ISO New England Adds First Data Center Load Forecast: 285 MW in Study
- Analysis Capacity & Wholesale Markets Policy & Markets
Marin Holt
- 0
- 6 minutes read
- Projects in formal study: 2 (up to 285 MW)
- Peak contribution in the 2040s: 130 MW
- Annual energy use, 2030-2035: 800 GWh
- Debuts in: 2026-2035 CELT report
ISO New England launched its first dedicated large-load demand forecast on May 1, 2026, and the early numbers are deliberately modest: the grid operator counts just two data center or crypto-mining projects far enough along to study, totaling at most 285 megawatts, and projects they will add only about 110 megawatts to regional peak demand through the 2030s.
The Data
The forecast debuts in ISO-NE’s 2026–2035 Capacity, Energy, Loads and Transmission report, known as CELT, and rests on surveys of New England transmission owners. Two proposed large loads sit in the formal study phase, with combined demand reaching 285 MW at the high end, and neither is expected to move overall system demand before the winter of 2027/2028. Over the longer horizon, ISO-NE projects large loads adding roughly 110 MW to summer and winter peaks in the 2030s and about 130 MW in the 2040s, while consuming around 800 gigawatt-hours of energy a year from 2030 to 2035 and climbing toward 1,000 GWh in the 2040s.
| Metric | 2030s | 2040s |
|---|---|---|
| Peak demand contribution | ~110 MW | ~130 MW |
| Annual energy use | ~800 GWh (2030–2035) | ~1,000 GWh |
| Projects in formal study | 2 projects, up to 285 MW combined | — |
Why It Matters
New England’s restraint is the story. While PJM and ERCOT field hundreds of gigawatts of speculative data center interconnection requests, ISO-NE’s verified pipeline is two projects and a few hundred megawatts — a reminder that headline queue figures elsewhere are mostly optionality, not committed load. Folding large loads into the CELT report also pulls them into the same planning machinery that drives transmission upgrades and the Forward Capacity Market, so a 285 MW campus now appears where planners price and reserve against it instead of discovering it late. Load Forecasting Supervisor Victoria Rojo was candid about how new the work is: “There’s a lot of work going on to get up to speed in this domain, and that’s going to continue as the forecast evolves.”
The Gap
The forecast measures what transmission owners report, which leaves anything pre-application invisible. A hyperscaler scouting sites in Maine or New Hampshire without a signed interconnection request does not appear, and Maine regulators are already moving to pause data center approvals before demand lands. The CELT figures also say nothing about how these loads behave electrically — whether they ride through voltage dips or trip offline during contingencies, the failure mode ERCOT flagged in Texas this spring. A megawatt total is a planning input, not a reliability model.
Critical Perspective
Skeptics will note the numbers look small precisely because the survey-based method lags reality. New England’s load-growth forecast has been revised modestly year to year, and forecasters across the country have undershot data center demand for three years running. One gigawatt-scale campus landing in the region would overwhelm the 130 MW figure for the 2040s overnight. The framework’s value is not its current totals — it is that New England finally has a defined place to put the next number when it arrives.
