Key Takeaways
Powered by lumidawealth.com
- Alphabet, Meta, and Microsoft spent ~$78B on capex last quarter, +89% YoY, mostly for AI data centers and GPUs.
- Stocks split: Meta and Microsoft fell after-hours on spend; Alphabet rose >6% on stronger cloud and AI traction.
- Microsoft posted a record $34.9B quarterly capex; demand still exceeds supply per CFO. Alphabet guides 2025 capex up to ~$93B with a “significant” increase in 2026.
- Backlogs support multi-year demand: Microsoft ~$392B commercial backlog; Google ~$155B. Meta’s payoff path is less clear without external cloud customers.
What Happened?
Alphabet, Meta, and Microsoft disclosed aggressive AI infrastructure outlays totaling about $78B in the quarter as they build data centers and procure accelerators. Alphabet reported stronger Gemini adoption (650M MAUs, +44% in three months) and Google Cloud growth (+34% YoY to $15.2B), raising 2025 capex to as much as ~$93B and signaling another step-up next year. Microsoft recorded a quarterly capex record at $34.9B while saying AI demand continues to outstrip capacity. Meta warned 2026 capex will be “notably larger” than 2025 and reiterated long-term AI ambitions despite Reality Labs losses. Shares diverged: Meta and Microsoft slipped; Alphabet gained.
Why It Matters?
Sustained hyperscale capex confirms a multi-year AI buildout that can drive semis, power, and data-center supply chains, but near-term free cash flow and margins face pressure if utilization lags. Alphabet and Microsoft have clearer monetization via cloud backlogs and AI services, providing visibility on returns; Meta’s ROI case is more dependent on ads and new devices, raising execution risk. Bubble concerns persist, yet order books and customer commitments imply durable demand. The capex race increases barriers to entry and may consolidate share around platforms with distribution, models, and compute at scale.
What’s Next?
Watch Amazon and Apple prints for cloud momentum and AI device signals. Track hyperscaler disclosures on site adds, accelerator procurement, and power constraints to gauge capacity timing. Monitor unit economics: AI service attach, cloud margins, and revenue per compute. For Meta, look for tangible AI ad lift, product adoption, and any move to resell excess compute. Policy and export controls remain swing factors for supply chains and China exposure.
