Wall Street hates Meta's AI spending guidance raise. We don't (www.cnbc.com)

Meta reported a strong quarter — revenue rose 26% YoY to $51.24B and adjusted EPS beat at $7.25 — but shares slid after management raised multi-year spending guidance and booked a one-time ~$16B tax transition charge tied to U.S. tax reform. CFO Susan Li said the tax hit is non-recurring and the company still expects meaningful cash tax savings going forward. The market’s real worry was Meta lifting the lower end of full-year capex to $70–72B (from $66–72B) and raising 2025 expense guidance to $116–118B, signaling heavier infrastructure, cloud and depreciation spend. Q4 revenue guidance ($56–59B) still sits above consensus, and Reality Labs losses narrowed, while ad metrics and engagement beat expectations. For the AI/ML community this is a calculated bet: Meta is front‑loading massive compute and data infrastructure to support model training, AI-powered ads and new product experiences. Management frames this as optionality — accelerated capacity positions Meta for faster advances (even “superintelligence” scenarios) while enabling immediate gains from higher compute use across apps. Key technical indicators: Meta AI >1B MAUs, Instagram 3B MAUs, Threads 150M DAUs, and a >$60B run rate for AI-driven ad tooling, plus Reels revenue momentum. The implication: more internal/external compute demand, broader deployment of large models across consumer products, and slower but likely positive long-term ROI — a major signal that hyperscalers will keep competing on raw compute and integrated ML services.