Forget Nvidia. This ETF Could Be the Next Big Winner From the AI Boom

Article argues AI’s power bottleneck could make electric utilities beneficiaries, highlighting Nvidia’s reported plan to support OpenAI’s Ohio data center with over $100B and eventual 8GW capacity. It frames the Vanguard Utilities ETF (VPU)—holding ~70 utilities such as NextEra, Duke, Southern, Constellation, and AEP—as a way to gain AI-linked electricity-demand exposure, citing a ~2.7% dividend yield. Key risks noted are utility regulation limiting returns on capex and community backlash against new data centers.

Analysis

The market is beginning to price AI electricity demand as a real load-growth story, but the first-order beneficiaries are not all equal. CEG has the cleanest near-term torque because wholesale power pricing can re-rate faster than regulated rate-base stories, while AEP/NEE are better through the slower capex and transmission cycle. SO/DUK likely see less immediate upside because the equity market will discount political friction and regulatory lag before it discounts incremental load.

The bigger second-order winner may be the grid and equipment ecosystem, but the utility equity capture is uneven: if data-center customers lock in direct PPAs, behind-the-meter generation, or self-build substations, the economic rent bypasses traditional utilities. That makes this theme less about demand volume and more about who owns the bottleneck assets with approved returns on capital. The key variable is whether state commissions allow enough ROE and timing recovery to turn capex into EPS rather than just more assets.

Contrarian risk: investors may be extrapolating a multi-year load wave into near-term earnings too aggressively. In the next 1-3 months, the main failure mode is sentiment rotation out of defensives once rates rise or AI capex growth decelerates; over 6-18 months, the real falsifier is utility under-earning versus allowed returns, or data centers self-supplying power and weakening utility capture. If that happens, VPU/XLU can stall even if the AI buildout itself remains intact.

Net: this is a valuation-sensitive thematic, not a blanket overweight. The best expression is selective long exposure to utilities with the best regulatory and capex conversion, rather than a broad ETF basket at this point.

Key Decisions for Investors

• Long NEE on pullbacks over the next 2-4 weeks as a quality growth utility proxy; prefer it over the broad utility ETF if the market is paying up for AI load exposure. Falsifier: any guidance that capex growth is not translating into rate-base expansion.
• Pair trade: long CEG / short VPU or XLU for 1-3 months. CEG has the most direct leverage to tightening power markets, while the ETFs will dilute that upside with slower-growth regulated names. Falsifier: utility commission language implying faster-than-expected ROE recovery or broad sector multiple expansion.
• Long AEP into 6-12 month grid-capex recognition, but size modestly; the thesis is EPS compounding from transmission buildout, not a quick re-rating. Falsifier: delayed rate cases or capex overruns that push ROIC below allowed returns.
• Use SO and DUK as relative shorts vs CEG/NEE if the market becomes indiscriminate. These are the names most vulnerable to the 'good story, slow monetization' problem. Falsifier: a clear acceleration in approved load additions or constructive state-regulatory outcomes.
• Watchlist alert: if VPU/XLU rallies another ~5-7% without corresponding earnings revisions, fade the move rather than chase it. The catalyst path is slow and should be validated by rate-base and EPS revisions, not just AI narrative momentum.