Mining Lab · Coinovo Academy · Running mining as a business
What a machine really costs
Whether a machine pays off comes down to three: efficiency in joules per terahash, the power price in cents per kilowatt hour, and the yield per terahash. Everything else is trimming. Anyone who knows these three needs no calculator, just a sheet of paper.
Yield = own hashrate divided by network hashrate x block reward x 144 blocks. The 144 are the blocks of one day at ten minutes apart. The pool fee comes off that, usually one or two percent.
Your share of the network times the daily issuance, that is the whole formula.
Consumption in kilowatt hours = watts divided by 1000 x 24. Times the power price that gives the daily cost. A machine at 3,500 W draws 84 kWh a day; at 30 cents that is 25.20 euros, every day, Sundays included.
Joules per terahash says how much power a machine needs for one guessing attempt. Hashrate alone says nothing: a machine with double the hashrate and double the consumption earns the same. Lower is better, and between 15 and 30 J/TH lies the difference between viable and subsidised.
Compare machines by J/TH, never by TH/s.
The contribution margin is daily yield minus power cost. It says whether the machine should run. Profit additionally deducts the purchase price and says whether it should have been bought. Those are two different questions and they can come out differently: a paid-off machine with a thin margin keeps running, a new one with the same margin is not bought.
More interesting than profit is the question: up to what power price does the machine carry itself? That is daily yield in euros divided by daily consumption in kWh. If your price is above it you are paying in, whatever the market does.
The break-even price is the most honest figure in mining.
Take a machine at 200 TH/s and 3,400 watts. That is 17 joules per terahash, a good figure but not a top one.
Consumption: 3,400 divided by 1000 times 24 hours is 81.6 kilowatt hours a day. At a tariff of 25 cents that costs 20.40 euros, every day.
Yield: say the network is at 950 EH/s and the daily payout 450 bitcoin. Your share is 200 TH/s divided by 950,000,000 TH/s, so 2.105 ten-millionths. Times 450 bitcoin gives 0.0000947 bitcoin a day. At 80,000 euros a bitcoin that is 7.58 euros. One percent pool fee comes off: 7.50 euros.
Contribution margin: 7.50 minus 20.40 is minus 12.90 euros a day. The machine burns money.
Break-even power price: 7.50 divided by 81.6 kilowatt hours is 9.2 cents. Below that tariff the machine carries itself, above it it does not. That is an industrial price, not a household one, and it is exactly why mining happens where power is industrial and cheap.
Suppose you found power at 6 cents. Then the day costs 4.90 euros and the margin is 7.50 minus 4.90, so 2.60 euros.
If the machine costs 2,400 euros, that is 2,400 divided by 2.60, so 923 days to payback, two and a half years. In that time at least one halving happens, cutting the yield in half, and the difficulty keeps rising.
So run the sum against yourself: halve the yield to 3.75 euros. Then the margin is minus 1.15 euros and the machine no longer carries itself. A purchase that only pays for itself after two and a half years but stops being profitable after one was a mistake.
The rule that follows: work out every machine with the yield AFTER the next halving. If it does not carry itself then, you do not buy it.
A payback of 400 days does not mean 400 days until profit. It holds only if price and difficulty stay the same, and neither ever does. The difficulty rises almost always, and the yield per terahash falls with it. Run the same machine with twenty percent less yield and see whether the calculation still holds.
Power supplies age, fans fail, maintenance costs time. A surcharge of five to ten percent on operating costs is not caution, it is experience. And a device that is down earns nothing while you wait for a replacement.
At home you pay household rates and hear the noise. With hosting you pay a price per kilowatt hour that includes maintenance and cooling, but you lose direct access and carry the provider's risk. Your own hall only pays off with many devices, because the connection and the cooling cost almost the same whatever the count.
If the contribution margin falls below zero, every further hour costs money. Then switching off is the right decision, even if the machine is not paid off. The purchase price is spent; it no longer belongs in any decision about tomorrow.
Money already spent is no argument for spending more.
Tasks, shifts and the daily hunt, plus an account that shows what has added up.
Opens straight away. No account, nothing to install.