Big US tech stocks swing as investors probe AI spend

Big US tech names were mixed after earnings as investors focused on rapidly rising AI spending and its uncertain payback. Meta lifted planned capex to as much as $145bn from $135bn and fell more than 5% after hours, while Microsoft dropped about 2% on weaker free cash flow of $15.8bn and Amazon fell 1.6% after softer next-quarter profit guidance. Alphabet rose nearly 6% after reporting profits up 30% and Google Cloud revenue up 63%, highlighting clearer AI monetization than peers.

Analysis

The key second-order effect is not “AI spending is high,” but that the market is beginning to discriminate between capex with visible monetization pathways and capex that still looks like a research subsidy. Alphabet is being rewarded because its AI spend is increasingly legible through cloud demand and infrastructure control; that makes the incremental dollar feel like a capacity expansion, not a science project. Meta is being punished because the spend is reading as open-ended operating leverage risk with no clear pricing model, which raises the probability of margin compression before any revenue offset arrives.

This creates a relative trade opportunity across the megacap complex. The winners are the firms that can internalize the full stack — models, silicon, and distribution — because they reduce dependence on third-party inference economics and preserve margin in a GPU-constrained world. That also has a spillover into the supply chain: custom silicon and cloud beneficiaries should outperform generic semiconductor exposure, while companies still reliant on rented compute may face slower FCF conversion than headline revenue growth implies.

The main risk is temporal mismatch: AI returns likely remain a 12-24 month story while the cash burn is immediate. In the next 1-2 quarters, free cash flow revisions and capex guideposts will matter more than revenue beats, so stocks with “tangible demand” can keep re-rating even if aggregate spend keeps rising. The consensus may be underestimating how much investor patience has narrowed; once a company is forced to defend spend without a near-term monetization bridge, multiple compression can happen faster than earnings deterioration.

Contrarian angle: the move may be overdone in the sense that high capex can still be value-accretive if the depreciation cycle and utilization ramp are managed correctly. But the burden of proof has shifted — capital allocation credibility now matters more than AI narrative breadth. Expect the next catalyst set to be guidance, not the current-quarter print: any hint of capex moderation or sharper monetization framing could trigger a sharp mean reversion in the losers.

Key Decisions for Investors

• Long GOOGL / short META for 4-8 weeks: own the name with visible AI demand pull-through and short the one where capex is still being sold as faith-based optionality; target 8-12% relative outperformance, stop if META provides a concrete monetization roadmap or GOOGL cloud growth decelerates sharply.
• Add MSFT on 1-3 month weakness only if free cash flow stabilizes: the setup is attractive for long-duration holders, but near-term downside persists while capex crowds out cash generation; use staggered entries and hedge with short-dated puts into the next earnings window.
• Underweight AMZN vs GOOGL for the next quarter: both benefit from AI infrastructure demand, but GOOGL currently has cleaner evidence of margin-accretive AI usage; pair should work if cloud AI demand remains the dominant signal and retail margin noise reappears.
• For tactical upside, buy GOOGL calls into any post-earnings consolidation: the market is rewarding proof-of-demand, and further guidance on AI/cloud utilization could compress implied vol faster than realized volatility over 30-60 days.
• Avoid chasing META until management quantifies monetization conversion: the risk/reward is asymmetric to the downside if capex keeps rising while revenue attribution remains vague; wait for a better entry after a 1-2 quarter reset or use put spreads if you need downside exposure.