When AI revenue rises but margins become harder to explain
Usage of AI features has grown, and additional charges have begun. But when the finance team asks, “So how much did AI contribute to revenue and costs, respectively?” the answer gets blurry. We know the overall revenue growth rate, but AI revenue is not separately visible, and hosting costs combine general usage with AI usage.
Figma’s second-quarter 2026 results show exactly this situation. Revenue rose 48% year over year, and it was the first full quarter to reflect AI credit monetization, yet the stock fell 16.52% in after-hours trading immediately after the announcement.
The takeaway is not “AI drove 48% growth, but costs also exploded.” The range of what can be confirmed from public figures is narrower. We can confirm that AI usage was converted into incremental charges, but revenue contribution and pure AI costs have not yet been separated.
What Figma proved is not ‘AI revenue’ but ‘conversion to paid usage’
Figma’s second-quarter revenue was $370.1 million, up 48% from $249.64 million a year earlier. The company described this as the first quarter to fully reflect AI credit monetization.
The usage signal is also clear. As of June 30, 2026, Figma had 15,964 customers with annual recurring revenue (ARR) of at least $10,000, up 34% year over year. More than 80% of this customer group consumed AI credits weekly. However, that 80% figure is based on the week during the quarter with the highest number of participating customers. It is neither an average across all weeks nor the usage rate among all paying customers, and the share of credit consumption that led to additional purchases was not disclosed.
What the results confirm: Figma charged for AI credits, and large customers actually consumed them.
What remains unknown: AI credit revenue, its share of total revenue, and the rate of additional purchases by customer.
Therefore, you should not attribute all 48% growth to AI. What these results establish is closer to “Figma has begun monetizing AI usage separately” than “AI drove Figma’s growth.”
There was cost pressure, but most of the R&D increase was labor expense
Cost signals closest to AI appear in cost of revenue before total R&D. Cost of revenue rose 117%, from $27.889 million a year earlier to $60.472 million, while GAAP gross margin declined from 89% to 84%.
According to the 10-Q, $27.10 million of the $32.583 million increase in cost of revenue came from technology infrastructure and hosting costs associated with growth in paid users and AI. But this figure includes both ordinary user growth and AI costs. It cannot be isolated and calculated as pure inference cost. You should also note that, compared with the prior quarter, GAAP gross margin actually recovered from 79% to 84%.
R&D expense rose 101%, from $83.052 million to $167.329 million. Yet $70.80 million of the $84.277 million increase was labor expense, including headcount growth, stock-based compensation, and related payroll taxes. AI-related technology infrastructure and hosting increased by $10.10 million, and the filing described this only as ‘primarily AI-related’; it did not classify the entire amount as AI cost.
| Observed figure | Interpretation you can verify | Interpretation to avoid |
|---|---|---|
| Revenue +48% | Company-wide revenue growth | AI generated all 48% growth |
| Cost of revenue +117% | Higher usage and AI-related infrastructure burden | The entire increase is AI inference cost |
| R&D +101% | Stock-based compensation and labor costs were a large share after the IPO | AI costs rose 101% |
The stock decline cannot be explained by AI costs alone either
At a particular point in after-hours trading following the results announcement, Figma’s stock was down 16.52%. That is different from the closing-return figure for regular trading, and price movement alone cannot establish a single reason for investors’ judgment.
There was real pressure on profitability. Non-GAAP operating margin fell from 16% in the first quarter to 10% in the second quarter. However, Figma said that sales and marketing investment related to its annual Config event affected second-quarter operating income. This non-GAAP change may reflect several items together, including infrastructure and AI costs for free users, labor expense excluding stock-based compensation, and Config-related sales and marketing expense.
On a GAAP basis, Figma reported an operating loss of $117.289 million, including $147.554 million in stock-based compensation expense. By contrast, same-quarter non-GAAP operating income excluding stock-based compensation and other items was $36.10 million, with a 10% operating margin. Rather than choosing one measure or mixing the two, show the differences in adjustments together so the nature of costs is not misunderstood.
Annual revenue guidance was raised by $40 million to $1.463 billion–$1.467 billion, but the midpoint of annual non-GAAP operating-margin guidance remained at 9%. In other words, higher growth guidance did not automatically mean better margin guidance.
An evidence sheet for reading AI monetization in your own product
The first action is not to calculate AI ROI with a single number, but to record usage, revenue, cost of revenue, operating expenses, and market reaction in separate fields. Using disclosures that break down the components of cost increases, such as Figma’s second-quarter 2026 10-Q, organize the information in the following order.
- Align the comparison periods.
Enter revenue, cost of revenue, gross margin, R&D, and sales and marketing expense for the current quarter and the year-ago quarter. Put sequential changes in a separate column so seasonality and year-over-year growth do not get mixed together. - Distinguish AI monetization from AI revenue.
Record the start of charging and credit usage rates as confirmed figures, while leaving separately undisclosed AI revenue and its contribution to total revenue as ‘not disclosed.’ - Preserve the stated scope of combined costs.
If a filing describes AI and general-usage hosting costs together, do not split them using an arbitrary ratio. Label them instead as ‘AI- and usage-related infrastructure.’ - Separate the nature of operating expenses.
Record labor expense, stock-based compensation, AI infrastructure, and event costs separately. Do not mix GAAP losses and non-GAAP profits in the same field. - Qualify stock-price reactions.
State whether the movement occurred during regular or after-hours trading and at what point in time. Do not settle on causes involving costs, guidance, or valuation when public information cannot separate them.
Success criterion: Instead of the one-line claim “AI did well,” you should be able to see both evidence and blank spaces: ‘credit charging began,’ ‘usage rate confirmed,’ ‘AI revenue not disclosed,’ ‘AI and general-usage costs combined,’ and ‘cause of stock decline not established.’ The core of this analysis is not filling those blanks with estimates.
Next quarter, watch two blanks before the growth rate
In Figma’s next results, first check whether AI credit revenue is disclosed separately and how much AI costs are separated within cost of revenue. Right now, we can see AI use and the start of charging, but there is not enough information to calculate unit economics.
The Figma case shows both the positive signal that AI can be monetized and the limitation that its profitability cannot yet be determined. Rather than connecting revenue growth, R&D growth, and stock-price decline in one line, separate what each figure actually measures to assess AI investment in your own product more accurately.
If you want to dig deeper
Figma Announces Second Quarter 2026 Financial Results This official release covers AI credit monetization, customer usage rates, quarterly results, and annual guidance. investor.figma.com
Figma, Inc. Quarterly Report on Form 10-Q The original filing lets you break down increases in cost of revenue and R&D into labor, stock-based compensation, and infrastructure costs. sec.gov
Figma Announces First Quarter 2026 Financial Results Official material needed to compare second-quarter margins with the prior quarter on the same basis. investor.figma.com
