Part IThe Story
There was no salesman. That was the strange part, the part that later became company lore and was retold so often it hardened into scripture. Some morning in 2003, in a Sydney office that two twenty-three-year-olds were paying for with what was left of a $10,000 credit-card limit, a fax machine had done its quiet overnight work, and lying in the tray was an order for a piece of bug-tracking software called Jira. The customer was American Airlines. The two founders looked at each other. "Did you talk to them?" one asked. "No, did you?" Nobody had. Nobody had flown to Dallas or Fort Worth, nobody had taken a procurement officer to lunch, nobody had sent a deck. An airline on the other side of the Pacific had found a product on a website run out of Australia, decided it was good, and bought it, apparently during its own business day, while the people who made it were asleep. "Holy shit," went the reported response. "American Airlines saw Jira on our website and bought it just like that!"
The sources do not record which of them pulled the page from the tray. It hardly matters. The fax contained, in compressed form, nearly everything Mike Cannon-Brookes would spend the next twenty-three years arguing: that distance could be an asset rather than a sentence; that a good enough thing would be bought rather than sold; that the most important events in his company would happen in a time zone he was not awake for; and that the Pacific Ocean, which had cut Australia off from the centres of money and power for two centuries, was now, for the first time, crossable at the speed of a phone line. It also contained a question he has never quite answered, and which runs beneath every chapter of his life — the billionaire years, the climate crusades, the Formula One livery, the falling-out with his best friend — like a fault line under a city. Who, exactly, was on the other end? And where, exactly, was he?
He was born in New Haven, Connecticut. He grew up in Taiwan and Hong Kong and at an English boarding school. He became, in his own words before an Australian Senate committee, "a patriotic Australian." The fax machine was a fitting first trophy for a man whose whole life had been a long-distance transmission.
By the Numbers
From a Credit Card to a Continent
$10,000Credit-card debt that financed Atlassian's start in 2002
A$48,000The founders' original ambition: a graduate's starting salary
$55MCash in the bank when Atlassian took its first outside round, $60M from Accel, in 2010
$4.37BMarket capitalisation at the Nasdaq IPO, 10 December 2015
$5.22BAtlassian revenue, fiscal 2025
300,000+Customers in more than 200 countries and territories
4,200 kmProposed length of Sun Cable's subsea link from the Northern Territory to Singapore
A childhood measured in frequent flyer points
The father came first, and it is worth spending a moment with him, because the son's life rhymes with his in ways the son rarely mentions. Michael Cannon-Brookes the elder was an English banker with a gift for planting flags. In 1984 he arrived in Australia from Hong Kong as managing director of Citibank, charged with securing the bank's Australian licence and building its presence from nothing. He later held senior roles at Westpac, ran IBM's banking division, and served as IBM's vice president of global strategy for growth markets from Shanghai; he was the first managing director of the law firm Freehills, which he helped turn from a traditional practice into a modern professional-services business. In July 2026 an organisation called Global Australians gave him a lifetime award, and its chief executive, Kate McQuestin, said that wherever Australians gathered — Singapore, Hong Kong, New York, London, Sydney — people told her stories about him: "A piece of advice he gave. A door he opened." Accepting it, the elder Cannon-Brookes said that "mentoring has never felt like a responsibility to me. It has always been a privilege." He is, in other words, a professional expatriate who made a vocation out of helping other expatriates belong somewhere.
The son was born on 17 November 1979, the youngest of three and the only boy, to this man and his wife, Helen. At six months he was in Taiwan. At three, Hong Kong. Then, when the family relocated to Sydney, he was sent to boarding school in England, and for several years — until his parents decided to settle — a primary-school child made the long-haul flight between England and Australia four times a year. Sit with that for a moment. Twenty-odd hours each way, four round trips annually, at an age when most children are learning to ride bicycles on their own street. The boy was not from anywhere. He was from the aisle seat.
What he got from it, characteristically, was a transaction. "Because of the flying, I had tons of Frequent Flyer points and back then points were worth something," he later told a reporter, "so I got the catalogue every year and got some shit." One year the catalogue offered a computer. He was eight or nine. His parents were Apple people with Macintoshes in the lounge room, but this one was his: an Amstrad PC20, "a real hunk of junk to be honest but . . . it sparked my interest in computers. It was a big, big deal." The first machine Mike Cannon-Brookes ever owned was paid for with the residue of distance — the loyalty points of a child shuttled across hemispheres. By 1991, before the World Wide Web existed in any public sense, he had talked his parents into putting the internet in the family home.
He finished school at Cranbrook, one of the most exclusive boys' schools in Sydney's eastern suburbs, a place of sandstone and ambition. It gave him, by one account, a certain manner. We'll come to who noticed.
The only one who replied
On the other side of Sydney — and in Sydney the other side means something, the city's class geography running roughly from the harbour mansions of the east to the brick-veneer suburbs of the west — an eleven-year-old named Scott Farquhar wanted a computer so badly he cried for it. He was born in December 1979 in Parramatta, the eldest of four children in a working-class family, and the recession of the early 1990s was biting. "I wanted a computer really badly because my friend had one and I knew you could play games on it," he recalled. "I remember emotionally blackmailing my parents and at one stage crying to get a computer – I feel horrible now, because my parents couldn't really afford one." A year or two later his father got hold of an old Wang through work. It was so obsolete it couldn't run Microsoft DOS. Farquhar spent nearly a year trying to make it play games. He never succeeded.
It is hard to invent a sharper pair of origin stories. One boy gets a computer free, from an airline catalogue, as the by-product of a privileged and peripatetic childhood. The other gets one secondhand, through his father's job, and it doesn't work, and he spends a year trying anyway. One went to Cranbrook. The other went to James Ruse Agricultural High School, a selective public school whose computing education he described as terrible to the point of non-existent. Both were born within weeks of each other in late 1979. Both, in 1998, at eighteen, landed in the same scholarship course in business information technology at the University of New South Wales — a cohort of about forty students who, as one of them put it, "came from all walks of life but we were all nerds who loved computers."
They met on the first day. They did not click. "People from Cranbrook," Farquhar said later, with the precision of a man who has been waiting years to say it, "had a bit of a swagger about them before they started university."
The scholarship was a co-op programme, which meant a series of six-month industry placements inside real companies. This turned out to be the formative experience — not the lectures but the placements, where the students were made to use the software that real companies used. Cannon-Brookes's verdict, delivered fourteen years later on a podcast: "it was just a horrible experience with the software we were made to use." File that. It is the seed of everything.
He was, already, a starter of things. With a classmate, Niki Scevak, he co-founded an online bookmarking tool called The Bookmark Box — "There were a lot of those," Kara Swisher reminded him, and he cheerfully agreed — and sold it to Blink.com in New York in 2000. "We had a little exit, it was a one-year journey; we learned a lot." He then did something nobody seems to remember about him: he "went and was a bit of a journalist for a while — wrote about tech and various other things." A brief interlude as an observer before the long career as the observed.
Then, in 2001, as graduation approached, he sent an email to his classmates asking whether anyone wanted to help him start a technology company. The forty-odd nerds who loved computers had options; the big consulting firms — IBM, PwC — were hiring, and the starting salaries were real. Scott Farquhar, the kid from Parramatta who had cried for a computer and once, for reasons unrecorded, wore a mohawk, was the only one who replied.
They would be best man at each other's weddings. They would buy houses next door to each other on the most expensive stretch of harbour in the country. They would share the title of chief executive for twenty-two years, an arrangement that the business press treated as a curiosity and that turned out to be the single most load-bearing structure in the company. And one day, in 2024, one of them would leave, and a book would be written about the reasons, and neither of them would talk to its authors.
But that is the end of a story whose beginning was a reply-all that only one person answered.
Forty-eight thousand dollars and a Titan
Every founding myth in technology is supposed to come with a vision. Cannon-Brookes has always, with something between modesty and mischief, refused to supply one. "Famously we didn't have one of those ambitious starting stories," he told Swisher in 2016. She suggested he could buy one in Silicon Valley. The real goal was arithmetic. "A lot of our colleagues went to work for big consulting companies — IBM, PWC — and our only goal was actually to pay ourselves as much as they were getting paid, which was $48,000 a year." Elsewhere the figure is quoted as $48,500; in a 2017 TEDx talk he added a second criterion — "not to have to wear a suit to work everyday." In a 2012 interview he gave the plainest version: "neither of us wanted to get a real job," and the ambition was "just a survive through year one and not look like fools."
The Australian press, which loves a self-deprecating rich man almost as much as it loves cutting a tall poppy down, christened them "the accidental billionaires," and the label stuck. It is half true. The ambition really was modest. The execution was not accidental at all.
They registered the name in 2001. It came from Atlas, the Titan condemned to hold up the sky — specifically, the founders have said, from the bronze statue of Atlas at Rockefeller Center in New York, a city neither of them yet worked in. The early tagline was a "high commitment to legendary service," and the Titan was meant as "kind of the original service to the world." Cannon-Brookes, ever the wordsmith, noted that they had "turned it into an adjective. Sort of an Atlassian effort." It is worth pausing on the choice. Two kids in a share house in Randwick picked, as their mascot, a figure whose defining characteristic is that he carries an impossible weight and is not permitted to put it down.
The share houses were in Randwick and Glebe. When they graduated to an office, it was above a sex shop. They financed the company on credit cards. They did not pay themselves for six to nine months — "Profitable from day one clearly means on day two you weren't paying yourself anything" — and then, for about two years, "a couple of hundred dollars a week sort of thing." The $48,000 was, for a while, a long way off.
What they were actually selling, at first, was labour. Atlassian began as a services business supporting other companies' customer-support teams, which meant the two founders took calls at all hours. It wasn't much fun. The bug-tracking software they used to manage the work was bad. The industry leader was an open-source tool called Bugzilla, so they built their own and gave it a name that was a joke at the incumbent's expense: Jira, a play on Gojira, the Japanese title of the original 1954 Godzilla. A monster to eat the bug. They quickly decided — and Cannon-Brookes still frames this as the company's first and most important lesson — that "building your own products beats fixing other people's products." Twenty-three years later he would remind the technology analyst Ben Thompson, who had once done a consulting stint for Atlassian, that "we started as a consulting and services business and then decided that software was a better business model." You went the scalability route, he told Thompson. So did we.
Jira 1.0 shipped in 2002. On 25 March 2004 they released a second product, Confluence, a corporate wiki pitched as "an application that was built to the requirements of an enterprise knowledge management system, without losing the essential, powerful simplicity of the wiki." Jira tracked the work; Confluence held the knowledge around the work; the two fed each other. "We had two rocket engines driving us along, not just one," Cannon-Brookes said. Within three years of founding, the company was profitable without a dollar of venture capital.
Not, he insists, out of virtue.
— Mike Cannon-Brookes, Recode Decode, 2016In 2002, A) it was nuclear winter for tech, B) we were two knuckleheads with no experience, and C) we were in Australia. We didn't even bother, to be honest, to try to get funding in the early days.
"It was either become profitable and grow, or die," he went on. "And so the ones that made it through that gauntlet have been very, very successful." Silicon Valley, from Sydney in 2002, "seemed a mile away" — he meant it figuratively, but the literal distance is about twelve thousand kilometres, and it was the literal distance that shaped the business.
Software should be bought, not sold
Here is the problem the fax machine solved. Most of Atlassian's potential customers were in the United States. The founders were in Australia. The conventional way to sell software to American companies in 2003 was to hire salespeople — expensive, mostly American, mostly in America — who would fly around and take the CIO to lunch. Atlassian could not afford them. And even if it could, an Australian company running a field sales force across the Pacific would be fighting with one arm tied behind its back, against incumbents who lived in the same zip codes as their buyers.
So they invested in the website instead. They wrote documentation good enough that a stranger could evaluate the product alone. They published the prices — in a world where enterprise software prices were a secret extracted through negotiation. And they set those prices low. Larry Marshall, then the Australian managing director of the venture firm Southern Cross Venture Partners, later described the trick with admiration: "They recognised very early on that on the web you could sell and you could target large corporations by pricing your product at a level where an engineer could just put it on their credit card and not have to get approval to buy it. I think they started at less than $3000 was the price back then." The engineer bought it; the engineer's colleagues used it; by the time anyone in procurement noticed, it was infrastructure. "Without the web you couldn't have done that," Marshall said, "you'd have to be knocking on doors."
Cannon-Brookes's slogan for this, repeated until it became the company's creed, was that Atlassian's software was "bought, not sold." It sounds like a marketing line. It was a confession of geography. They sold "on a team by team basis," he told Swisher; "we don't try to sell to an entire company at a single time." The model would later have a name — product-led growth — and a thousand companies built on its template. In 2003 it was what two knuckleheads in Australia did because they had no alternative.
They did try the alternative, once. Momentum was building, most customers were American, and it seemed like the moment for a presence there, so Atlassian opened a tiny office in New York. The leadership realised almost immediately that they had over-extended, and closed it. The company's own retrospective files this under a lesson — "The sooner you acknowledge a wrong turn, the sooner you'll get back on track" — and notes, with a dryness I find endearing, that it "was good practice in making tough calls – of which there would be many to follow." They reopened in the United States in San Francisco instead, which had a deeper pool of technical talent. The Rockefeller Center Atlas would have to be admired from a distance.
The numbers that followed seem, in retrospect, almost too clean. By 2011 Atlassian's revenue was $102 million, a 35 per cent increase on the prior year, with no salespeople at all; TechCrunch ran the absence as the headline. By 2012 it had about 500 staff, half of them in Sydney, with offices in San Francisco, Amsterdam, Gdansk, Porto Alegre and Kuala Lumpur, and had grown at roughly 47 per cent a year since 2007. That year, short of engineers, it drove a bus around Europe trying to hire fifteen of them in fifteen days. More than a thousand people applied. "There's more lines of code in a Ford than Facebook and Twitter combined," Cannon-Brookes said at the time, a line that was both a sales pitch and a theory of history: the world was becoming software, and the people making software needed somewhere to track their bugs.
In 2005 the company had started holding a quarterly hackathon called ShipIt — twenty-four hours in which business as usual stopped and anyone could build anything with anyone. It sounds like a perk. It was a pipeline. Jira Service Desk, later Jira Service Management, which would become Atlassian's fastest-growing product, began as a ShipIt project. Two years later, in 2007, the company wrote down its values, which remain the most quoted thing about it and are reproduced here exactly, profanity and typographical cussing intact: Open company, no bullshit. Build everything with heart and balance. Don't #@!% the customer. Play as a team. Be the change you seek.
In 2006 Ernst & Young named the two of them Australian Entrepreneur of the Year — the youngest people ever to win. Cannon-Brookes would later say that standing on that stage was one of the moments he felt most like a fraud.
"We don't want your money"
By the middle of the decade the venture capitalists came knocking, and the founders developed a routine. "We just always asked them the question, 'Why would we take money? We're growing, we're profitable, what would change here?'" Cannon-Brookes recalled. "And they'd have a whole host of reasons, and we wouldn't believe them and then we'd move on."
It went on like this for eight years. Then, in July 2010, Atlassian took $60 million from Accel Partners, at the time one of the largest venture investments ever made in an Australian company, and the details of the transaction are so contrary to the normal physics of venture capital that they're worth stating precisely. Atlassian had $55 million in the bank. It did not need the money. The founders wanted Accel to invest less — they "would have taken $5 million." Accel wanted to put in more. "It was a very strange negotiation, in reverse," Cannon-Brookes said, and dramatised it for Swisher: "'We don't want your money.' 'Yes, you do.'" The round was structured largely as a secondary — money that went to existing shareholders rather than into the company — and Accel took, in his words, "a pretty small piece."
Why bother at all? Because, seven or eight years in, after being "heads down for all that time," they had "sort of looked up" and realised they had built something special and needed help. They were thinking about going public. They were trying to build a board, and they discovered that a company that had never raised money was strangely hard to staff with directors. Experienced people would look at them and ask, "Oh, well how come you have no funding?" The absence of venture capital, which had been their great strength, read to the establishment as a missing credential. The Accel money was, in effect, a passport stamp.
The rest of the pre-IPO architecture followed the same logic: plant the flags the larger world requires, keep control. In July 2012 Doug Burgum, a software executive, became chairman. In April 2014 T. Rowe Price bought $150 million of shares from employees and early investors at a $3.3 billion valuation; the founders sold nothing and were believed to own about 78 per cent of the company between them. The same week, the University of New South Wales made Cannon-Brookes an adjunct professor, unpaid, and he used the occasion to say something that would echo through the next decade of his public life: "Eventually, Australia is going to run out of rocks to dig up. The future of the world is about technology."
In 2014 the company redomiciled its parent to the United Kingdom, a move Farquhar had to justify to a sceptical Australian press; the headquarters remained, in fact, in Sydney. On 10 December 2015 Atlassian listed on Nasdaq under the ticker TEAM, valued at $4.37 billion. Its fiscal 2015 sales had been $320 million, its net income $6.8 million. It was, for that year, a profitable software company going public, which was itself almost an anachronism.
The New York Times, writing about the founders a few years later, called Atlassian a "very boring software company." It had made them the first technology-startup billionaires in Australian history — the youngest people on the Forbes Australian rich list, household names in a country that had never had a household-name software founder — and the gap between those two facts was the whole of their situation. In America they were the people who made the bug tracker. At home they were proof of concept for a country.
Fortune, putting them on its 2016 40 Under 40 list at thirty-six, described the division of labour: Farquhar headed engineering and research; Cannon-Brookes ran product design and sales and marketing. The kid with the broken Wang ran the machine. The kid from the frequent-flyer catalogue ran the story.
The impostor in the business lounge
In June 2017, at TEDxSydney, in front of about four thousand people, the newly minted billionaire gave a talk about how he had felt like a fraud for fifteen years. It has since been played more than 3.6 million times on TED's site alone.
"Have you ever felt out of your depth? Like a fraud?" he asked, and then — in a phrase the press later had to asterisk — "And just kind of guessed/bulls****ed your way through the situation? Petrified that, at any time, someone was gonna call you on it?" He listed the moments. Interviewing Atlassian's first-ever HR employee, a function he knew nothing about. Accepting the Entrepreneur of the Year award in 2006. And meeting his wife.
That last story has a setting worth noticing. He met Annie Todd, an American fashion designer, in a Qantas lounge, on a flight from Sydney to San Francisco — the commuter's route between the two halves of his company. They married in 2010. The boy whose first computer came out of a Qantas points catalogue met the mother of his four children in a Qantas business lounge. The aisle seat, again.
His argument in the talk was not that impostor syndrome should be cured but that it should be read, like an instrument. He laid it out for a wire reporter afterward with unusual clarity:
— Mike Cannon-Brookes, to AAP, June 2017If you don't feel like an imposter, you're either within your limits - so you're not learning, you're doing something you know how to do and you're well in the comfort zone - or you are out of your depth and you don't know.
"I think some people freeze," he went on. "If you're aware of it, and you know what to do with it, that's the hardest thing." His prescription was to doubt the idea rather than the self — "questioning your ideas rather than yourself" — and to keep putting yourself into situations that provoked the feeling. "That's how I think certainly I get enjoyment in life, chuck yourself in the deep end and pretend you don't look like a flailing idiot." He named his heroes: the writer Neil Gaiman, Franklin D. Roosevelt, Alexander Hamilton. He confessed to "this weird obsession with US presidents." Of Roosevelt and Hamilton he said, "You find these people who have done an amazing amount of things - it wasn't like they did one thing, they just kept doing amazing things throughout."
Hamilton, of course, was the most famous immigrant in American founding mythology — born in the Caribbean, arrived with nothing, wrote his way into the centre. Roosevelt carried a weight. You do not have to push the parallel far. A man who was born in Connecticut, raised in Asian colonial capitals and an English dormitory, and who did not become Australian by birth, felt for fifteen years that he had wandered into a room he hadn't been invited to.
"This is a big journey," he said that week. "There's no ending, there's no destination here, and life's a bit the same way."
It's a line that reads as wisdom and functions as diagnosis. A person without a hometown has no destination either.
A hundred days or it's free
Within months of the TEDx talk, the impostor jumped into a pool he had no business swimming in.
In 2016 South Australia had suffered a statewide blackout, and by early 2017 the state's electricity system was a national political wound. Elon Musk's Tesla had begun offering grid-scale batteries. Cannon-Brookes, on Twitter, asked Musk how serious he was. Musk replied with a promise that became famous: Tesla would have a 100-megawatt system installed and operating within 100 days of contract signing, "or it is free." Cannon-Brookes pushed; the state's Labor government took up the offer. He had no stake in the battery and no expertise in grid stabilisation. What he had was attention, and the willingness to spend it.
The battery cost $90 million. It was delivered on time. Two years later, on a stage in Brisbane next to Al Gore, Cannon-Brookes reported that it had saved South Australia $33 million in stabilisation costs and was very profitable. He was smiling, an observer noted, which you don't often see people do when they talk about climate change. He was smiling because of the names the battery had been called. One senior politician had derided it as "the Big Banana of energy" — a reference to a roadside fibreglass tourist attraction — and another as "the Kim Kardashian of the energy world." He didn't name the speakers. He didn't have to: the first was Scott Morrison, by then prime minister; the second was the then energy minister, Matt Canavan. "Oh, it got a whole bunch of rude names," Cannon-Brookes told the audience of seven hundred.
What he valued in the battery wasn't the megawatts. It was the demonstration. "Once it's done, everyone goes, 'Oh, it's possible and it's profitable.' Then you get a whole rush of projects behind it, doing similar things." He called it a "lighthouse project." Note the metaphor: a lighthouse is a structure on the edge of a continent, built to tell ships far away where the land is.
In March 2018 he went to Canberra. The Senate had set up a committee on the future of work, and he appeared before it — "the first time I've done anything like this" — with an opening statement that reads, now, like a manifesto he had been drafting since the frequent-flyer years. He introduced himself: "I'm also a patriotic Australian, an entrepreneur, and a heavy technology investor. I'm also a dad. I want to see our country continue to grow and prosper so my kids can stay in Australia." Australia generated about 1 per cent of global GDP, he said; to stay rich, it needed to be a primary producer of 1 per cent of the world's technology, and it was nowhere near. "Do we want to be a primary manufacturer of technology, or a consumer of it?" Atlassian could start any new team of ten anywhere — "in Austin, in San Francisco, in New York, in Gdansk, in Kiev or… in Sydney." And then the sentence that every Australian technology lobbyist has quoted since: "As a country, we want to be a net exporter of products, not a net exporter of people."
The phrasing is worth reading twice. It came from a man who had himself been imported. My kids can stay in Australia. The boarding-school boy on the four-times-a-year flight had decided, at last, where he was from — and was now lobbying the parliament to make sure his own children never had to make the trip.
That same year, 2018, he bought Fairwater.
Exporting the sunshine
He had started buying houses in 2016 — a $7.05 million 1936 Leslie Wilkinson house called SeaDragon — and in September 2018 he bought the most expensive residential property in the country: Fairwater, a harbourside estate in Point Piper, for roughly A$100 million. The following day, according to the property press, he paid $17 million for Verona, another Wilkinson house, built in 1923, in Double Bay. In 2019 he bought a house near Fairwater for $12 million. Next door to Fairwater was Elaine, the Fairfax family's ancestral home since 1891, empty for nearly twenty years, which Scott Farquhar had bought in 2017 for somewhere between $71 million and $75 million depending on who was counting (the Fairfaxes, it was reported, had given him a sentimental discount). The two kids who had met in a scholarship cohort and disliked each other's swagger were now neighbours on the most expensive waterline in Australia.
He had, by then, also begun to invest seriously. In 2016 he set up a family office and called it Grok Ventures, after a word from a science-fiction novel. Through it he backed the improbable — the Australian Financial Review, over a 2019 lunch, catalogued investments "from self-driving vehicles to protein-from-maggots to help feed the world." He chaired the venture firm Blackbird. He sat on the boards of the self-driving company Zoox and the literacy nonprofit Room to Read. In 2019 he tweeted the sentences that marked the end of any pretence that he was a businessman with opinions rather than an activist with a business: "We face the greatest human challenge for generations and we're not improving our targets? We are using trickery and misleading about our national emissions. ... In the past, I believe Australia has always stepped up when asked. We've never dodged our global responsibilities, until now."
And he had found a project that matched the scale of his rhetoric. In September 2019, on the sidelines of the United Nations climate forum in New York — the city of the bronze Atlas — he announced he would put part of his personal fortune into a company called Sun Cable, which planned the world's biggest solar farm, its biggest battery, and a 3,000-kilometre cable to export the power to Asia. The budget was $25 billion. The battery, he boasted at a conference that year, would be "insanely large."
The plan grew in the telling. By 2022 the solar farm was to cover 12,000 hectares near Elliott in the Northern Territory — "a patch of nothing," one columnist wrote — generating 20 gigawatts. An overland line would carry power 800 kilometres to Darwin; from there a subsea cable, now 4,200 kilometres long, would run through Indonesian waters to Singapore, supplying about 15 per cent of the city-state's electricity. The price tag had risen to $30 billion or more. The cable alone, Bloomberg's David Fickling calculated, would be about six times longer than the current record-holder linking the UK and Norway under the shallow North Sea, and twice as deep as the deepest existing power cable, which runs from mainland Italy to Sardinia.
In March 2022 Cannon-Brookes and Andrew Forrest were the two largest contributors to a $210 million capital raise for the company. A word on Forrest, because he is the other Australian billionaire with a cable to sell. He is the iron-ore magnate known as "Twiggy," who built Fortescue by defying the incumbents of the Pilbara and who had once, more than a decade earlier, stood outside Parliament House in a high-vis vest beside Gina Rinehart to help kill a mining tax. A mining man who had turned toward green energy; a software man who had come to climate through a tweet to Elon Musk. Both of them, as one reporter put it, had "a robust sense of their own abilities and business judgement." The shareholder agreement gave each of them equal rights and equal vetoes.
It lasted less than a year. The funding was structured in tranches, released when Sun Cable's management — led by its founder, David Griffin — hit milestones. In November 2022 a milestone tied to Indonesian permits was missed. A $65 million instalment was held back while the two billionaires argued about the terms. On 11 January 2023 Sun Cable entered voluntary administration, citing "the absence of alignment with the objectives of all shareholders." The Sydney Morning Herald reached for Icarus. "There was always a whiff of singed feathers and burnt wax about the Sun Cable project," its columnist wrote. Fickling, in a piece titled for the billionaires' dreams that "won't come true," diagnosed "a bad case of billionaire's disease" and closed with a phrase every Australian schoolchild knows, from Geoffrey Blainey's history of the country: "A nation long ruled by the tyranny of distance will face the same constraints in the net-zero era."
That phrase lands differently on Cannon-Brookes than on anyone else. His entire company had been built on the proposition that the tyranny of distance had been repealed — that a fax could cross the Pacific overnight and an airline could buy software from a share house. Sun Cable was the same thesis, translated from bits into electrons. Fickling's point was that electrons are not bits. Cable doesn't scale like code.
Cannon-Brookes bought it anyway. In May 2023, in a closed bidding process, a consortium of Grok Ventures and the infrastructure investor Quinbrook won control of the collapsed company. Forrest's Squadron Energy had withdrawn, saying it was "unconvinced of the commercial viability" of sending power to Asia. Cannon-Brookes, as he so often does when the world tells him he's being unreasonable, issued a statement that was half policy and half pep talk:
— Mike Cannon-Brookes, on winning control of Sun Cable, May 2023We've always believed in the possibilities Sun Cable presents in exporting our boundless sunshine, and what it could mean for Australia. It's time to stretch our country's ambition. We need to take big swings if we are going to be a renewable energy superpower. So swing we will.
Exporting our boundless sunshine. The man had spent his childhood being exported. Now he wanted the country to export its sky.
Pens down
The other swing came first, and it was the one that made him, for a few months in 2022, the most talked-about businessman in Australia.
In June 2021 AGL Energy — the country's largest electricity generator and its single largest greenhouse emitter, responsible for about 8 per cent of national emissions — announced a plan to split in two. One company, AGL Australia, would hold the retail business and cleaner generation; the other, to be called Accel Energy (a name that must have caused a flicker of recognition in a man whose own first investor was Accel), would hold the giant coal plants. AGL's chairman, Peter Botten, told investors that "the winds of change in the electricity market have been substantially faster than many people have anticipated," and that the board was "very committed to turning this ship around." Cannon-Brookes looked at it and thought it "didn't make any sense." In fact, it was "going in the opposite direction of what we expected."
He assembled bankers, lawyers and accountants. The team called itself Project Arise. Then one of the advisers heard that Brookfield, the Canadian asset manager, was weighing a similar move — and here enters Mark Carney, the former governor of the Bank of England, who was by then running a new transition fund at Brookfield. A central banker who had spent years warning about climate risk from inside the financial system, joining a software founder who had spent years tweeting about it from outside. The press called them a "meeting of the minds." Their joint conclusion was that the quickest way to get coal out of the grid was to buy the company that owned the coal and close it from the inside. They would take AGL private, shut its coal plants by 2035 — far earlier than the company planned — and invest $20 billion in renewables and storage.
The first offer, in February 2022, was $7.50 a share, roughly $8 billion. AGL said it "materially undervalues the company." The consortium raised to $8.25, about $9 billion plus debt. The board rejected that too, over a weekend. On the Sunday night, Cannon-Brookes went to Twitter, where he had started the battery and where he would end this:
"The Brookfield-Grok consortium looking to take private & transform AGL is putting our pens down – with great sadness. This weekend, the board rejected our raised offer of $8.25. 46% more than the price of $5.55 about 90 days ago. Our path was the world's biggest decarbonisation project. From Aus. The board are proceeding with their demerger path. This path is a terrible outcome for shareholders, taxpayers, customers, Australia and the planet we all share."
From Aus. Two words, a full stop on each side, the patriot's signature on a failed deal.
Here is where the playbook gets interesting, because "pens down" turned out to mean "pens elsewhere." Weeks later Grok bought into AGL on the open market and became its largest shareholder, eventually with about 11.7 per cent. In May 2022 he wrote to all of AGL's nearly 150,000 shareholders telling them the demerger "misses one of Australia's biggest economic opportunities." Botten wrote back accusing him of "false" and "misleading" statements and demanding a correction "as an absolute priority." Grok replied that it was AGL that was "on the wrong side of the facts and the wrong side of history." Ahead of the scheme meeting set for 15 June, the demerger collapsed. That November his slate won a board fight. Reuters' headline called him what he had become: "Billionaire climate activist."
He did not get control. A year later, in November 2023, he was still writing letters — this one authored by Grok's chief executive, Jeremy Kwong-Law — complaining that AGL was locked in "old thinking," a spectator rather than a doer, with executive bonuses that let the company "stand still." But AGL's coal timetable had moved. You can lose the takeover and win the argument, if you are willing to keep showing up to the annual general meeting.
In October 2021, before any of this, he had pledged to donate and invest $1.5 billion in climate projects by 2030. Time put him on its first climate list in 2023. Conservative commentators accused him of using his fortune to bend national policy, which was precisely what he was doing; a former judge interviewed by the Sydney Morning Herald observed, with judicial calm, that at least he was doing it "in an open and transparent way."
Neighbours
In April 2022, in a Las Vegas nightclub, Scott Farquhar saved a man's life. It was reported by CNBC two months later, in the same summer that his co-founder was fighting AGL's board. It's one of those details that, laid next to another, starts to look like a portrait. One partner performing rescue in private. The other performing rescue in public, at national scale, with a microphone.
The co-CEO arrangement had always been an oddity that outsiders explained away. In practice it was a division of temperament as much as labour: Farquhar engineering and research, Cannon-Brookes product and the story. Farquhar was, on most accounts, more of the day-to-day; Cannon-Brookes leaned outward — to Canberra, to Musk, to Carney. It worked because each covered the other's absence. And then the absences began to lengthen.
Cannon-Brookes and Annie separated in July 2023, after thirteen years of marriage; the property press, which had chronicled every one of their purchases, now chronicled the asset split. The following April, Atlassian announced that Farquhar would step down as co-CEO at the end of August 2024 "to spend more time with his young family, improve the world via philanthropy, and help further the technology industry globally." Cannon-Brookes's formal statement was warm and exact: "The contribution Scott has made at Atlassian is impossible to quantify. Starting with just the two of us in 2001, to a global company of over 11,000 employees and over USD $4 billion in annual revenue, Atlassian would not be the company it is today without Scott. I am truly grateful to have had him by my side every day for the last 23 years." On social media he was looser: "Mate – I couldn't be prouder of what we have achieved together."
The Financial Review's headline on the day the arrangement ended was cooler: "Farquhar's Atlassian era passes without a grand farewell." Five weeks later, on 4 October 2024, Farquhar sold Elaine — the house next door to Fairwater — for A$130 million. In 2026 two Australian journalists, John Stensholt and Perry Williams, published Tech Bros – The Extraordinary Story of Atlassian through HarperCollins, an account of the company's rise and, as it was serialised and promoted, of "the eventual falling out between the founders." Neither founder cooperated. The book's own acknowledgements say so. A reviewer found the account of the rupture "largely third-hand and a little lifeless," which may be the most honest thing anyone can say about the end of a twenty-three-year friendship conducted mostly in private. I won't pretend to know more than the people who wrote the book.
By the Financial Review's 2025 Rich List — the first after his separation, no longer counting a joint fortune — Cannon-Brookes was thirteenth, at A$12.18 billion. Farquhar was fourth, at A$21.42 billion. For twelve years the two had climbed the lists in near lockstep, separated in some years by single places and tens of millions. Now there was a gap of nine places and nine billion dollars.
What Cannon-Brookes did in the months after he became sole chief executive is a list that reads like it was assembled by a novelist with a heavy hand. In February 2025 Atlassian signed a record multi-year title sponsorship with the Williams Formula One team, which became Atlassian Williams Racing. In March 2025 the Financial Review reported that the climate campaigner now travelled by private jet, a Bombardier Global 7500 — one of the longest-range business aircraft built, the kind of plane that turns the Sydney–San Francisco trip into something you can do without a lounge. He defended both: Forbes summarised his argument as "the need for pragmatic trade-offs in the pursuit of sustainability." By September 2025, a year into the solo era, the Financial Review catalogued "fires on all fronts": Atlassian's share price down almost 40 per cent in six months, a messy separation, staff upset about the culture and the artificial-intelligence ramp-up, the F1 team, the jet, and more than $US13 million in shares sold under a disclosed plan.
The boy from the Qantas catalogue had his own plane now. He no longer needed the lounge. He no longer met anyone there.
A steady hand on the tiller
The thing that made Atlassian's founders rich was, in a sense, never Atlassian's software but Atlassian's distribution — the bought-not-sold machine, the engineer with the credit card. Each of the company's great transitions since has been an attempt to keep that machine running while replacing its engine.
The first was the cloud. In October 2020 Atlassian announced the end of its Server products — the downloadable software, installed on customers' own machines, that had been the company's core for eighteen years. Sales would end in February 2021; support in February 2024. It was a cruel decision for customers who had built their operations on those servers, and a necessary one. By April 2024 Cannon-Brookes could declare that "Atlassian is a cloud-majority company," with more than 300,000 customers on its cloud products and a threefold increase in paid cloud seats in the three and a half years since the announcement. In 2025 the company said it would end on-premise support for Confluence entirely by 2029.
Not every turn worked. In September 2017 Atlassian launched Stride, a chat product built to compete with Slack. Less than a year later, on 26 July 2018, it sold the intellectual property for Stride and its older chat product, HipChat, to Slack — its competitor — took a small stake in Slack, and shut them both down. The lesson from the New York office, a decade earlier, applied: acknowledge the wrong turn early. In his 2025 vocabulary, such bets are "revisitable hunches," re-examined every quarter and abandoned "without apologies or embarrassment."
The second transition is the one he is still in. In April 2023 Atlassian announced "Atlassian Intelligence," built on OpenAI's models. In October 2023 it agreed to buy the video-messaging company Loom for $975 million. In April 2024 came Rovo, an AI tool for search and automation. In September 2025 it bought The Browser Company, makers of the Arc and Dia browsers, and agreed to pay $1 billion for DX, a company that measures developer productivity. The strategy, as he describes it, is deliberately unglamorous: Atlassian doesn't train its own foundation models but builds to be "multi-model on purpose," because "our job is adopting and delivering value quickly." The moat, he argues, is context — the "Teamwork Graph," built over eight years, which maps the relationships between people, projects, documents and decisions inside a company, and which by 2026 he described as indexing "150 billion connections." And the moat is design: "What differentiates is how it feels and works."
The market was not persuaded. By early 2026 the fear among American investors was that AI would let customers build their own Jira, and that the software engineers who were Atlassian's ultimate users were being laid off "in droves." Atlassian's shares fell 70 per cent in twelve months, to $US57. It was bumped from the Nasdaq 100. He and Farquhar paused their regular share sales. In March 2026 Atlassian said it would cut about 10 per cent of its workforce — 1,600 jobs — to pivot toward AI and enterprise sales. He announced it by video message. The man who had once preached Atlassian's warm, irreverent culture, the Sydney Morning Herald noted, was "cuddly no more." And the pivot itself contained a quiet heresy: enterprise sales. The company that had become famous for having no salespeople was hiring them.
There is a symmetry here that no one in Sydney seemed to remark upon. In March 2017, the month he started the battery campaign, Cannon-Brookes told an AFR business summit that pretending structural change wouldn't happen was absurd: "I think we should be honest and empathetic and say there are going to be a massive amount of jobs destroyed, right. There are going to be a massive amount of jobs created." Nine years later the structural change came for his own staff. He kept the optimism: "I am a personal believer, an optimist in AI. I think we are going to see many more jobs created as a result. The 'easy doom' narrative is pretty premature."
Then the quarter turned. Cloud revenue rose 29 per cent; the stock jumped 25 per cent in a single after-hours session and 50 per cent in three weeks, to nearly $US90. The Wedbush analyst Dan Ives said it "throws the software bear thesis out the window." Cannon-Brookes, asked how it felt, was all Atlassian understatement: "The only thing you can do is just have a steady hand on the tiller. Don't get too excited on one side or the other ... We have a lot of work to do. We have a fantastic business, and people are going to judge that." By September 2026, Semafor reported, the stock had risen 230 per cent from its April low while remaining 60 per cent below its late-2021 peak, and the company was valued at about $48 billion. The paper noted, without irony, that Atlassian had not posted an annual profit in a decade. In fiscal 2025 it had recorded revenue of $5.22 billion and a net loss of $257 million — a strange inversion for a company whose founding myth was that it had been profitable from day one because it had to be.
In 2026, too, the building was finished. Atlassian Central, at Tech Central in Sydney, is the world's tallest hybrid timber-and-steel tower; Cannon-Brookes wrote that its cost per square metre was roughly a third less than the old office. He stood on the thirty-ninth floor and remembered the share houses in Randwick and Glebe. "Almost 25 years ago," he began. Scott Farquhar left a comment under the post: "What a milestone." The two men had been asking for a technology precinct in Sydney for more than a decade, and Cannon-Brookes ended his post with the line that, more than anything he had ever said, finally answered the question of where he was from:
— Mike Cannon-Brookes, on the opening of Atlassian Central, 2026Atlassian Central is proof that when we work together, we can show the world what Australia's technology industry can be - not a satellite office for Silicon Valley, but a centre of gravity in its own right.
He no longer thinks the distance is a handicap. He thinks it is the point. Sydney, he told Semafor, is "a great microcosm" — a big, ordinary city of "banks, insurance companies, manufacturing companies, and grocery companies," fast to adopt technology but not driven by it. Silicon Valley has "froth and bubbles," he said, "amazing and naive at the same time," and "sometimes there's a lot of technology that's applicable between San Francisco and San Jose, and nowhere else in the world." Atlassian takes the Valley's ideas and checks whether they work in Sydney, "which is a proxy for every other big city in the world." One foot in, one foot out. The aisle seat as a strategy.
It does exact a price, and the price is paid in hours. Because Atlassian is headquartered in Australia, listed on Nasdaq, and sold everywhere, its chief executive keeps a timetable that splits the Sydney day into American pieces: work from five until seven, the children from seven until nine, work from nine until three. He was careful to tell the interviewer he is not one of those founders with a wellness routine — no journaling, no glass of water, no gym. Twenty-three years earlier the Americans had placed their order while he slept, and he had found it in the morning, a page curling in a fax tray. Now he does not sleep through their afternoon. At a quarter to five, in the dark, in Sydney, he gets up, makes an espresso, and goes to the computer.
Part IIThe Playbook
What follows is an attempt to extract from Cannon-Brookes's career the decisions that generalise — not the biography, which is unrepeatable, but the patterns that a founder, operator or investor could actually use. Several of these principles cut against standard Silicon Valley advice, and Cannon-Brookes himself has warned against treating Atlassian as a template: "We always say, don't copy us as a model. Well, copy us if you can, but don't use it as a case study on how every business should run." Take that caveat seriously. The value is less in the moves than in the reasoning behind them.
The principles
- 1.Let the constraint write the business model.
- 2.Price beneath the approval threshold.
- 3.Build the tool you were forced to hate.
- 4.Raise for partners, not for runway.
- 5.Use distance as a test market.
- 6.Read impostor syndrome as an instrument.
- 7.Close the wrong office quickly.
- 8.Build lighthouses, not monuments.
- 9.When you can't buy the system, own a piece and argue.
- 10.Split the crown along temperament.
- 11.Schedule invention.
- 12.Burn your own legacy before someone else does.
Principle 1
Let the constraint write the business model
Atlassian's most celebrated innovations — self-serve purchasing, published pricing, no sales force, profitability from the start — were not strategic insights so much as forced moves. In 2002 there was no venture money for "two knuckleheads with no experience" in Australia, so the company had to be profitable. There was no way to afford American salespeople, so the website had to do the selling. Cannon-Brookes's own summary is unsentimental: "It was either become profitable and grow, or die."
The non-obvious lesson is that the constraint did not merely shape the company; it gave it a structural advantage that better-funded competitors could not copy. A company that has raised a large round has to justify a sales force. A company that never could has to make the product sell itself, and once that discipline is built into the organisation, it compounds. By 2011 Atlassian was doing $102 million in revenue without salespeople. Competitors with the cash to do things the old way had little reason to learn the new way.
The danger, which Atlassian itself illustrates, is mistaking a forced move for a permanent virtue. In 2026 the company that pioneered "bought, not sold" announced layoffs tied to a pivot toward enterprise sales. The constraint wrote the model; it did not write the model forever.
Tactic: List the three things you cannot afford to do the conventional way, and design your go-to-market so that the absence of each becomes a feature customers notice.
Principle 2
Price beneath the approval threshold
Larry Marshall's description of early Atlassian pricing is the most practically useful sentence in the whole record: price at a level "where an engineer could just put it on their credit card and not have to get approval to buy it." At under $3,000, Jira did not need a procurement process. The engineer bought it, the team used it, and by the time the organisation noticed, removing it would have been more expensive than keeping it.
This is not discount pricing; it is a structural choice about who the buyer is. Cannon-Brookes described selling "on a team by team basis" rather than "to an entire company at a single time." The unit of sale was the team, and the team's budget had a ceiling set by someone else's expense policy. Find that ceiling and price under it, and you have converted an enterprise sale into a consumer purchase.
The American Airlines fax is the proof. A large company with a sophisticated procurement function bought Atlassian's software without anyone at Atlassian speaking to anyone at the airline. That only happens when the price and the evaluation experience together make talking to a salesperson unnecessary.
Tactic: Find out the dollar limit at which your target user can spend without asking permission, and build an entry tier that sits comfortably below it.
Principle 3
Build the tool you were forced to hate
The founders' co-op scholarship placed them in real companies for six-month stints, and the main thing they took away was contempt for the software those companies made their staff use: "it was just a horrible experience with the software we were made to use." Later, running a support-services business, they were annoyed enough by their bug tracker to write their own. Jira was born of grievance.
The pattern is worth naming because it inverts the usual "scratch your own itch" advice. It wasn't the founders' own itch, exactly — it was an itch they had been forced to experience inside other organisations, at the bottom of the hierarchy, as users without purchasing power. That vantage point is rare among people who later build enterprise software, most of whom meet these tools as buyers or executives. The user's-eye contempt produced products that users would buy for themselves, which fed directly into Principle 2.
Two decades later, Cannon-Brookes's vocabulary for this instinct is "constructive unreasonableness": "I get frustrated when things don't work the way they should. And then I try to change them."
Tactic: Write down the software you were made to use in your last three jobs and rank it by how much it humiliated the people with the least power to replace it.
Principle 4
Raise for partners, not for runway
Atlassian's first outside money is the strangest venture round in this profile: $60 million from Accel in 2010 when the company had $55 million in the bank, with the founders arguing they would rather take $5 million. They were not raising for cash. They were raising for credibility with the people they needed next — public-market investors and prospective board members, several of whom had looked at a profitable company without institutional backers and asked, "how come you have no funding?"
The principle is that capital is also a signal, and its value as a signal can exceed its value as money. Cannon-Brookes's general position on venture capital is carefully hedged: "I don't think it's a mistake, I think it's a mistake to have it as the default choice. It's a mistake to not think about why you're doing it and what change it's going to make to your business."
The corollary is in the structure: a largely secondary round, a "pretty small piece" of the company sold, and founders who by 2014 still held about 78 per cent. The negotiation ran in reverse because they could afford to walk away.
Tactic: Before any raise, write one sentence on what the investor's name will unlock that the money cannot, and size the round to that, not to the maximum available.
Principle 5
Use distance as a test market
For most of Australian business history, distance from the world's centres of capital was the defining handicap. Cannon-Brookes's later argument is that it can be an analytic instrument. Silicon Valley, he says, produces ideas that are "amazing and naive at the same time," and some technology "is applicable between San Francisco and San Jose, and nowhere else in the world." Sydney — big, corporate, quick to adopt but not technology-driven — is "a proxy for every other big city in the world." So Atlassian runs Valley ideas through Sydney as a filter.
The principle generalises beyond Australia. Any founder outside the dominant ecosystem has access to a kind of customer the insiders do not see clearly: the ordinary bank, the grocery chain, the manufacturer. Treating that market as a laboratory rather than a backwater produces products that travel. It is also, as the Senate testimony shows, a political argument — that a country can be "a net exporter of products, not a net exporter of people."
The cost is real, measured in a CEO who starts work at five in the morning to catch the American day. Distance is an advantage only if someone absorbs the time-zone tax.
Tactic: Pick one idea that is fashionable in your industry's capital city and test it with five customers in an unglamorous city before you build it.
Principle 6
Read impostor syndrome as an instrument
The TEDx talk, played millions of times, is usually received as reassurance. Its actual argument is more useful and more demanding: if you don't feel like an impostor, you are either inside your comfort zone and not learning, or so far out of your depth that you can't tell. The feeling is a gauge that tells you that you are at the edge of your competence, which is where growth happens. "Questioning your ideas rather than yourself" is the operating instruction.
His career is a sequence of deliberate deep ends: interviewing the company's first HR hire, accepting an award he felt he didn't deserve, challenging Elon Musk in public about grid batteries, testifying before a Senate committee for the first time, bidding for the country's biggest polluter. In each case the domain was one where he had no credentials and some leverage. The impostor feeling did not stop him; he treated it as confirmation he was in the right room.
The warning embedded in his own story is that the instrument must stay calibrated. Sun Cable's critics called it hubris and billionaire's disease. A man who seeks out the feeling of being out of his depth must also keep a way of distinguishing productive discomfort from simple overreach.
Tactic: Each quarter, name one situation where you felt like a fraud, and write down whether the doubt was about your ideas (fixable) or about yourself (ignore it).
Principle 7
Close the wrong office quickly
Atlassian's first American office, in New York, was closed almost as soon as it opened when leadership realised they had over-extended. Its chat products, HipChat and Stride, were sold to the competitor Slack in July 2018, less than a year after Stride launched — with Atlassian taking a small stake in the winner. Jitsi, a video tool acquired in 2015, was sold to 8x8 in 2018. The company's own retrospective frames the New York retreat as "good practice in making tough calls."
The non-obvious move is the Slack deal: not just exiting a losing market but converting the exit into an equity position in the company that beat you. It turns a defeat into an option on the winner's success, and spares customers a slow death of a product nobody would fund.
Cannon-Brookes's 2025 language for this habit is the "revisitable hunch" — a big bet re-examined every quarter and abandoned "without apologies or embarrassment." The emphasis on the absence of embarrassment is the key. Most organisations can identify a failing bet; few can kill one without someone losing face, so they don't.
Tactic: Put a review date on every major bet when you make it, and decide in advance what evidence would make you sell it to a competitor rather than shut it down.
Principle 8
Build lighthouses, not monuments
The South Australian battery was small by the standards of the grid problems it addressed — 100 megawatts, $90 million. Cannon-Brookes's point was that it didn't need to be large; it needed to be visible and to work. "Once it's done, everyone goes, 'Oh, it's possible and it's profitable.' Then you get a whole rush of projects behind it." A lighthouse proves the coast is navigable. A monument just impresses.
The battery was a lighthouse: fast, bounded, measurable, with a deadline — 100 days or free — that made success or failure legible to the public. Sun Cable, by contrast, drifted toward the monument: the world's biggest solar farm, biggest battery, longest and deepest cable, a budget that grew from $25 billion to over $30 billion. Its every superlative was a risk multiplied. Critics compared it to Icarus, and the project went into administration.
The two projects, run by the same man within six years of each other, are the clearest available test of this principle. One built a precedent. The other built a debate.
Tactic: Before starting a showcase project, cut its scope until it can be delivered by a public deadline you're willing to be embarrassed by.
Principle 9
When you can't buy the system, own a piece and argue
The AGL campaign failed as a takeover. Two offers, $7.50 and $8.25 a share, were rejected; Cannon-Brookes announced the consortium was "putting our pens down." But he did not leave. Grok bought into AGL on market and became its largest shareholder, eventually holding 11.7 per cent; he wrote to all the company's shareholders; the board's demerger collapsed before the vote; his slate later won a board fight. Without control, he changed the company's direction.
This is the activist-investor playbook applied by a founder, and its lesson for operators is that "buy or walk away" is a false binary. A meaningful minority stake, combined with a public argument and a credible alternative plan, can move an incumbent that refuses to be bought. It works best when, as here, the incumbent's own proposal is weak and other shareholders are already uneasy.
The approach has limits. More than a year later, Grok was still complaining publicly that AGL was moving too slowly. Influence without control is a long, grinding campaign, conducted one annual general meeting at a time.
Tactic: If an acquisition target rejects you, ask whether a 5 to 15 per cent stake and a public plan would achieve most of what control would.
Principle 10
Split the crown along temperament
For twenty-two years Atlassian had two chief executives, an arrangement most governance experts would have advised against. It worked because the split followed the founders' natures: Farquhar on engineering and research, Cannon-Brookes on product, design, sales and marketing — and, increasingly, on the outward-facing work of policy and public argument. Each covered the other's absences.
The lesson is not that co-CEOs are a good idea in general. It is that any leadership split must follow genuine differences in temperament and interest, not a political compromise between two people who want the same job. The Atlassian founders had arrived from opposite ends of Sydney, with opposite childhood computers, and had disliked each other at first. The difference was the asset.
The record also shows the cost of the structure's end. After Farquhar stepped down in August 2024, the sole CEO's first year was described as "messy," with fires "on all fronts." A structure that depends on two people's complementarity is fragile to the departure of either one, and the succession plan for a co-leadership is harder than for a single leader.
Tactic: If you share leadership, write down the specific decisions each partner owns outright, and rehearse now what happens to those decisions if one of you leaves.
Principle 11
Schedule invention
ShipIt, Atlassian's quarterly hackathon, began in 2005, when the company was small and growing fast enough that "business as usual" could easily have consumed every hour. The company's own lesson from that year is blunt: "Innovation needs dedicated time and space. It's not going to happen on your coffee break." Twenty-four hours, work on anything with anyone, a little competition.
What makes this more than a morale exercise is the yield. Jira Service Desk — later Jira Service Management, Atlassian's fastest-growing product — originated in ShipIt. A company that ran no internal venture fund and had no corporate research lab nevertheless produced a major product line from a recurring, cheap, scheduled ritual.
The general principle is that bottom-up invention in a growing company does not happen by default; it has to be put on the calendar, protected from being cancelled when a quarter gets tight, and connected to a path to production for the things it produces.
Tactic: Put a recurring 24-hour build window on the company calendar for the next four quarters, and assign one executive to shepherd the best result into the roadmap each time.
Principle 12
Burn your own legacy before someone else does
In October 2020 Atlassian announced the end of the downloadable Server products that had been its core for nearly two decades: sales ended in February 2021, support in February 2024. It was a decision that angered loyal customers and forced painful migrations. By April 2024 the company was "cloud-majority," with a threefold increase in paid cloud seats since the announcement. In 2025 it announced the end of on-premise Confluence by 2029.
The AI transition follows the same logic, with sharper edges: acquisitions of Loom, The Browser Company and DX, an AI layer built "multi-model on purpose," the 2026 layoffs of 1,600 people. Cannon-Brookes's analysis of AI is that customers who adopt it most heavily are "expanding the fastest for us," and that the people who should worry are not those replaced by AI but those replaced "by someone who's really good at using AI." He is applying the same rule to his own company.
The pattern across both transitions is that the incumbent product is ended by the company that built it, on a timetable it sets, rather than eroded by competitors on theirs. The cost — angry customers, a sharp stock decline, cultural damage — is real and was paid. The alternative was a slower and worse version of the same outcome.
Tactic: Set a public end-of-life date for the product you're most attached to, and use the announcement to force the migration plan you've been postponing.
A
Bought, Not Sold: The Atlassian Model Against the Grain
How Cannon-Brookes's choices compared to the conventional enterprise-software playbook of the 2000s.
| Conventional wisdom | The Atlassian approach | Status |
|---|---|---|
| Raise venture capital early to grow fast. | Bootstrapped for eight years; first round taken with $55M in the bank. | Validated |
| Hire a field sales force to sell to CIOs. | Self-serve website; sell to teams; price under the credit-card limit. | Revised in 2026 |
| Headquarter in Silicon Valley. | Stay in Sydney; use it as a proxy for every big city. | Validated |
| One CEO, clear chain of command. | Two co-CEOs for 22 years, split by temperament. | Ended 2024 |
| Protect legacy revenue. | End Server products; force cloud migration. | Validated |
| Build the biggest version of the vision. | Sun Cable: world's largest solar farm, battery and cable. | Risk |
T
Decision Timeline
The key moves that defined Cannon-Brookes's two decades at Atlassian and beyond.
2001
Cannon-Brookes emails his UNSW classmates about starting a company; only Farquhar replies.
2002
Jira 1.0 ships, financed with $10,000 of credit-card debt.
2003
An order from American Airlines arrives by fax, unsolicited.
2004
Confluence 1.0 released on 25 March: the second "rocket engine."
2005
ShipIt hackathons begin; the US office moves to San Francisco.
2010
$60M from Accel, with $55M already in the bank.
2015
Nasdaq IPO on 10 December at a $4.37B market capitalisation.
2017
Trello acquired for $425M; the Tesla battery challenge in South Australia.
2018
Chat products sold to Slack; Senate testimony; Fairwater purchased.
2020
End of Server products announced.
2022
AGL takeover bids rejected; Grok becomes AGL's largest shareholder.
2023
Sun Cable collapses, then is won by Grok; Loom acquired for $975M.
2024
Farquhar steps down; Cannon-Brookes becomes sole CEO on 1 September.
2026
About 1,600 jobs cut in an AI pivot; Atlassian Central opens in Sydney.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— Mike Cannon-Brookes, Recode Decode, 2016We just always asked them the question, "Why would we take money? We're growing, we're profitable, what would change here?" And they'd have a whole host of reasons, and we wouldn't believe them and then we'd move on.
— Mike Cannon-Brookes, testimony to the Australian Senate committee on the Future of Work and Workers, March 2018Quality – Australia does well here. Our technology graduates are highly desired around the world. However the lack of a local industry is challenging because our best ship off overseas for better opportunities. As a country, we want to be a net exporter of products, not a net exporter of people.
— Mike Cannon-Brookes, AAP, June 2017That's how I think certainly I get enjoyment in life, chuck yourself in the deep end and pretend you don't look like a flailing idiot.
— Mike Cannon-Brookes, Business Owners' Wisdom interview, 2012I think we won't be sort of celebrating until we feel like we've got less in front of us than behind us.
— Mike Cannon-Brookes, ahead of Team '26, May 2026I am a personal believer, an optimist in AI. I think we are going to see many more jobs created as a result. The "easy doom" narrative is pretty premature.
Maxims
- Start with a modest number. A goal of A$48,000 a year let two graduates survive year one without needing to look like visionaries.
- Let necessity design the model. No money, no experience and no proximity to Silicon Valley produced a business that rivals with more resources could not copy.
- Sell to the team, not the company. Price under the credit-card limit, and the enterprise will discover you on its own time — sometimes by fax.
- Take money as a credential, not a crutch. The best time to raise is when you'd rather the investor put in less.
- Treat the edge as a laboratory. An unfashionable city is a better test of whether an idea travels than the place that invented it.
- Doubt the idea, never the self. The feeling of being a fraud is evidence that you're learning; freezing is the only real failure.
- Kill wrong turns while they're still small. Close the office, sell the product to the rival, and keep the stake.
- Build lighthouses before cathedrals. A fast, bounded proof — 100 days or it's free — moves more capital than the grandest blueprint.
- Lose the takeover, win the argument. A minority stake and a public plan can turn an incumbent you could not buy.
- End your own legacy on your own schedule. The product you love most is the one a competitor will happily retire for you.
