Undervalued and Profitable: 3 AI Stocks That Still Fly Under the Radar

Equinix reported record annualized gross bookings in Q4 and guided to double-digit revenue growth for 2026, sending the stock up roughly 10% on the print. Digital Realty posted core FFO up 10% YoY in 2025 with an all-time high backlog, while Prologis delivered its best-ever quarter for lease signings as it pivots into data centers using extensive land holdings and lower borrowing costs. Valuations appear attractive relative to headline AI names: Equinix ~24x FFO (2% yield), Digital Realty similar multiple (~2.8% yield), and Prologis ~21x expected 2026 FFO (3.2% yield), making these three firms plausible ways to gain AI infrastructure exposure.

Analysis

Data-center exposure is a play on two linked convexities: hyperscaler AI capex (demand for racks and interconnects) and real-estate scarcity for powered, permitted sites. Expect the largest margin expansion where operators can (a) monetize cross-connect density and premium interconnection services, and (b) sign multi-year wholesale deals with price escalators — those revenue streams de-rate interest-rate sensitivity by converting growth into contracted cash flows over 3–7 years.

Second-order winners include heavy electrical infrastructure suppliers (transformers, UPS, gensets) and municipal permitting/utility upgrade contractors because lead times for grid upgrades often exceed building cycles; bottlenecks there create optionality and pricing power for landlords with shovel-ready land. Conversely, equipment vendors whose lead times shrink (modular, standardized pods) can cap new-build pricing and accelerate supply, pressuring returns if construction capacity grows faster than contracted demand.

Key risks are macro (rate normalization reversing valuation spreads), demand sequencing (a 6–18 month slowdown in hyperscaler bookings if chip cycles or model efficiency gains lower utilization), and local/regulatory constraints (carbon/utility curtailments that delay builds). Near-term catalysts to watch: quarterly booking conversion rates, wholesale backlog turn-to-revenue in the next 2–4 quarters, and municipal grid upgrade approvals — each can move FFO visibility and compress/expand spreads materially over 3–12 months.

Key Decisions for Investors

• Long DLR (12–24 month equity or LEAPS): Initiate a 2–4% portfolio position via shares or a diagonal LEAP call to capture wholesale backlog conversion; target 30–40% upside if FFO growth reconfirms and spreads tighten, with a 15–20% stop if bookings-to-revenue conversion misses for two consecutive quarters.
• Relative pair — Long PLD / Short EQIX (6–18 months): Size 1:1 to express capital-cost and land optionality over interconnection cyclicality. Use PLD Jan-2028 calls financed by EQIX Jan-2028 call sales (call spread) to limit downside; objective is 20–30% relative outperformance if financing costs stay lower for PLD and EQIX faces pricing pressure from direct hyperscaler builds.
• Credit/carry trade (3–5 years): Buy investment-grade bonds or preferreds of top-cap REITs with strong contracted cash flow (DLR/PLD) to harvest 3–6% real carry and potential 50–100bp spread compression if AI-driven capex sustains. Risk: rates shock or covenant deterioration — keep duration 4–7 years and hedged duration via interest-rate swaps if rates move >100bps.
• Tactical hedge (3–12 months): Buy short-dated puts on data-center equities or purchase power-utility exposure (grid upgrade/renewables names) as insurance against an AI demand pause. Allocate 0.5–1% to insurance instruments; expected cost <1.5% of portfolio for protection against a >20% drawdown in data-center names within 12 months.