Meta Q4 earnings preview: AI capex in focus
Meta reports after the bell Wednesday, and the core question is simple: does the AI buildout earn its keep? Shares are down more than 12% since October as the market digests a bigger, longer investment cycle in data centers and infrastructure.
Management lifted 2025 capital expenditures to $70-$72 billion (from $66-$72 billion) and flagged 2026 capex growth as "notably larger." That trajectory is the entire story for margins, free cash flow, and buyback capacity over the next 24 months.
Street setup
- Q4 EPS: $8.16 (consensus)
- Q4 revenue: $58.4 billion (consensus), up from $48.4 billion in Q4 2024
- Q4 spend: $21.9 billion, versus $14.4 billion in the year-ago quarter
- Reality Labs: ~$959 million revenue, ~$5.9 billion operating loss expected
What matters for finance
The scale of capex will push depreciation higher into 2025-2026. Expect pressure on gross margins and a tug-of-war between investment pace and free cash flow optics.
- Capex shape: Any color on phasing, useful life assumptions, and data center cohorts entering service.
- AI monetization: Timelines for model deployment and revenue lift in ads, messaging, and enterprise tooling.
- Opex discipline: Headcount, compute costs, and unit economics for training vs. inference.
- Capital returns: Buybacks relative to capex ramp; balance sheet flexibility.
- Reality Labs drag: Loss run-rate and how much is fixed versus variable.
Strategy and competitive context
Meta is spending aggressively alongside Amazon, Google, and Microsoft as AI workloads swell. The company increased its bets with a $14.3 billion purchase of 49% of Scale AI and named Alexandr Wang chief AI officer to lead Superintelligence Labs.
Execution risk is real: delays around its Llama 4 Behemoth model, and a reported consideration to shift the next flagship model to a proprietary approach from open weights. Meta also cut roles in its metaverse group, re-allocating spend into wearables like AI smart glasses.
Meanwhile, Google's Gemini 3 has seized the narrative. Over the last 12 months, Google's stock is up 66%, Meta is up 4%, and Amazon is up 1.8%-a clear read on investor patience for long-cycle AI capex versus near-term product traction.
Policy and legal overhangs
Calls to restrict social media access for users under 16 are rising-Australia has enacted a ban, and France is weighing similar steps. In the US, the FTC is appealing its loss in an antitrust case that argued Meta bought Instagram and WhatsApp to protect its position in personal social networking.
Pre-earnings checklist
- Guide: Does 2025 capex stay anchored at $70-$72B, and how "notably larger" is 2026 in dollars?
- Margins and D&A: Any update on server GPU useful lives and depreciation schedules.
- Ads demand: Signal on pricing and conversion lift from AI-driven targeting and creative.
- Model roadmap: Clear milestones for Llama 4 and whether a proprietary pivot changes partner ecosystems.
- Reality Labs: Path to narrowing losses and the role of smart glasses in near-term adoption.
- Capital returns: Buyback cadence relative to the capex ramp and free cash flow seasonality.
Bottom line
The print will be judged on two axes: revenue quality and capex credibility. If Meta shows accelerating monetization alongside a tight, phased capex plan, the market will give them time. If spend swells without clearer payoff timelines, expect the multiple to stay capped.
If you're building AI capability inside a finance org, here's a practical starting point: AI tools for finance.
Your membership also unlocks:
