In late September 2026, a Danish-Italian team reported that an industrial 70-cell solid oxide electrolysis stack had run for 25,000 hours. After conditioning, its voltage rose by just 0.05% per 1,000 hours over the remaining 22,000 hours, a tenth of the EU 2030 target of 0.5% per 1,000 hours (Hydrogen Central). The study, by DynElectro, SolydEra, Aalborg University and DTU, appeared in Nature Communications (Bioengineer).
The more useful number is in the same report. During a period of conventional DC operation, the same stack degraded by about 11.5% per 1,000 hours. Switching to the team's AC:DC operating mode cut that to under 1% during run-in (Hydrogen Central).
Same hardware, very different ageing. That is the core lesson for anyone sizing an electrolyzer plant: the electrolyzer degradation rate is not a fixed property of the technology. It depends on how you run the stack. And it feeds directly into your hydrogen cost.
What an electrolyzer degradation rate actually measures
A degradation rate tracks how fast cell voltage rises at a constant current. It is quoted in millivolts per thousand hours (mV/kh) or as a percentage per 1,000 hours. Higher voltage at the same current means more electricity for the same hydrogen output.
The US Department of Energy gives a clear benchmark for PEM stacks (DOE):
| PEM stack metric | 2022 status | 2026 target |
|---|---|---|
| Average degradation rate | 4.8 mV/kh (0.25%/1,000 h) | 2.3 mV/kh (0.13%/1,000 h) |
| Lifetime (operating hours) | 40,000 | 80,000 |
| Stack electrical efficiency | 51 kWh/kg H₂ (65% LHV) | 48 kWh/kg H₂ (69% LHV) |
A quick calculation shows why these figures matter. At 0.25% per 1,000 hours, a stack that runs 40,000 hours ends up about 10% above its starting voltage. That lines up with the end-of-life definition several studies use: "a 10% cell voltage increase" (Arnold et al.).
At constant current, power scales with voltage. So, as a rough illustration, a 10% voltage rise pushes a stack consuming 51 kWh/kg towards 56 kWh/kg by end of life, before any other losses.
Typical degradation rates by technology in 2026
The World Bank's ESMAP programme published a reference table for project planners in February 2026 (ESMAP):
| Technology | Stack durability (operating hours) | Degradation rate |
|---|---|---|
| Alkaline (AWE) | 60,000–90,000 | 0.1–0.25% per 1,000 h |
| PEM | 40,000–60,000 | 0.2–0.5% per 1,000 h |
| SOEC | 20,000–40,000 | 0.5–1.0% per 1,000 h |
The same report notes that "stack replacements in PEM systems can represent up to 15 percent of total system costs." For a fuller look at the AWE versus PEM trade-off, see our utility-scale comparison.
Read the September SOEC record against this table. It shows what one stack achieved under a specific operating strategy, at 750 °C and −0.5 A/cm² (Hydrogen Central). It is not yet a fleet average you can drop into a bankable model.
There is also a data gap. A 2025 study in Energy & Environmental Science notes that PEM durability tests longer than 3,000 hours are rare (Finger et al.). Ask how many hours sit behind any rate a supplier quotes.
Why stop-starts hurt more than steady hours
Datasheet rates usually assume steady operation. Plants fed by variable renewables do not run that way. Each full shutdown exposes the electrodes to conditions they were not designed to sit in.
Alkaline stacks
When a bipolar alkaline stack shuts down, reverse (shunt) currents flow as the electrodes return to their original chemical states. A 2025 study measured these at "on the order of 0.1–1% of the total current" in the middle cells of bipolar stacks. Under variable operation, a nickel-iron catalyst film in that study lost 98.2% of its nickel and 74.7% of its iron (Marquez et al.). A separate study of dynamic operation found the cathode suffered "irreversible oxidation of nickel to β-nickel hydroxide", leading to severe cathode degradation (Wang et al.).
PEM stacks
Shutdown is not harmless here either. Simulated shutdown holds in the Energy & Environmental Science study were linked to changes in the iridium catalyst's oxidation state, platinum dissolution from the porous transport layer coating, and platinum deposits inside the membrane (Finger et al.).
System level
A 2026 techno-economic thesis at TU Wien modelled a degrading PEM system under two strategies. When the plant followed hydrogen demand, frequent shutdowns "significantly reduce stack lifetime and increase hydrogen costs." When it followed the power supply, wind-powered systems reached the highest full-load hours and the lowest costs, while PV-only operation led to limited utilisation and uncompetitive LCOH (Kotter, TU Wien). Pairing solar with wind is one way to raise those hours, as we showed in our hybrid capacity factor analysis.
How degradation shows up in your LCOH
Degradation hits hydrogen cost through two channels: more electricity per kilogram, and stack replacements. ESMAP puts it plainly: "Efficiency losses and the cost and frequency of stack replacements are major OPEX components" (ESMAP).
A 2025 study from Leibniz University Hannover and EWE quantified the trade-off for an onsite green hydrogen supply chain (Arnold et al.):
- In the base case, the cheapest hydrogen came from replacing stacks once energy demand had risen 20%, which meant a 7-year replacement period.
- Across degradation scenarios of 2.5 to 12.5 µV/h, the optimal replacement period ranged from 14 years down to 5.
- Minimum LCOH ranged from 6.21 to 6.81 €/kg, a spread of 0.60 €/kg driven by degradation alone.
- Over 20 years of operation, that meant one stack replacement in the mildest scenario and three in the harshest.
Two caveats make this more relevant, not less. The authors question whether the common 10% end-of-life rule is techno-economically reasonable. And their model excludes start-stop cycles, so plants with frequent shutdowns could see higher figures.
Degradation also interacts with the levers we covered earlier: capacity factor and CAPEX versus discount rate. A cheaper stack that ages faster can lose its price advantage within one replacement cycle. This matters when comparing Chinese and Western electrolyzer quotes.
Five practical moves for developers and EPCs
- Ask for degradation at your load profile. Request the rate in mV/kh or %/1,000 h, the test duration behind it, and whether the test included cycling and shutdowns.
- Count starts, not only hours. Track full shutdowns as a design variable. Sizing renewables, storage and minimum load to avoid cold stops can protect stack life.
- Model degradation explicitly. Put a voltage-rise curve in your techno-economic model and test several end-of-life thresholds instead of hard-coding 10%.
- Negotiate the contract around it. ESMAP notes that most OEMs offer a standard two-year equipment warranty, with performance guarantees as an optional extra. Full service contracts may add about 3% of equipment CAPEX per year (ESMAP). Price a degradation guarantee into your bid comparison.
- Watch operating strategy research. The SOEC AC:DC result shows control strategy can change ageing by an order of magnitude on one stack. Treat it as a promising signal to follow, not yet a design basis.
Conclusion: make degradation a design input
The electrolyzer degradation rate decides how many kilowatt-hours each kilogram costs in year eight, and how many stacks you buy over a project's life. The latest research points one way: operating pattern matters as much as chemistry.
Projects that model starts, shutdowns and replacement timing early make better technology choices and stronger bankability cases. Those that copy a datasheet number into a spreadsheet risk discovering the gap during operation.
Hyzen helps engineering teams get these inputs right from day one. The platform already covers renewable input, electrolyzer sizing, water, gas conditioning, compression, storage and RFNBO checks, and Hyzen is building a full LCOH and techno-economic engine on top. Start with the free Site Screener to test your site's renewable profile and the full-load hours it can deliver.
References
[1] Hydrogen Central. DynElectro sets new world record for Solid Oxide electrolysis stack durability: 25,000 hours with near-zero degradation. Hydrogen Central (2026).
[2] Bioengineer.org. Alternating current trick keeps hydrogen electrolysis stack running for nearly three years (on Bilalis, Esposito et al., Nature Communications, DOI 10.1038/s41467-026-78001-1). Bioengineer.org (2026).
[3] US Department of Energy. Technical Targets for Proton Exchange Membrane Electrolysis. DOE Hydrogen and Fuel Cell Technologies Office.
[4] World Bank ESMAP. Electrolyzers for Hydrogen Production: Technical and Economic Characteristics. ESMAP Technical Report (2026).
[5] Arnold, M., Brandt, J., Tjarks, G., Vanselow, A., Hanke-Rauschenbach, R. Cost-optimized replacement strategies for water electrolysis systems affected by degradation. arXiv:2508.16370 (2025).
[6] Finger, S. et al. Degradation phenomena in PEMWE revealed by correlative electrochemical and nanostructure analysis. Energy & Environmental Science, 18, 9877–9894 (2025).
[7] Marquez, R. A. et al. Insights into catalyst degradation during alkaline water electrolysis under variable operation. Energy & Environmental Science (2025).
[8] Wang, G. et al. Probing Electrode Transformation under Dynamic Operation for Alkaline Water Electrolysis. Advanced Energy Materials, 15(28) (2025).
[9] Kotter, L. Degradation-aware techno-economic assessment of a PEM electrolysis system for renewable hydrogen production. Diploma thesis, TU Wien (2026).