1 week ago
27
Canva just told investors to expect less growth, and the reason is one of the most expensive problems in tech right now: making AI products that don’t bleed money.
The Australian design software giant revised its 2026 revenue growth forecast from 30% down to 20%, a decision CEO Melanie Perkins framed as a deliberate trade-off between speed and economic sustainability. The announcement came in the company’s Q2 2026 investor update, landing at a moment when the entire software industry is grappling with the same uncomfortable math: AI features that users love can be ruinously expensive to run.
The numbers behind the slowdown
Canva’s Q2 2026 revenue came in at $921.9 million, representing 25.2% year-over-year growth. The gap between 25% actual growth and the original 30% target reflects real operational friction.
The culprit, according to the company, is a combination of delayed product launches, slower distribution, and the resource-intensive process of rebuilding Canva’s AI architecture. Perkins indicated the company chose to deliberately slow product rollouts rather than ship AI features at unsustainable unit economics.
The company closed 2025 with annual recurring revenue of roughly $4 billion. It holds $1.47 billion in cash and carries a $42 billion valuation, giving it substantial runway to weather a period of deliberately slower growth. And it has remained profitable for nine consecutive years, a distinction that separates it from the vast majority of venture-backed software companies.
Building cheaper AI from the inside
The more interesting story is what Canva is doing to fix the cost problem. The company has managed to slash AI servicing costs by approximately 90%, a staggering reduction achieved through two main strategies: developing proprietary in-house models and acquiring AI startups.
One of those acquisitions was Leonardo.AI, an AI image and video generation company. Canva’s video model is now reportedly 17 times cheaper to run than comparable frontier models, while its image model is 30 times cheaper.
Canva’s user base of 265 million monthly actives makes this math particularly unforgiving. Even small per-query costs multiply into enormous line items at that scale. A feature that costs a fraction of a cent per use can easily translate into tens of millions of dollars per quarter when hundreds of millions of people are clicking the button.
What this means for the SaaS landscape
Perkins’ framing of this as a choice rather than a failure is notable. The company isn’t cutting its forecast because demand is weak. It’s cutting its forecast because it’s choosing to rebuild the economic foundations of its AI stack before scaling further. The 90% reduction in AI servicing costs suggests that strategy is already yielding results, even if it hasn’t yet fully flowed through to the top line.
For other SaaS companies navigating the same transition, Canva’s experience offers a template and a warning. The template: invest in proprietary models and acquire specialized AI teams to bring costs under control. The warning: even with $1.47 billion in cash and a profitable core business, the transition takes longer and costs more than your financial projections assume.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) ·