Figma Revenue Rose 43% While It Absorbed AI Agent Inference Costs

Figma grew revenue 43% year over year, according to a FourWeekMBA report, which also noted that the company covered the inference costs of its AI agents rather than passing model calls straight through to customers.

Read together, the two facts frame a question every software vendor is now answering. Generative features can lift seat prices and keep users inside the product longer, but every agent invocation carries a real inference bill, and that bill steepens with usage.

Figma's approach is friendly on the product side: design teams do not pay per generation or per agent action, so new capabilities get used instead of rationed, and they slot more easily into daily collaboration. The trade-off shows up in gross margin, where absorbed AI costs grow with adoption.

That is why revenue growth and AI costs belong in the same paragraph. A 43% increase says demand is still expanding; absorbing inference costs says management is currently prioritizing penetration and habit formation over near-term margin maximization.

For other SaaS companies this is a referenceable but hard-to-copy path. Keeping inference on your own books requires healthy cash flow and confidence in future pricing power. And once an agent becomes the default entry point, usage shifts from occasional to constant, which changes the cost structure again.

Two things are worth watching: whether Figma separates AI-attributable revenue from inference costs in its financial communication, and how pricing changes once bundled AI allowances run out. The first determines whether outsiders can judge the unit economics; the second determines what users will actually pay.