For Data Centers, Power Is the New Real Estate

AI-driven demand for data-center capacity continues to outstrip supply — CBRE flagged record-low vacancies in H2 2025 with customers pre-leasing into 2027+ — while the IEA expects global data-center energy use to double by 2030 and Pew projects US AI-related electricity use to rise ~133% by 2030. That energy bottleneck is prompting hyperscalers to secure generation (including modular nuclear deals involving Microsoft, Amazon and NextEra) and is creating investable opportunities across systems providers (Hewlett Packard Enterprise — forward P/E ~8, price-to-growth ~0.5), data-center REITs (Digital Realty, Equinix) and niche picks-and-shovels plays (Schneider Electric, Comfort Systems, EMCOR, Sterling, Stantec, AAON). Key risks remain execution and overbuild of planned facilities and local regulatory/power-supply pushback, so the podcast advocates diversified infrastructure- and services-oriented exposure rather than concentrated hyperscaler bets.

Analysis

Market structure: Winners are data-center kit and specialist services (HPE, AAON, FIX, EME, STRL, Schneider Electric) and coastal colo REITs with scarce capacity (Digital Realty DLR, Equinix EQIX). Losers are local utility/retail consumers facing higher rates and small generic landlords unable to monetize power access; hyperscalers (MSFT, AMZN, GOOGL) retain control but risk margin pressure if energy costs spike. Supply/demand: completions lag planned buildouts so near-term vacancy remains tight (CBRE H2 2025 lows), supporting pricing power for colo operators and contractors for 12–36 months.

Risk assessment: Tail risks include local permitting moratoria, major hyperscaler capex cuts, or SMR nuclear licensing delays — each could sharply compress demand or raise capex; probability moderate but impact high. Time horizons: immediate (days–weeks) = leasing/earnings reactions; short (3–12 months) = construction completions and PPA deals; long (1–5 years) = energy demand doubling by 2030 and grid/transmission bottlenecks. Hidden dependencies: transmission upgrades, water/cooling availability, transformer/substation lead times and renewable PPA pricing are key second-order constraints.

Trade implications: Tactical ideas: establish value-sized longs in HPE (HPE) and contractor specialists (AAON, FIX or EME) to capture durable hardware/service demand; overweight DLR/EQIX for scarce real-estate cashflows. Use options to lever duration — buy 12–24 month LEAP calls on HPE (or 6–12 month call spreads on DLR/EQIX) rather than short-term gamma. Entry: scale 25% now, add 25% on a pullback >8%, target 30–40% realized upside in 12–24 months or re-evaluate if pre-leasing rates drop below 60%.

Contrarian angles: Consensus fears of “overbuild” underweights the immediate energy bottleneck and transmission scarcity — the market may be underpricing select industrials and utilities with generation controls (NEE/NextEra) and modular nuclear optionality. HPE looks quantitatively cheap (forward P/E ~8 per commentary) and may rerate if order flow sustains; conversely, overexposure to speculative hyperscaler growth names without energy/resilience exposure is a mispricing risk. Unintended consequence: accelerating vertical energy acquisition by hyperscalers could reprice utility and renewable asset valuations over 1–3 years.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in Hewlett Packard Enterprise (HPE) over 12–24 months to capture server/network hardware demand; scale in 25% now and add 25% on any pullback >8%; trim on 30–40% price appreciation or if forward P/E rises above 12.
• Allocate 2–3% across data‑center REITs: 1.5% Digital Realty (DLR) and 1.5% Equinix (EQIX) as income-plus-growth plays; if uncomfortable with outright equity, buy 6–12 month call spreads to limit capital at risk and target 20–40% return if leasing trends continue.
• Buy a 1–1.5% position in picks-and-shovels names AAON (AAON) and one of FIX/EME (Comfort Systems FIX or Emcor EME) split 50/50 to play HVAC/electrical demand; hold 9–18 months and add if quarterly lease/IPD data shows sustained tight vacancy.
• Establish a 1% tactical long in NextEra (NEE) to get exposure to PPAs and potential SMR/nuclear optionality; exit or reduce if federal/state SMR approvals slip beyond 24 months or if utility capex guidance falls >15%.
• Implement a relative pair: long DLR (DLR) vs short American Tower (AMT) at equal notional (1% each) to express preference for data‑center real estate over tower growth; rebalance if DLR preleasing <60% or AMT announces materially higher data‑center exposure.