavy growth stocks got their discount rate back.
When a whole cohort re-rates on a rate print, the individual results are the excuse, not the cause. Some of these businesses have broken the AI bear case. Some just got a friendly tape on the morning they reported. The tape does not distinguish between the two. You have to.
What to watch
- The Q1 print in late October. Management guided first-quarter revenue to roughly $1.705–$1.715 billion, above consensus near $1.67 billion. A big beat says 13% was sandbagging. An in-line quarter says it wasn’t.
- Subscription ARR, not revenue. Revenue lags. ARR is the leading indicator of whether seats are actually compounding. Watch whether it tracks toward 18% or holds nearer 23%.
- Data Center. The legacy on-premises line is being migrated away deliberately, which flatters cloud growth and drags the total. Know which number a headline is quoting.
- Whether the cohort holds. If the software complex gives back this move on the next hot inflation print, Friday was a rates trade wearing an earnings costume.
The discipline is the same one that applies to every violent post-earnings move: separate what changed in the business from what changed in the story about it. Friday, Atlassian changed both. Only one is on the income statement.