25 to 30 years is a warranty term, not a failure date
Solar panels do not switch off at 25 years. They get gradually weaker, and a warranty expires. Those are two different events, and almost every page on this topic reports the second one as though it were the first.
What the industry calls “useful life” is the point at which a module has lost roughly 20% of its original output. The National Laboratory of the Rockies (the lab formerly known as NREL) puts it plainly: modules degrade at generally 0.5%–1% a year, and the retirement threshold is “often about 20% loss from their original output” — at which point, for modules built today, “it is probably 30 years.” Past that line the panel keeps producing. It just produces less, and the paperwork stops protecting you.
The Department of Energy’s current figure comes from a Berkeley Lab survey of US solar professionals: average operational lifespan has risen from around 20 years in 2007 to 25–35 years in 2025. That is a real increase, and it is not what the warranty says.
One thing to get out of the way: this article is about rooftop and ground-mount photovoltaic modules. Solar garden lights are a completely different product with a completely different answer, governed by battery chemistry rather than silicon degradation, and we cover how long solar lights last separately.
How fast panels actually degrade — and why the literature disagrees
The 0.5%-a-year figure everyone quotes traces to one paper: Jordan and Kurtz’s 2012 analytical review, which pooled nearly 2,000 reported degradation rates and found a median of 0.5%/year and a mean of 0.8%/year. That number has been copied for fourteen years. It is now being challenged.
| Study | What it measured | Median | Mean |
|---|---|---|---|
| Jordan & Kurtz, NREL, 2012 | ~2,000 reported rates, modules and systems, 40 years of field testing | 0.5%/yr | 0.8%/yr |
| Jordan et al., Compendium, 2016 | 11,000+ rates, ~200 studies, 40 countries (crystalline silicon only) | 0.5–0.6%/yr | 0.8–0.9%/yr |
| Theristis et al., Sandia/NREL/UCF, 2023 | 834 fielded modules, 13 types, 7 makers, 3 US climates, 5 years | 0.58%/yr | 0.62%/yr |
| NREL PV Fleet (Deline et al.), 2024 | 1,128 US systems, 8.5 GW, 24,000 inverter channels — system level | 0.75%/yr | — |
| Chen et al., Solar Energy Materials and Solar Cells, 2025 | Global compendium, all PV technologies | 1.00%/yr | 1.27%/yr |
The 2025 compendium is the outlier that matters. Its authors describe their global median of 1.00%/year and mean of 1.27%/year as an increase over the 2016 and 2013 numbers. That is double the figure the rest of the internet is still repeating, and we are not going to quietly pick the optimistic end.
Two things reconcile most of the gap, and neither is a reason to relax. First, module-level measurements in a lab are not system-level performance in a yard. The NREL fleet number of 0.75%/year includes soiling, inverter clipping and equipment downtime, which is why it sits above the module-level 0.58%. Second, heat matters more than anything else you control. The same NREL fleet work found degradation running from 0.5%/year in temperate zones to 0.88%/year in hot ones. A Garden Grove rooftop lives closer to the hot end than a Seattle one.
Compounded rather than subtracted, here is what those rates mean for a panel rated at 400 W today:
| Degradation rate | Output at year 25 | Output at year 30 | 400 W panel at year 30 |
|---|---|---|---|
| 0.50%/yr | 88% | 86% | 344 W |
| 0.75%/yr | 83% | 80% | 319 W |
| 1.00%/yr | 78% | 74% | 296 W |
| 1.27%/yr | 73% | 68% | 273 W |
The product warranty is shorter than the performance warranty
This is the detail that catches people out, and the top-ranking pages mention it without explaining it. A module carries two separate warranties of two different lengths, and the shorter one is the one that covers the panel physically falling apart.
- Performance warranty — guarantees output against a declining curve. Typically 25 years, sometimes 30.
- Product (workmanship) warranty — covers defects in materials and workmanship: a cracked frame, a failed junction box, a delaminating backsheet. Frequently 12 years on mainstream modules.
Qcells’ own datasheets state it directly. The Q.PEAK DUO M-G11 carries an “inclusive 12-year product warranty and 25-year linear performance warranty,” guaranteeing at least 98% of nominal power in year one, a maximum 0.5% decline per year after that, at least 93.5% at 10 years and at least 86% at 25 years. The L-G8.2 is the same 12-and-25 split at 0.54%/year and 85% at 25 years. Premium lines run longer at both ends — Qcells’ US flagship and Maxeon’s 40-year coverage are the exceptions, not the norm — but the default assumption should be that year 13 leaves you with output coverage and no defect coverage.
Note the 2% allowance in the first year. That is not padding. It is light-induced degradation, discussed below, written into the contract.
What actually fails
Outright module failure is rare. NREL’s operations and maintenance guidance cites a module failure rate of 0.025%–0.1% per year, and a 2024 NREL review found fewer than 1% of modules installed in the United States fail within their first five years. When something does go wrong, it is usually one of these.
The IEA-PVPS Task 13 field survey inspected 455 fielded silicon modules and recorded how often each defect turned up:
| Failure type | Occurrence in inspected Si modules | What it is |
|---|---|---|
| Moisture ingress | 19% | Water past the seal, usually at the junction box or module edge; drives corrosion and PID |
| Frame defects | 17% | Corrosion, joint separation, cracking, bending. Dominant in cold and snowy climates |
| Snail tracks | 12% | Discoloration of the front silver metallization. Appears in the first six months to three years |
| Backsheet defects | 9% | Bubbles over 5 mm, delamination, cracks, abrasion |
A few of these deserve naming individually, because they behave differently from steady wear.
Backsheet cracking. Around 2010 a low-cost polyamide backsheet (the “AAA” construction) entered the market. It passed the standard damp-heat and UV qualification tests, then began cracking in the field after five to ten years. The tests did not reproduce the mechanical stresses that caused it. This is the clearest example of a module passing certification and failing anyway.
Microcracks. These start during cell manufacturing, shipping, installation, and later from hail or someone standing on a module. They are near-universal and usually harmless: in one study cited by NREL, 84% of modules had at least one crack, but only 60% had cracks causing significant power loss. A crack only costs you output once repeated thermal and mechanical cycling wears through the metallization bridging the fragments. If you live in a hail belt, protecting the array from hail is worth doing before the fact, not after.
Potential induced degradation (PID). Several distinct mechanisms driven by the voltage difference between cells and the grounded frame. It is accelerated by moisture and by acetic acid, which is what EVA encapsulant produces as it breaks down. Modules using polyolefin encapsulant instead of EVA avoid that particular feedback loop.
Delamination. Encapsulant separating from glass or cells, most often over the junction box and at the edges, which then lets liquid water straight onto the cell metallization. It showed up in 11% of silicon modules within the first three years in the Task 13 data — an early-manufacturing defect more than an aging one.
Light-induced degradation (LID). A one-time drop in the first hours of sun exposure, within roughly the first 10 kWh/m² of light. Sandia, NREL and UCF measured it across module types: the worst was a mono-silicon module at −3.3%, a polycrystalline module from the same vendor lost 1.5%, and two PERC types lost 1.2%. It happens once, it is already priced into the warranty curve, and it is not a fault.
Silver is the common thread in several of these. Corrosion of the silver gridlines on the cell front is a documented degradation mechanism, and it is the same silver that makes recycling worth attempting — how much silver is in a solar panel turns out to matter at both ends of the panel’s life. For failures that show up as a dead or underperforming string rather than a visible defect, the common solar panel problems guide and how to test a panel with a multimeter will get you to a diagnosis.
The inverter goes first. Plan on it.
The panels are not the part you will replace. The inverter converts DC to AC, runs hot, contains electrolytic capacitors and switching semiconductors, and does not last 30 years.
NREL’s residential cost modeling assumes an inverter lifetime of 15 years inside a 30-year analysis period — one replacement, built into the economics from day one. The warranty market splits along the same line: NREL’s 2022 system cost benchmark models a commercial string inverter at a 12-year warranty against a residential microinverter at 25 years. String inverters are cheaper up front and get swapped once or twice; microinverters and optimizers cost more and are rated to go the distance. If you are choosing between architectures, SolarEdge versus Enphase is largely a decision about which of those two failure profiles you would rather own.
Inverters are also the main reason a system underproduces without anyone noticing. In NREL’s fleet data, median inverter availability across 1,128 US systems is 0.99, but at the 10th percentile it falls to 0.95 — one system in ten is losing 5% of its year to downtime. That is six times the annual degradation rate, and unlike degradation you can fix it. If your bill has crept up without your panels getting older, start there: why an electricity bill goes up with solar is more often an availability or rate-structure problem than a panel problem.
Batteries, if you have them, are on their own shorter clock again — see how long solar batteries last.
Year 30: keep them or pull them?
A 30-year-old array in decent condition is still making roughly 70–86% of its original power, depending on which degradation rate your climate and hardware landed on. For a 6 kW system, that is somewhere between 4.2 and 5.2 kW. It is not broken. The question is whether that roof space is worth more with newer panels on it.
Replace when one of these is true, not because a number rolled over:
- Measured output has fallen well below the warranty curve while the inverter and wiring test clean. That is a module problem, and while it is in the product warranty window it is someone else’s problem.
- You see burn marks, browned encapsulant, backsheet cracks or a compromised junction box. These are safety issues, not efficiency issues. Damaged modules are an arc-fault and shock risk.
- The roof needs replacing anyway. Removing and reinstalling a 30-year-old array costs nearly what removing and installing a new one does, so reroofing is the natural moment to repower.
- Your consumption has grown. Modern modules put roughly twice the watts in the same footprint, so the argument is usually about space, not about the old panels being worn out.
Before concluding anything from a low reading, rule out soiling. A dirty array in Southern California can lose several percent in a dry summer and give it all back after one wash. Cleaning rooftop panels safely is cheaper than a diagnosis.
The recycling gap is real, and it is an economics problem
Here is the number the cheerful pages leave out. NREL’s 2021 review of US PV recycling found that “less than 10% are being recycled in the United States.” The reason is not technical. Recycling processes for both silicon and cadmium telluride modules exist and work.
It is price. The Department of Energy’s PV End-of-Life Action Plan states it flatly: recycling costs a waste generator around $15–$45 per module, against a landfill fee of $1–$5 per module. NREL’s underlying figures are the same — module recycling in the US at $15–45, while non-hazardous landfill tipping at $26 a ton works out to under $1 a module. Nothing about that gap encourages the right outcome, which is why several states are now legislating recycling fees instead of waiting for the market.
Two practical consequences. If your array comes down, ask your installer specifically where the modules go and get it in writing — “we’ll handle disposal” means the landfill by default. And a working 30-year-old panel has more value reused than recycled: for a shed, a gate opener or a trickle charger, 80% of 250 W is plenty.
What actually extends panel life
Not much, honestly, and that is the good news. Modules are sealed, solid-state and have no moving parts. The short list:
- Buy for heat, not for peak efficiency. Degradation runs 0.5%/yr in temperate zones and 0.88%/yr in hot ones. Mounting that allows airflow behind the modules is worth more over 30 years than a fractional efficiency advantage on the datasheet.
- Read both warranty numbers before you sign. A 25-year performance warranty with a 12-year product warranty is a different product from one with 25 years on both.
- Monitor at the string or module level. Availability losses are larger than degradation losses and they are recoverable. Degradation is not.
- Keep them clean, and keep branches off them. Soiling is reversible; a branch strike that starts a crack is not.
- Do not walk on them. Standing on a module is one of the documented causes of cell cracking.
We build solar lighting rather than rooftop arrays, so the honest disclosure is that we do not sell you a panel at the end of this. If you want small solar that simply works outdoors for years, that is our end of the business — our solar pathway lights ship from Garden Grove, California.
Sources
- Jordan & Kurtz - Photovoltaic Degradation Rates, an Analytical Review (NREL/JA-5200-51664) docs.nlr.gov The origin of the 0.5%/yr figure everyone quotes: nearly 2,000 reported rates, mean 0.8%/yr, median 0.5%/yr
- Jordan et al. - Compendium of Photovoltaic Degradation Rates, Progress in Photovoltaics 24(7) research-hub.nlr.gov 11,000+ rates across roughly 200 studies and 40 countries, with crystalline silicon at a 0.5-0.6%/yr median, and the finding that hotter climates degrade faster
- Chen, Zuo & Chang - Compendium of degradation rates of global photovoltaic technology, Solar Energy Materials and Solar Cells 293 (2025) doi.org The dissenting 2025 figures - a global median of 1.00%/yr and mean of 1.27%, described by its authors as an increase over the 2016 and 2013 compendia
- Theristis et al. - Onymous early-life performance degradation analysis of recent photovoltaic module technologies (NREL/JA-5K00-82352) docs.nlr.gov 834 fielded modules across three US climates at a 0.58%/yr median, and the measured light-induced degradation figures of -3.3% mono, -1.5% poly and -1.2% PERC
- Deline et al. - Availability and Performance Loss Factors for U.S. PV Fleet Systems (NREL/TP-5K00-88769) docs.nlr.gov 1,128 US systems totalling 8.5 GW: a median system degradation of 0.75%/yr, 0.5%/yr temperate against 0.88%/yr hot, and inverter availability of 0.99 at the median but 0.95 at the tenth percentile
- Zuboy et al. - Getting Ahead of the Curve: Assessment of New PV Module Reliability Risks (NREL/JA-5K00-84131) docs.nlr.gov Polyamide backsheet cracking at five to ten years despite passing qualification testing; that 84% of modules carry a crack but only 60% lose significant power; and a US five-year failure rate under 1%
- IEA-PVPS Task 13 - Assessment of Photovoltaic Module Failures in the Field iea-pvps.org The failure-frequency table from 455 inspected silicon modules - moisture ingress 19%, frame defects 17%, snail tracks 12%, backsheet defects 9% - and delamination in 11% within the first three years
- US DOE - Photovoltaics End-of-Life Action Plan, and End-of-Life Management for Solar Photovoltaics energy.gov That recycling costs $15-$45 per module against $1-$5 to landfill it, and the Berkeley Lab survey finding an average operational lifespan of 25-35 years in 2025 against about 20 years in 2007
Questions people ask
How frequently do solar panels need to be replaced?
Most homeowners never replace a panel. Fewer than 1% of modules installed in the United States fail within their first five years, and the ongoing failure rate is around 0.025-0.1% per year. The component you should expect to replace is the inverter - NREL's residential modeling assumes a 15-year inverter lifetime inside a 30-year system life. Plan for one inverter swap and no panel swaps.
Why are people getting rid of solar panels?
Almost always for a reason other than worn-out panels. The three common ones are a roof that needs replacing (removal and reinstallation costs nearly as much as new panels, so people repower instead), a household whose electricity use has outgrown the array, and leased systems being cleared at a property sale. Panels degrading to uselessness is far down the list - a 30-year-old module is typically still making 70-86% of its original output.
Do solar panels stop working after 25 years?
No. Twenty-five years is the length of the performance warranty, not the life of the hardware. At the end of it a mainstream module is still guaranteed to produce at least 85-86% of its nameplate power, and it keeps producing after that at a slowly declining rate. The Department of Energy currently puts average operational lifespan at 25-35 years, up from about 20 years in 2007.
Can solar panels be recycled?
Technically yes, and in practice mostly not. Recycling processes for both silicon and cadmium telluride modules exist in the US, but fewer than 10% of American end-of-life modules are recycled. The reason is cost: $15-$45 per module to recycle against $1-$5 to landfill it. If your array comes down, ask your installer in writing where the modules are going, and consider reuse first - a 30-year-old panel at 80% output still runs a shed or a gate.
Does rain clean solar panels?
Partly. Rain removes loose dust well enough that most residential arrays never need washing, but it does not shift bird droppings, pollen film, lichen or the baked-on grime that builds up through a dry Southern California summer. Those cause uneven shading, which costs more output than an even layer of dust does. If production drops in a dry season and recovers after the first real storm, you had a soiling problem, not a degradation problem.
How long does a solar inverter last compared to the panels?
Roughly half as long. String inverters commonly carry 12-year warranties and NREL models a 15-year service life; microinverters are typically warranted for 25 years. Panels are warranted 25-30 years for performance. So in a 30-year system, expect the panels to outlive at least one inverter. Watch availability as well as age - in NREL's fleet data the worst decile of systems loses 5% of annual production to inverter downtime, several times the annual degradation rate.
Why is my electricity bill high when I have solar panels?
Age is rarely the cause. Panels lose well under 1% a year, so a system five years old should be within a few percent of new. The usual culprits are inverter or string outages that nobody noticed, a change in your utility's rate structure or net metering terms, soiling, new shading from a grown tree, or simply using more electricity than the array was sized for. Check your production monitoring against the same month last year before blaming the panels.
