Original research
American rooftop solar just had its sharpest fall in a decade, and half of it is one state
New residential net-metered connections by state, 2024 against 2023, from federal utility filings.
The finding
Key findings
- 1 New connections fell 30.6% in 2024
From 761,174 to 527,942: a loss of 233,232 connections against the prior year.
- 2 It is the largest fall in the eleven-year series
The only other down years were 2017 (-15.8%) and 2018 (-6.4%). 2024 is larger than both together in absolute and percentage terms.
- 3 California is 49% of the national decline
113,544 fewer connections in a single state, against a national fall of 233,232.
- 4 The seven largest declines are 89% of the total
California, Florida, Virginia, Arizona, Colorado, Texas, New Jersey. This is not a broad market contraction.
- 5 The steepest percentage falls are not the biggest states
Virginia fell 64.4% and Texas 54.4%, among states connecting at least 5,000 systems in 2023.
- 6 13 states grew, and one grew sharply
Massachusetts went from 15,722 to 30,323 connections, +92.9%: the only large state to move materially in the other direction.
- 7 2024 is a sharp fall, not a low level
527,942 connections is +6.4% against 2021 and -16.3% against 2022. The 30.6% headline is measured against a year inflated by a policy deadline.
- 8 The installed fleet still grew
A fall in additions is not a fall in installed systems. The US fleet reached 5,077,543 systems, up 11.6% on the year.
What happened
Between 2018 and 2023 the number of new residential net-metered solar connections in the United States rose every single year, from 285,609 to 761,174, more than doubling in five years. In 2024 it fell to 527,942.
A 30.6% fall is not a wobble. There have been only two other down years in the series: 2017 (-15.8%) and 2018 (-6.4%). 2024 is larger than both together, in absolute systems and in percentage terms.
The cumulative fleet still grew, because a fall in additions is not a fall in installed systems, 5,077,543 American households now have one. Confusing those two is the most common error in coverage of this data, and it runs in both directions: a decline in the growth rate gets reported as solar shrinking, and a growing total gets reported as evidence the market is fine.
New residential net-metered connections, national
| Year | New connections | Change on prior year | Cumulative fleet | Household share |
|---|---|---|---|---|
| 2015 | 316,574 | — | 958,850 | 0.74% |
| 2016 | 362,427 | +14.5% | 1,321,277 | 1.01% |
| 2017 | 305,006 | -15.8% | 1,626,283 | 1.23% |
| 2018 | 285,609 | -6.4% | 1,911,892 | 1.43% |
| 2019 | 371,810 | +30.2% | 2,283,702 | 1.69% |
| 2020 | 377,327 | +1.5% | 2,661,029 | 1.95% |
| 2021 | 496,400 | +31.6% | 3,157,429 | 2.28% |
| 2022 | 630,998 | +27.1% | 3,788,427 | 2.71% |
| 2023 | 761,174 | +20.6% | 4,549,601 | 3.22% |
| 2024 | 527,942 | -30.6% | 5,077,543 | 3.55% |
HyreSolar calculation from EIA-861. Connections are the year-on-year change in the cumulative count each utility reports.
Figures labelled HyreSolar calculation are computed by us from the EIA source files named below. EIA publishes the inputs; it does not publish these ratios.
The concentration is the story
The national number reads like a market-wide contraction. It is not. Of the 233,232 fewer systems connected in 2024, California accounts for 113,544, 49% of the total decline, from a state with 41% of the installed fleet.
Add Florida, Virginia, Arizona, Colorado, Texas, New Jersey and you have 89% of the national fall in seven states. The remaining 31 declining states account for the rest between them.
Where the decline happened
Mapped, the pattern is not regional. States that fell hardest sit next to states that grew. Virginia fell 64.4% while neighbouring states held roughly flat; Massachusetts grew +92.9% in the same year New Jersey fell -22.0%.
That is what you would expect if the driver were state-level policy and financing conditions rather than anything national. It is not what you would expect from an interest-rate story alone, which would hit every state at once.
The ten largest declines, in absolute terms
Ranked by the number of connections lost rather than the percentage, the table is dominated by large states, which is what makes the concentration finding meaningful rather than trivial.
Rises and falls, side by side
Shown around zero, the asymmetry is clear: the declines are far larger than the gains, and the gains are concentrated in a small number of Northeastern and Midwestern states with active state incentive programmes.
New connections by state, every jurisdiction
| State | Connections 2023 | Connections 2024 | Change | Change % | Fleet at end 2024 |
|---|---|---|---|---|---|
| California | 278,068 | 164,524 | -113,544 | -40.8% | 2,090,983 |
| Florida | 61,479 | 39,146 | -22,333 | -36.3% | 289,774 |
| Virginia | 31,011 | 11,038 | -19,973 | -64.4% | 69,553 |
| Arizona | 38,717 | 21,892 | -16,825 | -43.5% | 317,033 |
| Colorado | 34,631 | 19,057 | -15,574 | -45.0% | 179,751 |
| Texas | 24,823 | 11,317 | -13,506 | -54.4% | 133,501 |
| New Jersey | 22,316 | 17,401 | -4,915 | -22.0% | 199,822 |
| New York | 27,943 | 23,177 | -4,766 | -17.1% | 226,979 |
| New Mexico | 9,320 | 4,581 | -4,739 | -50.8% | 61,209 |
| Connecticut | 17,699 | 13,142 | -4,557 | -25.7% | 103,824 |
| North Carolina | 10,617 | 6,421 | -4,196 | -39.5% | 57,393 |
| Pennsylvania | 16,765 | 13,009 | -3,756 | -22.4% | 81,375 |
| Washington | 11,204 | 8,471 | -2,733 | -24.4% | 59,235 |
| Oregon | 10,073 | 7,600 | -2,473 | -24.6% | 52,600 |
| New Hampshire | 5,555 | 3,372 | -2,183 | -39.3% | 22,503 |
| South Carolina | 4,044 | 2,040 | -2,004 | -49.6% | 40,055 |
| Oklahoma | 5,873 | 3,947 | -1,926 | -32.8% | 16,755 |
| Arkansas | 4,661 | 2,750 | -1,911 | -41.0% | 19,146 |
| Rhode Island | 5,003 | 3,297 | -1,706 | -34.1% | 17,321 |
| Idaho | 4,319 | 2,631 | -1,688 | -39.1% | 22,899 |
| Nevada | 20,844 | 19,182 | -1,662 | -8.0% | 134,257 |
| Missouri | 5,316 | 3,848 | -1,468 | -27.6% | 30,022 |
| Minnesota | 5,244 | 3,861 | -1,383 | -26.4% | 22,621 |
| Iowa | 3,795 | 2,488 | -1,307 | -34.4% | 16,392 |
| Vermont | 385 | -664 | -1,049 | -272.5% | 9,193 |
| Kentucky | 2,344 | 1,385 | -959 | -40.9% | 10,069 |
| Michigan | 3,586 | 2,633 | -953 | -26.6% | 23,456 |
| Delaware | 1,944 | 1,126 | -818 | -42.1% | 14,087 |
| Maryland | 9,110 | 8,361 | -749 | -8.2% | 108,428 |
| Kansas | 2,334 | 1,618 | -716 | -30.7% | 8,583 |
| Wisconsin | 3,741 | 3,237 | -504 | -13.5% | 18,581 |
| Nebraska | 665 | 270 | -395 | -59.4% | 2,807 |
| Indiana | 1,000 | 657 | -343 | -34.3% | 11,045 |
| West Virginia | 767 | 594 | -173 | -22.6% | 3,535 |
| Alaska | 388 | 245 | -143 | -36.9% | 2,706 |
| South Dakota | 181 | 165 | -16 | -8.8% | 469 |
| Tennessee | 21 | 8 | -13 | -61.9% | 86 |
| Alabama | 15 | 8 | -7 | -46.7% | 76 |
| North Dakota | 9 | 13 | 4 | +44.4% | 81 |
| Ohio | 5,104 | 5,113 | 9 | +0.2% | 27,520 |
| Wyoming | 397 | 570 | 173 | +43.6% | 2,913 |
| Louisiana | 2,770 | 2,957 | 187 | +6.8% | 34,417 |
| Mississippi | 235 | 608 | 373 | +158.7% | 1,789 |
| District of Columbia | 2,490 | 3,017 | 527 | +21.2% | 17,780 |
| Utah | 7,367 | 8,025 | 658 | +8.9% | 81,115 |
| Hawaii | 4,062 | 4,776 | 714 | +17.6% | 98,418 |
| Montana | 2,030 | 2,806 | 776 | +38.2% | 9,151 |
| Georgia | 6,079 | 7,150 | 1,071 | +17.6% | 18,612 |
| Maine | 2,224 | 4,348 | 2,124 | +95.5% | 14,867 |
| Illinois | 26,884 | 30,401 | 3,517 | +13.1% | 103,145 |
| Massachusetts | 15,722 | 30,323 | 14,601 | +92.9% | 189,611 |
All jurisdictions with a figure for both years, ordered by absolute change. HyreSolar calculation from EIA-861.
Journalists are welcome to lift the row for their state. We ask for a link to this page, because the method and its limits are stated here.
Measured against the right baseline
Comparing 2024 against 2023 alone overstates the fall, because 2023 is not a normal year. It contains the systems pulled forward by California’s April 2023 grandfathering deadline, and it is the highest figure in the series by a wide margin.
A fairer reference is the pre-announcement trend. The US connected 496,400 residential systems in 2021 and 630,998 in 2022. Against 2021, 2024’s 527,942 is +6.4%; against 2022 it is -16.3%.
So the honest range is this: 2024 is down 30.6% on an inflated year, roughly 16.3% on 2022, and still above 2021. All three comparisons are true. Publishing only the largest one would be the easy story and the wrong one, and we would rather state the range than pick the number that makes the best headline.
The decline by region
| Region | Connections 2023 | Connections 2024 | Change | Change % | States that fell |
|---|---|---|---|---|---|
| Northeast | 113,612 | 107,405 | -6,207 | -5.5% | 7 of 9 |
| Midwest | 57,859 | 54,304 | -3,555 | -6.1% | 9 of 12 |
| South | 168,283 | 101,873 | -66,410 | -39.5% | 13 of 17 |
| West | 421,420 | 264,360 | -157,060 | -37.3% | 9 of 13 |
HyreSolar calculation from EIA-861, aggregated to US Census Bureau regions. The District of Columbia is counted in the South.
Every region fell. The West, which contains California, fell hardest at -37.3%; the Northeast fell least at -5.5%.
Why California, and why now
California changed the economics of residential solar on 15 April 2023, when the California Public Utilities Commission’s net billing tariff (universally called NEM 3.0) replaced retail-rate net metering for new customers of the three big investor-owned utilities. Exports are now compensated at avoided-cost values well below the retail rate.
Applications filed before that date were grandfathered onto the old tariff, which pulled an enormous volume of installations forward into late 2022 and early 2023. So 2023 is inflated by the rush and 2024 is depressed by its absence. Both effects are real, and reading either year alone overstates the case.
The utility-level filings let us watch it happen inside each service territory rather than inferring it from a statewide total. We do that here: San Diego Gas & Electric, the smallest of the three, fell hardest, which is what net billing would produce, because SDG&E has the highest retail rates and therefore the most export value to lose.
What else was happening in 2024
Plausible contributors we cannot measure here
Financing costs. Residential solar is overwhelmingly financed, and loan pricing moved sharply over 2023–24. This dataset contains nothing about interest rates or loan terms.
Installer insolvencies. Several large national installers failed or restructured during the period, stranding pipelines. EIA-861 counts interconnections, not the companies that made them.
Federal credit timing. Changes to the residential clean energy credit altered the cash-purchase calculation for systems placed in service in different years.
What the data can and cannot separate
The geographic concentration argues against a purely national cause: an interest-rate shock would depress every state, and 13 states grew.
The timing and the utility-level pattern in California fit the tariff change closely. That is strong circumstantial evidence for NEM 3.0 in California specifically.
It does not extend to the other half of the decline. States with no comparable tariff change account for 51% of the fall, and this analysis cannot apportion that between financing, installer failures and state programme changes.
And where it grew
| State | Connections 2023 | Connections 2024 | Change | Change % |
|---|---|---|---|---|
| Massachusetts | 15,722 | 30,323 | +14,601 | +92.9% |
| Illinois | 26,884 | 30,401 | +3,517 | +13.1% |
| Maine | 2,224 | 4,348 | +2,124 | +95.5% |
| Georgia | 6,079 | 7,150 | +1,071 | +17.6% |
| Montana | 2,030 | 2,806 | +776 | +38.2% |
| Hawaii | 4,062 | 4,776 | +714 | +17.6% |
| Utah | 7,367 | 8,025 | +658 | +8.9% |
| District of Columbia | 2,490 | 3,017 | +527 | +21.2% |
| Mississippi | 235 | 608 | +373 | +158.7% |
| Louisiana | 2,770 | 2,957 | +187 | +6.8% |
| Wyoming | 397 | 570 | +173 | +43.6% |
| Ohio | 5,104 | 5,113 | +9 | +0.2% |
| North Dakota | 9 | 13 | +4 | +44.4% |
Every jurisdiction where new connections rose. HyreSolar calculation from EIA-861.
Methodology
How connections are counted
EIA-861 is a stock survey. It asks each utility how many net-metered residential customers it has at year end, not how many it connected during the year. New connections are therefore derived by differencing consecutive years. This is the standard approach and the only one the survey supports.
It inherits three things. A utility that reclassifies accounts shows a jump. A utility that fails to file shows a collapse. And a system that is disconnected, or moves off net metering, reduces the apparent additions. None of these is separable within the data.
The Vermont anomaly, reported not smoothed
Vermont’s reported count falls by 664 in 2024, which almost certainly reflects a reporting or reclassification change rather than systems being removed from roofs. We publish it unsmoothed and flag it here. Silently dropping the row would make the national total wrong; silently correcting it would make the method unreproducible.
Inclusion floor on percentage rankings
The steepest-percentage table is restricted to states that connected at least 5,000 systems in 2023. Without a floor, a state going from 40 connections to 12 produces a −70% headline that means nothing.
What is excluded
Residential PV that is not net-metered does not appear. That matters most in Texas, where 64% of reported residential PV capacity sits outside net metering, so the Texas decline shown here is a decline in net-metered connections and should not be read as the whole Texan market. The size of that gap, by state, is measured here.
The 2025 early release is excluded. It covers a partial set of filers and using it would show a spurious collapse of more than half.
What we are not claiming
We measured what happened, not why. NEM 3.0 is the documented policy change with the right timing and the right geography to explain the California figure, and the utility-level pattern is consistent with it. Interest rates, installer insolvencies and federal credit timing all plausibly contributed elsewhere. This analysis cannot apportion between them, and does not try.
Terms used on this page
- New connection
- The year-on-year increase in a utility’s reported count of net-metered residential customers. An approximation of newly interconnected systems, net of disconnections.
- NEM 3.0 / net billing
- California’s successor tariff to retail-rate net metering, adopted in CPUC Decision 22-12-056 and effective for interconnection applications from 15 April 2023.
- Grandfathering
- The provision keeping customers who applied before a tariff deadline on the old terms, here, for 20 years. It is what caused applications to be pulled forward into 2023.
- Installed fleet
- The cumulative number of systems connected. It grew in 2024 even though additions fell, because additions remained positive.
Citation, reuse and corrections
How to cite this study
Full citation. HyreSolar Research, “American rooftop solar just had its sharpest fall in a decade”, September 2026. Analysis of US Energy Information Administration Form EIA-861, 2014–2024. Available at https://hyresolar.com/research/rooftop-solar-slowdown/
In text. “according to a HyreSolar analysis of federal utility filings” — with a link to this page.
In a chart or table. “Source: HyreSolar analysis of EIA-861 (2024)”.
What you may reuse
The underlying data is a public US government dataset and carries no restriction. The analysis, rankings, derived ratios and charts on this page are ours, and you are welcome to reproduce them — including the charts — for editorial, academic and non-commercial purposes with attribution and a link to this page.
We ask for the link rather than a bare mention because the methodology and the limitations live here. A figure quoted without them is easy to misread, and several of the numbers on this page carry conditions that change what they mean.
Who produced this
The HyreSolar research desk. We do not attach an individual byline to these studies, because the work is a scripted analysis of a public federal dataset rather than an authored opinion, and a personal byline would imply a kind of authorship that is not what happened here. What is accountable instead is the method: the source files are named, the arithmetic is stated, the extraction is scripted, and the validation is published.
HyreSolar is an independent analysis and matching service. We are not an installer, a lender or a utility, and no installer pays for placement, ranking or mention in this research. See the editorial policy.
How this study is built
Annual Form EIA-861 workbooks for 2014–2024 are downloaded from EIA and parsed by script into a single dataset. Every figure on this page — in the prose, in the tables and in every mark on every chart — is read from that dataset at build time. Nothing is typed by hand.
That is not a stylistic preference. It means a number in a sentence and the same number in the table beneath it cannot drift apart, a chart cannot disagree with its own caption, and next year's EIA release updates the entire study by regenerating one file rather than by someone editing 4,000 words and hoping they caught every instance.
The workbooks are not consistent between years — sheet names change, header rows move, a measure is renamed, a column appears in one year only, and one large utility is filed under two different spellings. The extraction addresses columns by their header meaning rather than their position, and keys utilities on their EIA number rather than their name, because every one of those inconsistencies silently produces wrong output if ignored.
Corrections
If you find an error, tell us and we will fix it on the page with a dated note rather than silently. That includes disagreements about method: the inputs are public and the arithmetic is stated, so the argument can be had on the evidence.
Update schedule
EIA publishes final Form EIA-861 data for a year in approximately October of the following year. This study is rebuilt against the new release and republished at the same URL, so links do not break and the accumulated citations stay attached to the current numbers.
Questions
Did rooftop solar decline in 2024?
Is this because of NEM 3.0?
Which state fell the most?
Did any state grow?
Was this an interest-rate effect?
Does this mean rooftop solar is in decline?
When will the next year of data be available?
Written and audited by
HyreSolar Research
Primary-source research, data analysis and fact checking
We are a research desk, not a sales floor. We read the statute, the tariff, the code section, the federal filing or the manufacturer data sheet ourselves, and we publish the figure with the document it came from and the date we retrieved it. Where a number cannot be traced to a primary source, we publish the shorter page and say what we could not verify. That rule has cost us whole sections, and it is the reason the rest can be trusted.
- 160
- primary sources read and cited
- 220
- figures with a retrieval date
- 115
- federal and state government sources
- 66
- researched pages published
How this desk works
- Primary sources only. Statutes from the legislature’s own publishing system, federal data from the agency that collects it, code text from the adopted edition, manufacturer claims from the data sheet. We do not cite an article that cites a source; we go and read the source.
- Every figure carries its provenance. A named document and the date we retrieved it, so you can check it and so you know how old it is. Retrieval dates are not decoration: an EIA rate from May is a different fact from an EIA rate from August.
- We publish what we could not verify. Every research page carries a section naming the things we tried to establish and could not, and why. A paywalled standard, a state website that refused the request, a manufacturer that publishes no figure at all.
- We separate measurement from modelling from our own reasoning, and label which is which on the page. A laboratory measurement, an assumption inside a modelling tool and our own inference are three different kinds of claim and they are never presented as one.
- We do not sell solar, and we take no payment for placement, ranking or a favourable mention. Nobody buys a position on this site.
Data as of EIA-861 2024 final release. Authorship on this site is organisational: the analysis belongs to the desk rather than to a named individual, and we do not publish credentials we do not hold. Our editorial policy sets out how we source, date and correct what we publish.
Sources & retrieval dates
- US EIA, Form EIA-861 — Net Metering (annual files, 2014–2024) — Utility-level net-metered capacity, installations and energy sold back by state and customer sector, plus PV-paired battery capacity and installations from 2023. Downloaded as the published annual ZIP archives. Retrieved 2 September 2026.
- US EIA, Form EIA-861 — Sales to Ultimate Customers (annual files, 2014–2024) — Utility-level residential revenue, sales and customer counts by state, used to derive the average residential price and to count the households a state actually meters. Retrieved 2 September 2026.
- US EIA, Form EIA-861 — Distributed Generation that is not Net Metered (2024) — Residential photovoltaic capacity served under buyback, feed-in and utility-owned arrangements rather than net metering. Capacity only; this schedule collects no installation count. Retrieved 2 September 2026.
- US EIA, Average Price of Electricity to Ultimate Customers by End-Use Sector — EIA’s own published state price series, used only to validate our derivation. Agreement across all 357 overlapping state-years is within 0.005¢/kWh, i.e. EIA’s own rounding. Retrieved 2 September 2026.
Does solar still pay where you live?
A national slowdown is not a verdict on your roof. The model uses your utility’s current tariff.
HyreSolar is an independent analysis and matching service. We are not an installer, lender or utility. When a reader asks to be introduced, installers may pay us a referral fee. That fee never buys ranking, scores or placement in research. Our editorial policy sets out the rules.