Atlassian Data Center EOL: Dates and Planning

Which Atlassian Data Center end-of-life deadline should drive your plan?

March 28, 2029 is end of life for affected products. March 30, 2028 ends expansion purchases for existing customers. If your migration plan needs new licenses or more capacity, work backward from that earlier milestone.

Source: Atlassian's transition timeline, checked October 19, 2026. The 2028 planning target is Avaratak's recommendation, not a universal migration completion requirement.

Everyone in your organization is about to write down the wrong date.

The date they will write down is March 28, 2029 — the day Atlassian says Data Center reaches end of life, licenses expire, and instances become read-only. It sounds like a three-year runway. It reads like a problem for a future budget cycle. And treating it that way is the most expensive mistake a Data Center customer can make this year, because the dates that actually decide your migration all arrive earlier.

The timeline Atlassian has published

Atlassian announced the wind-down of Data Center in September 2025, and the schedule is deliberately phased. Three dates matter, all per Atlassian's official end-of-life timeline.

March 30, 2026 has already passed: new customers can no longer purchase Data Center subscriptions or Data Center Marketplace apps. March 30, 2028 is the operational wall for everyone else: the last day existing customers can buy new Data Center licenses, purchase Data Center Marketplace apps, or expand existing licenses. And March 28, 2029 is end of life: Data Center product licenses and their associated Marketplace app licenses expire, and instances become read-only. Atlassian says it will continue technical support and critical security fixes until that date, renewals are prorated to end on it, and extended maintenance beyond it is available only by exception.

The affected scope includes Jira Software, Jira Service Management, Confluence, and Crowd Data Center and associated apps. Bitbucket and Bamboo Data Center are exceptions: Atlassian says both will continue beyond 2029 through its hybrid licensing approach. Jira Align Data Center is also outside this end-of-life scope.

Why 2028 is the real deadline

Here is the reframe I keep drawing on whiteboards: March 2029 is when the lights go read-only. March 2028 is when your estate freezes. After that date you cannot add a user tier, cannot buy the Marketplace app your new compliance requirement suddenly demands, cannot expand anything. If your headcount is growing — and I hope it is — you would be running a growing business on a licensing model that legally cannot grow with you, for a full year.

Now work the project math backward from there. A well-run enterprise migration takes real time once funded: assessment, data remediation, an app strategy, several test migrations, and the change management nobody budgets enough for. We made the longer version of this argument when we wrote about why the best migrations are boring, and why Atlassian's 2026 transition tooling rewards assessing early: migrations fail on the unremarkable Tuesday six weeks before cutover, not on the weekend itself. Put those pieces together and the honest finish line is not March 2029. It is somewhere in 2028, with budget conversations that belong in this year's planning cycle — not next year's.

There is also a scarcity problem nobody puts in the FAQ. Every Data Center customer is staring at the same fixed date, which means migration expertise — internal and external — gets progressively harder to book as the calendar compresses. The organizations that move in 2026 and 2027 get to choose their pace. The ones that start in 2028 get whatever pace is left.

Your actual options, honestly stated

For most organizations the destination is Atlassian Cloud, and the case is stronger than it was even two years ago — not least because the platform's AI investments, Rovo included, are cloud-only, a point we unpacked when we covered how the Teamwork Graph closes the context gap between AI and your business. Atlassian's Ascend program bundles migration incentives, and some of them are time-boxed, so check the current terms before you assume they will still be on the table in 2028.

But a trusted advisor should say the other parts out loud too. If your constraint is regulatory or data-related, evaluate Atlassian's cloud data residency and compliance posture against your actual obligations — verify, don't assume, in either direction. For government workloads, our IL5 and FedRAMP submission timeline explains the distinction between planned submissions and authorization. If Bitbucket is the center of your Data Center estate, the dual license changes your calculus specifically. And if you genuinely believe you cannot move, that is a conversation to open with Atlassian now, while there is time to be creative, rather than in 2028, when there isn't.

The part that eats schedules: apps

In every enterprise migration I have seen, Marketplace apps are where the timeline goes to die — not because they are hard, but because their complexity surfaces late. So make the app inventory week-one work: which apps have cloud equivalents, which have migration paths, which are quietly load-bearing, and which nobody has opened since 2023.

While you are in there, mind the licensing mechanics on the other side. In cloud, every Marketplace app is licensed at the same user tier as its host product, and crossing a tier boundary reprices the whole portfolio at once — the details are in our guide to Atlassian licensing and how to stop overpaying. A migration is the single best moment you will ever get to shed shelfware, and for the largest organizations it is also the right moment to look at commercial structure — Atlassian's Flex model exists precisely for enterprises whose adoption refuses to hold still.

What I would do this quarter

Four moves, in order. Run the free Portfolio Insights assessment against your instance and let the bad news arrive while it costs nothing. Build the app inventory and score it against cloud availability. Model your cloud licensing at real headcount — including the tier you will actually land in, not the one you hope for. Then pick your migration window deliberately, working backward from a 2028 finish, and put the budget line in this year's plan.

None of that commits you to anything. All of it converts a discovery problem — the expensive kind, discovered on a Saturday — into a scheduling problem, which is the kind organizations are genuinely good at solving.

The read-only date is Atlassian's. The decision date is yours, and it is a lot closer than the countdown clock suggests.

If you're a Data Center shop, I'm curious: what does your app inventory look like — and which part of your estate do you suspect will be the real schedule risk?

Related reading

• Jira Cloud Migration Costs and Partner Selection
• Bitbucket and the Atlassian Software Collection