Skip to content
US installed cost $3.75 per wattPower 17.45¢/kWh · July 2026
SolarPanelCosts.comRun the numbers
How this is built

Every source, and every thing it cannot tell you

A payback figure is not a fact about the world. It is the output of a model, and a reader who cannot see the model has no way to judge whether to believe the number. So here is the model.

Sources

Where each number comes from

What people paid
Lawrence Berkeley National Laboratory, U.S. Distributed Solar and Storage Data, 2026 edition

Median gross installed price per watt DC for host-owned residential systems installed in 2024-2025, from 205,297 records. Formerly published as Tracking the Sun. Berkeley Lab assembles it from state incentive programme and utility interconnection records, so these are prices on contracts that closed.

What a roof produces
PVGIS v5.2, European Commission Joint Research Centre

Annual kWh per kW installed, modelled at the largest city in each state. For North America PVGIS reads NSRDB, which is NREL's own National Solar Radiation Database, so the underlying radiation is the same data PVWatts uses.

What electricity costs
US Energy Information Administration, Electric Power Monthly, table 5.6.A

Average residential retail price by state, July 2026, refreshed monthly.

Incentive programmes
DSIRE, N.C. Clean Energy Technology Center, NC State University

Index of state, utility and local programmes reaching rooftop photovoltaics. Licensed CC BY-SA 4.0; this site reproduces the factual index only and links out for the detail.

Tax law
Public Law 119-21, 26 U.S.C. 25D and 48E, and IRS Notice 2025-42

Read from the enacted text and the US Code rather than from secondary coverage, because a great deal of published solar content still describes the pre-2025 rules.

Assumptions

What we assumed, and why

AssumptionReason
0.5% output loss per yearModern module warranties promise no worse than this and most beat it slightly.
2.5% annual electricity price risesBelow the last decade's actual trend, deliberately. It is the assumption that makes solar look worse, which is the right direction for a site that is not selling it.
25 year horizonThe industry-standard production warranty, so the longest period a buyer has any contractual reason to believe in.
20 degree tilt, facing due south, 14% system lossesA 4-in-12 roof pitch is 18.4 degrees. The loss figure is the PVWatts default, covering soiling, wiring, mismatch, inverter inefficiency and age.
400 W modulesUsed only to translate a system size into a panel count, which is how installers describe systems to households.
Generation is worth the retail rateTrue under full net metering. Where a state has moved off it and we know, a haircut is applied and the state page says so. This is the assumption most likely to flatter solar.
Limits

What this data genuinely cannot tell you

The price data is not national

This is the big one. Berkeley Lab builds its dataset from the administrative records of state incentive programmes and utility interconnection queues. A state that runs no such programme files nothing, so it does not appear. Only 11 states have enough reported installations to publish a local median, and CA alone accounts for 78% of the sample.

That is why this site quotes $3.75 a watt as the national figure rather than $4.25. The first is the median of the state medians, where each covered state counts once. The second is the median of every reported system, which is mostly CA and an expensive market at that. Where a state has no local data, its page uses the national median and tells you it is doing so, rather than quietly presenting a national number under a local heading.

Reported prices lag

Administrators report on long and uneven schedules, so the most recent year in the file is partially reported rather than genuinely smaller. The figures here pool two installation years to get enough systems per state for a median to mean anything, and every page prints the span rather than implying the prices are from this morning.

A median is not your quote

Half of buyers paid more than these figures and half paid less, and the spread between the 25th and 75th percentile is wide. Roof complexity, a panel upgrade, a long conduit run, a difficult permit office, or simply which company reached you first all move the number more than the choice of module does.

We model a good roof

Due south, unshaded, at a normal pitch. Real roofs face other directions and have trees and chimneys on them. An east or west facing array loses roughly 15%, and shading can cost far more than that. Nothing on this site substitutes for a site survey.

We do not know your tax position

State credits, property tax exemptions and sales tax exemptions all depend on circumstances this site cannot see. Nothing here is tax advice.

Corrections

If something here is wrong

Every figure on this site is traceable to a source listed above, and the scripts that fetch and compute them refuse to write anything at all rather than publish a partial or implausible result. If you find a number that does not survive checking, tell us and it gets corrected, with a note saying what changed.