1 hour ago
2
Ten years before she became a billionaire, in 2011, a 24-year-old Melanie Perkins built a software prototype called Canvas Chef. The idea was simple: a prompt box where you described what you wanted to design, and it appeared on-screen.
Perkins pitched it to more than 100 investors. They all said no.
Fifteen years later, Canva is Australia’s most successful technology start-up. The company, an online design platform that lets anyone produce professional-looking presentations, social posts and marketing materials, was built by Perkins and her now-husband Cliff Obrecht – today two of Australia’s richest people.
Canva has 265 million monthly users, $US4 billion ($5.6 billion) in annual revenue, and a market valuation sitting at around $100 billion. This week, at one of the world’s largest football stadiums in Los Angeles, the company unveiled what its co-founders are calling its most significant product launch to date – Canva AI 2.0.
It is, at its core, exactly what Canvas Chef was supposed to be on a much larger scale. It’s a prompt box where you describe what you want, and it appears.
“We believe this will be the most significant moment in Canva’s history,” Perkins says. “We are extraordinarily excited to introduce this to the world.”
The timing is either perfect or catastrophic, depending on who you ask.
Canva AI 2.0 arrives in the middle of what the technology industry has started calling the “SaaSpocalypse” – a violent repricing of the software companies that defined the last decade. Atlassian’s market capitalisation has cratered to $26 billion. Figma, the designer’s darling that listed at $US19.3 billion last July, has shed roughly a third of its value. Google’s AI website builder, Stitch, triggered sell-offs across the sector within hours of its announcement. And even Anthropic’s Cowork product – a desktop assistant targeted at everyday white-collar workers – has been cited as a contributor to billions wiped from Australian software stocks.
The world’s tech billionaires, and their investors, are rattled. Canva co-founder Obrecht recently issued and then quickly deleted a post to employees about the velocity at which AI firms were working, saying his staff should match the pace.
Dan Romanoff, a senior equity analyst at Morningstar who covers software companies, says Canva executives have good reason to be spooked. “I think Canva is more susceptible to AI threats than Adobe given its lower-end customer base,” he says.
“Given how valuations have collapsed for both Adobe and Figma, it’s hard to imagine how Canva’s valuation has been sustained.”
Rory O’Driscoll, a veteran software investor at Scale Venture Partners, also sees cause for concern amid the confetti and on-stage raps that have characterised its events at SoFi Stadium.
“The market will throw out all the babies with the bathwater before picking back up the babies it wants to keep,” he says. “Canva, as a founder-led, innovative company that is profitable and has customer love, can do that. But they have to get it done.”
Inside Canva, Obrecht has been getting it done. He says he’s been preparing for this moment for two years. And now he finally gets to talk about it.
The work wasn’t visible from the outside. No major new features have been launched, and no splashy acquisitions announced until recently. The internal slogan was four words: “More AI, less UI.” The engineering effort involved rebuilding Canva from the ground up – rewiring every feature in the platform so AI agents could access and trigger them directly, without a human clicking anything.
“It’s been a long time wearing baggy clothes in the gym,” Obrecht told this masthead in an interview ahead of Create. “We’ve been working hard before we get to go to the beach and rip our shirt off.”
Canva has been here before. Between 2016 and 2018, the company undertook a full rewrite of its codebase – a project employees described at the time as “hellish” – during which it shipped almost nothing new. The result was the foundation that powered Canva’s next six years of growth.
Obrecht sees the current rebuild as the same bet. “We’ve had to go slow to go fast,” he says. “And now we’re back to moving like a go-kart, not like a big cruise liner.”
The product that emerged this week is ambitious, and will determine whether Canva thrives amid the AI turmoil or is lost at sea.
Canva AI 2.0 introduces conversational design (natural language in, finished work out), agentic orchestration (describe a goal and Canva co-ordinates its tools to deliver it), connectors to Slack, Gmail, Google Calendar and others, and what Canva calls “living memory” – persistent context that keeps projects on brand without manual input. An orchestration layer built partly on Simtheory, an AI agent management company Canva acquired earlier this month, ties it all together.
Cameron Adams, Canva’s co-founder and chief product officer, says the hardest single piece was the Canva Design Model, a proprietary AI system that understands how design works and can generate and edit layered, editable compositions rather than flat images.
“That has been well over two years of development,” Adams says. “Nothing that you see in Canva AI 2.0 could have been done without it.”
The re-architecture required every person in the company to contribute, and will – in Obrecht’s words – flip the company from being a “design platform with AI tools” to an “AI platform with design tools”.
“People don’t wake up in the morning going, ‘I want to create a social post,’” Obrecht says. “They wake up saying, ‘I want to grow my business.’ When the capabilities are available to help them achieve their complete goals, you need to move there.”
But there is a quieter problem that no amount of new features can solve on its own, and it was articulated earlier this year by someone who has no interest in Canva’s failure.
Jason Lemkin, founder of the influential software investor community SaaStr, wrote a piece this month in which he described his own drift away from the product. He’s been a paying customer for more than eight years. He hasn’t opened Canva in months. It happened gradually: thumbnails went to one AI tool, video clipping to another, sales decks to a third. Each replacement was purpose-built and was slightly better than Canva at its narrow task.
“Each of these tools does one thing well really well,” Lemkin wrote. “And just like that, my personal Canva usage went to zero. Not because I made a decision to leave. I just stopped going there.”
His team still uses Canva daily. So by every standard metric, they’re a retained customer. But the person who drove the original purchase has disengaged – and that dynamic, Lemkin argues, is invisible in aggregate growth numbers.
“The most dangerous time to have a stealth churn problem is when your growth is so strong you’d never think to look for one.”
Obrecht would never frame the launch in these terms. But Canva AI 2.0’s entire architecture – the workflow ownership, the connectors, the shift from individual features to end-to-end goals – is basically designed to answer Lemkin’s problem.
Adams also argues that what Canva is launching this week cannot be replicated by the wave of AI coding tools now flooding the market.
“I think it’s pretty impossible to vibe code Canva,” Adams says. “When you think of the lifecycle of that software – who’s maintaining it, who’s fixing bugs, who’s keeping up – you end up with 1000 people using that system, and who’s making sure it’s not going to break, that it doesn’t have security holes in it? That is a whole piece of creating software that often gets glossed over.”
The pressing question that sits behind all of this is when Canva will finally be listed on the public stock market. Expected in 2027, it is likely to make instant millionaires of some of its 5500 employees and be the biggest event in recent Australian technology history.
Canva is trading at roughly 12 to 13 times trailing revenue in secondary markets. Romanoff, at Morningstar, is sceptical. “I doubt there is a ton of appetite for a design software IPO, especially when many investors are already writing the main player in this space completely off,” he says. He also flags a distinction that will matter to public investors: Canva’s nine consecutive years of profitability are on a free cash flow basis, which is a different and more relaxed threshold than generally accepted accounting principles.
Obrecht confirmed, in what appears to be the first public acknowledgement of the strategy, that the staged rollout of Canva AI 2.0 – initially to just one million test users – is explicitly designed to generate data on a new business model before going public.
“We are moving from charging for features to charging essentially for AI credits,” he says. “It’s really a growth-first margin decision, and given the magnitude of this launch, it’s going to shift that usage significantly. So we’re very much waiting on the data. You want to do that as a private company.”
Unlike traditional software, where the marginal cost of serving another user is negligible, AI features carry real compute costs. Canva needs to understand what that price is – and what users will tolerate – before Wall Street gets to set expectations.
“I actually would way rather launch in a market that isn’t super-frothy,” Obrecht says, “because then you can launch with a valuation that’s fair. A depressed market doesn’t scare us at all.”
Adams says there’s a key difference between Canva and the companies that have seen their valuations fall to pieces.
“I think the SaaSpocalypse is a reflection on companies who really don’t have their AI story figured out,” he says. “For us, because we’ve been working in it for so many years now, and we’ve laid the proper foundations, we’re paying the dividends for it.”
Rick Baker, partner at Blackbird Ventures and an early Canva investor, says the AI pivot makes perfect sense.
“This is Canva stepping up to become one of the major forces in AI globally and sets it up beautifully for the next phase of growth as it heads towards an IPO.”
Before an IPO, the real test will be the millions of users and how they react to this week’s announcements.
“Canva is still the king of ‘good enough, fast enough’ design,” AI analyst Tanveer Ahmad says. “It is an ‘80 per cent solution’. It does 80 per cent of what Photoshop does, 80 per cent of what Premiere does, and 80 per cent of what Google Docs does.
“For most marketers, that’s perfect”.
On the ground in Los Angeles, Obrecht is clearly in his element, doing what he has been doing for more than a decade: talking about the company he and Perkins started from her mother’s living room in Perth, with the energy of someone who cannot quite separate himself from the product.
He and his co-founders are on little sleep, having now rehearsed step-by-step the event that will make or break the company’s future.
“It’s never been funner,” he says. “But it’s never been harder.”
Three years ago, Canva was growing headcount at 50 per cent a year. Last year, it was below 10 per cent. This year, below five per cent. Obrecht says it’s this decision that means he can rule out a headline in six months that the company is laying off chunks of its workforce, which has already happened at Atlassian, Block, and what will surely be countless more software companies. The efficiency gains are coming from AI tools, not from cutting people, he says. “Rather than this gas-on, hit-reverse, lay-off kind of mentality, we’ve seen this coming, and we’ve just eased our foot off the accelerator.”
Whether the market will reward that discipline depends on what happens next. The AI credits will be consumed or they won’t. The workflows will hold or they’ll fragment.
Somewhere, in all of that, Canvas Chef will get its answer.
The Business Briefing newsletter delivers major stories, exclusive coverage and expert opinion. Sign up to get it every weekday morning.
