Some Atlassian AI Features Cost a Thousand Times Less to Run as Gross Margin Rose
Some of Atlassian's features stayed on older models that got very cheap and still work really well.
20VC with Harry Stebbings · Anthropic's Super Bowl Ad: Who Won & Lost? | Sierra Hits $150M ARR: Is Customer Support Too Crowded?Mike Cannon-Brookes, the co-founder of Atlassian, the software company behind workplace tools like Jira and Confluence, says some of "our features are a thousand times cheaper to run than when we introduce them today," while its gross margin rose over six or seven quarters of heavy AI use.
He says the company's inference costs, the computing cost of running AI each time a customer uses it, have fallen, not risen.
Many investors and commentators feared those costs would crush software companies. Cannon-Brookes says Atlassian built a large team to manage the costs, adopts new models quickly, and keeps some features on older models that get very cheap. Those features still work really well as AI features, he says. Atlassian's AI costs have fallen, not risen, even as it deployed a lot of AI.
What was expensive on day one became far cheaper to run.
Questions people ask
What is gross margin?
It is the share of revenue left after the direct cost of delivering a product, including hosting and AI computing costs. If gross margin rises while a company uses more AI, it often suggests the AI is getting cheaper faster than usage is growing, though other factors can also lift margin.
Why would an older AI model cost less to use?
Providers often lower prices over time as efficiency improves and competition grows. Atlassian's approach was to move routine features onto older models once it learned they still worked really well there.