Actual results may differ materially as a result of various factors, including those set forth in today's earnings press release and in our quarterly report on Form 10-Q filed with the SEC. A reconciliation of GAAP to non-GAAP measures is included in today's earnings press release. The earnings press release and an accompanying investor presentation are available on our website at investor.atmeta.com. We're also seeing good progress with AI for ad creative, with a meaningful percent of our ad revenue now coming from campaigns using one of our generative AI features.

Advancements in our recommendation systems have improved quality so much that it has led to a 5% increase in time spent on Facebook and 6% on Instagram just this quarter. The percent of people using Meta AI is growing, and we are seeing new users' AI retention increase, too, which is a good sign for that continued use. We launched the Meta Quest 3S Xbox Edition last month, and we're seeing record interest in cloud gaming. Q2 total revenue was $47.5 billion, up 22% on both a reported and constant currency basis.

Marketing and sales increased 9%, primarily due to an increase in professional services related to our ongoing platform integrity efforts, as well as marketing costs partially offset by lower employee compensation. Second quarter operating income was $20.4 billion, representing a 43% operating margin. Our tax rate for the quarter was 11%, which reflects excess tax benefits from share-based compensation due to the increase in our share price versus prior periods. Capital expenditures, including principal payments on finance leases, were $17 billion, driven by investments in servers, data centers, and network infrastructure.

What went well
  • Total revenue of $47.5B, up 22% YoY on both a reported and constant-currency basis, driven by AI-unlocked efficiency gains across the ad system.
  • Family of Apps ad revenue of $46.6B, up 21% (22% constant currency), with the online commerce vertical the largest contributor to growth.
  • New AI-powered ads recommendation model drove roughly 5% more ad conversions on Instagram and 3% on Facebook.
  • Engagement gains from recommendation improvements: time spent up 5% on Facebook and 6% on Instagram in the quarter; Instagram and US Facebook video time each up more than 20% YoY.
  • Meta AI surpassed more than 1 billion monthly actives, available in over 200 countries and territories.
  • Ray-Ban Meta glasses sales accelerated with demand outstripping supply; launched Oakley Meta HSTN performance AI glasses.
  • Operating income of $20.4B at a 43% operating margin; net income $18.3B ($7.14 EPS); Family of Apps operating margin of 53%.
  • G&A fell 27% YoY on lower legal-related costs.
What went wrong
  • Average price-per-ad growth slowed modestly from Q1 to +9% due to accelerated impression growth (skewed to lower-monetizing regions).
  • Reality Labs operating loss of $4.5B; segment revenue only $370M (up 5%) as AI-glasses growth was partly offset by lower Quest sales.
  • Total expenses rose 12% YoY to $27.1B on higher infrastructure and R&D (+23%) costs; management flagged a sharp acceleration in depreciation in 2026.
  • Free cash flow fell to $8.5B amid $17B CapEx and $15.1B of non-marketable equity investments (including the Scale AI minority stake).
  • Ray-Ban Meta supply constrained on most popular SKUs despite production increases.

Guidance Changes

MetricPeriodCurrent guidance
Total expenses growthFY2026Infrastructure expected to be the single largest contributor to expense growth, driven by a sharp acceleration in depreciation; employee compensation the next largest driver (full-year cost of 2025 AI hires); no dollar figure given
Capital expendituresFY2026Expected to grow, with the big driver being scaling GenAI training capacity (servers, networking, data centers), plus continued core-AI investment; no dollar figure confirmed by management
Data center financingFY2026Expect to self-finance a large share; exploring co-development with financial partners to attract external financing (no finalized transactions)
Reality Labs2H2025Working to ramp Ray-Ban Meta supply to better meet demand later in the year

Performance Breakdown

MetricYoYNote
Total revenue +22% $47.5B; up 22% reported and constant currency, driven by strong ad performance from AI efficiency gains.
Family of Apps ad revenue +21% $46.6B (+22% constant currency); online commerce the largest contributor; growth strongest in Europe (+24%) and rest of world (+23%), North America +21%, Asia-Pacific +18%.
Ad impressions +11% Growth mainly driven by Asia-Pacific; engagement tailwinds on Facebook and Instagram plus ad-load optimizations on Facebook.
Average price per ad +9% Increased advertiser demand from improved ad performance; growth slowed modestly vs Q1 due to accelerated impression growth.
Family of Apps other revenue +50% $583M; WhatsApp paid-messaging revenue growth and Meta Verified subscriptions.
Total expenses +12% $27.1B; cost of revenue +16% and R&D +23% (infrastructure and compensation), marketing & sales +9%, partly offset by G&A -27% on lower legal costs.
Operating income $20.4B (43% margin) Strong ad revenue growth outpacing 12% expense growth.
Reality Labs revenue +5% $370M; increased AI-glasses sales partly offset by lower Quest sales; expenses $4.9B (+1%); operating loss $4.5B.
Net income / EPS $18.3B / $7.14 Aided by an 11% tax rate reflecting excess tax benefits from share-based comp on higher share price.
Capital expenditures $17B Investments in servers, data centers and network infrastructure (incl. finance-lease principal).
Free cash flow $8.5B Elevated CapEx and $15.1B non-marketable equity investments; repurchased $9.8B stock and paid $1.3B dividends.
Headcount -1% QoQ Over 75,900 employees; earlier performance-related reductions dropped out, partly offset by hiring in monetization, infrastructure, Reality Labs, AI and compliance.

Earnings Call Themes & Trends

TopicPrevious mentionCurrent periodTrend
Superintelligence and Meta Superintelligence LabsFive AI opportunities framed last quarter (ads, engaging experiences, business messaging, Meta AI, AI devices)Established MSL (foundations, product, FAIR + new frontier lab led by Alexander Wang, Nat Friedman, Shengjia Zhao); progress on Llama 4.1/4.2 and next-gen models.
AI infrastructure / compute build-outLarge CapEx rampBuilding multi-gigawatt clusters: Prometheus (~1GW+) online next year, Hyperion scaling up to 5GW, plus multiple Titan clusters; front-loading compute.
AI-powered advertisingAndromeda retrieval architecture introduced 2H2024New recommendation model expanded to new surfaces; ~5% more conversions on Instagram, 3% on Facebook; meaningful share of ad revenue from GenAI creative features.
Engagement and videoRecommendation-driven engagement growthTime spent +5% Facebook, +6% Instagram this quarter; Instagram video time +20% YoY; over 2/3 of US Instagram recommended content now original posts.
Meta AIReach growingMore than 1 billion monthly actives; largest query driver is WhatsApp; expanding into content discovery and translation/dubbing.
New surface monetization (Threads, WhatsApp)Pre-monetizationAds introduced in Threads feed and WhatsApp Updates tab; low ad supply near-term, not a meaningful revenue contributor for the next few years.
AI glasses / Reality LabsRay-Ban Meta momentumSales accelerating with demand outstripping supply; Oakley Meta HSTN launched; more to share at Connect on Sept 17.

Q&A Summary

Eric Sheridan (Goldman Sachs): Key learnings from the AI strategy over the last 3-6 months that informed talent/compute shifts, and how to think about OpEx/CapEx over the next 12-18 months?
Zuckerberg: the most aggressive AI-progress assumptions have been most accurate; investing in elite talent-dense teams and a leading compute fleet, and driving the tech through all products. Li: 2026 budgeting not yet started, but infrastructure will be the single largest 2026 expense-growth driver (sharp depreciation acceleration), then employee compensation from AI hires; 2026 CapEx driven by scaling GenAI training capacity.
Brian Nowak (Morgan Stanley): Changed technological gating factors for superintelligence over the next 24 months, and factors that could drive further core-engagement lifts over the next 18 months?
Zuckerberg: self-improvement is a key research area; small, talent-dense teams are optimal for frontier research. Li: near-term core-recommendation focus on session-adaptive recommendations, helping smaller creators break out, interest exploration, scaling up models and incorporating LLMs, plus efficiency to protect ROI.
Doug Anmuth (JPMorgan): Has open-source thinking changed, and can Meta continue to self-finance 2026 CapEx (implied >$100B)?
Zuckerberg: thinking on open source largely unchanged - will keep releasing leading open models but not everything, with added safety considerations as models approach superintelligence. Li: expect to finance a large share themselves, but exploring co-developing data centers with financial partners to attract external financing.
Justin Post (Bank of America): Is the compute mostly for internal use or could there be external/business-model opportunities, and how do you think about ROI on CapEx?
Li: focused on internal use cases (core AI, ads ranking, training frontier models, future inference); not currently pursuing external use. Core-AI ROI is strong and well-measured; GenAI is much earlier on the return curve and not expected to be a meaningful revenue driver this year or next; building infrastructure with fungibility in mind, ordering servers as needed.
Youssef Squali (Truist): Where are glasses on the path to a computing platform, and how will SBC progress with all the AI hires?
Zuckerberg: very excited about glasses progress; believes glasses are the ideal AI form factor and that lacking AI glasses could be a cognitive disadvantage in the future. Li: increased comp/SBC from 2025 AI hires reflected in the revised expense outlook; focused on limiting dilution via the buyback program that offsets equity comp, plus dividends.

More on Meta Platforms, Inc.

Reported 2025-07-30 · figures from the Meta Platforms, Inc. Q2 2025 earnings call.

See how VectorShift works for your firm

Request Demo