AI is Driving Utilities to Spend a Record $240 Billion in 2026. Buy These Stocks to Capitalize on the Power Surge.

Industry watchers estimate utilities may spend up to $240B in 2026 to meet AI-driven electricity demand as power needs rise sharply (10% growth from 2005-2025 vs. ~60% expected from 2025-2045), pushing up power prices and triggering consumer/regulatory pushback. The article argues regulated utilities could face rate-approval constraints, favoring alternatives like Bloom Energy’s $20B total backlog (but ~29x P/S and no sustained profits) versus more valuation- and dividend-friendly options such as Brookfield Renewable (≈4.5% yield, 1.6x P/S) and NextEra (≈2.8% yield; proposed Dominion acquisition; ~9% earnings growth and ~6% dividend growth annually). Overall, AI demand is a clear tailwind, but investors are advised to be selective due to valuation and rate-regulation risk.

Analysis

The market is likely to overpay for the most visible “pick-and-shovel” name while underappreciating the financing and execution penalty that comes with utility-scale load growth. If data-center demand forces utilities into a multi-year capex supercycle, the first-order winner is not the regulated utility itself but the asset owners that can lock in contracted cash flows without waiting for rate-case approval. That argues for a preference for contracted renewables and behind-the-meter power over pure regulated exposure, especially where utility earnings are still hostage to regulatory lag.

The second-order risk is that power scarcity becomes a tax on AI deployment, not just on utilities. If interconnection queues and local opposition persist, hyperscalers can still win by contracting private generation, but smaller data-center builders and colocation operators face delayed go-live dates, higher carrying costs, and tighter project finance terms. That creates a relative-benefit trade in large balance-sheet winners like MSFT/GOOGL versus more rate-sensitive or capacity-constrained AI infrastructure plays, with the caveat that valuation discipline matters because the “AI power” narrative is now crowded.

Contrarianly, the most obvious momentum name may be the weakest risk/reward: backlog growth does not equal durable equity value if gross margins, working capital, or financing costs normalize poorly. Conversely, NEE and BEP/BEPC look like better compounding vehicles if they can monetize long-duration contracts, but the key watch item is whether capital costs stay high enough to compress returns on new projects. The thesis is falsified if regulators allow rapid pass-through of AI-driven capex and if contracted power pricing fails to stay ahead of rising development costs.

Key Decisions for Investors

• Fade BE strength into rallies: initiate a tactical short or buy put spreads over 1-3 months, because the stock is pricing backlog conversion and margin durability that still need proof. Falsify if management prints accelerating gross margin expansion plus repeat orders without heavier dilution or debt.
• Long BEP or BEPC versus a basket of regulated utilities over 6-12 months. The trade benefits if contracted power remains the cleanest way to monetize AI load, while rate-regulated names remain stuck in approval delay; cut if regulators start granting faster passthrough and utility multiples re-rate.
• Pair trade: long NEE / short a broad regulated-utility proxy if AI-driven capex is being misunderstood as a pure risk. NEE has the better mix of contracted renewables and utility scale, but the short leg protects against the market rewarding the entire sector indiscriminately.
• Add D only as a catalyst-driven watchlist long, not a core position, until regulatory and acquisition timing are clearer. The upside is Virginia data-center optionality; the risk is that deal/regulatory friction and capex intensity cap near-term rerating.