Microsoft and Google face $650bn AI spending test as sell-off opens fresh entry point

Wedbush highlights a structural AI-driven infrastructure expansion with more than $650bn of 2026 capex committed by Microsoft, Alphabet, Amazon and Meta, arguing the hyperscalers are redesigning platforms for AI-first workloads that create higher switching costs and long-lived assets. Microsoft is flagged as an underestimated winner as AI monetisation ramps — Wedbush cites a ~$625bn backlog with ~45% risk exposure, a possible $100bn OpenAI funding round at >$800bn valuation, and 30% growth in Microsoft’s non-OpenAI backlog (Anthropic-led). Alphabet is noted for very large 2026 capex of $175–185bn that will pressure near-term margins but underpin faster cloud and ad monetisation as Gemini and AI-native workloads drive uptake. The note frames near-term FCF noise and margin compression as transitory and positions Microsoft and Alphabet as long-duration infrastructure leverage plays for investors looking past the 2026 spending bulge.

Analysis

Market structure: The $650bn 2026 hyperscaler AI capex cycle (MSFT, GOOGL, AMZN, META) reallocates cloud economics toward long-lived, AI-optimised assets, raising switching costs and deepening enterprise lock‑in. Winners: MSFT (enterprise distribution + Microsoft 365 embed) and GOOGL (search + Gemini-backed ad/cloud lift); losers: smaller cloud resellers, legacy on‑prem vendors, and software with low differentiation. Expect outsized demand for GPUs, datacenter power/cooling, and semiconductor equipment (ASML, LRCX) over 12–36 months.

Risk assessment: Near‑term FCF noise and margin compression are real — GOOGL capex $175–185bn and concentrated MSFT exposure (≈45% of a $625bn backlog, ≈$281bn conversion risk) could disappoint if OpenAI funding stalls. Tail risks include stricter export controls on accelerators, a failed $100bn OpenAI raise within 90 days, or multi‑year underutilisation of purpose‑built racks. Monitor Nvidia supply, cloud gross margins, and quarterly capex cadence as 0–6 month catalysts.

Trade implications: Tactical longs: overweight MSFT and GOOGL via buy-and-hold core sized 2–4% each, scaling into 10–20% further pullbacks over 3–6 months; add semicap equipment (ASML, LRCX) 1–2% for exposure to hardware demand. Pairs: long MSFT / short AMZN (2:1 notional) to express superior enterprise monetisation; use 6–12 month call spreads on MSFT/GOOGL to cap premium. Trim consumer ad/levers (META) and non‑specialist cloud plays (small integrators).

Contrarian angles: Consensus underweights execution risk and third‑party model dependency (OpenAI/Anthropic), and may overstate near‑term pricing power — history (2010s cloud capex) shows multi‑year lag to margin inflection. Possible mispricing: MSFT’s sell‑off likely overstates backlog conversion risk; conversely, GOOGL’s capex noise may understate long‑run ad/cloud upside. Watch grid/energy regulation (18–36 months) as an unintended constraint on expansion.

Key Decisions for Investors

• Establish a 3% long position in MSFT now, add another 2% on any further 10–15% pullback; hedge with a 1% allocation to 6–12 month MSFT 20–30% OTM call spreads to capture AI monetisation while limiting premium exposure.
• Establish a 2.5% long position in GOOGL (class A or C) sized for 3–6 month thesis; reduce by 50% if quarterly capex guidance rises >10% QoQ without commensurate cloud revenue acceleration or if GCP revenue growth fails to exceed 20% YoY in the next two quarters.
• Open a relative-value pair: long MSFT (2% notional) / short AMZN (1% notional) over 3–9 months to express differential in enterprise AI monetisation; close if AMZN outperforms MSFT by >15% in 60 days or if AMZN announces major AWS enterprise wins.
• Rotate 1.5–2% into semiconductor equipment (ASML, LRCX) and NVDA exposure: buy 6–12 month call spreads on NVDA to participate in sustained GPU demand; sell or reduce cyclical consumer/retail tech exposure (META, selected retail names) by 2–3% where multiple compression is driven by discretionary ad weakness.
• Set hard monitoring triggers: if OpenAI funding ≥$100bn at >$800bn valuation completes within 90 days, add 50% to MSFT/GOOGL positions; if it fails or if either company’s cloud margin falls >300bp QoQ, pare positions by 30–50% within 30 days.