US power demand to climb through 2027 on data centers, EIA says

EIA projects US electricity consumption will rise to 4,269 billion kWh in 2026 and 4,399 billion kWh in 2027 (from 4,195 billion kWh in 2025), driven by data centers for AI/crypto and further electrification. Commercial power sales are forecast to surpass residential in 2026 (1,550 vs. 1,508 billion kWh). Fuel mix shifts modestly, with coal down from 17% (2025) to 15% (2026–27) while renewables increase to 27% by 2027 and natural gas holds at 40%.

Analysis

The investable signal is not simply “more electricity,” but a reallocation of margin to the bottlenecks that can deliver firm power and expand the grid. Roughly 200 billion kWh of incremental demand across 2025-27 is material, but the first money flows to transmission, transformers, switchgear, turbine OEMs, and merchant generators with existing interconnection; regulated utilities will see the benefit later through rate-base growth, not immediately through EPS.

The gas read-through is more nuanced than it looks: flat generation share with rising absolute volumes means upstream gas is not the clean winner, while pipeline and storage names with exposure to power-burn and industrial load get a steadier demand floor. Coal is the obvious structural loser, but the bigger second-order loser may be hyperscalers if localized power scarcity raises delivered electricity costs faster than AI monetization improves.

Contrarian point: the market tends to over-own the broad “AI power” trade and under-own the constrained supply chain behind it. Lead times on transformers, turbines, and interconnects can keep this theme alive for 6-18 months even if demand estimates wobble; the near-term reversal risk is a lower-than-expected data-center build pace, falling power prices, or regulatory delays that push out capex. For NGS and UUUU specifically, this is not yet a direct catalyst set—both are better treated as watch items until there is evidence of higher field activity or nuclear contracting, not as immediate beneficiaries.

Key Decisions for Investors

• Long PWR / short XLU for 1-3 months: best expression of grid bottleneck spending versus slow-moving regulated utility earnings; exit if utility capex guidance disappoints or rates fall sharply.
• Buy ETN or HUBB on pullbacks and pair against broad market utilities: 2-4 quarter setup from transformer, switchgear, and data-center electrical capex; thesis breaks if order books soften or backlog growth slows.
• Long CEG or VST on weakness for 6-18 months: merchant power and nuclear-backed firm supply should capture rising commercial load and tighter regional capacity; cut if forward power curves roll over or data-center project timelines slip.
• Short coal exposure via BTU or CEIX over 3-6 months: declining share plus ongoing fuel substitution creates a cleaner secular short than broad energy; thesis weakens if gas spikes enough to force coal burn back into dispatch.
• No trade on NGS or UUUU yet; keep as alerts only until there is verifiable evidence of volume acceleration, contract wins, or policy catalysts that convert the macro demand story into company-level earnings.