Meta Q2 2026: Ad Machine Strong, Capex Spooks the Street

Meta’s Q2 2026 earnings landed on July 29 with a revenue beat, an EPS miss, and an after-hours selloff of roughly 8-10% — and almost every headline collapsed two unrelated stories into one. The ad business did not crack; it accelerated. What cracked was free cash flow, and the two have almost nothing to do with each other.

Here are the two stories, separated. Story one: advertising revenue grew 27% year over year to $59.4 billion, on impressions up 14% and average price per ad up 12% — both versus Q2 2025. Story two: free cash flow fell about 91% year over year to $784 million, because $31.1 billion of quarterly capital expenditure consumed roughly 98% of the quarter’s $31.86 billion in operating cash flow. One is a demand story. The other is a financing story.

This breakdown reconciles both with Meta’s own numbers — what was reported, which percentages describe which denominators, where the $2.4 billion legal charge sits, how the FY26 capex guide moved, and how the print compares with Alphabet’s and Microsoft’s same-week reports. Everything below is sourced from Meta’s press release and earnings-call transcript, both dated July 29, 2026, plus attributed press coverage.

The headline numbers, from Meta’s official press release and the earnings-call transcript: total revenue of $60.8 billion, up 28% year over year (27% on a constant-currency basis), against an analyst consensus in the roughly $60.2-60.3 billion range per CNBC’s LSEG and StreetAccount citations. Diluted EPS came in at $6.18, down 13% from $7.14 a year earlier and roughly 14-15% below the ~$7.22 consensus estimate CNBC cited. Net income fell 14% year over year to $15.8 billion.

Engagement narrowly missed: family daily active people reached 3.60 billion, up 3% year over year and up from 3.56 billion three months earlier, against a StreetAccount estimate of 3.61 billion. Reality Labs was better than feared — revenue of $431 million (up 16% YoY on AI-glasses strength) with a $4.62 billion operating loss, versus the $5.07 billion loss analysts had modeled, per CNBC. And guidance leaned soft: Q3 2026 revenue of $61-64 billion, whose $62.5 billion midpoint sat about 1% below the ~$63.15 billion consensus cited at the time of the print, assuming a roughly 1% currency headwind.

Strip out everything else and the advertising machine had an excellent quarter. Family of Apps ad revenue reached $59.4 billion, up 27% year over year (26% constant currency). CFO Susan Li attributed the volume side — ad impressions up 14% YoY — to growth in engagement and users plus ad-load optimizations, “healthy across all regions.” The price side — average price per ad up 12% YoY — she attributed to ad-performance gains, better macro conditions relative to Q2 2025, and currency tailwinds, partially offset by strong impression growth from lower-monetizing surfaces and regions.

The AI tooling behind those pricing gains is where Meta’s own claims concentrate — and they should be labeled as Meta’s claims, since no independent benchmark exists. Per Li on the call: Advantage+, the end-to-end automation suite, crossed a $75 billion annual ad-revenue run-rate this quarter; a new Meta Generative Recommender ad-matching model plus the GEM ranking model drove an 8.3% increase in ad clicks and a 15.7% uplift in conversions on Facebook in early deployment; and more than 9 million small businesses now use at least one generative-AI creative tool. Non-ad revenue is also stirring: Family of Apps “other revenue” crossed $1 billion for the first time, up 73% year over year, driven primarily by WhatsApp paid messaging and subscriptions.

For the advertiser-side budgeting math — what +12% price per ad and +14% impression growth, both year over year, actually do to your planning model for the rest of 2026 — see our companion breakdown of Meta ads pricing after Q2. Worth holding in mind alongside the growth story: the same ad engine driving these numbers remains under separate brand-safety and ad-quality scrutiny that advertisers should factor into placement decisions.

That chart is the quarter in one image. Every revenue line grew at a double-digit clip year over year — and every cost line grew faster. Research and development alone rose 67% YoY to $21.7 billion, and total costs and expenses rose 55% YoY to $42.0 billion, a figure that includes $3.6 billion of one-time charges covered in Section 04. The market did not sell Meta because the ads business slowed. It sold Meta because the cost of building what comes next grew twice as fast as the business paying for it.

Nearly every outlet reported “net income down 14%” and “free cash flow down 91%” — both year over year — as if they were the same collapse. They are not, and conflating them is how investors and operators alike misread this quarter. Net income fell mostly because of one-time charges: $2.4 billion for legal proceedings plus $1.18 billion in severance. Free cash flow fell because capital expenditure — $31.1 billion in a single quarter — consumed roughly 98% of the $31.86 billion Meta generated in operating cash flow. Different numerators, different denominators, different causes.

Susan Li said the reconciling sentence herself, verbatim on the call: “Second quarter GAAP operating income was $18.8 billion, representing an 8% decline year-over-year and a 31% operating margin. Excluding the Q2 legal charges and severance expenses, our second quarter operating income would have increased 9% year-over-year.” That is the whole P&L story in two sentences: as reported, operating income fell 8% YoY and the operating margin — operating income as a share of revenue — compressed to 31% from 43% a year earlier; excluding the charges, by Meta’s own stated math, operating income would have grown 9% YoY.

Read the two halves of that table separately and the quarter makes sense. The P&L story is a temporary one: charges of this size do not recur every quarter, and Meta still expects full-year 2026 operating income to land above full-year 2025’s. The cash-flow story is structural: as long as quarterly capex runs at or near operating cash flow, free cash flow stays pinned near zero — and that is a choice Meta is making, quarter after quarter, on purpose.

Two one-time items dragged the quarter’s profitability. First, a $2.4 billion charge related to legal proceedings, sitting inside a general-and-administrative line that ballooned to $5.6 billion from $2.7 billion a year earlier. Second, $1.18 billion in severance connected to the May 2026 headcount reduction, which affected roughly 8,000 employees — most expected to be off the books by the end of Q3 2026. Total headcount stood at 75,472 at quarter-end, down about 1% year over year per the press release, and down 3% from Q1 per Li’s remarks — two different comparison windows, not a contradiction.

The consequence flowed straight into guidance: Meta raised the low end of its full-year 2026 total-expense outlook to $165-169 billion specifically to incorporate the $2.4 billion charge, and lifted its tax-rate guidance for remaining quarters to 15-17% from 13-16%. Even so, Li repeated that Meta continues to expect full-year 2026 operating income to come in above 2025’s — the company’s own signal that it views the Q2 profit dent as episodic, not structural.

The number that moved the stock was forward-looking. Li, verbatim: “We anticipate 2026 capital expenditures, including principal payments on finance leases, to be in the range of $130-145 billion, narrowed from our prior outlook of $125-145 billion.” Technically a narrowing; practically, the floor rose $5 billion while the ceiling held. After a quarter in which capex already consumed roughly 98% of operating cash flow, the market read the revision as “more spend,” not “less risk.”

The financing picture underneath is shifting in ways worth stating plainly. Meta ended the quarter with $90.3 billion in cash and marketable securities against $83.7 billion in debt, after roughly $24.9 billion of net long-term debt issuance in the first half of 2026. Quarterly dividends of $1.353 billion exceeded the quarter’s entire $784 million of free cash flow, and Meta repurchased no shares. On the call, Li told analysts Meta has been “evolving our capital structure in recent years to include a greater mix of debt as we work to bring down our cost of capital” — and pointed to partner capital as another lever. The day before earnings, Meta announced a strategic venture with BlackRock to develop a 1-gigawatt data center in El Paso, Texas — reportedly a ~$14 billion project with BlackRock holding an 80% stake to Meta’s 20%, per press accounts of the deal structure.

Where does the money go? Servers, data centers, and network infrastructure, per Li — and Meta’s compute book extends well beyond its own buildings. The company’s multi-vendor chip commitments, including its AMD deal, are part of the same diversification story we covered in our analysis of the AMD-Anthropic compute deal, which put Meta’s AMD commitment at 6GW (February 2026) inside AMD’s roughly 14GW announced compute book.

Meta’s print did not land in a vacuum. Alphabet reported a week earlier — we broke that quarter down in our Alphabet Q2 2026 analysis — and Microsoft reported its FY26 Q4 the same evening as Meta. Putting the three on one page with a consistent metric set is where the pattern shows: the market is no longer punishing AI capex in the abstract. It is punishing capex that lacks a visible revenue line scaling alongside it.

Sources: Meta figures from the July 29 press release and call transcript; Alphabet figures from its July 22 report as covered in our analysis, with the free-cash-flow claim attributed to CNBC; Microsoft’s Azure figures from Microsoft’s own newsroom release. The full Microsoft print — Copilot seats, segment detail, and its own capex optics — gets its own treatment in our Microsoft FY26 Q4 breakdown.

The structural read, echoed across CNBC’s and TechTimes’ coverage: Microsoft and Alphabet can point investors to a cloud line that scales with AI capex. Meta’s AI spending currently surfaces mostly as a stronger ad-ranking engine — real money, but embedded inside an existing revenue line rather than visible as a new one. Same-evening, same-macro, opposite stock reactions. That contrast, more than any single Meta number, explains the selloff.

The most important strategic statement of the call was not in the prepared remarks — it came in the Q&A, in Zuckerberg’s answer to Morgan Stanley’s Brian Nowak on Meta’s compute strategy. The subtext of the answer: why Meta isn’t simply renting out its data centers the way Microsoft and Amazon do.

Most coverage quoted Zuckerberg’s shorter line about getting “a lot of offers for compute at a significant premium over what we paid for it” and missed the actual logic: Meta believes the margin on selling intelligence — models, agents, outcomes — will stay durably higher than the margin on renting raw compute. That is the thesis that explains the capex. In the press release, Zuckerberg framed it as AI “accelerating our core business today, powering our next generation of products, and opening the door to entirely new enterprise opportunities.” On the call, he named three buckets for where that enterprise opportunity lives:

Our projection, stated as ours: if Family of Apps “other revenue” (the line that just crossed $1 billion in a quarter, up 73% year over year) becomes the disclosure vehicle for enterprise AI revenue, that single line will decide whether the market re-rates Meta’s capex the way it re-rated Microsoft’s. Until a revenue line visibly scales with the spend, quarters like this one — strong ads, punished stock — are likely to repeat.

Here is the disconnect that matters if you buy media rather than stocks: investors punished Meta while every operational signal an advertiser cares about improved. Inventory grew (impressions +14% YoY), pricing strengthened (price per ad +12% YoY), and the automation layer taking over campaign construction crossed — by Meta’s own account — a $75 billion annual run-rate. A falling stock price changes none of your auction math. Rising ad prices change all of it.

The practical follow-through: pressure-test your Meta spend against these Q2 signals before the Q3 print resets the numbers again. Our paid media team runs exactly this kind of platform-earnings-to-budget translation for clients each quarter, and our analytics practice builds the measurement layer that keeps platform-reported lift honest against your own conversion data.