Figma grows 43% but trades near 6x sales, and AI costs aren't why

Figma (NYSE: FIG) grew revenue 43% from a year earlier to $391.4 million in the third quarter, $17.4 million above the midpoint of its own guidance, and raised its 2026 forecast again. The shares still tumbled 7% in extended trading on November 4, CNBC reported, and were at $20.82 before the open on November 5, according to Investing.com. At that price Figma's enterprise value is about 6.0 times its annualized third-quarter revenue, by SaasBuzz's calculation from the company's filings: the same price per point of growth as HubSpot, which grew 17%.

Our take: the market is right to pay less for Figma than its growth suggests, but the reason most coverage gives, the cost of AI credits, is the weakest part of the case. The lower gross margin, which came as AI and hosting costs rose, cost Figma about $17 million of gross profit against last year's margin, 4% of revenue. The numbers that earn the discount are the ones under contract: billings, by our calculation, grew 24%, remaining performance obligations fell from June, and net dollar retention dropped from 139% in March to 131%.

Jason Lemkin, who runs SaaStr and invests in early-stage B2B software companies, put the puzzle the way many readers saw it the morning after the report:

Is Figma cheap at about 7 times revenue?

Lemkin's 7x holds on the market value before the results. At the November 4 close of $22.13, on Nasdaq's data, Figma's 535.6 million shares, the count on the cover of its 10-Q, were worth $11.9 billion, 7.6 times the third quarter's revenue run rate of $1.57 billion. Take away the $1.72 billion of cash and marketable securities, against no debt, and the enterprise value was 6.5 times. At the pre-open price it was 6.0 times.

Set against software companies that have reported the same quarter, that is cheap for the growth. We divided each company's multiple by its growth rate, a rough way to ask how much the market pays for each point of growth:

Source: SaasBuzz analysis of each company's third-quarter results release and 10-Q cover share count, and Nasdaq closing prices on November 4, 2026. Enterprise value is market value less cash and investments plus debt, divided by the quarter's revenue times four. Datadog's and ServiceNow's prices were after their results; Figma's and HubSpot's were before theirs.

Datadog, the closest match on growth, is valued at about twice Figma's multiple on growth nearly eight points lower. ServiceNow lost nearly as much gross margin as Figma did, falling to 77.5% from 81% on a non-GAAP basis, and still costs more per point of growth. So margin alone does not explain the gap. Figma is being priced as if its growth will not last, which is where the contract numbers come in.

How much did AI credits cost Figma's margin?

Less than the reaction suggests. Figma's non-GAAP gross margin, which leaves out stock pay, was 89.7% in the second quarter of 2025, its first report as a public company, and has drifted down since, by our calculation from each results release:

Source: SaasBuzz calculation of non-GAAP gross profit over revenue from Figma's results releases on SEC EDGAR, August 2025 to November 2026.

Had Figma kept a year-earlier margin of 86.4%, it would have made about $17 million more gross profit in the quarter; at the 89.7% of mid-2025, about $30 million more. The 10-Q says technical infrastructure and hosting costs "relating to AI and increased usage" added $29.5 million to cost of revenue from a year earlier, with $14.5 million more of AI-related costs in research and development and $8.8 million of hosting for free users in sales and marketing.

Some of it was a choice. Praveer Melwani, the chief financial officer, said in Figma's prepared remarks that "we made the decision in Q3 to cover the cost of inference for the Figma agent ahead of its GA launch on October 6th." The agent now consumes paid credits. In August Figma also repriced its credits "to include more credits in each add-on", and Melwani told CNBC that "you're potentially trading off near-term revenue, investing in your customers." Against that, Figma said it more than doubled the share of agent work run on its own post-trained model between July 1 and October 31 and cut the average cost per agent message. Gross profit dollars still grew 36%.

Dylan Field, the chief executive and co-founder, listed that gross profit growth, not the margin, in his own summary of the quarter:

Why did Figma stock fall after beating earnings?

Look past the income statement. Figma does not report billings, but revenue plus the change in deferred revenue, a common stand-in, was $391.2 million in the quarter against $314.6 million a year earlier, growth of 24%, by our calculation from the cash flow statements. Deferred revenue has barely moved since March, at $626.6 million on September 30. Remaining performance obligations, revenue under contract not yet recognized, were $677.5 million, up 31% from a year earlier but down from $687.0 million in June and $682.3 million in March, according to Figma's filings.

Net dollar retention, which measures how much more customers with more than $10,000 of annual recurring revenue spend after a year, was 131%, after 139% in March and 136% in June. The fourth-quarter forecast of $402 million to $404 million implies 33% growth, against 43% now.

The strongest objection is that Figma's own changes muddy these numbers. Its largest customers can now draw down AI spend under flexible agreements, which accounted for 80% of eligible add-ons bought from the August pricing update through October, the company said, and that can change when and how much it bills up front. Figma has also guided low all year: it beat the midpoint of its revenue guidance by $10.8 million, $17.4 million, $21.1 million and $17.4 million in the last four quarters. Both points are fair, and they are why we think the discount is too deep at 6x against HubSpot's growth. But a company asking to be valued on 43% growth needs its forward numbers to show it, and this quarter they did not.

Are Figma's customers cutting seats?

Not on the company's figures. About two-thirds of customers with more than $10,000 in annual recurring revenue "added full seats compared to their prior renewal", gross retention "remained in the mid-to-high 90s", and those customers grew 28% in number to 16,566, Figma said. The fear that AI agents shrink software seats, part of the argument behind October's software selloff, has not shown up in Figma's base.

What would make the market treat Figma as an AI winner?

Figma gives usage, not AI revenue. By October 31, 65% of its $10,000-plus customers used the Figma agent weekly and over 80% consumed AI credits, and it named three customers that raised their AI credit commitments to between more than six and nearly eight times their first purchase. Until it discloses what credits bring in, investors are left to judge AI by the costs they can see and the revenue they cannot.

The next report will test it. Figma paid for the agent's inference itself until its October 6 launch, and most of its products now consume paid credits, so gross margin should show whether that cost now carries revenue. If billings and remaining performance obligations grow with revenue again, the case for a multiple closer to Datadog's gets much stronger. If they keep lagging, 6x is the market reading the contracts correctly, whatever happens to the margin. Our count of October's software stocks has the rest of the sector's moves.