Not so long ago, visual design platform Canva was the bright new kid on the block. But the advent of AI has taken the gloss off the 13-year-old vendor's meteoric rise, even as it continues to accumulate new users and customers. Its mission to democratize access to visual design still has strong appeal in a market where the tools have traditionally been hard to learn, expensive to use, and don't easily connect together — in the enterprise market as well as among individuals and smaller businesses. But AI agents provide a new way to dismantle these same barriers, taking on the task of figuring out what tools to use and how to join them together to achieve the results users are looking for. This means that, even while Canva itself embraces AI to streamline and enhance its own capabilities, it's competing for market share and user mindshare with a new generation of AI-native vendors.

This week, diginomica caught up with Cameron Adams, Canva co-founder and Chief Product Officer, during a visit to London for the latest in its series of AI Vision events. Canva has been no laggard in responding to AI, and earlier this year it launched a complete re-architecture of its platform to put agentic AI at its heart. The goal of Canva AI 2.0, the company said at the time, was to democratize access to AI capabilities in the same way it had done for visual design. But that ambitious move came at a price, as Adams explains:

When we launched it in April, there was one thing we didn't foresee. This is kind of a humble brag, but we had too many people want to use the product. So we've got 250 million people that are using Canva. When we announce something like this, millions of people swarm on it. And what we saw as they started using Canva AI 2.0 was that it was more engaging — people who used Canva AI 2.0 were three times more likely to use AI.

So they were using AI three times more because of the experience that they were getting, and we had 75 million people using our AI tools. But this was causing an immense load on our architecture, on our systems, and frankly, on our costs. It was incredibly expensive to bring someone into Canva AI 2.0, because it cost us more to generate stuff, they were using it more, and it was this vicious cycle of increasing cost.

The ensuing pause in the roll-out of the new AI platform while Canva sought to get these unanticipated costs under control meant a sharp slow-down in its expected revenue growth and caused jitters among investors. According to the Australian Financial Review, the company's Q2 update to investors in early August downgraded its full-year growth projections from 30% down to 20% — although that's still a very healthy growth rate from the $4 billion in ARR that it reached at the end of 2025. The title then reported that this led to some investors looking to offload shares in the secondary market at a steep discount to its huge $42 billion USD valuation. Many investors, of course, are keen to see the company launch an IPO, but Adams says its founders are in no rush to take that much-anticipated step.

Meanwhile, the company has been hard at work resolving those AI cost challenges, as he explains:

The demand exploded before the economics were ready, and so we paused the roll-out in order to get our systems in order. We needed to do two things there. We needed to make our AI systems more efficient, and we needed to make our AI systems cost less.

The team at Canva have done an amazing job over the last four months to really focus on this from a number of different vectors — looking at the models we use, whether they're in-house or they're external models, looking at the types of work that people want to do with their AI, optimizing models for that, figuring out cost savings we can use by running on different servers and compute around the world.

The happy story is that we've managed to get down our compute and our costs and still deliver the exact same amazing AI experience. We've actually managed to reduce the cost per task by 90%, which then enables us to roll out to even more users, and also to include free users in Canva AI 2.0.

While all this has been going on, Canva has continued to expand its core functionality, with 100 updates since the start of the year across the tools that make up its flagship Visual Suite, and the launch of a new ProSuite that brings together its family of tools aimed at professional designers — the Affinity design platform, Cavalry animiation tool, Flourish data visualizer and Leonardo generative AI platform, accompanied by a raft of feature updates and improved workflows and automations. Next up is the long-awaited iPad version of the updated Affinity platform, which will arrive in November.

The addition of ProSuite rounds out Canva's offering to the enterprise market, providing a suite of tools for creative experts that complements and integrates with the general-purpose Visual Suite. The company says its enterprise business has doubled in the past year, and a growing roster of enterprise customers are happy to tell their stories. But will that growth continue when AI assistants and agents provide an alternative means of accessing previously complex tasks? While Adams acknowledges there's a trend towards the use of foundation model agents such as Anthropic Claude, OpenAI's ChatGPT or Google Gemini as the entry point for accessing functions and data in the enterprise, he believes that Canva still provides essential context and workflows that enterprise users need to take their work forward. He tells me:

Someone will ask AI for something and it will give them the first design, but they need to take that further. They need to share that with their team. They need to change something in that design. And figuring out the flows, which we have done getting from Claude or ChatGPT or Gemini into Canva, is a large part of the experience that we think about...

This platform, where AI, people and creativity meet, is proving to be a winner. Enterprise teams are still choosing Canva to house all of their work on, to drive their marketing, their creative, through to then surface all of that brand to the rest of their teams across the company, whether it's a sales team or their internal HR team, their comms teams. They're still choosing that because Canva is the best place to do that work. Sure, they have different on-ramps now that can get into the Canva flow, but having that true creative operating system is still proving to be immensely valuable.

Canva also believes that there's growing dissatisfaction with the generic visual style that AI foundation models typically produce. AI-generated images tend to have a recognizable style that betrays a lack of individuality and authenticity. Adams comments:

I'm sure you've seen a lot of commentary about types of content that your generic AI tools can create, and not all of it's positive. Often the content doesn't feel like it's yours. It doesn't feel authentic. It feels like the average of content that you can get out of the market. Everything is starting to look pretty much the same... When you want to stand out and you see your content [look generic], where do you go? What do you do? ...

We've been working on fusing that creative platform with the world's best AI in order to create this new wave of creativity that represents individuals, creates real content, and actually engages your audience.

Adams cites five elements Canva has built into its AI platform that he believes will help its customers' creative work stand out:

• Individual context. He says: "It needs to understand what someone is trying to make. So being able to take in all your context, your ideas, and work with you to produce the content from it."
• Memory. He explains: "It needs to understand the work you've already done... If you've created 10 sales decks in Canva, you should be able to create the 11th faster than you created the 10th."
• Collaboration. "We know that people aren't just working with AI; they're working with their teams. They're working inside their organizations. They're working how they've always worked, and it's that beauty of human relationship and collaboration that produces the best work. So we need a platform that recognizes that and makes sure that you can still work with the person across the hall or in another city, just as you would with any other product."
• Organization context. "You need to know the organization that you're a part of, so it's not just your work, but all the work across your organization that feeds into what you create next."
• Quality. "The quality bar has to remain high. It can't be like that [generic] flyer that gets pumped out 500,000 times a day. It needs to be authentic. It needs to be high quality. It needs to be you."

Canva is also committed to making sure that AI augments human creativity rather than attempting to replace or mechanize it. He adds:

The goal isn't just more content. We don't want people just to create more and more designs. It's really about content that can only come from you, that places people at the center of creation. It's not just about that one prompt that you type into the prompt box. It's a real workflow of working on a document, working on a presentation, getting your ideas, sharing it with your teammates, having AI tweak it, scaling it out to multiple different audiences and different formats — it's a very messy and big workflow. And we work on improving every step within that workflow, not just the first box that you type. That's how we see the future of creative AI.

Canva's summer setbacks are typical of the challenges every SaaS vendor faces from AI. It's particularly telling to see an up-and-coming challenger like Canva face these difficulties. The vendor is still growing fast, particularly in the enterprise market, and has built up a formidable pool of AI talent that has significantly overhauled its core platform to take maximum advantage of the technology. In short, it's done everything it should, and yet it's still had to ease off the accelerator after being caught out by tokenomics and the difficulty of predicting and controlling inference overheads. So that's the first challenge — adjusting to the unfamiliar economic landscape of AI.

Canva will likely bounce back, but the next challenge is that new wave of competition from AI-native vendors. The danger of 'stealth churn' was recently identified by SaaS expert Jason Lemkin, founder of the SaaStr conference. This is the phenomenon where early adopters like Lemkin start turning to AI-native alternatives, even while their colleagues continue to use Canva. The broader trend means that the vendor seems to continue to do well, even when it's losing its most passionate advocates. Lemkin writes:

When your topline is growing like that, when new users are pouring in and B2B enterprise adoption is accelerating, you cannot see stealth churn in your numbers. It’s invisible. The new revenue more than masks the quiet disengagement happening at the edges... But those power users are often the leading indicator for where the broader market goes 12-18 months later.

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.

The final challenge is that how a vendor sets out its value proposition has to change. Canva's traditional narrative of an easy-to-use platform with joined-up workflows as an alternative to complex, expensive and fragmented tools certainly still resonates in the broader market. But customers can now achieve similar results with AI-native alternatives, in which an AI agent interprets the user's requests and orchestrates the necessary resources on their behalf. The differentiation is no longer in the user interface, but in what happens behind the scenes. In the enterprise AI market, vendors have to emphasize what it is about their platform that ensures AI delivers the best outcomes.

Canva seems to have the right ingredients in place, with its platform providing context, memory and team collaboration that all contribute to a strong AI story. But it doesn't yet seem to have figured out the best way to weave those ingredients into its narrative in an era when the user experience is increasingly a conversation with an AI agent rather than having to grapple with a screenful of disparate software tools.

That Canva faces these challenges is probably a salutary warning for every vendor in the industry of what's ahead. My sense is that Canva is better placed than most to weather these storms. Meanwhile enterprises need to be alert to these factors as they evaluate the vendors they choose to work with. Economic models are in flux, user adoption may be giving off misleading signals, and vendors are still learning how to best present their AI credentials. There's no reliable way to pick out winners when all of these parameters remain so ill-defined.