Alphabet and Meta Delivered Worrisome News for Investors -- and It Has Trillion-Dollar Implications

Alphabet plans $180 billion to $190 billion of AI-driven capex this year, while Meta lifted its AI spending forecast to $125 billion to $145 billion. Early AI deployment has boosted Google Cloud sales 63% in the March quarter and improved Meta's ad tools, but the tradeoff is that both companies spent $0 on buybacks in Q1 2026 versus $15.1 billion and $12.8 billion in Q1 2025. The article argues that rising AI investment could crowd out one of Wall Street's biggest tailwinds: share repurchases.

Analysis

The market is increasingly treating AI capex as a balance-sheet arms race, not a normal growth investment cycle. That matters because the first-order winners are still the hyperscalers, but the second-order winners may be the infrastructure bottlenecks—power, networking, memory, and GPU assembly—where spend gets monetized before any software ROI is proven. If capex stays elevated through the next 2-3 quarters, incremental free cash flow will likely migrate away from repurchase programs and toward suppliers with contractual pricing power.

The more interesting signal is not that buybacks disappear, but that the buyback floor under mega-cap tech weakens at the exact time index ownership is most concentrated. That can increase volatility in GOOGL and META even if fundamentals keep improving, because a large marginal buyer is being replaced by a less predictable capex cycle. In other words, investors may get the earnings upside from AI before they get the capital return support that historically cushioned drawdowns.

A contrarian risk is that the market may be underestimating how quickly AI monetization can offset capex, especially for ad-tech and cloud. If model deployment lifts revenue per user or cloud margin mix faster than expected, the feared crowding-out of buybacks becomes a temporary bridge rather than a structural impairment. The key timing issue is months, not days: the next two earnings seasons will determine whether this is a capex peak or a new, durable spending plateau.

The clearest hidden beneficiary is NVDA, but not just from GPU unit demand—also from the need for recurring refreshes, networking attach, and memory intensity as model size and inference traffic scale. Intel is more of a relative loser unless it can capture a credible share of the AI accelerator or foundry stack, because rising industry capex without product relevance can still leave it stranded. The broader implication is a capex rotation away from capital returns and toward compute landlords, which is constructive for the picks-and-shovels complex but not for passive holders expecting tech to keep buying itself.

Key Decisions for Investors

• Stay long GOOGL and META on a 6-12 month horizon, but size for higher near-term volatility: the core thesis is operating leverage, not buyback support; upside remains intact, but drawdowns may be sharper if capex guidance stays elevated.
• Add to NVDA on pullbacks as the cleaner second-order beneficiary of sustained AI spend; risk/reward is favorable over 3-9 months because every incremental hyperscaler dollar likely expands the installed-base refresh cycle and networking attach.
• Underweight INTC versus NVDA on a 3-6 month view: if AI capex continues to expand, capital will keep flowing to the ecosystem with pricing power and product pull-through, not the laggard trying to regain relevance.
• Pair trade: long NVDA / short a basket of mature mega-cap cash return proxies that are funding AI with reduced repurchases; the thesis is that capital return scarcity will compress relative multiples in the market’s former “self-funding compounders.”
• Use earnings season as the catalyst window: if GOOGL/META show capex up and buybacks flat again, consider a tactical long-vol or reduced directional exposure, since the market may start repricing the loss of repurchase support before fundamentals fully reflect AI monetization.