AI Power Demand Doubles By 2027: Here's Ways To Position

The first-order winner is not the utility sector as a whole, but the parts of the power stack with the shortest conversion from demand to cash flow: grid equipment, transformers, switchgear, EPC, and independent power producers with existing assets. Names like ETN, PWR, HUBB, CEG, and NRG should monetize AI load growth faster than regulated utilities because they can reprice backlog and capacity now, while utility EPS usually lags behind load announcements by 2-4 rate cases.

The main second-order effect is that higher load growth tightens the supply of everything upstream: interconnection gear, gas pipelines, and dispatchable generation. That creates a favorable setup for companies with scarce assets and balance-sheet flexibility, but it also raises the risk that utilities fund capex with expensive equity or debt, which can mute the advertised growth. For XLU, the market may overestimate the purity of the growth story: rate-base expansion is helpful only if regulators allow timely recovery and financing costs do not absorb the uplift.

Contrarian view: the consensus is probably too early on the utility re-rating. AI power demand is real, but most of the earnings impact arrives over 6-18 months, not next quarter, and the immediate beneficiaries are more likely suppliers than regulated distribution names. The key falsifier is any slowdown in data-center build plans, a drop in hyperscaler capex, or a sharp rise in long-end rates that forces utilities to issue more equity than expected. If that happens, the trade shifts from growth multiple expansion to a capital-intensive trap.