How long does a power station last — hours, or years?
Troubleshooting Updated 2026-08-08
The question splits in two and the answers are unrelated. Over years, the number on the box is a cycle count — 3,000 to 6,000 across the records in this library — and it is only meaningful next to the capacity threshold it was measured to. Two brands here both state 4,000 cycles: one counts down to 80% of original capacity, the other to 70%, so the same number is not the same promise.
Over a single charge the answer is a division, and it needs fields most manufacturers leave out. Only 10 of the 29 station records here state everything that division requires, which is why this site prints an hour figure for that group and refuses to estimate one for the rest.
A cycle count without its threshold is not a specification
Three machines, three cycle claims, read with their thresholds
Each pair of rows comes off one manufacturer document. Read them as pairs — a cycle count and the capacity it is counted down to are one statement, and separating them is how "4,000" ends up looking like a fair comparison.
The same headline number as the row two below it.
See the record →The looser of the two thresholds in this library, which makes the cycle count above easier to reach.
Identical headline figure, different test endpoint.
See the record →Ten points stricter than the row above, on the same number of cycles.
The longest claim in the library, and it is measured to the stricter of the two thresholds.
See the record →Official specverified 2026-08-01station records · battery block
Nothing here is a trick played by any of the three brands: each of them prints its threshold beside its number, and every figure above is off the manufacturer’s own page. The comparison is what goes wrong, because the headline travels and the threshold does not. A machine rated for 4,000 cycles down to 70% has agreed to lose almost a third of its capacity by the end of that count; one rated for the same 4,000 cycles down to 80% has agreed to lose a fifth. Neither will fall over on the day it gets there. A third shape turns up in this library too: a record that states a decade of service life in words and gives no cycle count at all. Years cannot be converted into cycles without knowing how often a machine is cycled, so that record simply has no figure in this column, and no figure is what it gets.
What those counts look like in years, and what the warranty says
Cycles divided by a cycling rate, against the term the maker guarantees
The first two rows are our arithmetic on the counts above, at a deliberately aggressive rate of one full cycle every day. The warranty rows are the manufacturers’ own.
Nobody cycles a backup machine daily, so read this as a floor rather than a forecast.
The record's 6,000 cycles divided by 365 cycles a year. The cycle count is the manufacturer's; the rate is an assumption we state, and the division is ours.
The record's 4,000 cycles divided by the same 365 cycles a year, to the looser 70% threshold. Same arithmetic, same assumption, different endpoint.
The number with a company behind it, and it is a fraction of the cycle-life arithmetic above.
Stated as a base term plus an extension for buying direct.
Official spec + Derived by WattPairverified 2026-08-01station records · battery block
That gap is the practical finding. Even at a cycling rate no household backup machine will ever see, the cycle counts land two to three times beyond the warranty term — so for anything used a handful of times a year, the cycle number is not the binding constraint and should not be the deciding line on a spec sheet. What it is genuinely useful for is the opposite case: a machine cycled hard every day on a van roof or a balcony, where the count and its threshold start to describe real years.
And the other meaning: how long it lasts on one charge
The same sentence is just as often about a single evening, and that answer is a division rather than a rating: usable capacity, after depth of discharge and inverter efficiency, over the load plus the machine’s own self-consumption. Every one of those four terms has to come from somewhere, and this library holds a complete set for only 10 of the 29 station records. On the rest this site refuses the division rather than filling the gaps with a plausible constant, which is why a runtime figure appears on some records and not on others.