Morgan Stanley ups tech giants’ capex forecast to $805B for 2026

Morgan Stanley lifted its 2026 capex forecast for Amazon, Alphabet, Meta, Microsoft, and Oracle to $805 billion, nearly double the prior year, signaling a larger wave of AI infrastructure investment. The revision is supportive for Microsoft and Meta, with prediction markets pricing Microsoft as the largest company by December 2026 at just 0.8% YES and Meta at 100% YES for $740 by April 2026. The news is constructive for tech capex expectations but is more likely to affect individual mega-cap names than the broader market.

Analysis

The real signal is not “more AI spending” but a widening capital intensity gap inside megacap tech. The winners are the firms best able to convert incremental capex into a durable workload advantage: MSFT and META, with ORCL as a smaller but interesting beneficiary if sovereign/enterprise AI demand stays hot. The second-order loser is any hyperscaler that has to defend share with matching spend but lacks equivalent monetization density; that typically shows up first as margin compression, then as multiple compression, especially if market pricing starts treating capex as an arms race rather than optional growth spend.

The market is likely underestimating the supply-chain beneficiaries and overestimating the duration of the spending elasticity. A 2026-2027 capex step-up should pull forward demand for power equipment, networking, advanced packaging, and certain semiconductor tooling names before it fully translates into revenue for the hyperscalers. But this setup is fragile: if AI workload growth decelerates, or if CFO scrutiny shifts from “build now” to “prove ROI,” the group can re-rate quickly because the market is effectively paying for future utilization that is not yet visible in earnings.

The prediction-market pricing looks most stretched on META, which implies the crowd has already anchored to the bullish narrative and may be ignoring execution risk around monetization efficiency and regulatory overhang. MSFT appears more mispriced on a relative basis because it can convert capex into higher-value enterprise workflows faster than consumer-ad-driven models, but even there the upside is likely more about relative performance than absolute leadership by late 2026. The contrarian view is that this capex surge is bullish for suppliers and only conditionally bullish for the hyperscalers themselves; the broad index implication may be positive, but the equity alpha is likely in picks-and-shovels rather than in the most crowded AI leaders.

Key Decisions for Investors

• Long ORCL / short a basket of the most crowded AI megacaps for 3-6 months: express the view that incremental AI infrastructure spend still has more upside in remaining catch-up beneficiaries than in already-expensive leaders; keep the short leg beta-adjusted.
• Buy MSFT on a 2-4 week pullback, target a relative outperformance trade versus META through year-end 2026; thesis is faster monetization conversion and lower execution friction, with downside if cloud growth slows materially.
• Initiate a tactical long in AI infrastructure suppliers or semicap equipment for 6-12 months as the cleanest second-order beneficiary of the capex step-up; use options if you want defined risk because the trade can run ahead of revenue recognition.
• Avoid chasing META upside at spot levels implied by the prediction market; if long, fund it versus a basket of other hyperscalers because the market appears to be pricing perfect execution and may not be compensating regulatory or ROI risk.
• Set a catalyst watch on the next two earnings seasons: if capex guidance rises but revenue per AI dollar does not inflect, rotate out of hyperscaler longs and into suppliers immediately; that is the point where the trade becomes a margin story rather than a growth story.