Meta Q3 2026 Results: A $9.89 Billion Legal Charge Meets Negative Free Cash Flow and New Class C Shares

Data cutoff: October 21, 2026, 8:00 p.m. ET. Latest completed U.S. equity session: Wednesday, October 21, 2026. Results released after that session.

One quarter, three different financial stories

Meta's October 21 results present a familiar investment challenge in an unusually sharp form: an advertising platform can be growing strongly, recognizing an exceptional legal cost and expanding physical AI infrastructure all at once. Looking at only one profit or cash-flow number would leave readers with an incomplete answer about the business's trajectory. The quarter brought 26% revenue growth, a multi-billion-dollar legal charge, negative reported free cash flow and a proposal for a new nonvoting class of stock. Each has a different implication for current earnings and the financing of future growth.

The company reported third-quarter revenue of $64.601 billion, up from $51.242 billion a year earlier. GAAP operating income fell to $12.988 billion from $20.535 billion, taking the operating margin from 40% to 20%. Net income nonetheless increased to $9.860 billion from $2.709 billion, with diluted EPS rising to $3.83 from $1.05. The year-earlier tax rate was unusually high; the 264% net-income growth therefore should not be presented as a matching improvement in operating profit. Comparability matters as much as magnitude.

The advertising engine remains productive

Meta reported average daily active people of 3.64 billion across its Family of Apps during September, a 3% year-over-year increase. Ad impressions rose 10% and average price per advertisement rose 14%. Those metrics support the proposition that monetization gains are coming from more than audience expansion. They do not identify what percentage of the price improvement resulted from AI recommendations, auction competition or changes in advertising mix, so attributing the entire gain to any one product would overstate the disclosure.

Advertising economics matter because they fund much of the group's infrastructure ambition. A business with billions of users can improve sales without comparable user growth if engagement, pricing and conversion efficiency improve. But average price per ad is an aggregate figure. It can rise because advertisers bid more, because higher-priced geography or formats take a larger share, or because improved targeting makes impressions more valuable. The company needs continued advertiser return on spending to defend that pricing when the economic cycle weakens.

Meta's newer Muse products and AI initiatives attract attention, but the release did not isolate a clean, comparable revenue or profit line for each product. Investors should be careful not to invent one. The most reliable currently visible commercial evidence sits in the long-established advertising system, while newer consumer-agent business models remain an emerging layer of the valuation debate.

The legal expense must be understood precisely

Meta disclosed that its agreement with 52 U.S. state attorneys general, announced in August 2026 and concerning teen mental health and social media, provided for payments of up to approximately $18.59 billion. In Q3, the company recognized a $9.89 billion charge in costs and expenses related to the agreement. The charge is not the same as saying the full amount was paid in cash during the quarter, and the maximum agreement amount is not a newly announced $18.59 billion Q3 expense.

The company provided an illustrative reconciliation: excluding the specific Q3 charge, operating income would have been $22.88 billion instead of $12.99 billion and diluted EPS would have been $7.40 rather than $3.83. Those are management's charge-excluded figures, not a substitute GAAP result. They are useful in assessing the advertising and operating-cost trend, but investors cannot simply erase the underlying obligations from enterprise value. Litigation costs represent a real use of resources even if recognized in one accounting period while paid in another.

More regulatory matters remain active. A historical settlement may reduce one category of uncertainty without removing ongoing privacy, youth-safety or antitrust risks. In modeling Meta, the relevant questions are the expected timing of cash payments, the possibility of additional charges, and the product or operational changes required by legal agreements. Treating every legal charge as either entirely recurring or entirely irrelevant would misstate the situation.

Negative free cash flow is not an accounting error

Capital expenditures, including principal payments on finance leases under Meta's disclosed convention, reached $34.91 billion in the quarter. Operating cash flow was $32.79 billion, leaving non-GAAP free cash flow of negative $2.12 billion on the company's definition. This is the clearest expression of the capital intensity of the AI strategy. Advertising generated enormous operating cash, yet the investment program consumed more than that cash over the three-month period.

For 2026, Meta narrowed its capital-expenditure outlook to $135 billion–$145 billion from $130 billion–$145 billion. Narrowing the low end is not equivalent to an entirely new $145 billion commitment, but it confirms that the company expects an extraordinary spending year. Cash, equivalents and marketable securities stood at $82.45 billion at September 30. Liquidity provides flexibility, but a large cash balance should not be used to dismiss returns on invested capital. Equipment, power and facilities have finite useful lives, and depreciation eventually moves through earnings even if payment occurred earlier.

The crucial comparison is not simply AI spending against the current quarter's revenue. Infrastructure built now may support several years of services, advertising optimization and products. The right test is incremental economic profit over the life of those assets, after operating costs, power, maintenance, depreciation and replacement. If AI improves advertising yields sufficiently, the return can be attractive; if model and compute pricing decline while utilization disappoints, the same capacity can become a burden.

New Class C shares introduce a governance variable

Alongside earnings, Meta announced board approval for a new publicly listed, nonvoting Class C stock, with an intended corporate charter effective date of January 19, 2027. The company envisages a one-time dividend of one Class C share for each outstanding Class A and Class B share, contingent on the amendment becoming effective. The record and payment dates had not been determined at the cutoff. Meta said the economic effect on the Class A price should resemble a two-for-one stock split, while voting power would not change.

This announcement is not a cash dividend and should not be treated as an immediate doubling of shareholder wealth. Economic ownership must be assessed across the securities a holder receives. Meta says the structure is intended to assist capital raising, compensation and possible acquisitions while preserving founder-led control. That can improve financial flexibility, but it also intensifies questions about voting rights, future share issuance and how management accountability works when economic exposure and voting influence differ.

Market debate and what changes next

Meta guided fourth-quarter revenue to $73 billion–$76 billion and expects 2026 total expenses of $175 billion–$179 billion, now including the Q3 legal charge. It stated that, excluding that charge, 2026 operating income should exceed 2025's. Reality Labs losses are expected at $18 billion–$19 billion for 2026. These are management forecasts; they are neither a completed Q4 nor evidence that Reality Labs is already profitable.

The bullish view emphasizes 26% sales growth, rising ad prices and a distribution network that can use AI internally and sell new AI products. The bearish view emphasizes accelerating capital requirements, recurring compliance burdens and governance constraints from nonvoting shares. An after-hours price reaction is an observation about investor expectations, not proof of either interpretation. Thursday, October 22's regular trading will offer a more complete response to the earnings call and the new stock proposal.

Beyond that session, watch Meta's formal information statement, subsequent disclosure of Class C record and payment dates, advertiser pricing and engagement, future capital expenditure and the pace at which operating cash recovers relative to investment. Without a confirmed date, a future earnings release should remain undated rather than assigned a fabricated calendar entry. The defining test is whether Meta can sustain advertising momentum while lifting the cash return on its much larger AI asset base.

Conclusion

Meta's results demonstrate a productive advertising business alongside three substantial shareholder questions. A $9.89 billion legal charge reduced reported operating profit, a $34.91 billion investment quarter pushed free cash flow below zero, and nonvoting Class C shares could alter financing and governance. None of these developments nullifies 26% revenue growth. Together, they make future cash generation per existing economic share the most useful measure of whether Meta's AI expansion is creating durable value.