Canva's Growth Cut Is a Distribution Problem Wearing a Cost-of-Goods Costume

Canva entered 2026 growing 30% and now expects to finish the year nearer 20%, a disclosure attributed to co-founder Melanie Perkins in a SaaStr recap of the 20VC x SaaStr podcast with Harry Stebbings, Jason Lemkin and Rory O'Driscoll ^[1][2]. Part of the stated reason is that the AI features Canva is shipping cost real money, and subsidising frontier model calls across a prosumer base is expensive ^[3].

That is a reduction of roughly one third in the growth rate ^[1]. The temptation is to file it under gross margin, which is where AI serving costs usually get filed. O'Driscoll's read on the podcast was that the margin math is a second-order business model issue, and that the question that matters is whether 30 to 20 is on the way to 10 ^[4]. Cost of goods is a number you can engineer down. A demand curve that is being re-routed somewhere else is not.

The scale numbers make the stakes concrete. Canva did roughly $3B in GAAP revenue last year and is described as private at about $3.6B growing 20% ^[5][6]. The two comparables cited are Adobe at $23B growing 12% and trading at three to four times revenue, and Figma public at $1.4B growing 40% ^[7][8]. All three, per the discussion, face the same question: whether AI is a feature you incorporate or a new thing that makes you obsolete ^[9].

Figma took its version of the hit in public. According to the recap, Dylan Field was explicit that gross margins will be meaningfully impaired as agentic products get used, and the stock traded down 20% ^[10]. O'Driscoll's argument for why Figma is nonetheless better protected has nothing to do with model costs: Figma sells to enterprises coordinating groups of people building software, so even fully automated creative work leaves corporate process and workflow to monetise, while Canva's prosumer motion is exactly where a typed request and an immediate output substitutes best ^[11].

The sharpest datapoint in the transcript is not financial. Lemkin says SaaStr churned Canva and Notion, having built its own ad server and creative generation network running on agents, and that the agent never once suggested Canva ^[12][13]. Stebbings tied that to what he heard from the President of Uber, whose stated biggest fear is disaggregation of the UI, with an assistant routing a request to whichever provider is cheapest ^[14]. Losing a user to a competitor is a pricing and product problem. Not appearing in the consideration set of the layer making the choice is a distribution problem, and it does not show up in net revenue retention until late.

On price, Lemkin lands Canva at around $12B today, citing 20% growth on roughly $4B ARR against public comps, decelerating rather than accelerating ^[15]. That is about three times ARR ^[2]. O'Driscoll's pushback: Datadog, Cloudflare and JFrog grow in the mid-20s with 20%-plus operating margins and trade at 15 to 17 times NTM, and the difference is that nobody asks whether those companies exist in five years ^[16]. The spread between three times and 15 to 17 times is five to six turns of revenue, all of it existential discount ^[3]. Either way, the recap notes liquidity only arrives at the end of a journey at least 12 months long, and Stebbings' note to LPs holding Canva is that when the private window shuts, getting anything done at scale becomes very difficult ^[17][18].

Watch whether 20% holds through the year or slides, whether Canva or Adobe start disclosing model-serving cost as a unit economic rather than an aside, and what secondaries print while that 12-month window stays closed.