Must-Buy Non-Tech Stocks for 2026 Amid AI-Driven Data Center Boom

Four of the “magnificent 7” plan roughly $380 billion of AI infrastructure capex in 2025 (a 54% YoY increase) with further increases expected in 2026, driving strong demand for data-center-related services and equipment. Zacks highlights five non-technology plays to capitalize on that demand: Comfort Systems (FIX: next-year rev +14.7%, EPS +16.4%; Zacks EPS est. +21.1% in 60 days), Vertiv (VRT: rev +20.7%, EPS +26.3%; NVIDIA partnership cited), Sterling Infrastructure (STRL: rev +19.1%, EPS +14.6%; Q3-2025 E-Infrastructure revenue $417.1M, total backlog $2.6B), Dominion Energy (D: rev +6%, EPS +5.9%; MOU with Amazon on SMR development) and Alcoa (AA: rev +3.1%, EPS +3.1%; aluminum exposure to cooling/structures and potential site sales).

Analysis

Market structure: The $380B lift in 2025 capex (+54% YoY) by hyperscalers (NVDA/AMZN-linked builds) creates a long-duration demand shock for precision cooling, UPS/power, power-hungry metals (aluminum) and construction services — winners are VRT, STRL, FIX and AA; legacy commodity-exposed suppliers and commoditized HVAC contractors risk margin compression. Pricing power will be strongest for modular, GPU-generation-ahead power/thermal vendors (Vertiv) and turnkey E-infra integrators (Sterling); expect multi-year multi-year backlog visibility to compress time-to-contract and support 10–30% higher ASPs in 2026–27 for specialized gear.

Risk assessment: Tail risks include a 20–40% hyperscaler capex pause from macro recession or revenue shock, regulatory limits on incentives for data-center siting (local power/permits), or a sudden drop in aluminum prices if Chinese capacity ramps — each could cut revenues 15–35% for exposed names in 6–18 months. Near-term (days–weeks) volatility will track NVDA/AMZN commentary and power-price prints; medium-term (3–12 months) risks are backlog cancellations and energy policy; long-term (2–5 years) risks include vertical integration by hyperscalers cutting vendor margin.

Trade implications: Direct idea — overweight VRT (infrastructure + NVIDIA partnership) and STRL (E-infra backlog up 97% YoY) with 12–18 month horizons; use LEAP call spreads to cap cost. Defensively, hold D (Dominion) for regulated cash flows as a 2–3% ballast and consider AA as a 6–12 month commodity play if LME aluminum >+10% y/y. Hedge portfolio beta with short-term power-price or industrials exposure and buy 3–6 month puts on high-beta contractors to limit drawdowns.

Contrarian angles: Consensus underestimates energy constraints and site-permit friction — expect select hyperscalers to monetize closed heavy-power industrial sites (Alcoa remarketing) which benefits AA short-term but could reduce STRL/VRT service TAM if techs insource buildouts. History: telecom buildouts (2000s) produced localized oversupply and margin squeezes after irrational build; a 20–30% pullback scenario is credible if capex growth normalizes to <20% YoY. Look for mispricings where stock price already reflects perfect execution; prefer companies with >2yrs backlog (STRL) and product differentiation (VRT) over momentum-only winners (FIX) when entering.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in VRT via 12–18 month LEAP call spreads (buy Jan 2027 calls, sell a higher strike) to target +30–50% upside while capping premium; reassess at NVDA architecture announcements or if VRT backlog growth <+30% QoQ.
• Initiate a 1.5–2% long in STRL equity (or equivalent options) targeting +40% in 9–12 months given E‑Infrastructure backlog (+97% YoY); size position with a protective 3–6 month put or a stop-loss at -20% from entry.
• Allocate 1% to AA as a commodity exposure (buy shares or 6–12 month call spread) if LME aluminum is >+5% vs 6-month average or if US industrial power discounts persist; exit if aluminum futures drop >15% or if US power costs jump >20%.
• Hold a 2–3% defensive position in D for regulated cash flow and dividend yield; trim to lock gains if long-term SMR milestones slip beyond 12 months or if regulatory approvals stall (>6 months delay).
• Implement a pair trade: long STRL (1.5%) vs short XHB or a broad construction ETF (1.5%) to isolate data‑center E‑infrastructure growth; unwind if STRL backlog growth slows to <20% YoY or XHB outperforms by >10% over 60 days.