When Mike Cannon-Brookes and Scott Farquhar rang the Nasdaq bell in December 2015, it looked like the start of something inevitable.
Their company, Atlassian, wasn’t just another software float, it was a proof point for Australia: a company built in Sydney could sit at the same table as Silicon Valley’s best.
Its business model selling subscriptions to software looked like the future at a time when manufacturing things looked a bit like the past.
Ten years later, and the world is in reverse. Atlassian is being removed from the Nasdaq 100 – a list of the 100 largest companies on the premier exchange for tech companies – to be replaced by Sandisk, a hardware manufacturer known for its memory cards and flash drives that long predates the dotcom bubble and the iPhone.
The ejection is a symbolic handover and an illustration of just how brutally AI is tearing through the software sector.
Sandisk enters with a market cap around $US100 billion ($140 billion); Atlassian exits at roughly $US17 billion. Just a year ago, Sandisk shares traded near $US33. They’re now around $US905 – a 2700 per cent run that would have turned a $35,000 punt into $1 million. Over the same window, Atlassian has lost roughly two-thirds of its value, with the stock down 85 per cent from its 2021 peak, despite the company posting its first billion-dollar cloud quarter.
That’s not to say Atlassian’s business is broken. By most operational measures, it’s the strongest it has ever been. What’s broken is the thesis investors spent a decade buying into.
Atlassian was supposed to embody the enterprise software era which was built around recurring revenue from software that cost almost nothing to distribute to more users. But the market has decided that even strong software franchises can be vulnerable if AI changes the economics of coding.
The roughest version of the anti-Atlassian thesis is that investors now fear AI will make its software easier to clone, while simultaneously eliminating many of the software engineering roles that use it. That is why the stock has been punished even while the company has posted strong revenue numbers.
People briefed by Atlassian’s biggest customers and backers say the company’s bull case is simple: it has time. Atlassian’s most important customers are big ships that take a long time to turn around.
If a major retailer or financial institution wants to change software, it can take years and cost billions. Just ask companies such as the ASX or ANZ, which have blown vast sums of money on technology projects. That’s why when many big companies are considering changing their technology tools and considering vendors’ offerings on a scale based on their features, they often award an incumbent twice as many points as a newcomer to factor in the difficulty of switching.
It’s an advantage that could protect Atlassian from being dumped by its biggest customers for 18 to 36 months. Meanwhile, Atlassian hasn’t been standing still. It has quietly been buying up AI companies such as Secoda, a little-known firm that helps companies use AI to answer questions from their own proprietary data.
“People are begging Atlassian and [another pre-AI big tech company] Salesforce to catch up,” said one person briefed on the market’s position.
Cannon-Brookes makes a version of the same argument: last week he called the “vibe-code your own software” thesis “nonsense” among serious buyers, and pointed out that only 40 per cent of Atlassian’s users are developers anyway.
Sandisk, by contrast, doesn’t need to tell a story. It just needs to ship storage. Every data centre training a frontier model needs what Nvidia’s Jensen Huang has called “the working memory of the world’s AIs” – storage that AI models pull from in real time.
The semiconductor company has raised its “price per bit” by more than 100 per cent in the past year and can’t make the stuff fast enough.
That kind of growth has brought opportunists. Last Wednesday, the shoe company Allbirds, which symbolised the Obama-era Silicon Valley with its wool running shoes (really), responsible business pledges and a $US4 billion public offering – announced it was selling itself to a brand-licensing outfit and reinventing the residual shell as “NewBird AI”. Backed by a $US50 million investment, the company plans to buy graphics processing units (GPUs) and lease them to developers.
The stock, which had cratered before the announcement, jumped 600 per cent in a day. By Thursday, it had already retreated 36 per cent.
When a failing wool runner company can juice its stock by hoarding computer chips, we are firmly in the euphoric phase of the hardware cycle.
The Atlassian-to-Sandisk swap and the Allbirds-to-NewBird pivot look like opposite stories, but they’re actually the same one. A 40-year-old storage company and a zombie sneaker brand buying GPUs now occupy the same narrative slot that enterprise software companies held just a year ago.
Cannon-Brookes may well be proved right over a five-year horizon, and history may look back on this swap as the exact moment the market temporarily lost its mind. But, at least for now, the hierarchy has changed: investors want either direct AI leverage or a critical infrastructure role. Everything else is yesterday’s news.
The Business Briefing newsletter delivers major stories, exclusive coverage and expert opinion. Sign up to get it every weekday morning.
David Swan is the technology editor for The Age and The Sydney Morning Herald. He was previously technology editor for The Australian newspaper.Connect via X or email.
Nick Bonyhady is the business editor of The Sydney Morning Herald and The Age. He is a former deputy federal editor, technology editor and industrial relations reporter.Connect via X or email.