ASIC miner heat reuse: what the heat is actually worth

Every watt an ASIC miner draws leaves the box as heat, so a 3,250 W Antminer S19 is an 11,089 BTU/h heater that happens to earn bitcoin. Whether ASIC miner heat reuse is worth doing comes down to one question, and it is not a question about the miner: what were you heating with before? Against electric baseboards or a resistance furnace, miner heat is cheaper than the heat you are already buying, and the mining revenue is the discount. Against natural gas or a heat pump it is not close, and no amount of hashrate closes the gap. Everything below is the arithmetic behind those two sentences, done on machines we actually have on the shelf.

What a million BTU of heat costs, by source

A million BTU is roughly what a small, well-insulated house burns through in a cold week. Pricing every heat source in the same unit is the only way to compare them honestly, because a therm, a kilowatt-hour and a terahash are not comparable in their own units.

Electricity is priced at 18.44 ¢/kWh, the EIA US residential average, which is the figure the rest of this site uses. Gas is priced at $14.37 per thousand cubic feet, the average of the EIA residential series for December 2025, January 2026 and February 2026, converted at 1.037 million BTU per Mcf. Both move monthly and both vary enormously by state; put your own numbers in before you decide anything.

The number that would have flattered us, and why it is not in the table

The most recent residential gas price EIA has published is May 2026 at $19.83 per Mcf. Using it would have made miner heat look about 38% better against gas than the table above shows. It would also have been wrong: residential gas costs more per Mcf in summer because the fixed charges on the bill are spread across almost no volume, and nobody heats a house in May. The winter months are what you would actually pay during the season you would actually run a miner for heat, so those are the months in the table.

What the mining credit is really worth, per machine

Here is the part that makes miner heat different from every other heat source: the machine pays you back some of the electricity. The credit per kilowatt-hour is exactly the machine’s break-even power price — if a miner breaks even at 5 ¢/kWh, it is handing you 5 ¢ back on every kilowatt-hour you feed it, whatever you actually pay. Subtract that from your tariff and you have what the heat really costs.

Read the last two columns together. Every one of those machines produces heat more cheaply than a resistance heater does, because a resistance heater credits you nothing — that is the whole of the case for mining for heat, and it is a real case. None of them gets near gas or a heat pump. And note what the effective-rate column is not saying: at 18.44 ¢/kWh none of these machines is profitable as a miner, which is exactly what our ranking of every ASIC by the power price it breaks even at says. Buying one for heat is a different transaction. You are buying heat, and the mining is the rebate.

Figures derived 2026-08-15, at a hashprice of $31.26 per PH/s per day with bitcoin at $63,009 and network difficulty at 127.48 T. Hashprice moves every day and difficulty retargets roughly fortnightly, so the mining side of this table has a shelf life measured in weeks. The heat side does not move at all. If you want the method rather than the snapshot, it is in how to calculate mining profitability.

Every machine we stock, seen as a heater

The same catalogue, sorted by the only thing that matters when you are buying heat: what a kilowatt of heat capacity costs to buy. Wall power and price are read live from the shop, so this table cannot drift from what we are actually selling. The heat column is watts × 3.412, which is a unit conversion rather than an estimate — a miner does no mechanical work and stores nothing, so everything that goes in comes back out as heat.

Used machines → · What each cooling type makes you build → · Compare every machine on efficiency → · Browse everything in stock →

The spread in that last column is the finding. The cheapest heat capacity on the shelf and the most expensive differ by a factor of nearly fifty, and the heat coming out of both is identical — a BTU is a BTU whether the silicon that made it was fabricated in 2020 or last year.

When the heat is the product, buy the used machine

This is where mining for heat inverts the ordinary advice. Normally we would tell you to buy on efficiency, because J/TH is what decides whether a miner earns anything. Size the two candidates by heat instead of by budget and the argument changes shape.

• A used Antminer S19 at $159.99 gives you 29.23 TH per kilowatt of heat, and that kilowatt of heat capacity costs $49.23.
• A new S21 Pro 245T at $4,200 gives you 66.67 TH per kilowatt of heat, and the same kilowatt costs $1,142.86.

The new machine genuinely is better: per kilowatt of heat it earns about $1.17 a day more than the old one, which is why its heat costs $28.60 per million BTU against the S19’s $42.89. But it costs $1,093.63 more per kilowatt of heat to buy, and at that rate the premium takes 935 days of running to repay — two years and seven months, at 24 hours a day, every day.

A heater does not run 24 hours a day, every day. It runs during your heating season, and that is the number that decides this:

So the position we will defend: if you are buying a machine to heat something, buy the cheap used one. Nine years is longer than an S21 Pro will plausibly still be worth running, and you would be paying that premium up front for a benefit that only accrues in the months you want heat. Spend the $4,000 difference on insulation, or on twenty-five more used machines, or keep it. The honest counter-argument is the one case where it reverses: if the machine is going to run all year anyway and the heat is a winter bonus rather than the purpose, you are back to buying on efficiency and the new machine wins on the ordinary arithmetic. Which of those two you are doing is worth deciding before you buy, not after.

The used shelf is here, and the lot-scale version — grading, freight, what happens when a unit arrives dead — is on buying used miners in bulk.

Our own listing gives the wrong advice for this one use

The S19 above is titled “2,800 W Low-Power VNish Tune”, and for almost every buyer that tune is the point of the machine: it drops chip voltage and frequency to pull wall draw down from the factory 3,250 W. If you are buying it for heat, do not run that profile. At 2,800 W the machine makes 9,554 BTU/h instead of 11,089 — you have thrown away 14% of the thing you came for, and given up hashrate to do it. Efficiency tuning is a strategy for people whose problem is the power bill. When heat is the product, the power bill is the product too. Tell us what you are doing with it and we will tell you which firmware profile that argues for.

Air or hydro decides what you can actually do with the heat

An air-cooled miner gives you a jet of hot, dusty, very loud air out of one end. That heats the room it is in, and with ducting it can heat the room next door, and that is about the limit of it. A hydro machine gives you something qualitatively different: hot water in a closed loop, at a temperature and a flow rate you can plumb into a buffer tank, a hydronic floor, a greenhouse bench or a heat exchanger. Almost every serious heat-reuse installation in the world is a liquid one for exactly this reason.

Two honest caveats on that table. Every one of those machines is quote-only and none is a domestic proposition — the smallest carries 16,361 BTU/h and the largest 36,167, and most want three-phase power. And a hydro miner does not ship with a loop: the pump, the buffer, the heat exchanger and the rejection path are all yours to build, which what each cooling type actually requires covers in detail. If you want liquid heat at a domestic scale, the answer today is not on this shelf.

What actually stops people

We sell this hardware and we get these calls, so this list is what people discover after they buy rather than before:

• The noise. A full-size miner runs at about 75 dB. That is not a background appliance, it is workshop noise, and no enclosure makes it liveable in an occupied room. Our noise page has the measured ladder by machine, and it is the single most common reason a heat-reuse plan gets abandoned.
• The circuit. Everything except the small machines wants 200–277 V and a dedicated breaker. It will not run on a North American 120 V outlet, which a 1,500 W space heater will. The power requirements page has the breaker table.
• It has no thermostat. A miner is on at full output or it is off. It cannot modulate, so it will overheat a small room and then you will open a window, which is heat you paid for going outside.
• Summer. For the months you do not want heat, you are paying for the electricity and then paying again to remove the heat. The heating-season table above is the whole argument; the other half of the year is the other half of the bill.
• Dust and humidity. A miner pulls the room’s air through its heatsinks continuously. Lint, pet hair and sawdust are what turn a heatsink into a blanket, and why mining hardware fails is what happens next.

At lot scale, heat reuse is the case for old machines

The operations that buy used miners specifically for heat — greenhouses, drying sheds, workshops, pool and process heat — are buying the one thing an obsolete machine still does as well as a new one. Their arithmetic is the table above repeated at scale: heat capacity is the requirement, capital per kilowatt is the constraint, and the hashrate is whatever it turns out to be. It is also the most honest remaining use for a generation of hardware that cannot clear its break-even rate on an open tariff, which is why we keep stocking it rather than scrapping it.

We are not going to print a lot size, a lead time or a tier price on this page that we have not confirmed we can fulfil for your order. If you are sizing a heat installation and want to know what is actually on the floor, tell us the kilowatts of heat you need and where it is going, and we will tell you what we have. Start at the bulk used page, or the contact page. If you already have machines and the heat project is over, we buy used miners back.

Frequently asked questions

Can you actually heat a house with a Bitcoin miner?

Yes, physically — a 3,250 W miner is an 11,089 BTU/h heater, about 2.2 times a full-size 1,500 W space heater. Whether it is a good idea depends entirely on what you heat with now. It beats electric resistance heat. It does not beat natural gas or a heat pump. And it needs a 240 V circuit and somewhere that tolerates 75 dB, which rules out most houses before the economics get a say.

How much heat does an ASIC miner produce?

Essentially all of its electrical draw. Watts × 3.412 gives BTU per hour, and it is an identity rather than an estimate, because a miner does no mechanical work and stores no energy. An Avalon Q at 1,674 W makes 5,712 BTU/h; a used S19 at 3,250 W makes 11,089; an S21 Pro at 3,675 W makes 12,539.

Is mining for heat actually cheaper than just running a heater?

Yes, and by a knowable amount. Both convert electricity to heat at essentially 100%, so they cost the same to run — but the miner credits you its mining revenue, which per kilowatt-hour is exactly its break-even power price. At the 15 August 2026 hashprice a used S19 credits about 3.81 ¢/kWh, turning an 18.44 ¢ tariff into an effective 14.63 ¢. You pay for that discount in capital, noise and a dedicated circuit.

Which miner is best for heat reuse?

The cheapest one per kilowatt of heat that you can physically site, which on our shelf today is a used S19 at about $49 per kilowatt of heat capacity. The efficiency premium on a new machine takes around nine years to repay on a 2,500-hour heating season, which is longer than the machine is worth running. If noise or a 120 V-only circuit is the constraint rather than money, the answer is a small home-class machine instead, and that is a different conversation.

Can I use miner heat for hot water or a greenhouse?

Practically, only with a hydro-cooled machine, which gives you a closed water loop you can plumb into a heat exchanger. Air-cooled miners give you hot air, and extracting usable water heat from that is a project with poor returns. Every hydro machine we stock is quote-only, carries 16,000–36,000 BTU/h and mostly wants three-phase power, so this is a commercial installation rather than a domestic one.

What happens in summer?

You pay twice: once for the electricity and again for the air conditioning fighting it. That is why the heating-season hours matter more than any other number on this page. If the machine is only worth running for its heat, it should be switched off outside the heating season — and an idle machine keeps its resale value while a machine running below break-even converts hardware you own into electricity you paid for.

Where to go next

• ASIC miner cooling: air, hydro and immersion — the heat each machine makes, from the other side: what you have to build to throw it away.
• ASIC miner power requirements — voltage, breakers and why 120 V is not an option.
• ASIC miner noise, measured — how loud each machine in the catalogue actually is.
• Compare ASIC miners — every machine sorted by efficiency and running cost.
• Solar-powered Bitcoin mining — the other way to make the power cheap instead of making the heat useful.
• How to buy a used ASIC miner — what to check before you wire money for second-hand hardware.