Atlassian hits $1 billion quarterly cloud revenue - and here's why it thinks AI creates more work, not less

The anxiety gripping enterprise software right now is not subtle. The theory du jour is that if AI makes developers more productive, companies will need fewer seats, spend less on tooling, and vendors like Atlassian will suffer. Wall Street has been fickle with the sector, and Atlassian has not been spared. 

But Co-Founder and CEO Mike Cannon-Brookes is making a data-backed argument that the opposite is happening. Customers using AI code generation tools create 5 percent more tasks in Jira, have 5 percent higher monthly active users, and expand their Jira seats 5 percent faster than those who do not. The more AI these companies adopt, the more Atlassian they use. 

That is the central thread running through Atlassian's second quarter fiscal year 2026 earnings. The company has surpassed $6 billion in Annual Run Rate revenue, delivered its first ever $1 billion cloud revenue quarter (up 26% year-over-year), and grown Remaining Performance Obligations (RPO) 44% year-over-year to $3.8 billion. It closed a record number of deals above $1 million in Annual Contract Value (ACV), nearly doubling year-over-year.

More software, more collaboration, more Atlassian

During the latest earnings call, Cannon-Brookes addresses the market jitters head on:

In these type of times when there's a lot of noise, we can forget the fundamentals. Enterprise customers want a platform they can trust.

He then unpacks the reasoning behind the 5 percent statistic. Coding speed accounts for roughly 20% to 30% of a developer's overall job, he argues, so a productivity gain in code generation translates to perhaps a 10% to 15% improvement across the whole engineering organization. But that improvement does not mean teams finish their roadmaps and go home. They come up with more things to build. More software means more complex architectures, more services to operate, more governance to enforce, and more coordination between the humans responsible for all of it. As he explains:

You're also creating a lot more technology, a lot more software and services, which makes your architecture more complicated. It gives you more things to manage [...] If you create more software, you're going to have more management, more overhead, more collaboration.

He also notes that the companies adopting AI coding tools tend to be the most growth-oriented in the first place – the ones pushing boundaries and expanding fastest. Those are the companies Atlassian wants as its leading indicators.

AI products and the pricing question

Rovo – Atlassian's AI capability layer – has surpassed five million monthly active users, with millions of agentic workflows running monthly. Underpinning it is the Teamwork Graph, now containing more than 100 billion objects and connections across first and third-party tools, which provides the context that makes Rovo's search, chat, and agent experiences relevant to what customers are actually working on.

The monetization vehicle is the Teamwork Collection – Atlassian's premium bundle including enhanced AI credits alongside Jira, Confluence, and Loom. In under nine months, more than 1,000 customers have upgraded and purchased over one million seats, with double-digit seat expansion rates compared to standalone buyers.

On pricing model concerns, Cannon-Brookes is direct – customers prefer predictable seat-based pricing, and Atlassian's job is to manage AI costs within that envelope. Chief Financial Officer Joe Binz points to continued gross margin improvement even while delivering Rovo at scale, which suggests the engineering teams are keeping inference costs under control.

The Service Collection remains Atlassian's fastest-growing product at scale: more than 65,000 customers, 50% of the Fortune 500, enterprise growth exceeding 60% year-over-year. More than two-thirds of Service Collection customers now use it for non-IT functions – Human Resources, Finance, workplace management. More than 40% of agentic workflows built on the platform in the past six months have been in Service Collection customers, and the company launched Customer Service Management within the quarter to open up further use cases.

Agents, integrations, and expanding the surface area

When asked about Anthropic's Cowork as a potential competitive threat, Cannon-Brookes sees it from a different angle. Anthropic is a partner, he notes, and new AI tools are additional consumers of the Teamwork Graph's context layer, not replacements for it. Atlassian's Model Context Protocol (MCP) server means tools like Cowork draw on and contribute back to the Teamwork Graph – making the data layer more valuable, not less.

Jira now supports assigning work to Rovo agents, custom-built agents, or agents from external platforms. As Cannon-Brookes explains: 

Our customers don't use models, they use applications. We don't sell chips, we sell apps.

Although it does not come up on the call, Atlassian's acquisition of The Browser Company is worth noting in this context. At Team '25 Europe last year, The Browser Company's Chief Executive Officer Josh Miller and Atlassian's Distinguished Product Manager Sherif Mansour outlined how the combination would pair design-led software – Miller's browser Dia intercepts existing habits rather than demanding new ones – with the governance infrastructure that makes enterprise AI trustworthy. If Cannon-Brookes's integration-first philosophy holds, Dia becomes another surface through which the Teamwork Graph reaches users, in the same way MCP serves external agent platforms. It extends Atlassian's reach without requiring customers to change how they work – which, as Miller observed at the time, is the hardest problem in software adoption.

Cloud migrations contributed a mid-to-high single-digit impact on cloud revenue growth in Q2. Cloud Net Revenue Retention (NRR) has ticked above 120% for the third consecutive quarter, and RPO growth continues to outpace cloud revenue growth – a forward indicator, since those are multiyear commitments. Binz reiterates the medium-term outlook of 20%-plus compounded annual revenue growth through fiscal year 2027, alongside a 25%-plus non-GAAP operating margin commitment. He flags that Data Center revenue will be down meaningfully next year on tough comparatives, but expresses confidence that cloud growth will compensate.

My take

Mizuho's Gregg Moskowitz describes the current software market as a "twilight zone" – a company reports a strong quarter, raises guidance, and still sees shares fall. It is a fair characterization of the disconnect between Atlassian's operational performance and how the market is responding.

Cannon-Brookes is presenting evidence that, for collaboration and workflow tooling, AI adoption correlates with more usage, not less. Is that 5% figure conclusive on its own? No – it comes from self-selected early adopters who are likely growth-oriented anyway, and he acknowledges as much. But the direction is consistent with the broader point that AI-generated output still needs to be managed, governed, tested, and deployed by humans working in teams. The operational evidence – RPO acceleration, NRR trajectory, enterprise deal volume – is difficult to argue with. The market will catch up, or it won't. Atlassian doesn't seem to be waiting around to find out.