Q&A with Canva CEO Melanie Perkins on the company's growth in enterprise, competing with AI labs, token pricing, investing in its own models, and more
Context & Ripple Effects
Earlier coverage tracked Canva’s evolution from an Adobe challenger into a global creative-software platform, with growing enterprise adoption, a developer ecosystem, and substantially larger scale in users and annualized sales. The latest interview places AI-model investment and token pricing inside that broader expansion.
The strategic issue is not simply adding AI features: Canva is positioning its own product distribution and enterprise relationships against AI labs that increasingly offer creative capabilities directly.
First-order effects
- Canva is committing resources to proprietary AI models while framing token pricing as part of how customers access AI capabilities.
- Enterprise customers gain a supplier that is trying to package AI creation tools within an existing design workflow rather than sending users to standalone AI labs.
Second-order effects
- Canva’s move raises pressure on AI labs and creative-software rivals to compete on workflow integration, enterprise administration, and pricing—not only model output quality.
- Token-based access makes the cost of AI features a more explicit product decision for Canva customers, increasing the importance of demonstrating measurable value per use.
Third-order effects
- If creative platforms continue building or tailoring their own models, AI differentiation may shift toward distribution, integrated workflows, and customer relationships rather than residing solely with model providers.
- Enterprise AI purchasing could increasingly consolidate around existing work platforms that bundle governance, creative tools, and metered AI usage; the balance between bundled access and token charges remains unsettled.
The trend: This is one instance of established software platforms using proprietary or tailored AI models and usage-based pricing to defend workflow ownership as AI labs move closer to end users.