Most people tell the Atlassian story from the IPO bell. Start earlier, with an email that almost nobody answered.
In 2001 Mike Cannon-Brookes, a university student at UNSW, sent a note to his classmates asking if anyone wanted to start a technology company. One person wrote back: Scott Farquhar. That single reply is the whole origin of a business now worth tens of billions.
He was sent to boarding school in England before landing at Cranbrook in Sydney, and then UNSW on a co-op scholarship.
The reason they gave later was strikingly modest. The two have said they wanted to earn what a big employer would pay a graduate, about A$48,000, without working for anyone else. No grand mission. A salary, and the freedom to skip the interviews.
Cannon-Brookes was not a typical Sydney kid. Born in New Haven, Connecticut, to a British banking executive and his wife, he was six months old when the family moved to Taiwan, then Hong Kong at three. He was sent to boarding school in England before landing at Cranbrook in Sydney, and then UNSW on a co-op scholarship. What almost nobody remembers is that this was not his first company. Before Atlassian he had co-founded The Bookmark Box, a bookmark-management tool, with classmate Niki Scevak, and sold it to Blink.com in 2000. He had already learned that a small web product could find a buyer.
Atlassian began in 2002 on about A$10,000 of credit-card debt. They had no investors and no sales staff. The early business was supporting other companies' customer service teams, which demanded round-the-clock availability. Frustrated by the bug-tracking software on offer, they built their own, called Jira, and then made the unusual decision to stop doing services and sell it. Confluence followed in 2004.
The rule they set was that the software should be bought, not sold. A customer could find it, try it and pay on the website. That choice kept costs tiny and let them run for years without outside money. When they finally took capital, in July 2010, it was US$60 million from Accel Partners, and by then revenue was heading toward US$100 million.
The price of growth came later. A Confluence flaw in 2023, a data exposure in 2019, and layoffs of about 500 in 2023 and roughly 10% of staff in March 2026 showed that a company this big is no longer a two-student experiment. Farquhar stepped back as co-CEO in August 2024, leaving Cannon-Brookes alone at the top. The company reported US$5.22 billion in revenue for fiscal 2025, though still a net loss of US$257 million.
The lesson is in the first line of the story. The company was started by someone who had already sold one small thing, and who asked a room of people for a partner, got one answer, and took it.