# Canva Slashed Its 2026 Growth Forecast After AI Costs Blew Past Expectations > Source: > Published: 2026-08-12 23:15:45+00:00 *Canva's AI story is not a simple growth story anymore. The company is showing you the hard part of generative software: the product gets better only if the cost of serving it stays under control.* You don't have to look far to see why Canva is pushing so hard into its own AI stack. The company has more than 265 million monthly users, more than 31 million paid users, and $4 billion in annualized revenue, according to figures Canva co-founder and COO Cliff Obrecht gave TechCrunch earlier this year. That is real scale. It also means every expensive AI request arrives multiplied by millions. That is the trap. A feature that looks clever in a launch demo can become a margin problem when ordinary users start asking it to make videos, edit presentations, pull in workplace data and build whole campaigns from a prompt. Canva's answer arrived at Canva Create 2026 in Los Angeles on April 16, where the company introduced Canva AI 2.0 as a research preview. BusinessWorld reported that Obrecht described the release as a shift from a design platform with AI tools to an AI platform with design tools. That line matters because it changes what Canva is selling. It is no longer only trying to help you make a poster or a deck faster. It wants to sit at the start of the work itself, taking a rough idea, a meeting transcript, an email thread or a campaign brief and turning it into something editable inside Canva. Perkins called the suite Canva's biggest launch ever in the same briefing, according to BusinessWorld. That would sound like ordinary launch-day language if the company had not put hard numbers underneath it. Obrecht said Canva AI usage tripled over the last year, while the company's business-to-business segment reached $500 million in annualized revenue. A product used casually by consumers is one thing. A product embedded in company workflows is something else. ## The cost problem is the real story Canva's own launch material says Canva AI 2.0 is built around a design model that can generate fully layered, editable work rather than flat images. That distinction is not cosmetic. If you use Canva for real work, you need to change the headline, swap an image, move a button and keep the brand rules intact. A locked image is a toy. An editable design is software. But editable AI costs money to serve. That is why the efficiency claims are more important than the product adjectives. IT Brief reported that Canva said its Lucid Origin image-generation model is five times faster and 30 times cheaper than comparable frontier alternatives, while its image-to-video model is seven times faster and 17 times cheaper. Canva also pointed to a Proteus style transfer model that it said was twice as fast and 23 times cheaper. Those are the numbers to watch. If they hold up at scale, Canva can offer more AI inside ordinary subscriptions without letting third-party model bills eat the business. If they do not, the company faces the same ugly choice every software company faces now: charge more, ration usage, or accept lower margins. Frankly, that is where the AI boom gets less romantic. Someone has to pay for the inference. ## Figma shows the same pressure Canva is not fighting this math alone. Figma's first-quarter 2026 filing with the SEC said revenue rose 46% year over year, but cost of revenue jumped 253%. The company attributed $33.7 million of that increase to technical infrastructure and hosting costs tied to AI and higher paid-user usage. That is a plain warning from a public company filing, not a gloomy analyst note. The Motley Fool reported in July that Figma shares fell 51.6% in the first half of 2026 despite strong operating results, with investor concern focused partly on AI-native competition. The filing shows another piece of the same story: even when customers keep paying, AI can make the cost side move faster than investors like. Growth is still good. Uncontrolled growth is not. Canva has one advantage Figma does not have in quite the same way: it is still private, so it can make expensive infrastructure choices before the public market grades every quarter. It also bought Leonardo AI in 2024, giving it a stronger base for its own creative models instead of relying only on outside systems. That acquisition looks less like a product tuck-in now and more like insurance. For you, the useful lesson is not that Canva has solved AI economics. It has not proved that yet. The lesson is that the serious software companies are already moving past the launch-demo phase and into the margin phase. They need models that are cheaper, faster and specific enough to do the actual job users pay for. Canva AI 2.0 may turn out to be a strong product. The harder test is whether Canva can keep making AI feel abundant without letting the bill run the company. That is the part every SaaS founder should be watching. **Also read:** [Phoebe Gates's Phia Knew About Cookie Stuffing Months Before It Admitted It](https://startupfortune.com/phoebe-gatess-phia-knew-about-cookie-stuffing-months-before-it-admitted-it/) • [How to Use Long Tail Keywords to Dominate SEO in 2026](https://startupfortune.com/how-to-use-long-tail-keywords-to-dominate-seo-in-2026/) • [How Founder Secondary Sales Actually Work and Who Has to Say Yes](https://startupfortune.com/how-founder-secondary-sales-actually-work-and-who-has-to-say-yes/)