It's also not fake-cheap dashboard SaaS.
Figma is a vulnerable-but-salvageable company with a credible path to winner status — if it proves it becomes the agent-native visual product-creation layer, not merely a seat-based design tool with AI features.
AI multiplies the prototypes, variants, and experiments flowing through product teams. That helps Figma if it becomes the shared canvas. AI also lets PMs and engineers go straight from prompt to code in Lovable, v0, Cursor, Claude Code. That weakens it.
Whether Figma wins depends on which dominates. The strategy is to make the canvas the place where both forces converge — humans and agents both editing the same shared object.
Every PM, founder, marketer, and engineer can now generate prototypes. Make weekly active users were up +70% QoQ. More than 75% of customers above $10k ARR consume AI credits weekly.
PMs go straight from prompt to working app. Engineers iterate UI in code. Vercel's v0 ships production deployments from prompts. Anthropic's Claude Design (April 2026) targets visual creation. None of them require a Figma file.
The bull case isn't "Figma adds AI." That's commodity. The bull case is that AI creates so much software, product teams need a shared, visual, governable place where humans and agents can inspect, refine, version, approve, and systematize what gets created.
Figma's bet is that it becomes that place. Today's price says the market doubts it.
Figma priced its IPO at $33 in 2025, traded above $100 on day one — implying about $50B of market cap — and is now below $20. The market is pricing four things at once: IPO valuation excess, AI seat compression, AI inference costs, and post-IPO dilution.
Two clusters of metrics matter. The first three are bullish — enterprise customer expansion is real, broad, and accelerating. The fourth is bearish, and it's the most important data point in this entire analysis.
Three of those four would be the headline of any other software IPO. The fourth is the AI signal.
The 9-point compression over two years is the early AI tell. 136% net dollar retention is the offset. The next two years will decide which signal compounds.
The valuation argument is a scenario problem. At 5.5x EV/FY26 revenue, the market is no longer pricing Figma like a scarce 40% grower with 90% gross margins and endless seat expansion. It's pricing deceleration, AI-cost pressure, and uncertainty about the terminal model. But it isn't pricing impairment either.
Tap each scenario. The chart redraws to show the revenue path, FCF margin, multiple range, and resulting equity outcome. The "what must be true" column is the lens — fact-check the world against it, not against a target price.
The honest read: today's price doesn't demand the bull case. It demands the base case to prove out within four to six quarters. The first measurement window is Q1 2026 earnings on May 14, 2026 — the first read after AI-credit enforcement began in March.
Figma already won that fight. The pressure is from AI-native tools attacking different parts of the design-to-code chain — and from model providers themselves, who increasingly want to own visual creation rather than power it.
Open each row to see how it attacks, how Figma defends, and what I think.
Brand, Creative Suite distribution, existing designer relationships. Adobe's bundle is still the default in agencies and education.
Already won most of collaborative interface design. The Adobe deal was abandoned over regulatory concerns; Adobe XD never recovered as a credible alternative.
Prompt-to-app, prompt-to-prototype, working code, deployment. They absorb the incremental work that would have grown Figma's seat count.
Owns design-system context, cross-functional review, and enterprise governance. Make is Figma's bet that those advantages travel into the prompt-to-app workflow.
Engineers iterate UI directly in code. Visual editors move into the IDE. Cursor's "one agent across every surface" — GitHub, Slack, Linear, JetBrains — is uncomfortably close to Figma's strategic territory.
The Dev Mode MCP server feeds design context into agents. Done well, Figma becomes the system of record for visual context that agents need; done poorly, it's the artifact agents stop reading.
Agents may become the workspace. Claude Design (April 2026) shows model providers want to own visual creation, not merely power it. The chat window becomes the canvas.
Stays neutral across models. Becomes the enterprise system of record that Anthropic, OpenAI, and Google all integrate into — rather than competing with each.
Lower price, open file formats, code-native design systems. Engineering-led teams increasingly say the source of truth is Storybook, not the canvas.
Collaboration depth, ecosystem, enterprise admin, plugin community, embedded workflow memory. Hard to recreate; harder to switch off.
The story-card method: name the qualitative change you're betting on, list the observable signals that would say it's working, and the tripwires that would say it's breaking. Watch the signals after they're set, not before.
Six dimensions, scored. Founder-led quality and financial resilience are the load-bearing strengths. Pricing-power durability is the load-bearing question.
An excellent founder-led company, real customer love, strong financial base, credible AI strategy — but still a hybrid transition story with unresolved pricing-model risk.