Big Tech’s AI spending is depriving investors of juicy payouts

Goldman Sachs says the largest AI hyperscalers are set to spend $755 billion on capex this year, an 83% increase, as AI infrastructure ambitions and a memory-chip crunch drive spending higher. To fund that investment, the group cut buybacks by nearly two-thirds in Q1, and Goldman expects S&P 500 buyback growth to slow to just 3% this year. The article points to a mild headwind for shareholder returns rather than a broad market shock.

Analysis

The key second-order effect is that AI capex is not just a margin story; it is a capital allocation regime shift that lowers the probability of “automatic buyer” support under the mega-cap complex. When free cash flow is redirected from repurchases to data-center buildouts, the market loses one of the most price-insensitive sources of demand, which can amplify downside on any earnings disappointment or multiple compression event. That matters most for names whose equity story has been partially underwritten by recurring buybacks rather than accelerating operating leverage.

The bigger beneficiary set is the industrial and infrastructure layer behind the buildout: power equipment, grid interconnect, thermal management, fiber, and select semiconductor memory suppliers with tight supply. The immediate constraint is less about AI demand fading and more about bottlenecks pushing spend into less efficient channels, which can lift revenue for suppliers while compressing returns on incremental hyperscaler capex. Over the next 6-18 months, the market is likely to reward the picks-and-shovels names with cleaner backlog visibility while punishing hyperscalers if capex continues to outrun monetization.

The risk to the negative view is that the spending wave may still be early enough that investors tolerate weaker capital returns as long as top-line growth remains intact. But if ad demand, cloud growth, or enterprise AI uptake slows even modestly, the valuation gap could re-rate quickly because the cash-return cushion is gone. A reversal would likely need either a hard capex pause or a clear demonstration that AI infrastructure is producing faster incremental revenue per dollar deployed.

The contrarian setup is that the buyback cut could be misunderstood as a bearish signal when it is partly a forced rebalancing of balance sheets, not a wholesale deterioration in fundamentals. That suggests the best short may not be the hyperscalers outright, but relative exposure to the most buyback-sensitive names versus those with stronger operating leverage or more diversified cash returns. The move looks more underpriced in index-level effects than in single-name fundamentals, because buybacks have been a hidden support for large-cap tech breadth.

Key Decisions for Investors

• Short an equal-weight basket of AMZN/GOOG/META/MSFT/ORCL on any 1-2% post-news bounce; 1-3 month horizon. Risk/reward favors a 5-8% drawdown if capex guidance stays sticky and buybacks remain suppressed, with tight risk if managements explicitly re-accelerate capital returns.
• Go long power-infrastructure and grid-enablement names versus hyperscalers for a 6-12 month trade; best expressed as long a basket of electrical equipment and data-center infrastructure suppliers vs short MSFT/GOOG. Upside comes from capex monetization shifting to suppliers even as hyperscaler ROIC compresses.
• Buy downside protection on a mega-cap tech ETF or the most buyback-sensitive hyperscaler into the next earnings cycle; 2-4 month tenor. The asymmetry improves because reduced repurchase support raises gap risk on any guidance miss.
• Pair trade: long memory/storage beneficiaries versus short a low-growth cash-return tech name over 3-6 months. The thesis is that supply bottlenecks and AI demand keep component pricing firm while capital-return-heavy names lose marginal bid support.
• If you want outright exposure, own the picks-and-shovels beneficiaries, not the capital allocators; use dips to accumulate rather than chasing hyperscaler rallies. The trade works best while capex revisions are still being ratcheted upward and before the market prices in slower incremental returns.