How AI Costs and High Demand Slashed Canva Growth Forecast
Canva cut its expected revenue growth rate by a third to 20% after the unexpectedly high cost of delivering artificial intelligence features prompted the design-software company to slow its rollout, according to reporting by Inkl and Yahoo Finance. The margin compression exposes deep structural cost challenges facing the broader software-as-a-service sector.
The Unit Economics Breakdown Behind the Growth Revision
For years, Canva distinguished itself within the venture ecosystem by scaling rapidly while maintaining profitability. That trajectory shifted as enterprise users embraced generative tools at rates that overwhelmed initial architecture assumptions. According to email statements provided to Fortune by Canva CEO and co-founder Melanie Perkins, user demand for new AI capabilities “significantly exceeded” internal forecasts. This forced executives to deliberately tap the brakes on deployment to rebuild backend systems.
Perkins noted that users created three times as many designs under Canva AI 2.0 compared to previous iterations. While the agentic upgrade validated product-market fit, it simultaneously triggered severe cost burdens associated with continuous compute utilization. To address this, Canva implemented infrastructure overhauls that reduced the cost per completed task by nearly 90% since April. Yet, the sheer volume of inference requests necessitated a downward revision of annual revenue growth targets from historical trajectories down to 20%.
Similar fiscal strains are visible across the competitive landscape. Figma, widely tracked as a public-market parallel to Canva, reported a steep drop in its free-cash-flow margin to 14% in the second quarter, down from 27% in the first quarter. Figma’s forward projections similarly reflected cooling momentum, with third-quarter revenue growth forecast at 36%, marking a notable deceleration from the 48% clip recorded in June.
Inference Expenses Versus Zero Marginal Cost Software
The core financial challenge stems from a fundamental break in traditional software unit economics. Derek Hernandez, senior research analyst covering the intersection of SaaS and AI at PitchBook, explained the structural shift to Fortune:
AI has stripped SaaS of its traditional zero marginal cost advantage, which historically served as the primary growth engine for much of the software industry. Because customers now demand far more advanced capabilities, modern technology has turned usage expenses into a widespread hurdle for every single one of these businesses.
Hernandez categorized the financial mechanics using an automotive analogy. Software training mirrors the capital-intensive manufacturing process of building a vehicle, whereas inference acts as the recurring fuel and maintenance cost incurred during actual operation. Because enterprise customers demand dynamic, real-time AI generation rather than static computations, these inference expenses recur with every single user prompt. When these usage costs scale faster than subscription price adjustments, operating margins inevitably compress.
Managing these expanding computational expenses requires rigorous corporate restructuring and careful valuation management.
Strategic Realignments Ahead of Potential Public Markets Entry
The timing of Canva’s strategic pause carries direct implications for its corporate timeline. Previous reports indicated that an employee share sale valued the design platform at $42 billion, with market observers eyeing a potential initial public offering window in 2026. PitchBook analysts now suggest that timeline may shift toward next year as management prioritizes margin stabilization over raw top-line acceleration.
By slowing broader deployments until unit economics stabilize, Canva aims to present a resilient financial profile to institutional investors. Protecting profitability in an era of expensive inference requires disciplined cost governance and technical re-architecture.
The software industry can no longer rely on the infinite leverage of traditional cloud distribution. As inference costs continue to reshape quarterly earnings reports across the sector, long-term winners will be defined by their ability to optimize backend architectures before scaling advanced feature sets to global user bases.