In September 2025 Atlassian announced the end of its self-hosted Data Center products. New customers couldn't buy them from 30 March 2026, existing customers can't expand from 30 March 2028, and the products become read-only on 28 March 2029. Atlassian admits cloud costs more per user, so it offers first-year discounts, long migration trials and credits to move early.
Key takeaways
- Three dates: no new Data Center sales from 30 March 2026, no expansions from 30 March 2028, end of life on 28 March 2029.
- Atlassian's own page says the per-user cost of cloud is higher than Data Center.
- It pays customers to move early: 10% to 20% off the first year of Cloud Enterprise for purchases by June 2027, long trials, credits and dual licensing.
- FY2026 revenue was US$6.57 billion, up 26%, with cloud revenue up 31% in Q4.
I worked in IT support in the late '90s and early 2000s.
Back then, software was a box. You paid a big upfront fee, then a setup fee, then maybe something extra for support and upgrades. Every few years a new version came out and you paid again. Subscription wasn't really a concept.
I tell that story in Chapter One of The Subscription Playbook to show how recent the software subscription really is.
Atlassian, the Sydney-founded company behind Jira and Confluence, lived that whole shift. And in September 2025, it set the date for the last step.
What did Atlassian announce?
Atlassian had already killed its "Server" products. This time it announced the end of the self-hosted Data Center editions too, the versions big companies run on their own infrastructure.
| Date | What happens |
|---|---|
| 30 March 2026 | New customers can no longer buy Data Center subscriptions or Data Center Marketplace apps |
| 30 March 2028 | Existing customers can no longer buy new subscriptions, expansions or upgrades |
| 28 March 2029 | End of life. Subscriptions expire and Data Center products become read-only |
Bitbucket and Bamboo Data Center are exceptions.
That's three and a half years from announcement to end of life. Long, staged and public.
Did Atlassian admit cloud costs more?
Yes, and that's the bit I find most interesting.
Atlassian's own end-of-life page acknowledges that the per-user cost of cloud is higher than Data Center. On its community forum, some administrators said the move to cloud premium tiers meant cost increases of "nearly 50%."
So Atlassian didn't pretend. Instead, it paid customers to move:
- 10% to 20% off the first year of Cloud Enterprise for customers who buy by June 2027;
- migration trials as long as the customer's remaining Data Center term;
- credits for unused maintenance;
- dual licensing for large customers, so they can run both while they move.
That's a migration with the cost spelled out and the transition paid for.
Why will customers move, even at a higher price?
Because the moat is enormous.
In the book's MOATS framework, S stands for Switching. I used TechnologyOne as the example: its enterprise software is so embedded in its customers that its churn was under 1% in 2020. (Here's what TechnologyOne did next.) Customers don't make snap decisions to leave software like that. They see the price, rethink, and talk to the vendor.
Jira and Confluence are the same. Years of tickets, workflows, documentation, integrations and habits. The book calls this the "learning trap." You could move to a competitor. But you'd have to migrate everything and retrain everyone.
Faced with that, a higher cloud bill with a first-year discount looks like the easy option.
My take
Running two business models at once is one of the most expensive things a company can do. I see it in small businesses all the time: a "legacy" plan nobody's allowed to touch, alongside the new one. Atlassian shows the grown-up way to end it. Pick a date far enough out, publish it, pay people to go early, and don't flinch.
What happened to Atlassian's numbers?
The strategy is showing up in the results.
| Measure (fiscal 2026, to 30 June 2026) | Result |
|---|---|
| Revenue | US$6.57 billion, up 26% |
| Q4 cloud revenue | US$1.21 billion, up 31% |
| Subscription ARR | US$6.6 billion, up 23% |
| Profitability | GAAP profitable in Q4 |
Cloud is growing faster than the business as a whole. That's the migration doing its job, and the higher per-user price on cloud plays a part. If you want to understand why investors love that kind of recurring growth, see annual recurring revenue (ARR).
How deep is your moat?
Score your business on the five MOATS factors, including how hard it is for customers to switch.
Where could this go wrong?
Two places.
Customers who resent being pushed. Some will leave rather than pay more. Atlassian is betting that number is small. A business with a weaker moat shouldn't make the same bet without testing it first.
Moving too fast. A six-month deadline on deeply embedded software would have been a disaster. The length of the runway is what makes it acceptable. If you're thinking about a forced move or price rise, the price increase guide covers notice and communication.
What can you steal from Atlassian?
You don't need to run enterprise software. Any business with an old plan, an old service or an old way of doing things can use this.
- Retire legacy offers with dates two to three years out (or months, for simpler services), and publish them well in advance.
- Stage it. First stop selling to new customers. Then stop expansions. Then end it.
- Pay for early migration. Discounts, credits and free overlap periods.
- Be upfront when the new option costs more. Atlassian says it plainly on its own page.
- Make the new option better, not just newer. Customers forgive a higher price far more easily when they can see why.
For a broader look at the model, read the SaaS business model guide and our Atlassian pricing teardown.
What to do next
List every plan, package or legacy price you still support. For each one, ask: what does keeping it cost me in time, systems and confusion?
Then check how hard it would be for those customers to leave if you moved them. That tells you how much runway and incentive they'll need.
MOATS ScorecardScore your business on the five moat factors.Open the free tool →There's more on building switching costs in dig your moat.
Frequently asked questions
Why would a company force customers off a product that still makes money?
Running two models costs a lot. Two codebases, two support teams, two sets of security work, and two pricing systems. Consolidating onto one subscription platform lets a company ship faster and earn more per customer over time. The trade-off is short-term friction and some customers leaving, which is why long notice and incentives matter.
How much notice should you give before retiring a plan?
It depends on how embedded the product is. Atlassian gave existing customers about two and a half years before expansions stop, and three and a half years before end of life, because large companies run their operations on Jira and Confluence. A small business retiring a service plan might need three to twelve months. Longer is kinder, and usually cheaper.
Should you tell customers the new option costs more?
Yes. Atlassian says it plainly on its own end-of-life page. Customers will work it out anyway, and being upfront builds trust. Pair the honesty with clear reasons (what they gain) and real help with the cost, such as discounts, credits or a free overlap period.
What happens to customers who don't migrate?
For Atlassian, Data Center products become read-only at end of life, so customers can see their data but can't keep working in it. For a smaller business, it usually means the old plan simply ends. Either way, tell people exactly what happens on the final date, and remind them several times before it arrives.
Get both books free
The Subscription Playbook (ebook + audiobook) and Feed a Starving Crowd, instantly. In Australia? Claim the free hardcover.
Robert Coorey, MBA
Robert is the author of The Subscription Playbook and Feed a Starving Crowd and a Director at BPAG Accountants in Sydney. More about Robert.
Sources
- Atlassian, Data Center End of Life (accessed Oct 2026)
- Adaptavist, Atlassian sunsets Data Center in March 2029 (10 Sep 2025)
- Atlassian Community, Good-Bye Data Center discussion (Sep 2025)
- Atlassian, Q4 and FY2026 results (6 Aug 2026, SEC Exhibit 99.1)
Facts and prices checked 27 October 2026. Spotted something out of date? Tell us.