3 / 4 · 5 min
Break-even power price and payback
The buyer's main number is not income but the price per kilowatt-hour at which the machine breaks even. It takes one line to compute and it decides everything.
A hardware seller talks about income, but income is a difference, and you pay its second half out of your own pocket every month. The same machine feeds a family in a country with cheap power and runs at a loss in a flat on a city tariff, and luck has nothing to do with it: one number sits between those cases — the price per kilowatt-hour at which revenue exactly equals the electricity bill. We call it the break-even price and put it next to income on every mining page.
On the left is the power price at which net income is zero. On the right is revenue at today's coin price and today's difficulty, divided by the daily consumption. Anything cheaper than the threshold leaves you the difference; anything dearer you pay out of pocket.
An Antminer S21 on today's numbers
The machine does 200 TH/s and draws 3500 W. At the current Bitcoin network hashrate and a reward of 3.15 BTC per block it mines 0.00009832 BTC a day — already net of the typical 1% pool fee. At a Bitcoin price of $80,000 that is $7.87 of revenue. Consumption: 3.5 kW × 24 hours = 84 kWh a day. Break-even: 7.87 ÷ 84 = $0.09 per kilowatt-hour. On power at $0.05 the machine nets 7.87 − 4.20 = $3.67 a day. On a city tariff of $0.10 it brings minus half a dollar a day, and no payback will ever arrive, however long you wait.
Payback reads as “no faster than”
This machine costs $1499 in our catalogue. Divided by $3.67 net, that is 409 days. It is not a date but an upper bound on luck: the calculation assumes network difficulty, coin price and your tariff all stay exactly as they are today for all 409 days. That has never once happened in the history of mining. So payback reads as “no faster than that”, not “in exactly that”.
Multiplying today's income by 365
The most expensive beginner's mistake looks harmless: “$3.67 a day is $1340 a year, so the machine pays back and then some”. Hidden in that multiplication is the assumption that the network will not grow. It always grows while mining pays: every new miner shrinks your share. Do not compute an annual income — compute several scenarios, and see in which of them you are still in profit.
Difficulty risk points one way only
Suppose network hashrate grows 3% a month — a calm pace by Bitcoin standards. Over twelve months that is a factor of 1.43, and your share is divided by it: the same machine will mine 70% of today's amount. Revenue falls from $7.87 to $5.51 while the electricity bill does not move at all — at $0.05 you are left with $1.31 a day instead of $3.67. The coin price may cover that, or may not; difficulty growth does not ask.
What else eats income besides power
- Pool fee — usually 0.5–2%; our calculator uses 1% by default.
- Downtime: reboots, overheating, dropped links. 95% uptime is 5% less mined, while electricity is paid for the full day.
- Losses in the power supply and wiring: the socket gives up more than the machine's spec sheet says.
- Cooling and noise: an ASIC is 70–80 dB, it does not live in a flat, and hosting charges its own price per kilowatt.
- Wear and resale value: the machine gets cheaper along with the income it produces.
Open mining, type in your own electricity price — it is remembered — and look at the break-even column. Anything whose threshold is below your tariff does not suit you at any coin price.
Open miningA machine draws 3000 W and brings $9 of revenue a day. What is its break-even price and what does it mean on a $0.12 tariff?
Consumption: 3 kW × 24 = 72 kWh. Break-even: 9 ÷ 72 = $0.125 per kilowatt-hour. On a $0.12 tariff the machine is still in profit, but only by $0.36 a day — any move in difficulty or price pushes it under. Such a purchase is judged by the margin to the threshold, not by income.
A seller promises payback in 8 months. Which two questions should you ask?
First: at what electricity price was it computed — if at $0.03 and yours is $0.10, the number is not about you. Second: does it allow for difficulty growth — almost never, which means the real term is longer, and the faster the network grows the longer it gets.