Knowledge · Vision · Engineering
SKAR RESEARCH BRIEF 003 · ENERGY & INFRASTRUCTURE

Planning for
Load Growth.

A scenario-based decision framework for understanding rapidly rising U.S. data-center electricity demand.

Published July 202610-minute readIndependent research
THE QUESTION

How should planners act when demand growth is large but the forecast range is wide?

SKAR VIEW

The correct response is not to choose a single forecast. It is to identify decisions that remain useful across low, central, and high demand conditions—and to stage commitments around observable triggers.

The U.S. Department of Energy’s 2024 data-center energy-use report, produced by Lawrence Berkeley National Laboratory, describes a sharp change in the electricity-demand outlook. Artificial intelligence is one driver, but the planning problem also includes cloud services, conventional data-center growth, domestic manufacturing, and broader electrification.

01 · The baseline has already moved

DOE reported that U.S. data-center electricity use increased from 58 TWh in 2014 to 176 TWh in 2023. Data centers represented approximately 4.4% of total U.S. electricity consumption in 2023. That historic increase matters because new projects enter a system that is already managing regional constraints, interconnection queues, equipment lead times, and reliability requirements.

176 TWhestimated U.S. data-center use in 2023
4.4%share of total U.S. electricity consumption

02 · The 2028 range is the planning problem

DOE’s reported projection spans 325 to 580 TWh by 2028, equivalent to approximately 6.7% to 12% of total U.S. electricity. The range is not noise to be averaged away. It reflects uncertainty in deployment pace, computing demand, efficiency, regional concentration, and infrastructure response.

LOW CASE325 TWhPrepare capacity with staged commitments.
MIDPOINT*452.5 TWhUse only as a comparison case, not a forecast.
HIGH CASE580 TWhTest transmission, generation, and equipment constraints.

*Arithmetic midpoint of DOE’s published range, calculated by SKAR for scenario comparison.

03 · A trigger-based decision framework

No-regret actions

Improve load visibility, study regional constraints, validate equipment lead times, and define demand-response options.

Trigger points

Link later commitments to contracted load, interconnection milestones, utilization, and observed regional demand.

Option value

Preserve alternatives in site design, phasing, generation mix, storage, and flexible operating schedules.

A good plan distinguishes irreversible decisions from reversible ones. Land acquisition, major transmission, and generation commitments have different lead times and exit costs than modular equipment, staged construction, or flexible load agreements. The analytical task is to align each commitment with the uncertainty it resolves.

04 · Limitations

National electricity-use projections cannot answer a regional interconnection or facility-design question. Local resource adequacy, transmission constraints, weather, water availability, permitting, load shape, redundancy requirements, and technology efficiency must be modeled separately. This brief is a decision framework, not a power-system study.

Primary sources

  1. U.S. Department of Energy, “DOE Releases New Report Evaluating Increase in Electricity Demand from Data Centers,” December 20, 2024.
  2. U.S. Department of Energy, “Clean Energy Resources to Meet Data Center Electricity Demand.”