A machine that stops covering its own electricity is not a machine that needs patience. It is a decision, and there are exactly six things you can do about it. This page is that decision — what each option costs, what each one moves, and when each one wins. If what you want is the arithmetic itself, worked line by line on a machine we stock, that is how to calculate mining profitability; this page assumes you have run it and did not like the answer.
The one line you need before the rest makes sense:
break-even electricity rate = hashprice ÷ (24 × J/TH)
Hashrate is not in it — it cancels. Efficiency, not size, decides whether a machine survives your power price, which is why a 245 TH/s machine and a 95 TH/s machine of the same efficiency die at exactly the same tariff. The ASIC miner buying guide derives it.

First, confirm you are actually below it
Measured 15 August 2026: hashprice $31.32 per PH/s per day, network difficulty 127.48 T, network hashrate ~900 EH/s, BTC $63,072, next retarget estimated −2.98%. US average residential electricity 18.44¢/kWh (EIA, May 2026). At that hashprice the shortcut is 130.5 ÷ J/TH:
| Machine | Stock power | J/TH | Break-even electricity rate |
|---|---|---|---|
| Antminer S19 95TH | 3,250 W | 34.2 | 3.82¢/kWh |
| Antminer S19 Pro 110TH | 3,250 W | 29.5 | 4.42¢/kWh |
| Antminer S19k Pro 115TH | 2,645 W | 23.0 | 5.67¢/kWh |
| Antminer S19 XP 141TH | 3,010 W | 21.35 | 6.11¢/kWh |
| Canaan Avalon Q 90T | 1,674 W | 18.6 | 7.02¢/kWh |
| Antminer S21 Pro 245T | 3,675 W | 15.0 | 8.70¢/kWh |
Use your delivered rate, not the energy line on the bill — delivery, demand charges and any fuel adjustment are all real cents per kilowatt-hour. Operators talk themselves onto the wrong side of this line more often by using the wrong tariff than by getting the physics wrong. And note that difficulty has fallen 1.7% over the last twelve months, so a fleet that was marginal in October 2025 may have moved back above the line without you doing anything.
The six options, ranked by how much they actually move the number
In descending order of effect. The first two are worth more than the other four combined.
1. Change what you pay for electricity
Nothing else is close. Two cents per kilowatt-hour is worth more than every firmware and maintenance improvement put together, and it is the only variable most operators can genuinely negotiate. On a 3,250 W S19 running continuously, 2¢/kWh is $1.56 a day, $570 a year, per machine — against total revenue of about $2.98 a day. For a fleet of fifty that is $28,500 a year for a phone call and a contract. Ask about off-peak and interruptible tariffs specifically: a machine you can curtail is worth a different rate from one you cannot.
2. Move the machines instead of replacing them
This is the option most operators never price, and for a fleet it is usually the best one. An S19 at 34.2 J/TH is scrap at 6¢/kWh and a perfectly good asset at 3¢. The machine is not dead; it is in the wrong building. Freight on a pallet of miners is a known, four-figure number; replacing the same pallet with current-generation hardware is a five- or six-figure one. Hosting is the packaged version of the same move, and the honest comparison is hosting’s all-in rate against your own delivered rate, not against the energy line. What a site actually costs to stand up in the first place is itemised in mining setup costs explained.
3. Undervolt or run a lower-power tune
Trades hashrate for efficiency, which is the correct trade when power price is the binding constraint — and the wrong one when it is not. A tune that gives up 8% of hashrate for 15% of the watts moves break-even in the right direction; a “boost” tune that adds hashrate at worse J/TH moves it the wrong way, which is why we publish the stock figure next to the tuned one on every listing. It voids what warranty a used machine has. Free to try, reversible, and worth an afternoon before anything more expensive — and it is one of the few places where the folk wisdom is wrong in a checkable way, which we go through in five mining myths checked against the numbers.
4. Fix uptime before buying anything
Usually the largest unclaimed gain on an existing fleet and it costs nothing. A machine that is down 6% of the month has given up 6% of its revenue while its break-even calculation assumed 100%. Bad PSUs, thermal throttling, a marginal breaker, one dead hashboard silently cutting a machine to two-thirds — diagnosing which of those you have is a day’s work and it is the cheapest hashrate you will ever add.
5. Switch off and hold
An idle machine keeps its resale value; a machine running below break-even converts equipment you own into electricity you paid for. Capital already spent is not a reason to keep running — that money is gone whether the machine hashes or not. Switching off is a decision, not a failure, and it is the right one when margin on power alone is negative and there is no retarget in sight that fixes it.
6. Sell, and time it on the announcement
The steepest fall in a used machine’s value comes when the next generation is announced, not when it ships and not gradually with age. If a fleet is within a cent of your power price, that announcement is the moment it is worth the most it will ever be worth again. What the market actually pays, machine by machine, is in ASIC depreciation and US tax treatment — and the short version is that price tracks efficiency, not age.
What each option costs, and when it wins
| Option | What it costs | What it typically moves | When it wins |
|---|---|---|---|
| Renegotiate power | Time | 1–3¢/kWh, everything | Always try first |
| Move or host the fleet | Freight, downtime | Several ¢/kWh | Machines are healthy, site is wrong |
| Undervolt | An afternoon; any remaining warranty | 5–15% of watts | Marginal by under a cent |
| Fix uptime | Parts and labour | Whatever you are losing | Before any purchase |
| Switch off | Revenue, none of it profitable | Stops the bleed | Negative on power, no recovery in sight |
| Sell | The asset | Recovers capital | Announcement pending, or site cannot change |
| Buy newer hardware | The most of any option | Efficiency only | Power is capped and cannot be expanded |
Notice which one is last. Replacing hardware is the option operators reach for first and it is the most expensive way to buy a cent of headroom — an S19k Pro at 23 J/TH survives 5.67¢ and an S21 Pro at 15.0 J/TH survives 8.70¢, so the upgrade buys three cents for roughly eleven times the price per unit. It is the right answer only when your power is physically capped: if you have 200 A and no more, efficiency is the only way left to buy hashrate. With power to spare and a capital limit, more used machines beat fewer new ones every time.
Frequently asked questions
What is mining break-even?
It is two thresholds, not one. The first is the electricity rate at which a machine’s revenue exactly equals its power bill, calculated as hashprice divided by 24 times the machine’s efficiency in joules per terahash. The second is the point at which cumulative margin has repaid what you spent on the hardware and installation. A machine that fails the first test never reaches the second, no matter how long it runs.
How do I calculate my mining break-even point?
Start with the break-even electricity rate: hashprice divided by 24 times J/TH. If your all-in rate is below that, work out daily margin as hashrate in PH/s times hashprice, minus kilowatts times 24 times your rate, then divide total capital by that daily margin for payback in days. Include freight, electrical work and a spare power supply in the capital figure, and use your delivered electricity rate rather than the energy line on the bill. The worked version is in the calculation guide.
Why does my break-even point keep changing?
Because hashprice moves for two independent reasons: the Bitcoin price, and total network hashrate. More machines joining the network shrinks your fixed share of it, so revenue can fall while the coin price rises. Difficulty retargets every 2,016 blocks and the block subsidy halves roughly every four years, currently 3.125 BTC since April 2024. It can also move in your favour — difficulty is 1.7% lower than it was a year ago and 18% off its October 2025 peak, and what actually drives the coin price is a separate question from what drives your revenue.
What electricity rate does a used Antminer S19 need?
Below about 3.82¢/kWh at a hashprice of $31.32 per PH/s per day, measured 15 August 2026. A stock 95TH S19 runs at 34.2 J/TH, so it is the least tolerant machine on our shelf and the first to stop earning when hashprice falls. A used S19k Pro at 23 J/TH survives to roughly 5.67¢, which is why it is the better buy for anyone whose power is not genuinely industrial.
Is it better to sell an unprofitable miner or switch it off?
Switch off first, decide second — an idle machine keeps its value and stops the loss immediately, so nothing is lost by taking a week over the decision. Sell if your site cannot change and a new generation has just been announced, because that is when the price is highest. Hold if a difficulty retarget or a coin-price move could plausibly put you back above the line, or if you can move the machines somewhere cheaper: freight on a pallet is far less than replacing it.
Does break-even differ between large and small operations?
The formula is identical; the inputs are not. Large operations negotiate a much lower delivered rate, which is the variable that dominates the result, and they amortise electrical work, spares and freight across many machines. That is the whole advantage — not better machines, cheaper power and cheaper overhead per unit. It is also why a fleet that is marginal for a single-unit buyer can be genuinely profitable for someone buying a lot of the same machines.
The machines these numbers get run against
The machines these numbers actually get run against:
- Lot of 40 Used Bitmain Antminer S19 95TH/s Miners
- Bitdeer SealMiner A2 Pro Air 255TH/s Bitcoin Miner – 14.9 J/TH, January Batch
- Bitmain Antminer S19K Pro 120T – Efficient Bitcoin Miner
More in Bitcoin miner, or the full ASIC miner inventory. The arithmetic behind all of this — break-even electricity price from J/TH — is set out in our ASIC miner buying guide.