Figma Beat Revenue but Lost 18% — The Market Isn't Worried About AI Costs. It's Worried About What Figma Is Becoming.

Figma's stock fell 18% on the day after the company reported Q2 2026 revenue of $370.1 million — a 48% year-over-year jump that beat analyst consensus of $352 million. Non-GAAP EPS of $0.08 doubled the $0.04 estimate. The headline numbers were strong. The market sold anyway.

The headline blame is AI costs. Gross margins declined, as AI compute expenses grew. Those are real signals, and they deserve attention. But the market's punishment here runs deeper than quarterly margin pressure.

The architecture shift the market is pricing in

Dylan Field said it on the earnings call, almost as a thesis statement for the company's next five years: "As code gets commoditized and value moves up the stack, the opportunity is building the canvas for full-stack creation." He then walked through Code Layers, the Figma agent, generative plugins, and Skills — all products designed to move Figma from a design tool to a platform where AI agents consume the application at scale.

Put plainly, Figma is no longer selling design seats. It's selling AI consumption infrastructure for product teams.

That is not a lateral move. It's an entirely different cost structure. Design tools are high-margin SaaS — software running on shared cloud infrastructure, charged per seat, with gross margins typically above 85%. AI agent infrastructure is closer to cloud compute — every AI credit consumed carries a model inference cost that flows directly through cost of revenue. That's why gross margins compressed in the first full quarter of AI credit monetization.

The adoption numbers suggest this transition is real, not rhetorical.

This is what separates Figma from Canva and Adobe. Canva's AI bets are optimized for marketing creatives and one-off content generation. Adobe Firefly is bolted into a legacy desktop toolchain. Figma's architecture — a browser-based collaborative canvas — is the natural operating surface for multi-agent product development workflows. The platform is already where designers live. The question is whether developers and AI agents will live there too.

The margin math that changed in Q2

The cost of this transition shows up in three places:

The stock-based compensation story is its own signal. That's not just dilution — it's management buying growth and talent at scale when the market itself is skeptical. Stock-based compensation surged because Figma needed engineers to build AI infrastructure and salespeople to monetize it.

The balance sheet is not under stress. But the trajectory is clear: revenue is growing 48%, while the cash conversion rate from revenue to operating cash flow has compressed sharply.

What would change the thesis

The long-term bull case remains intact. If Figma's canvas becomes the default surface where product teams — designers, developers, and AI agents — interact, the total addressable market expands far beyond design seats. Software margins on AI consumption credits could eventually exceed the original seat-margin model, once model costs compress and scale effects kick in.

However, the near-term return profile is uncertain. Three specific things would convince me the architecture transition is on track:

• Gross margins stabilize above 87% while AI credit revenue continues to accelerate — proving that model inference costs are scaling down faster than credit consumption is scaling up
• Q3 revenue hits $375 million or higher — suggesting the implied growth rate is a trough, not a trend
• AI credit revenue as a percentage of total revenue crosses 15% and carries its own gross margin above 70% — showing the new revenue stream is profitable on its own terms, not just a cost center subsidizing seat growth

Conversely, the thesis breaks if gross margins fall below 80% while revenue growth decelerates below 30%. That would signal AI infrastructure costs are eating value faster than the platform is capturing it. Or if AI credit adoption flattens — if weekly adoption among enterprise customers stops growing — then the investment thesis collapses regardless of margins.

Where capital goes

The debate is not whether Figma remains an important platform. The debate is whether the architecture transition — from design tool to AI consumption infrastructure — justifies the current cost trajectory.

The stock has fallen from a 52-week high of $91 to $23 today. At $12.3 billion market cap, it trades at roughly 9.6 times trailing sales. That's not cheap for a company with negative GAAP earnings and shrinking margins, but it's not absurd for a 48%-growth platform either. The forward P/E of 283x tells you what the earnings story looks like right now — which is nothing. The question is whether earnings reappear as AI margins stabilize.

I believe Figma is on the right side of a real architectural shift — design is moving from human-created artifacts to agent-consumed canvases, and the company with the best collaborative canvas wins that transition. But being on the right side of a transition doesn't mean the stock makes money in the transition itself. The margin compression and the massive increase in operating leverage all point to a period of earnings uncertainty that could stretch into 2027.

In my opinion, this is not a position for large allocation right now. The product architecture thesis is compelling, but the financial mechanics are still resolving. A smaller position — something in the 2–3% range that can absorb volatility without dragging the portfolio — makes more sense than the 10% position a revenue-beat headline might tempt you into. Wait for evidence that gross margins are stabilizing before committing more capital.

The break condition is clear: if Q3 shows margins above 87% and AI credit revenue crossing 15% of total revenue, the story shifts from "expensive transition" to "inflected platform." Until then, I'd rather deploy capital where the architecture advantage has already translated into financial proof.

Demand is robust. The risk is just that the transition period is longer and more expensive than the market priced at $91.