Forum

AI Is Not Free: What Canva’s Slowdown Tells Designers

A utility pole and power lines silhouetted against a vivid pink and orange sunset

For a few years, generative AI features in creative software have felt almost free to the people using them. A button appears, an image or layout comes out, and the cost stays invisible. On 12 August, one of the most successful design companies in the world made that cost visible.

Canva told Fortune that it had cut its expected revenue growth rate for the year by a third, to 20 percent, and slowed the rollout of its AI features after the cost of delivering them turned out higher than expected. Co-founder and chief executive Melanie Perkins said demand for the new features had “significantly exceeded” the company’s expectations. Canva said it would rebuild parts of its architecture and reduce unit costs before rolling out more broadly, and that it had already cut the cost per AI task by nearly 90 percent since launching Canva AI 2.0 in April.

This is a company that grew quickly and made money, now saying that AI broke one of its assumptions. Designers should pay attention.

The economics behind the button

Traditional design software has close to zero marginal cost: once it is built, another user costs very little. Generative AI is different. Every image, layout and edit runs on expensive computing hardware, and heavy users cost far more to serve than light ones.

That has practical consequences for the tools designers rely on:

  • Credits and quotas are here to stay. Expect generative features to be metered, tiered or rate-limited, even in subscriptions that feel unlimited.
  • Features may arrive slowly or unevenly. A tool may announce a capability long before it is available to everyone.
  • Prices will move. When costs fall, prices may follow; when demand spikes, limits may tighten.
  • Free tiers will be the first to change.

Not only Canva’s problem

Canva is unlikely to be the only company facing this arithmetic; it is simply unusually candid about it. Any service that offers generative features to a large base of users pays for every request. What differs is how openly companies explain the trade-offs to the people who depend on their tools, and that candor gives customers a reason to plan.

Design for efficiency

For working designers, the lesson is not to avoid AI features but to use them deliberately. Generating fifty variations because it is easy has a cost, even if the bill goes to someone else for now. Thoughtful briefs, fewer and better prompts, and editing rather than regenerating are habits that save time today and money later.

There is also an environmental dimension. Computing at this scale uses a great deal of energy. Studios that care about the footprint of their work can include AI usage in the same conversations they already have about printing, travel and hosting.

What it means for small studios

Large agencies can negotiate enterprise contracts. Freelancers and small studios are more exposed to sudden changes in pricing and limits. A few protections help:

  • Avoid building a whole service around a single AI feature you do not control.
  • Track how much generative capacity a typical project actually uses.
  • Build fees that reflect the real cost of tools, including AI credits, rather than absorbing them silently.
  • Keep a fallback workflow for when a feature is throttled or withdrawn.

A view for entrants and educators

Students often meet AI tools through generous free trials and educational plans. It is worth teaching the economics alongside the techniques, so that new designers understand what their tools cost to run and why access can change. The International AI Award asks entrants to explain why they chose a particular model or tool. Cost, availability and sustainability are legitimate parts of that answer, alongside quality.

The takeaway

Canva’s announcement is not a sign that AI in design is failing. Its users clearly want these features. It is a sign that the invisible costs are becoming visible, and that the economics of creative software are changing. Designers who understand that shift, and work with it rather than assuming abundance, will have more stable practices and fewer surprises.