Introduction: Revenue beats, profit disappoints, and spending swells
Meta (META) delivered a mixed third quarter: revenue outperformed while EPS undershot expectations, as a one-time tax charge, heavier infrastructure buildouts, and expanded hiring for AI talent weighed on profitability. The stock fell more than 6% after-hours as management paired the print with higher capital expenditure plans—not just this year, but with a clear warning that 2026 will be even heavier. The message is unmistakable: Meta is leaning hard into Meta AI investments across models, compute, and products, even if that means a bumpier margin path in the near term.
Q3 at a glance: Top-line strong, EPS dinged by taxes and capex
- EPS: $1.05 vs. $6.72 expected (Bloomberg consensus), reflecting a one-time tax-related charge and higher opex/capex tied to AI.
- Revenue: $51.24B vs. $49.6B expected; Family of Apps revenue hit $50.77B (vs. $48.6B est.).
- Stock reaction: Shares fell >6% post-print as investors digested the cost trajectory.
Takeaway: The advertising engine is humming, but the P&L optics are dominated by front-loaded investment—by design.
Capex ramps: 2025 up again—and 2026 gets “notably larger”
Management raised 2025 capex to $70B–$72B (prior $66B–$72B), and flagged even steeper growth in 2026.
CFO Susan Li outlined what’s driving the increases:
- Infrastructure costs (AI data centers, incremental cloud, depreciation) as the primary growth driver
- Compensation as the No. 2 driver—especially a full-year impact from 2025 hiring and additional technical hires in AI-critical areas
Why it matters: This is a classic “invest through the cycle” posture. Meta is building a durable compute moat that should translate into better ad relevance, faster product iteration, and stickier engagement across apps.
Inside the spend: Big checks to accelerate Meta’s AI roadmap
Recent, disclosed allocations underscore the scale and urgency of Meta AI investments:
- $14.3B in Scale AI and appointing its CEO as Chief AI Officer
- $1.5B+ for a new El Paso, TX data center
- $27B financing with Blue Owl Capital for the Hyperion data center in Louisiana
- Ongoing talent acquisition from leading AI labs and platform rivals
Strategic read: Meta isn’t trying to sell enterprise AI like hyperscalers. It’s embedding AI to raise ad yield, boost creator tools, drive algorithmic discovery, and elevate user utility across social, messaging, and devices.
Why this path (and why now): The Meta playbook vs. hyperscalers
Unlike Amazon, Google, and Microsoft—who monetize AI via enterprise cloud—Meta’s model is consumer-first:
- Monetization loop: Better models → more relevant content & ads → higher ROI for advertisers → more spend → more data/feedback → better models.
- Data advantage: Billions of daily interactions across Facebook, Instagram, WhatsApp, Threads, and Reels form a uniquely rich corpus for model training and ranking improvements.
- Distribution flywheel: AI-powered features (creative tools, recommendations, safety systems) can scale across the Family of Apps with minimal incremental friction.
Investor lens: The trade-off is near-term margin compression for a potentially superior long-run ROIC if AI upgrades expand engagement and ad effectiveness at global scale.
Hardware & on-device AI: Momentum, but questions remain
Meta is infusing AI into its Ray-Ban Meta smart glasses, Quest 3, and the Meta Ray-Ban Display concept. While these products showcase compelling on-device AI use cases (hands-free capture, real-time assistance, multimodal experiences), some analysts remain cautious on the clarity of consumer outcomes and killer use cases in XR.
One device analyst noted concern that leadership hasn’t fully articulated “what problems are uniquely solved” by Meta’s XR + AI stack and what remarkable experiences will convert skeptics into daily users. Translation: the narrative needs sharpening, even as the tech gets stronger.
Ads resilience vs. AI search risk: the balancing act
Meta’s results confirm ad demand remains healthy, outpacing expectations. That said, the AI search/browser wave poses a structural overhang for attention and discovery. The company’s counter is familiar: make the feed smarter, accelerate content relevance, and lower creative friction for advertisers and creators through AI tools—so that Meta’s walled garden feels indispensable, regardless of off-platform shifts.
Expense anatomy: What’s pressuring EPS now
- One-time tax charge: The main reason EPS diverged so sharply from consensus.
- Depreciation + cloud: AI infra is capex-heavy; the D&A curve steepens.
- Comp: A full year of 2025 hires hits 2026 expense lines; ongoing AI-specialist premiums keep unit costs elevated.
Net effect: Operating leverage is delayed, not denied—provided engagement and ad ROI continue to trend up as the AI stack matures.
Guidance signals: Read the fine print
- 2025 capex nudged higher; 2026 capex growth “notably larger.”
- Total expenses to grow “significantly faster” in 2026 than 2025, led by infra and comp.
- Management continues to frame spend as multi-year and returns-driven (model performance, ad efficacy, safety systems, and product velocity).
What to watch next:
- Ad ROI improvements tied to model upgrades;
- Engagement per user across Reels/Stories/Video;
- Creator adoption of AI creative tools;
- Unit cost per inference/training trendlines;
- Proof points that capex translates into better personalization and higher conversion.
Bottom line for investors
- Bull case: Meta’s massive, focused AI reinvestment compounds its network effects across the Family of Apps, lifts ad yield, and seeds new consumer AI experiences.
- Bear case: Heavy spend meets uncertain XR adoption and intensifying competition for attention (including AI-native browsing and agents).
- Base case: Volatile quarters as accounting and capex run hot, but fundamentals improve if AI continues boosting relevance and monetization.
Short-term pain, long-term platform play
Meta’s quarter tells a consistent story: the ad engine works, the AI ambition is escalating, and management is choosing scale today to earn optionality tomorrow. EPS was hit by a tax item and the weight of an AI supercycle that demands extraordinary capital. The company’s guide—“notably larger” 2026 spending—signals confidence that Meta AI investments will translate into superior ranking, richer formats, smarter safety, and ultimately better monetization.
For shareholders, the trade is clear: near-term margin pressure for a shot at structural advantages across content, commerce, and communication—if Meta can turn unprecedented capex into everyday utility consumers love.