Forget Chips: AI Is Now a Power Trade. These 2 Dividend-Paying Industrials Prove It.

NextEra Energy (valued >$178B) could become the world’s largest regulated utility if its $67B all-stock merger with Dominion Energy closes in 2H 2027, pending regulators. The deal is described as boosting NextEra to 110GW of power (including much nuclear), supporting structural electricity demand from AI data centers, electrification, and industrial reshoring. The article also highlights dividend strength: NextEra raised its quarterly dividend by 10% in 2026 and has increased it for 31 straight years (yield ~2.8%), while Oneok’s dividend yield is ~4.8% with >78% cumulative growth over the past decade.

Analysis

The market is likely to overpay for the most obvious “AI power” names while underpricing the infrastructure bottleneck that actually monetizes the trend: interconnection rights, permitted sites, and firm delivery. That makes NEE and OKE qualitatively better positioned than generic utilities because they can sell reliability, not just electrons or molecules. The second-order winner is not necessarily the lowest-cost power producer; it is the company that can sign long-duration contracts before competitors can build transmission or permit new supply.

For NEE, the real upside is not the dividend story but the possibility that it becomes a gatekeeper for large-load demand. The main risk is that this narrative gets capitalized well before cash flow is visible, while utility equity issuance, higher-for-longer rates, or a slower approval path on the Dominion deal compress the multiple. D has event optionality, but the 2027 close date means the market will likely trade the stock on regulatory headlines rather than fundamentals for the next 12-18 months.

OKE is the cleaner near-term monetization vehicle because its fee-based model can capture incremental load growth without needing a full re-rating of the power sector. The contrarian miss is that much of the data-center buildout may be served by on-site generation, nuclear, or direct utility PPAs, limiting how much incremental volume truly flows through pipelines. If signed load commitments disappoint over the next 1-3 quarters, the AI-infrastructure premium could fade quickly; if they accelerate, the rerating should first show up in the toll collectors and then in the higher-duration utilities.

Key Decisions for Investors

• Prefer OKE over broad utility exposure: long OKE / short XLU for a 1-3 month relative-value trade, targeting AI-load monetization with less rate-regulatory duration risk. Falsify if OKE fails to convert data-center discussions into signed fee-based contracts.
• Use NEE only on weakness, not chase strength: wait for a pullback or for evidence of actual PPA backlog conversion before adding. The risk/reward is better if the stock de-rates on rates or merger skepticism and then rerates on confirmed load wins.
• Treat D as a regulatory optionality watch item, not a high-conviction long: long D only if the market offers a wider-than-normal spread versus regulated utility peers and regulatory approvals remain on track. Falsify if state/federal review language turns restrictive.
• If looking for a hedge against the AI-power optimism trade, short overextended renewable-development names or higher-duration utility proxies that lack firm data-center contracts. The thesis breaks if they announce multi-year, take-or-pay load deals.
• Set a catalyst alert for the next 1-3 quarters of hyperscaler load announcements: if large-load signings slow, reduce exposure to both NEE and OKE, since the market will likely have already priced in the narrative before cash flow follows.