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

Equinix, Digital Realty and Prologis are positioned to benefit from large AI-driven data-center investment: Equinix reported record annualized gross bookings and guided to double-digit revenue growth for 2026, Digital Realty delivered +10% core FFO in 2025 with an all-time backlog, and Prologis posted its best-ever quarter for lease signings. Valuation metrics look attractive versus headline AI names: Equinix ~24x FFO with a 2.0% yield, Digital Realty similar FFO multiple with a 2.8% yield, and Prologis ~21x expected 2026 FFO with a 3.2% yield.

Analysis

Data-center landlords are benefiting from an AI-driven capex wave, but the durable winners will be the owners who control three non-obvious bottlenecks: immediate interconnection density, shovel-ready land/permits, and stable low-cost power arrangements. Owners that can monetize interconnections command stickier revenue per rack and can reprice faster as AI tenants demand colocated peers and low-latency fabrics; conversely, owners lacking that density will increasingly compete on price, compressing wholesale margins.

Second-order supply constraints — medium-voltage upgrades, large-capacity UPS/generators, and specialized mechanical contractors — create a 12–36 month window where balance-sheet scale and permitting relationships convert into real economic optionality. That advantages diversified landlords with global land inventories and low funding costs: they can underwrite longer construction tails and capture outsized IRR on incremental AI deployments while smaller players face multi-quarter delays and cost overruns.

Key risks cluster around three vectors: (1) a hyperscaler capex reset if AI hardware cycles stall or unit economics for inference change, (2) localized power and permitting pushback that can strand projects, and (3) interest-rate driven re-rating of REIT multiples. Near term (months) watch backlog/booking flow and utility contract renewals; medium term (12–36 months) the market will re-price winners vs. those that mis-timed supply additions. The consensus is underestimating embedded industrial-to-datacenter conversion optionality and overestimating pure chip-alpha as the sole avenue to capture AI growth — real estate owners are the leverage point to physical deployment risk, not just a passive beneficiary.

Key Decisions for Investors

• Core long REIT sleeve: Accumulate DLR (size 1–3% portfolio) with 12–36 month horizon; hedge tail risk with a 12-month 7–12% OTM put (10–20% notional) to protect against a capex reset. Rationale: buy optionality to capture delayed supply bottlenecks while limiting downside from macro-driven REIT de-rating.
• Tactical spread to express interconnect moat: Buy an EQIX multi-year bull-call spread (calendar 18–36 months) sized 0.5–1% portfolio, funded by selling near-term NVDA call spreads. This rotates short-term chip volatility into a higher-income, capped-upside bet on colocator re-rating; expected reward is asymmetric if bookings accelerate, capped loss equals premium outlay.
• Optionality overweight in real-estate-to-data conversion: Add PLD (size 1–2% portfolio) with 18–36 month view and sell near-term covered calls to fund incremental purchases. Thesis: land + permitting optionality compresses time-to-market for AI deployments; covered calls improve carry while retaining upside to conversion-driven rent re-pricing.
• Risk management guardrail: If bookings or utility-contract disclosures miss expectations over a single quarter, trim REIT exposure by 30% and redeploy into short-dated cash or short volatility; conversely, add to winners on confirmed multi-quarter booking beats. Set tactical stop-loss/hurt threshold at a 20–25% drawdown per position unless hedged.