Berkeley Lab Puts 1-5 MW Solar at a 45-Month Median, While New York’s Mid-Size Queue Fell to 7.1 Months
- Median to permission, 1-5 MW solar: 45 months
- New York mid-size median, 2024: 7.1 months
- Applications analyzed: 2.7 million
- Projects above 100 kW still in process: 76%
- States with upgrade cost data: 4
Solar projects of 1 to 5 MW took a median of 45 months to reach permission to operate between 2020 and 2025, against two months for residential systems under 30 kW. Lawrence Berkeley National Laboratory published those figures in September 2026 in “Interconnection Timelines and Costs for Distributed Energy Projects in the United States (2000-2025),” a study of more than 2.7 million applications funded by the Department of Energy’s Interconnection Innovation Exchange. The slower number travelled further than the trend underneath it. In New York, the state with the longest large-project queue, median timelines for mid-size projects of 50 kW to 1 MW fell from 22.7 months in 2020 to 7.1 months in 2024.
The Data
Berkeley Lab assembled the dataset from utilities, public utility commissions, state energy offices and Interconnection.fyi. Solar accounts for about two-thirds of the sampled capacity at 77.5 GW, hybrid projects for 13.8 GW and standalone storage for 8.8 GW. Systems under 30 kW make up 97.3% of applications by count. Projects of 1 to 5 MW hold the largest share of capacity at 37.4%.
Size drives the wait. Storage overall reached permission to operate in a median of five months and hybrid projects in four, but storage of 1 to 5 MW took 26 months. Between 2020 and 2025, 46% of projects under 30 kW cleared in 31 to 90 days, while fewer than 1% of projects from 100 kW to 10 MW did. Across the same period 8% of projects under 100 kW were still in process, against 76% of projects above 100 kW and 89% of those between 500 kW and 5 MW.
Why It Matters
The Interconnection Innovation Exchange set 2030 targets of a one-day median from request to agreement for small systems, with 99% of those applications executing an agreement. Against a two-month median for sub-30 kW solar, that target is a long way off even in the fastest segment. For anyone filing above 100 kW, the operative figure is not the 45-month median but the 76% still-in-process share, because it describes a queue with no exit date attached. The practical reading is to source timeline assumptions from the most recent year in your own state’s table, and to treat any national cost figure as a New York artifact until another state reports.
What It Means
The aggregate medians are backward-looking, and the state tables show it. Berkeley Lab reports a 16.0-month median for mid-size projects in New York across 2020 to 2024, but the yearly column runs 22.7, 21.0, 16.7, 12.6 and 7.1. Massachusetts falls from 19.0 months to 8.6 over the same years, and Rhode Island from 21.4 to 10.7. Arizona is the exception and moved the other way, from 7.4 months to 9.4. A developer pricing an interconnection delay off the pooled figure would be budgeting for a queue that three of these four states no longer run.
| State | 2020 | 2024 |
|---|---|---|
| New York | 22.7 | 7.1 |
| Rhode Island | 21.4 | 10.7 |
| Massachusetts | 19.0 | 8.6 |
| Arizona | 7.4 | 9.4 |
The same caution applies to the headline large-project number. New York’s median for projects above 1 MW reads 45.3 months across 2020 to 2024, yet its yearly figures run 48.8, then 42.8, then 28.4, with 2023 and 2024 left blank for small sample size. Rhode Island cleared projects above 1 MW in a median of 7.1 months and Arizona in 10.2. Berkeley Lab also found the days needed to execute an interconnection service agreement fell 26% on average across all project sizes between 2020 and 2025.
The Gap
Berkeley Lab is direct about its data gaps, and the dataset overview table is the part worth reading twice. Application data covers 24 states and 188 utilities. Timeline data from application to permission to operate narrows to 15 states and 28 utilities. Timeline data to an interconnection service agreement narrows again to 8 states and 36 utilities, and New Jersey alone supplies 94% of that sample. California contributes 65% of all applications, followed by New York at 11%, Arizona at 10% and New Jersey at 9%. California is excluded from the agreement analysis because the state waives a pre-construction agreement for small projects.
Upgrade cost data is thinner still, and reading it as a price is a mistake. Berkeley Lab holds usable cost data for four states, Colorado, Connecticut, New York and Rhode Island, and New York supplies more than 99% of it. Average upgrade costs run from $0/kW for solar under 30 kW to $383/kW for solar of 5 to 10 MW. The report attributes the state variation largely to cost-sharing practice rather than to engineering, noting that New York customers paid no upgrade costs on 99% of projects. The national cost range therefore describes one state’s cost-allocation rule.
Sources
- Lawrence Berkeley National Laboratory, “Interconnection Timelines and Costs for Distributed Energy Projects in the United States (2000-2025),” September 2026
- Berkeley Lab Energy Technologies Area, publication record for Interconnection Timelines and Costs
- pv magazine USA, “Distributed solar interconnection timelines stretch to nearly four years for larger projects”
Critical Perspective
The 45-month figure measures more than the utility. Berkeley Lab defines time to permission as the months from application submittal to the utility’s approval, and states plainly that the window includes construction and inspection. A developer’s own build schedule therefore sits inside the number quoted as an interconnection delay, and the report does not separate the two.
The trend reading above carries a caveat the report supplies itself. On its application-year analysis for larger projects, Berkeley Lab writes that sample size is insufficient and that the analysis does not determine whether timelines in recent years are decreasing. The year columns in the state tables are populated for mid-size projects and they do fall sharply, but they record what those states logged rather than a demonstrated national trend. Arizona moving the other way across the same years is the useful reminder.
Coverage is the deeper limit. A dataset described as national rests on New Jersey for 94% of its agreement timelines and on New York for more than 99% of its cost observations, while California supplies 65% of the applications and is excluded from the agreement analysis altogether. Berkeley Lab says all of this openly and calls standardized reporting essential. Until more states report, the honest description is a detailed account of a few jurisdictions rather than a picture of the country.
