The print is not the hard part. Meta put up $64.601 billion of revenue in Q3 2026, up 26% year over year on both a reported and constant-currency basis. Family of Apps advertising alone was $63.061 billion. Daily active people hit 3.64 billion. Impressions rose 10%; average price per ad rose 14%. That is still an advertising company that can raise both volume and price in the same quarter.
What the market is being asked to underwrite is everything stacked around that machine: a $9.89 billion charge that cut operating margin from 40% a year ago to 20%; free cash flow that flipped to −$2.12 billion on $34.91 billion of capex including finance-lease principal; Muse — Meta's personal agent — as a compute-heavy product that is not yet global; a new Class C non-voting stock framed as a capital-raising tool with no specific financing plan attached; and a legal calendar that Li explicitly said still includes youth-related trials in the United States in the coming months.
1. The legal charge is the margin story — and it is not the last word
GAAP operating income was $12.988 billion, down 37% year over year. The press release is blunt about why: the August 2026 agreement with 52 state attorneys general over teen mental health and social media provides for payments of up to approximately $18.59 billion, and Q3 includes a $9.89 billion charge related to that agreement.
Strip the charge and the quarter looks different. Ex-charge operating income would have been $22.88 billion; diluted EPS $3.83 would have been $7.40; the 24% effective tax rate would have been 17%. Li told the call that ex-legal, operating income would have risen 11% year over year. Full-year 2026 expenses were raised to $175–179 billion specifically to fold in the charge. Excluding it, management still expects 2026 operating income above 2025.
That is useful for reading the core P&L. It is not a get-out-of-jail card for the cash and legal path. The settlement's total payments can run to ~$18.59 billion. Li also warned that youth-related scrutiny continues in several markets and that "a number of youth-related trials" are scheduled in the U.S. in the coming months. The $9.89 billion is a large, visible line item. The overhang is the remaining trials and any further settlements or judgments that have not yet been priced.
2. Ads still do the work; AI is already in the monetization stack
The engagement and monetization color on the call is specific enough to take seriously.
Instagram time spent per user hit an all-time high and grew double digits. Facebook Reels time spent hit all-time highs globally and in the U.S. Threads overtook X in monthly actives; time spent per user grew over 20% year over year. Partnership Ads crossed a $15 billion annual revenue run-rate (+97% YoY). Nearly 10 million advertisers used at least one gen-AI creative tool. Meta One passed 15 million subscriptions and trials.
On the ads-tech side, Li walked through conversion lifts that are small in isolation and material in aggregate: GEM expansion driving a 9.6% increase in app ad conversions and 1.7% for offsite conversion campaigns; Andromeda improvements adding 1.4% conversions on Facebook; generative retrieval on Facebook Feed nearly 2% conversions in targeted segments. Muse-family models are cutting per-unit content-understanding costs by more than 80% across Instagram, Threads, Facebook, and ads in offline evaluations.
None of that requires the personal-superintelligence pitch. It is the boring AI story: ranking, retrieval, creative tools, partnership inventory — the stuff that shows up in ad price and conversion. That funded the quarter. FoA other revenue (WhatsApp paid messaging and subscriptions) was $1.097 billion, up 59%. Reality Labs was $443 million, down 6% on lower Quest sales offset by AI glasses. RL remains a drag: FY26 losses guided to $18–19 billion, narrowing to around $15 billion in 2027 as spend is reallocated "primarily" to AI.
3. Muse: product-market fit claims vs unit economics the market must underwrite
Zuckerberg's Muse section is the part of the call that sounds like a new company bolted onto the old one.
More than 10 million people in the U.S. and Canada set up Muse in roughly six weeks. About 25% of setups use it on a given day — more than twice, he said, Threads' starting DAU rate or what Meta believes other AI apps started at. Shopping is the top use case: more than a quarter of Muse users, average purchase over $200. App Store #1 almost daily. Muse Wonder sits on the paid tier; free defaults to Muse Spark. Business model: free for most with a commerce take-rate paid by businesses; paid tokens/subscriptions for advanced models; business/enterprise subscriptions; optional hooks into Meta content and ads. Enterprise stack via CJ Desai: Meta Enterprise Platform, API, Muse Code.
Then the constraint that matters for the stock. Muse is not global yet. Zuckerberg said the biggest constraint so far is compute intensity: "we have not launched it globally yet," partly for localization and connectors, "but it's partially that we want to make sure that we do this efficiently." Target: improve compute cost efficiency by more than 10x in the next year; build capacity to serve more than a billion people over the next few years; profitable once revenue lines are at scale if efficiency lands. Li echoed the same 10x framing and said total serving costs will rise as the user base expands.
So the underwrite is not "does anyone use Muse?" Early usage looks real. The underwrite is whether commerce take-rates and paid tiers scale fast enough to cover serving costs before (or as) Meta burns through the efficiency roadmap — and whether agentic commerce friction (merchants protecting customer relationships, permissioned agent-to-agent protocols) slows the shopping wedge that currently looks strongest. Li previewed a Personal Agent Protocol with partners including Sierra, Walmart, Stripe, and Shopify; she also said near-term many experiences will remain AI-assisted rather than fully autonomous. That is a careful CFO way of saying monetization at scale is not a 2026 story.
4. Capex, FCF, and Class C: the funding question without a financing ask
Q3 capex including finance-lease principal: $34.91 billion. Free cash flow: −$2.12 billion. Cash and marketable securities: $82.45 billion. Headcount: 66,775, down 15% year over year (Li also noted down 12% from Q2 after the May 2026 reduction). FY26 capex guided to $135–145 billion, narrowed from $130–145 billion — the floor was raised, not the ceiling cut. Q4 revenue: $73–76 billion with about a 1% FX headwind.
On 2027, Li refused a budget number. "We are quite early in developing our budget and capacity plans for next year." Expect aggressive investment behind the core business and AI. "Strong overall revenue growth throughout 2027" — not a full-year figure, and she later clarified that new AI businesses may begin contributing incremental revenue in 2027 but the near-term focus is product-market fit. No 2027 expense or capex guide until the Q4 call.
Into that vacuum lands Class C. Effective around January 19, 2027: a new publicly listed non-voting Class C common stock, one-for-one stock dividend to Class A and Class B holders, economic effect of a two-for-one split, voting power unchanged. Founder Alignment Agreement with Zuckerberg. Special Committee language: tool for capital raising, employee compensation, acquisitions, and other corporate purposes while preserving the founder-led voting structure. Li on the call: "We don't have any specific financing plans to announce at this time," but they thought it "prudent to establish the appropriate capital structure framework" for a diverse pool of funding sources.
Read that without the IR gloss. Meta just printed negative quarterly FCF on AI and infra spend, raised the capex floor, declined to quantify 2027 spend, and created a non-voting share class that can be used to raise capital without diluting Zuckerberg's control. That may be prudent governance for a multi-year AI build. It is also a signal that equity (or equity-like) funding sits on the menu even while the ads engine is growing mid-20s. Markets that treat Class C as a free split miss the capital-structure optionality.
5. Valuation: cheap vs Meta's own history — if you trust the earnings path
As of late October 2026, the multiple is not expensive relative to Meta's own past, with the usual caveat that prices move and PE sources disagree slightly.
- Yahoo Finance (as of ~10/23/2026 on the quote page; price context roughly $634–643 around 10/23–10/26 — prices move): trailing P/E ~21.92, forward P/E ~18.94.
- GuruFocus: forward PE 19.00 as of Oct 26, 2026.
- FullRatio: trailing PE 21.65 as of Oct 23, 2026 ($642.62 / TTM EPS $29.68); 5-year avg ~23.23; 10-year avg ~25.63.
Trailing sits a bit below Meta's own 5- and 10-year averages on FullRatio; forward sits near 19 on Yahoo and GuruFocus. A FinanceCharts snapshot as of Oct 16 (pre-print) showed trailing ~26.71 / forward ~19.72 — note the date mismatch; prefer Yahoo + FullRatio for current and FullRatio for history. No percentiles invented here.
The valuation argument in either direction is not the multiple. It is whether 2027 earnings survive (a) Muse serving costs before the 10x efficiency claim lands, (b) legal cash that may exceed the Q3 charge, and (c) whatever capex and opex path Li eventually guides. A ~19x forward multiple on an ads franchise growing mid-20s is not demanding if that path holds. It is demanding if Muse becomes a multi-year margin sink and Class C is used for large equity raises while RL still loses ~$15 billion.
What would change this view
More constructive: Muse unit economics disclosed with take-rates or contribution margins that show a path to profitability before global scale; 2027 guide that keeps operating income growing while AI spend rises; youth trials that settle without material new charges beyond what is already reserved; Class C used primarily for compensation/M&A at modest scale rather than a large primary raise.
More negative: Serving-cost commentary that walks back the 10x efficiency target or delays global Muse indefinitely; another large legal charge or adverse youth-trial outcome; 2027 budget that implies sustained negative FCF without a clear ads offset; a Class C primary offering sized large enough to reframe the capital structure as equity-funded AI build rather than FCF-funded.
If a note cannot say what would make it wrong, it does not get published.
Sources
- Meta Reports Third Quarter 2026 Results (Oct 21, 2026)
- Q3 2026 Earnings Call event page
- Official earnings call transcript (PDF); Motley Fool / IR transcript text used for quote checks
- Follow-up Q&A call transcript (same IR folder / Q4cdn series), optional color
- Valuation: Yahoo Finance META key statistics, GuruFocus forward PE, FullRatio META PE — as-of dates as cited above
Not investment advice. Community research viewpoint only. No position sizes.