Summary

  • Atlassian reported fiscal-2026 Cloud revenue of US$4.411 billion and Data Center revenue of US$1.831 billion. Its fiscal-2027 guide calls for Cloud growth of about 25.5% and a Data Center decline of about 17%.
  • Applying those approximate rates to the audited bases yields a Cloud increase of roughly US$1.125 billion and a Data Center reversal of roughly US$311 million. The reversal absorbs about 27.7% of the Cloud dollar gain before about US$40 million of Marketplace growth produces the 13% total-revenue guide.
  • The Data Center decline is not a disclosed churn or migration number. It combines the expiry of a fiscal-2026 revenue-recognition benefit, customer purchases pulled forward from fiscal 2027, continuing Cloud migrations and weaker Data Center seat expansion.
  • Cloud and Data Center run on different recognition and cost curves. A migration can preserve the customer relationship while lowering first-year revenue and gross margin, which makes Subscription ARR useful counterevidence—but not a substitute for revenue, RPO or cash.

The deployment waterfall

Atlassian's fiscal-2027 targets appear to offer three different speeds. Cloud revenue should grow about 25.5%. Data Center revenue should fall about 17%. Marketplace and other revenue should grow about 12%. Total company revenue, after all three, should rise about 13%.

Those rates cannot be added, subtracted or averaged because they begin from different balances. The fiscal-2026 Form 10-K supplies the denominators: US$4.410627 billion of Cloud revenue, US$1.830941 billion of Data Center revenue and US$330.740 million of Marketplace and other revenue. Together they produced US$6.572308 billion.

Applying the approximate guidance mechanically gives a useful reconstruction. Cloud reaches about US$5.535 billion, an increase of US$1.125 billion. Data Center falls to about US$1.520 billion, a decline of US$311 million. Marketplace and other reaches about US$370 million, adding about US$40 million. The result is approximately US$7.425 billion, or 12.98% growth before rounding.

This is not a company-published line-item forecast. It is the arithmetic implied by Atlassian's approximate growth rates and audited starting points. That provenance matters because the guide may move, each rate is rounded and the actual deployment mix can differ.

The reconstruction still reveals the economic weight of the transition. The US$311 million Data Center reversal is equivalent to 27.7% of the US$1.125 billion Cloud gain. Almost twenty-eight cents of every gross Cloud growth dollar is absorbed before Marketplace contributes. Cloud mix would rise from 67.1% to roughly 74.5%; Data Center would fall from 27.9% to about 20.5%.

The headline therefore is not that Atlassian's Cloud business grows only 13%. It is that a fast-growing Cloud line must carry the comparison cost created by a shrinking, differently recognized on-premises line.

Why fiscal 2026 was front-loaded

The reversal starts with a product decision. On 8 September 2025, Atlassian announced that Data Center would reach end of life in March 2029. The investor update also changed the revenue allocation for future Data Center subscriptions.

A Data Center contract contains a term licence and support. Revenue allocated to the licence is recognized upfront when the licence is delivered. Support revenue is recognized over the subscription term. Once the expected product life became shorter, Atlassian assigned more of the contract value to the upfront licence and less to ongoing maintenance for subscriptions entered into after the announcement.

The customer did not need to buy another dollar for that accounting allocation to change the timing of recognized revenue. Atlassian said the change would mechanically increase fiscal-2026 Data Center and total revenue and raise GAAP and non-GAAP margins. It also said Cloud revenue, Marketplace revenue, operating cash flow and free cash flow would not change from the allocation itself.

The first evidence appeared immediately. The fiscal-Q1 shareholder letter said the EOL recognition effect contributed about 1.6 percentage points to total-revenue growth and 5.7 points to Data Center revenue growth in the quarter. By year-end, Atlassian said the same timing effect had benefited fiscal-2026 non-GAAP operating margin by about four percentage points.

Recognition was only one force. Fiscal-2026 Data Center revenue increased 25% to US$1.831 billion even as customers migrated to Cloud. Atlassian attributes the strength to pricing, greater upfront term-licence recognition, retention and customers pulling purchases and expansions into fiscal 2026. Its fiscal-Q3 letter had already warned that purchasing activity was moving forward from later periods.

The old line therefore entered fiscal 2027 with an unusually high comparison. Part of the prior year's revenue had been recognized sooner; part of the purchasing had occurred sooner. A transition can be commercially successful and still leave an accounting hollow behind it.

What the US$311 million is—and is not

Atlassian names four causes for the guided 17% Data Center decline. Fiscal 2027 laps the EOL revenue-recognition benefit. Customers pulled some purchasing from fiscal 2027 into fiscal 2026. More customers are migrating to Cloud. Remaining Data Center customers are moderating seat expansion while they prepare to move. Pricing, renewals and hybrid deployments are expected to offset part of the fall.

That list prevents an attractive but false label. The derived US$311 million is not a disclosed churn amount. It is not a count of customers that left, a cash outflow, a reduction in RPO or an invoice sent to the migration programme. It is the approximate change in one GAAP deployment-revenue line under a four-part management explanation.

The same discipline applies to the US$1.125 billion Cloud increase. Atlassian expects migrations to supply mid-to-high single-digit percentage points of Cloud revenue growth, but it does not publish the migration dollars. Cloud also grows through paid-seat expansion, cross-selling, higher-value editions, Teamwork and Service Collections, Rovo-related adoption, new customers and acquisition effects.

The two gross changes are not independent demand. One enterprise can leave Data Center and enter Cloud, creating a negative movement in one ledger and a positive movement in another. Adding both as separate customer wins would count the same relationship twice. Treating the Data Center decline as lost business would erase the relationship from the other side.

This is why the waterfall is more useful than a single growth percentage. It preserves the transfer between deployments while leaving unexplained movements unexplained.

Migration changes the recognition and cost curve

Atlassian's Cloud subscriptions are generally recognized ratably as the service is provided. Data Center combines upfront licence revenue with ratable support. The 10-K says first-year Cloud revenue is typically lower than Data Center revenue. Moving a customer between deployments can therefore reduce current revenue even if contract value, retention or long-term economics remain sound.

Costs move too. Data Center customers host the software. In Cloud, Atlassian carries infrastructure, support and growing AI usage. The company expects fiscal-2027 GAAP gross margin of about 84.5%, slightly below the calculated 84.83% in fiscal 2026, as Cloud mix and Rovo hosting costs offset infrastructure and support efficiencies.

The Cloud gain is consequently not pure operating leverage. It purchases a different delivery obligation. A customer receives continuous service, product updates, security operations and access to Cloud-only capabilities; Atlassian receives a ratable revenue stream while assuming more of the operating surface.

Marketplace adds a second-order effect. Atlassian currently takes a lower share from third-party Cloud app sales than from Data Center apps because it is encouraging development on Forge. The reconstructed Marketplace increase is only about US$40 million, and a successful migration can still move app commerce onto a lower-take-rate surface.

None of that makes migration unattractive. Cloud can deepen product adoption, reduce version fragmentation, improve continuous delivery and make Rovo and the Teamwork Graph available to customers that could not use them on Data Center. It does mean the migration case must be proved after hosting cost, support, incentives and revenue timing—not merely by showing a higher Cloud mix.

ARR, RPO and cash are separate witnesses

Atlassian proposes Subscription ARR as the clearer measure during the Data Center distortion. It ended fiscal 2026 at US$6.606 billion, up 23%, and management guides to about 18% growth in fiscal 2027. ARR annualizes active Cloud and Data Center subscriptions at a point in time, so the upfront-versus-ratable revenue pattern does not affect it in the same way.

ARR is useful, but not clean enough to answer every migration question. It combines both deployments. A Data Center subscription can still sit inside ARR before migration; a Cloud subscription enters the same stock after migration. Contract start and end dates, renewal, upgrades and invoice timing also affect the measure. Atlassian explicitly says ARR should be viewed independently of GAAP revenue.

RPO supplies another witness. Year-end remaining performance obligations were US$4.817 billion, up 44%, while current RPO grew 27%. The fiscal-Q4 shareholder letter connects that growth to larger, longer enterprise agreements. But RPO is contracted revenue not yet recognized. About 65% is expected within twelve months and the rest later. It cannot be inserted into the fiscal-2027 revenue bridge, treated as cash or used as proof that a migration is complete.

Cash is the fourth witness. Operating cash flow was US$1.353 billion and free cash flow US$1.319 billion in fiscal 2026. The EOL allocation did not create cash by itself, and the 10-K warns that contract terms, billing and collection run on their own schedules. A strong transition should eventually make the Cloud relationship visible across revenue, ARR, renewals and cash, not just one of them.

The deadline is part of the economics

The Atlassian Ascend timetable has already removed new Data Center sales to new customers. Sales and expansions to existing customers are due to stop in March 2028. Support generally ends in March 2029, with an extended-maintenance path for certain customers.

Those dates create bargaining power and execution risk. Atlassian can offer migration incentives, price Data Center, expand isolated and regulated Cloud capabilities and decide how much engineering capacity remains on the old product. Customers can renew, migrate, use a hybrid path, reduce seats or select a rival, but their option set narrows as support dates approach.

The customers most likely to remain late are not necessarily the least valuable. They may carry data-residency, isolation, performance, integration or change-control requirements that make migration harder. If Atlassian closes those gaps, the remaining cohort can become a high-value pool of prospective Cloud conversions. If it cannot, the same deadline can become a substitution event.

That is the acceptance test behind the US$311 million reversal. The revenue has already moved on the accounting clock. The customer relationship still has to cross.

Sources