265 Million People Still Love Canva. The Market Took $11 Billion Off Anyway.

Canva is the most affectionately held product in enterprise software. Not the most respected or the most technically admired. The most liked. Ninety-five percent of the Fortune 500 are on it. Two hundred and sixty-five million people opened it last month. Teachers, nonprofits, small councils, church bulletins, and an entire generation of marketers who learned layout inside it. Net promoter scores in the high fifties, which for a productivity tool is close to fan behavior.

Last week that company lost a great deal of paper value in a matter of days.

Blackbird and AirTree, its earliest and largest local backers, had external valuers mark it down 17 percent, from 42 billion US dollars to 34.9 billion, a cut of 7.1 billion US or roughly 10 billion Australian. Canva's own independent valuer, running the annual employee share issue, went further, trimming about 20 percent to 31 billion US from 38.9 billion a year earlier, which is more than 11 billion Australian dollars off the internal number. Private-share broker Hiive was quoting stock on the secondary market at an implied 30 billion, a 29 percent discount. The 2027 Nasdaq listing, already pushed back once, now reads as a date rather than a plan.

Anyone who has spent the last two years being told that brand is the one durable asset in an AI market should sit with this one, because Canva is the strongest available test of that claim and the result is more complicated than either side of the argument wants it to be.

What actually got marked down

The instinct on a brand desk is to look for the mistake. There isn't one, at least not one a marketing organization could have prevented.

Revenue in the June quarter was 921.9 million US dollars, up 25.2 percent year on year. The company has been profitable for nine consecutive years, which puts it in a category of approximately nobody among venture-backed software companies at its scale. It closed the quarter with 1.47 billion in cash.

What moved was the forecast. Melanie Perkins told shareholders in her Q2 update that Canva was cutting expected revenue growth from 30 percent to 20 percent, and she was unusually plain about the cause. The average cost of serving an AI task had run too high. The platform had been leaning too heavily on frontier models. The company chose to slow product releases and rebuild the architecture underneath rather than keep shipping into bad unit economics. She presented it as a deliberate decision to fix the math before scaling it.

That is a good decision, made in public, by a management team with a nine-year record of not flattering its own numbers. The market repriced the company anyway.

So the question is not what went wrong. It is which asset actually repriced, and it was not the brand. Two hundred and sixty-five million people did not think less of Canva last week. What repriced was the story premium sitting on top of the brand: the part of the number that was never about the installed base and always about the slope of the line in front of it.

We ran this same arithmetic two weeks ago on Airtable, which raised 1.4 billion dollars and sold for 1.285 billion. Airtable was profitable, growing above 20 percent, credibly repositioned around AI, and it cleared at 2.7 times revenue. Canva is a larger, healthier, far more loved version of the same experiment, and the market is returning roughly the same verdict. On the 20VC podcast, SaaS investor Jason Lemkin said the quiet number out loud: twelve billion. Rory O'Driscoll of Scale Venture supplied the arithmetic behind it, pointing out that Adobe grows at 12 percent and trades at three to four times revenue.

Canva was loved for relief, and relief always gets cheaper

Here is the part the beloved-brands-are-safe thesis skips.

There are two kinds of brand love and they have wildly different durability. One is love of identity. Trader Joe's, Patagonia, Nintendo, Apple on its better days. The customer is buying a statement about who they are, and the affection survives the product getting worse or dearer, because leaving costs them something socially.

The other is love of relief. The customer is not making a statement. They are grateful. Someone removed a specific weekly humiliation from their life.

Canva is the greatest relief brand of the last fifteen years. The feeling it manufactured was never I am a Canva person. It was thank god I don't have to open Photoshop. Thank god I don't have to file a ticket with design. Thank god I don't have to explain to my boss why the deck looks like that.

Relief has no defense against something more relieving. The gratitude was never attached to Canva. It was attached to the absence of friction, and it transfers instantly, and without guilt, to whatever removes more of it. Lemkin's diagnosis is the whole problem in five words: agents "just routed us around Canva." Not beaten. Routed around. Nobody churns in anger. They simply stop needing the step, and a step that isn't taken generates no session, no habit, and no renewal conversation.

The audit is one sentence long and most teams will not enjoy running it. Is the affection you measure a statement your customers are making, or a relief they are feeling? Net promoter score cannot tell the difference. It scores both identically, right up until the week it doesn't.

The promise was priced on zero marginal cost

There is a second thing buried in Perkins's explanation that deserves more attention than the valuation headline got.

Canva's promise, stated plainly, is that professional design is free and effortless for everybody. That promise was economically possible because of a specific property of the old product. A template is a fixed cost. Draw it once, amortize it across two hundred and sixty-five million people, and serving the two hundred and sixty-five millionth design costs approximately nothing. Freemium works. Generosity is cheap. The free education tier, the seven hundred thousand nonprofits, the billion dollars of donated product value: all of it was affordable because the marginal unit was free.

Generative AI puts a meter on the promise. Every image, every Magic Studio call, every agentic edit costs real money to serve, per user, per action, indefinitely. Canva was running something on the order of 800 million AI tool uses a month. That is not a fixed cost spread across a base. That is a bill that arrives every month and grows with love.

So Canva ran into something more awkward than a competitor. Its brand promise and its cost structure came apart. The more faithfully it delivers the thing people adore it for, in the new medium, the more expensive it becomes to be Canva. Perkins stopped and rebuilt, and Blackbird's Rick Baker says the cost of serving AI is now down 90 percent, which if it holds is the most consequential fact in the entire news cycle and the one least likely to be quoted.

The question for everyone else is unpleasant in proportion to how good your positioning was. What does your brand promise cost to keep, per use, now that the feature customers want most is metered? A great deal of positioning written between 2015 and 2023 quietly assumed the answer was zero, forever.

The employer brand takes the first real damage

Note the timing on that internal number. The independent valuation cutting about 20 percent exists to satisfy US tax rules ahead of an annual share issue going to more than five thousand employees. Australian capital gains rules are shifting in a way that could leave staff taxed against a recovery they have not banked.

Same pattern as Airtable, different mechanism. Late-stage equity has been the most effective employer-brand instrument in technology for fifteen years. Take less cash, join the story, believe the line goes up. It is a brand product in the strict sense: a promise about a future outcome, sold to a sophisticated audience, priced entirely on narrative.

Canva did not mismanage its way here, and that is exactly why it lands. A story about a badly run company teaches nothing. A story about a company that has been profitable for nine straight years, grew 25 percent last quarter, and told its own investors the truth about its own cost problem teaches every candidate in the market that the instrument was mispriced all along. Recruiting loses pricing power on the narrative half of the package, and no amount of employer-brand content wins it back.

The strangest part is that the assistants are recommending it

The simple version of this story is wrong, and the reason is worth the whole article.

While Lemkin argues that agents are routing around Canva, one of Canva's fastest-growing acquisition channels is agents. Users held more than 26 million conversations with the Canva app inside ChatGPT by October 2025. Sacra's read on the company flags LLM referral traffic as a new inbound channel outright. Assistants are being asked to make something and are handing the job to Canva by name.

That is brand equity performing precisely the function brand equity is supposed to perform, in the one venue nobody built a strategy for. A model asked to recommend a design tool reaches for the name with the most mental availability in its training data and its retrieval, and in this category that name is Canva. Thirteen years of being the default answer to how do I make a poster turns out to be a retrieval asset. Canva's own B2B CMO argued in June that brand is now infrastructure. She was more right than she knew, and about a different layer than she meant.

So both things hold. The brand is being routed around inside the workflow and reached for inside the recommendation. Which of the two compounds faster is the live question in the business, and it is a brand question, not an engineering one.

What it speaks to

The comfortable reading of this week is that AI came for a beloved brand, so nobody is safe. The useful reading is narrower and less flattering.

Canva's brand is not damaged. It is arguably the healthiest asset on the books. What got repriced is everything stacked on top of it: the growth slope, the story premium, the assumption that scale plus affection produces a durable multiple. Strip the narrative out, the way the Airtable comparable did in public a fortnight ago, and what remains is the name, the habit, and the cost of leaving. For a company with 265 million monthly users and around four billion dollars of annualized revenue, the market's working guess at that residual currently runs from twelve billion to thirty-five, and the width of that spread is the actual story.

Being loved and being needed are not the same asset, and only one of them shows up in a multiple. Canva holds an extraordinary amount of the first. The open question, for Canva and for every brand whose promise was we make the hard thing easy, is how much of the second survives a technology that makes the hard thing free.

Nobody is safe. The exposure is not random, though, and it is measurable. Most brand teams will spend the next two years measuring affection instead.