8%.

That was Figma's stock drop on the day Google Stitch launched. The decline expanded to about 13% within two days; Figma stock has cumulatively fallen about 35% since 2025.

Stitch's core selling point is "Vibe Design" -- you don't need to draw wireframes, drag components, or adjust pixels. You just talk. "Give me three menu options, Airbnb style," and Stitch directly generates interactive UI and frontend code. Free, distributed through the Google Workspace ecosystem (hundreds of millions of enterprise users across Gmail, Docs, Drive).

This isn't trivial competition for Figma -- this is Google using the "free + bundling + distribution" triple play, going straight for Figma's moat. But Figma didn't sit idle.

Figma's Counterattack: Code Layer + Agent

On June 25, Figma released perhaps the biggest update in its history. Three core changes deserve attention:

First, the code layer enters the collaboration canvas. Teams can directly clone code repositories within Figma and test flows at the design layer; the wall between design and development is further dismantled. This isn't a simple "design-to-code" plugin -- code itself becomes part of the design canvas, with designers and developers collaborating in the same space.

Second, Design Agent is embedded in the canvas, not a sidebar. This is the most brilliant aspect of Figma's AI strategy. Previous AI design tools (including Figma's own early AI features) placed ChatGPT-like dialog boxes in the sidebar -- you describe a need, AI generates a design, you copy-paste it back. This creates massive "Toggle Tax": frequent switching between design tools and AI tools causes context loss; efficiency drops rather than rises. Design Agent embeds design-system-aware Agents directly in the canvas, with design systems (component libraries, tokens, variables) serving as AI's exclusive context window. The more mature a team's design system, the greater the AI gains.

Third, Parallel Prompting and control priority. Agents can simultaneously generate multiple style/direction options for selection, but AI generation serves only as a starting point -- Figma deliberately emphasizes "humans make the final decisions," preventing AI from overstepping. This is a precise reading of designer psychology: designers don't fear AI helping; they fear AI designing for them.

The Paradigm Shift in Design Tools

The Figma vs Stitch war is essentially a collision of two design tool philosophies.

Stitch represents the "zero barrier" route: enabling people who can't design to create UI with voice. Its target users aren't professional designers, but product managers, entrepreneurs, small business owners -- people with ideas who can't use Figma. Free is its weapon; the Google ecosystem is its distribution channel.

Figma represents the "augment the professional" route: making professional designers and design teams more efficient. Its moat is design systems -- enterprise component libraries, tokens, variables, design specifications. These are organizational assets built over years; the deeper AI understands these assets, the greater the value.

The Endgame of AI Design Isn't Replacement

The UX Tools Spring 2026 survey (1,478 designers) shows 53% of teams now rank "evaluable AI generation efficiency of tools" as a core selection criterion. This indicates designers have shifted from anxiety about "will AI replace me" to pragmatism about "which AI tool actually helps me be more efficient."

The AI-ification of design tools isn't a zero-sum game. Figma's 2025 revenue was $1.06 billion (+41% YoY), showing that even amid intensifying competition, the design tools market is still growing. AI isn't eliminating designers' jobs; it's redefining them -- from "artists" who draw boxes and align pixels, to "design directors" who define design systems, guide AI, and control quality.

Sketch to Figma was a collaboration revolution; Figma to AI-native design tools is a productivity revolution. In 2026, this revolution is just heating up.