Home / Blog / October 2026

Guide · Oct 2, 2026 · 6 min read

What uptime means for a mining machine, and what a day offline costs

Uptime is the share of the year your machine is actually hashing. At 99% that is still 3.65 days off. For an Antminer S21 XP, one day off is 0.000129 bitcoin that nobody pays back, so the figure is worth knowing before you choose where a machine lives.

Firsthand card: "What uptime means for a mining machine, and what a day offline costs", with 3.65 days marked as the time a machine is offline in a year at 99% uptime

Uptime is the share of the year your machine is hashing

A bitcoin miner earns only while it is running. Every second it spends guessing at the puzzle counts toward its share of the day's rewards, and every second it is off counts for nothing. Uptime is how much of the time it was on, written as a percentage.

A year has 8,760 hours. A machine that hashed for 8,672 of them had 99% uptime. The same machine hashing for 7,884 hours had 90%. Both are described with the same word, and the gap between them is a month of output.

Uptime is the one number on a hosting page that the machine's spec sheet cannot tell you.

99% still means three and a half days off

Percentages hide how much time is behind them, so turn them into days. At 99% uptime a machine is off for 3.65 days a year, which is 87.6 hours. At 95% it is off for 18.25 days. At 90% it is off for 36.5 days, more than a month.

Those gaps do not arrive as tidy days. They arrive as a few hours here after a power cut, a day there after a cooling fault, and a long stretch if a part has to be replaced. The percentage is the sum of all of them over a year.

A day off costs one day's coins, and nothing makes it up

Take an Antminer S21 XP, rated at 270 terahashes a second, which is 0.27 petahash. On 1 October 2026 a petahash earned 0.00047799 bitcoin a day on our hashprice chart, so the machine mined about 0.000129 bitcoin that day, worth $10.79 at the day's close of $83,571.

One hour offline is a twenty-fourth of that: about 0.0000054 bitcoin, or $0.45. One day is 0.000129 bitcoin. Over a year, 1% of downtime is 0.000471 bitcoin, about $39. At 95% uptime the loss is 0.002355 bitcoin, about $197. At 90% it is 0.004711 bitcoin, about $394.

The coins are gone for good. Hashprice is your machine's slice of the block rewards paid out in that stretch of time, and a block is won once. If your machine was off when it was found, the reward went to the machines that were on, and the network does not hold it back for you to collect later.

A bigger machine loses more in absolute terms and the same in proportion. An Antminer S21 XP Hyd, rated at 473 terahashes a second, mined about 0.000226 bitcoin on 1 October, $18.89 at the same close. Its 3.65 days at 99% uptime cost about 0.000825 bitcoin, roughly $69. Double the hashrate, double the loss, and the same 1% of the year.

Chart: What a petahash earned each day, in satoshis. Every day your machine is offline forfeits that day's height of the line. Open the chart

What takes a machine offline

Most downtime comes from the same short list: the site loses power, the cooling fails or struggles in heat, the internet connection drops, the machine itself has a fault, or it has to be switched off for maintenance. A hashboard that fails does not always stop a machine; it can lose part of its hashrate and carry on, which is its own quiet kind of downtime.

Sites differ in how often each of these happens. A site on a stable grid with its own backup has fewer power cuts. A site in a hot climate has to work harder on cooling. That is why uptime is a property of the site, and why it belongs in your choice of where a machine lives. The repair guide covers what happens when it is the machine that fails.

Site uptime and machine uptime are not the same number

A site's uptime counts the hours its power, cooling and network were working. Your machine's uptime also counts its own faults. A site at 99% that keeps your machine on the repair bench for a week leaves that machine at about 97% for the year: 365 days, less 3.65 days of site downtime and 7 days of repair, over 365.

Hashrate matters as well as being on. A machine hashing at half its rating for a day has produced half a day of coins, so a daily average of delivered hashrate against the rating is the fairest measure of how well a machine is doing. The pool's graph shows that directly, which is why it is the first place to look.

Check your own machine in the pool dashboard

You do not have to take anyone's word for uptime. The pool your machine mines in shows its hashrate over time. A machine that is running looks like a steady line near its rated hashrate. A machine that stopped looks like a gap, and a machine with a failed board looks like a lower step.

Read the average over a week rather than a single reading. The pool's hashrate figure is an estimate built from the shares your machine submits, and it can wobble by as much as a fifth from one reading to the next even on a healthy machine. A gap of hours is real. A dip of a few percent for a few minutes is not.

A cheaper site can be the right site

Higher uptime is not automatically better. Power is the biggest running cost, and a site with cheaper power and lower uptime can still be the better home for an older, less efficient machine. It mines fewer coins and each one costs less to make. A machine crosses to cheap power once explains when that trade is worth making.

The reverse is also true. While a petahash earns well and the machine is efficient, every hour offline costs more, so the site with the highest uptime is where a young machine earns the most. You can see each site's current rate and uptime on our sites page.

Put a number on the trade. A machine at a site with 90% uptime mines 9.1% fewer coins than the same machine at a site with 99%. The cheaper site has to earn that shortfall back through the price of its power, and an older machine, which pays more power per coin, has more to gain from a lower rate than a new one.

What to ask before you put a machine anywhere

Ask for the site's uptime over the last twelve to twenty-four months, not the last week, so one good month cannot flatter it. Ask how downtime is counted: whether a machine that is on but hashing at half speed counts as up. Ask what happens to a machine that is down for a long time, who fixes it and who pays for parts.

Then check the answer against the pool dashboard once the machine is running. A host that is open about its uptime is also easy to check, and the first month of your own graph will show you whether the number holds.

What it means for a miner

Uptime is a multiplier on everything else you worked out. A machine's efficiency and its power rate decide what each coin costs to make. Uptime decides how many coins it makes at all. Over a year, the gap between 99% and 90% on an S21 XP is about 0.0042 bitcoin, roughly 33 days of the machine's output at today's hashprice.

Pick the site by its record, expect a few days off a year even at the best of them, and read your own graph by the week. Mining output depends on price, difficulty and uptime, and is not guaranteed.

Disclosure

Firsthand Bitcoin sells and hosts mining hardware, including this machine. No manufacturer, distributor or affiliate programme paid for or reviewed this page and we take no commission on the links above. Historical figures are computed from daily bitcoin price and network hashprice, each day valued at its own prices. Nothing here is investment advice.