Figma Q2 2026 Earnings: $148M Stock Comp, $112M GAAP Loss

Figma grew revenue 48% to $370.1 million in the quarter ended June 30, 2026 — its third straight quarter of acceleration and, in the CFO's words, its "first full quarter of AI credit monetization." It also reported a GAAP net loss of $112.2 million and, in the same release, non-GAAP net income of $42.6 million. Nearly all of the $154.7 million gap between those two bottom lines is one expense: $147.6 million of stock-based compensation, spread across every functional line of the income statement. A year earlier that line was $7.3 million. The company did not suddenly start paying its engineers in stock; the IPO changed when the accounting is allowed to recognize it. This post traces the gap line by line in a public Beancount ledger — three audited years plus the quarter — and follows the subscription cash that arrives as deferred revenue before it becomes revenue.

The Headline Numbers​

Figma reports on a calendar year. Q2 2026 is the three months ended June 30, 2026. GAAP figures below come from the August 5, 2026 earnings release (Form 8-K Exhibit 99.1) and the Form 10-Q for the same quarter; the non-GAAP rows are the company's own reconciliation in that release.

Read the table in two halves. The top half is a fast-growing software company whose gross profit grew slower than revenue (40% vs 48%) and whose GAAP operating line swung from a $2.1 million profit to a $117.3 million loss. The bottom half says why: stock compensation went from a rounding error to 40% of revenue. Add it back and GAAP operating income before stock comp rose from $9.4 million to $30.3 million — the underlying operating business improved while the reported one collapsed.

Free cash flow fell 12% even as non-GAAP net income more than doubled, and that divergence has a cause too: Q2 carried Config, the annual user conference, and the six-month cash flow includes a $56.1 million 2025 bonus payment. Operating cash flow was $60.9 million, a 16% margin.

Revenue Deep Dive​

Figma does not report segments. It reports revenue by the billing address of the customer, and the quarter's growth was broad:

International customers are now more than half of revenue and grew faster than the US. The release does not describe international growth as ahead of plan, so neither do we — but a 54% share makes Figma less of a US-cycle story than most US software companies.

The customer metrics carry the thesis. Figma had 15,964 paid customers with more than $10,000 of annual recurring revenue (+34%) and 1,635 above $100,000 (+46%). Net Dollar Retention was 136% — the average large customer's spend rose by more than a third in a year. Management's own language, verbatim from the release:

• Robust demand / record quarter: CFO Praveer Melwani — "Q2 was a record quarter and our first full quarter of AI credit monetization."
• Expansion from existing customers: "Net Dollar Retention Rate remained strong at 136% as customers expanded both seats and AI credit add-ons."
• Raised outlook: "Figma raises full year revenue guidance, reflecting sustained seat expansion and AI adoption." Full-year 2026 revenue is now guided to $1.463–$1.467 billion (+39% at the midpoint), a $40 million raise; Q3 to $373–$375 million.
• Product ramp: as of July 31, 2026, "over 50% of Paid Customers with more than $10,000 in ARR were already using the Figma agent on a weekly basis," and over 80% of those customers were consuming AI credits weekly in the quarter.

Tie that to the ledger. Revenue is one credit on Income:Revenue (−370.083 MUSD in Beancount's sign convention). AI credits are not a separate account and should not be: the 10-Q does not disaggregate them, and they bill as add-ons to the same subscriptions. What the filings do show is where AI shows up on the cost side, which is the next section.

Two of the seven themes we scan every release for are conspicuously missing. There is no "demand exceeds supply" and no language about sustained price increases — even though AI credit limits, enforced since March 2026, are in effect a price on heavier usage. The 10-Q is franker than the release: after enforcement began, Figma "observed elevated customer support volume, instances of customer dissatisfaction expressed through public and social channels, and reduced usage by certain customers."

The Margin Story​

Two different things are moving here, and the ledger keeps them apart.

Gross margin is the AI story. Cost of revenue rose 117% to $60.5 million against 48% revenue growth. The 10-Q attributes $27.1 million of the $32.6 million increase to "technical infrastructure and hosting costs relating to AI and increased usage of our platform by paid users." Gross margin fell five points year over year, from 88.8% to 83.7%. It recovered four points from Q1's 79.4%, in the same quarter that Figma first charged for AI credits for a full three months — consistent with credit pricing starting to cover inference cost, though the 10-Q does not break AI revenue out to prove it. Stock comp barely touches this line: $2.0 million, 3% of cost of revenue.

Operating margin is the stock-compensation story. R&D more than doubled to $167.3 million, of which $61.8 million (37%) is stock comp. G&A rose 169% to $104.7 million, of which $64.9 million (62%) is stock comp — nearly all of the $65.8 million increase. Sales and marketing, $154.9 million, carries $18.8 million of stock comp plus the Config spend and $13.5 million more of hosting for free users, which Figma books as a marketing cost.

Non-GAAP operating margin fell from Q1's 15.6% to 9.8% as Config landed; guidance for the full year is a 9% non-GAAP operating margin at the midpoint. That is the number to watch: it excludes stock comp, so it isolates whether AI costs and growth spending are outrunning revenue.

The One Big Question: How Much of the Loss Is Real?​

The release reconciles the two bottom lines itself. Every row below is the company's:

Stock comp is 93% of the positive adjustments. The rest is small: bitcoin and a bitcoin ETF that Figma holds on its balance sheet lost value, an acquisition's intangibles amortized, and the employer's share of payroll tax on vesting shares — which, unlike the stock comp itself, is paid in cash.

Where the stock comp sits, by line, against the same quarter a year earlier:

Why was the year-ago number so small? Most pre-IPO Figma RSUs had two vesting conditions: time served and a liquidity event. Until the July 2025 IPO satisfied the second, GAAP recognized almost nothing. The IPO released it all at once — the FY2025 10-K records a one-time cumulative $975.7 million charge for RSUs whose service condition had already been met. FY2024 had its own version: an $801.2 million charge when Figma removed the performance condition on vested RSUs so employees could sell in the 2024 tender offer. Both are modeled as filed, inside the functional lines, not pulled into a special account.

So the honest reading of the loss is neither "Figma is losing $112 million a quarter" nor "the real profit is $43 million." Stock comp is a real cost — it is paid in shares instead of cash, and shareholders bear it as dilution. The cash side is visible too: Figma paid $45.5 million in Q2 ($161.6 million in the first half) to cover employees' taxes on net-settled RSUs, a financing outflow that converts part of the stock comp into cash after all. What the ledger shows is that the step change from Q2 2025 to Q2 2026 is overwhelmingly a recognition-timing change. Ex-SBC, the operating business earned more this year than last.

Figma's history adds one irony. FY2023 is the only profitable year in the ledger, with $737.8 million of net income — because Adobe paid a $1.0 billion termination fee when the two companies abandoned their merger in December 2023. The ledger books that fee as its own labeled Income:OtherNet posting, so FY2023's operating loss of $73.5 million stays visible underneath it. The would-be acquirer's own books are in our Adobe ledger.

Deferred Revenue: Billings Before Revenue​

Figma sells subscriptions with monthly or annual terms and, per the 10-K, "typically invoice[s] our customers in advance." Cash arrives, lands on the balance sheet as deferred revenue — a current liability, as filed — and is recognized as revenue over the subscription term. The ledger asserts that liability at every period end:

Deferred revenue equals about five months of current revenue ($626.8 million against a $1.48 billion annualized quarterly run rate), and the 10-Q says 72% of Q2's revenue came out of the March 31 balance. Remaining performance obligations — contracted but not yet recognized, whether billed or not — were $687.0 million, up from $647.9 million at year end.

Billings are not a GAAP line, but they can be calculated from the cash flow statement: revenue plus the change in deferred revenue. Our calculation, not a company metric:

Calculated billings grew slower than revenue in Q2 — 34% against 48% — because deferred revenue slipped $0.9 million in the quarter after rising $26.5 million in Q2 2025. Quarterly billings for a company that moved to new pricing, packaging and billing models in March 2025 are noisy, and monthly AI credit add-ons and pay-as-you-go usage bill on a far shorter cycle than an annual seat, which structurally shortens the deferred balance. But this is the line to track: if AI credits keep growing as a share of revenue, recognized revenue will increasingly arrive without first sitting in deferred revenue — good for how fast revenue appears, less good for visibility into the next quarter. Our CrowdStrike post walks through the same bridge for a subscription business with a much larger deferred book.

Tracking a $2.4B company in plain text​

Modeling Figma in Beancount forces every thousand dollars to reconcile: assets equal liabilities plus equity at each period end, and each income statement sums to zero, with net income offset into Equity:Adjustments and retained earnings set by the balance assertion. That is how we model every company. Figma files in thousands, so the ledger keeps three decimals of its millions unit.

The Q2 income-statement transaction from the ledger (Beancount signs: income credits negative, expenses positive; the posting comments carry each line's stock comp from the filing's footnote):

; FY2026Q2 Income Statement — FY2026 Q2 (three months ended June 30, 2026; unaudited)
; Net income (loss): -112.152 MUSD.
; Check: -370.083 + 60.472 + 167.329 + 154.856 + 104.715 + -7.614 + 2.477 + -112.152 = 0
2026-06-30 * "Figma, Inc." "FY2026Q2 Income Statement"
  Income:Revenue                                      -370.083 MUSD  ; revenue, +48% YoY (Q2 2025: 249,640)
  Expenses:CostOfRevenue                                60.472 MUSD  ; cost of revenue; includes SBC 2,036
  Expenses:ResearchAndDevelopment                      167.329 MUSD  ; research and development; includes SBC 61,777
  Expenses:SellingGeneralAdministrative                154.856 MUSD  ; sales and marketing; includes SBC 18,815
  Expenses:SellingGeneralAdministrative                104.715 MUSD  ; general and administrative; includes SBC 64,926
  Income:OtherNet                                       -7.614 MUSD  ; other income, net
  Expenses:IncomeTax                                     2.477 MUSD  ; provision for income taxes
  Equity:Adjustments                                  -112.152 MUSD  ; net income (loss) offset; retained earnings set by balance assertion

The non-GAAP bridge is deliberately not in the transaction. It lives as a comment above it in the ledger file, because a ledger records what happened, and "net income excluding stock comp" is an opinion about what should count.

The balance-sheet lines that carry this post's two stories — the subscription liability, and the equity account where stock comp accumulates:

; @m49-fragment balance-only
2025-12-31 balance Liabilities:Current:DeferredRevenue              -595.334 ~ 0 MUSD
2026-06-30 balance Liabilities:Current:DeferredRevenue              -626.783 ~ 0 MUSD
2025-12-31 balance Equity:CommonStockAndAPIC                       -2950.011 ~ 0 MUSD
2026-06-30 balance Equity:CommonStockAndAPIC                       -3178.407 ~ 0 MUSD

Paid-in capital rose $228.4 million in six months while total equity fell $32.7 million. That is stock comp's double entry in miniature: the expense drives the accumulated deficit down, the same amount comes back up through paid-in capital, and the dilution sits in the share count rather than in cash. Since the end of 2023, paid-in capital (including the preferred stock that converted at the IPO) has grown from $502.8 million to $3.18 billion, while the $540.1 million of retained earnings the Adobe fee left behind turned into a $1.70 billion accumulated deficit.

The Multi-Year Arc​

Figma listed in July 2025. Its IPO prospectus carries audited statements for FY2023 and FY2024 and the FY2025 10-K the third year, so the ledger starts at FY2023 — no earlier year is modeled from unaudited figures.

Revenue grew 48% in FY2024 and 41% in FY2025; the quarter's 48% is a re-acceleration on a larger base. Net income swings wildly — +$738 million, −$732 million, −$1.25 billion — but none of those swings comes from the core business. FY2023 is the Adobe fee. FY2024 is the tender-offer RSU release. FY2025 is the IPO catch-up charge. On the company's own non-GAAP basis, operating income was $27.1 million, $127.2 million and $129.5 million over the three years. Every period in the ledger ties to the filing's total assets, total liabilities and net income. Figma carries no debt: the debt accounts are open and empty.

The Verdict: Bull vs. Bear​

Bull Case

• Revenue growth has accelerated for three straight quarters to 48%, and full-year guidance went up $40 million to $1.463–$1.467 billion.
• 136% Net Dollar Retention, driven by "both seats and AI credit add-ons," means existing customers fund most of the growth.
• Gross margin recovered from 79.4% in Q1 to 83.7% in Q2 — the first evidence that AI credit pricing can pay for AI inference.
• Before stock comp, GAAP operating income more than tripled year over year ($9.4 million → $30.3 million), and the company holds $1.67 billion of cash and marketable securities with no debt.
• The biggest stock-comp shocks ($801 million in FY2024, $976 million at the IPO) were one-time catch-ups; the run-rate is now visible and should fall as a share of revenue.

Bear Case

• Stock comp is still 40% of revenue, and Figma spent $161.6 million of cash in the first half on employee taxes for net-settled RSUs — the "non-cash" charge has a cash tail.
• Gross margin is still five points below a year ago; the release's confident language about AI adoption is not yet matched by a return to 88–89% margins.
• The 10-Q reports "reduced usage by certain customers" and "elevated customer support volume" after AI credit limits were enforced. The release's "robust demand" narrative does not quantify that churn risk.
• Calculated billings grew 34% in Q2 against 48% revenue growth, and deferred revenue fell slightly in the quarter — the forward-looking liability is growing slower than the income statement.
• Non-GAAP operating margin is guided to just 9% for 2026, below FY2025's 12.3%: growth is being bought with AI hosting and marketing spend.

Our Take

Figma's Q2 is a better quarter than its GAAP loss and a worse one than its non-GAAP profit. The $112 million loss is mostly an accounting echo of how pre-IPO RSUs vest, and the ledger shows the operating business improving underneath it. But the $43 million profit asks you to ignore a stock-comp bill equal to 40% of revenue, plus the cash Figma spends covering its employees' taxes on it. The two numbers that decide the next year are both on the ledger: gross margin, which tells you whether AI credits pay for their own inference, and deferred revenue, which tells you whether that new revenue is being contracted in advance or only billed as it is used. Watch those, not the adjusted bottom line.