Figma bets on human judgment at Config 2026 while the AI powering its canvas belongs to someone else

The detail that deserves more attention is the supplier-becoming-competitor dynamic: Figma is paying API costs to at least one model provider that is actively building in the design tool category, which means Figma is effectively subsidizing a rival's R&D runway while having no contractual protection against that rival undercutting it on price or features.

This is largely disconnected from recent activity in our archive, as we have no prior coverage to anchor it to. That said, it belongs squarely in the broader conversation about SaaS platforms that bolted AI onto existing products without securing the intelligence layer, a pattern visible across productivity, creative, and developer tooling categories. The structural problem Figma faces is not unique to design: any platform that resells third-party inference without a differentiated data flywheel or proprietary fine-tuning is exposed to the same margin compression and competitive leakage. Figma's canvas expansion actually sharpens the risk rather than hedging it, because a larger surface area of AI-dependent features means greater API cost exposure at scale.

Watch whether Figma announces a direct model partnership, an acquisition in the inference or fine-tuning space, or a proprietary training initiative within the next 12 months. Absence of any such move by Config 2027 would confirm that Figma is accepting structural dependency as a permanent condition rather than a transitional one.