Figma Q2 revenue jumps 48% to $370M, AI agent use hits 50%

Figma's second quarter numbers look like a company hitting its stride and a company spending everything it earns, at the same time. Revenue came in at $370.1 million, up 48% year on year, the third consecutive quarter of accelerated growth, according to Figma's release on Stocktitan. The company raised its full-year outlook by $40 million to a range of $1.463 to $1.467 billion, and the AI adoption metrics read as real rather than vaporware: over 50% of paying customers with more than $10K in ARR are using the Figma agent weekly, and over 80% of that same cohort consume AI credits weekly.

Then the market read the rest of the release and the stock fell after hours. Third quarter guidance landed at $373 to $375 million, roughly 1% sequential growth from a quarter that grew 48%, and management held operating income guidance flat while raising revenue, per coverage of the earnings call at BigGo Finance. Non-GAAP operating margin was 10% for the quarter, and the implied full-year non-GAAP operating margin sits around 8% at the midpoint, below the 12% achieved in 2025, as AI and go-to-market investments accelerate.

Why this matters if you are not holding the stock: this is one of the first clean looks at what AI monetization actually costs a software company that has been running at an 84% gross margin. Figma is telling customers to burn credits, the credit revenue is showing up, and the compute behind it is not free. Cost of revenue rose faster than sales, reducing both GAAP and non-GAAP gross margins by five percentage points. The agent adoption numbers are the good news, and the margin picture is the price of it.

The honest caveats are worth flagging. The release does not break out how much of that 48% growth came from seat expansion versus AI credit consumption, and it does not say when management expects margins to recover. It is one quarter of monetization data, not a trend, and AI-native design tools are moving in the same direction, so Figma's moat argument now has to include the agent, not only the canvas. What the reporting does not give you is the unit economics of a single credit, or how sticky weekly agent use actually is once the novelty passes.

For a founder or product leader watching this: the shift from seat-priced SaaS to usage-priced AI is not a free upgrade, and even a company with 136% net dollar retention and record customer growth has to eat the compute bill before the operating leverage shows up.