Atlassian's privacy hierarchy
Atlassian will soon train its AI on customer data. Unless they pay enough.
Last week Atlassian chief executive and co-founder, Mike Cannon-Brookes, declared artificial intelligence was the "best thing to ever happen" to the $US38 billion software giant after the company narrowed its statutory losses by almost 80 per cent to $US54 million for the 12 months to June 30.
Atlassian is one of many software as a service companies battling to find its place in a world where newer, smaller AI businesses claim to deliver a superior product at a fraction of the cost. Cannon-Brookes has cause to celebrate the narrative shifting after a seemingly constant stream of negative headlines since the departure of co-founder and former co-CEO Scott Farquhar two years ago.
But the company is hardly out of the woods. Revenue may have surged 26 per cent to $US6.6 billion over the year but Atlassian shares are still trading 20 per cent lower than their peak price of $US184 in 2025, before the onset of the so-called SaaSpocalypse. And Atlassian has not delivered a full year GAAP profit for shareholders since its first year as a listed company a decade ago.
Cannon-Brookes told investors three months ago that this will all change in the 2027 financial year. He’s already moved to reduce the balance of stock-based compensation – one of the main reasons behind its decade of losses [[ The company recognised $US1.6 billion in stock based compensation in FY26 which doesn’t account for another $US279.5 million in restructuring costs following successive layoffs.]]. He also says it’s time the market realises Atlassian has "25 years of context no one else has".
It will certainly realise soon, because Atlassian has worked out how to cash that context in. Atlassian builds tools for project management, software development and collaboration. Next week, it will start siphoning data from the bulk of its 300,000 customers to improve its own artificial intelligence features.
