KNOWLEDGE BASE
← Back to Blog
28 Jul 2026 · digital-heating · engineering · economics
The COP Problem: When a Heat Pump Beats a Miner, and When It Doesn't
A commercial heat pump delivers 3 to 4 kWh of heat per kWh of electricity. A miner delivers 1. Here is the arithmetic on when the Bitcoin revenue covers that gap, and when it doesn't.
The strongest objection to Bitcoin heat reuse is not noise, or volatility, or regulation. It is a single number: coefficient of performance. Any engineer assessing your proposal already knows it. Here is the arithmetic, run honestly, including the cases where we lose.
The objection
A commercial hot water heat pump has a COP of roughly 3 to 4. Put in 1 kWh of electricity, get out 3 to 4 kWh of heat. It does this by moving heat that already exists rather than creating it. EECA puts commercial hot water heat pumps at a COP of around 3 to 4, against 0.75 to 0.95 for a fossil fuel boiler.
A Bitcoin miner has a COP of 1. Every watt in becomes a watt of heat out. That is the whole basis of digital heating, and it is also its weakness.
So on heat alone, a heat pump beats a miner by a factor of three to four. Not marginally. Comprehensively. Anyone selling you a miner as a straight heating upgrade over a modern heat pump is either not doing the maths or hoping you won't.
The miner's case rests entirely on the second output. The question is not whether the miner is a worse heater. It is. The question is whether the Bitcoin revenue covers the gap.
The arithmetic
To deliver the same amount of useful heat:
Read that as two different tricks. The heat pump divides your power bill by its COP. The miner divides nothing, but it hands part of the bill back to you as Bitcoin.
So the contest is simple. The miner only wins if the Bitcoin it earns is worth more than the saving the heat pump would have made. Set the two costs equal and solve:
Put a number on that. A heat pump with a COP of 3.5 does the same heating job on about 29% of the electricity, so it saves you the other 71%. Allow that a real miner loop captures around 90% of its heat rather than every last watt of it, and the target moves slightly. The miner has to earn about 74 cents of Bitcoin for every dollar of electricity it burns, just to draw level with the heat pump.
Where is your delivered electricity price, is mining revenue per kWh of electricity consumed, and is the fraction of miner electricity you actually capture as useful heat. A 90% capture rate is realistic for a well designed immersion to water loop once you account for pump power, tank losses and the heat that escapes the room rather than the circuit.
What a miner actually earns per kWh right now
Hashprice is the industry's revenue yardstick: what one petahash per second of computing power earns in a day, before you pay for any of the electricity. It moves with Bitcoin's price, network difficulty and transaction fees, so it is the single number that tells a miner what a day of work is currently worth.
Today it sits at roughly US$32 per PH/s per day. At NZD/USD around 0.58, that is about NZ$55.
Mining hardware efficiency is quoted in joules per terahash (J/TH), and the units line up conveniently: a petahash of hashrate running at 15 J/TH draws exactly 15 kW. Over 24 hours that is 360 kWh consumed to earn NZ$55, or 15.4 cents of Bitcoin per kWh burned. Same sum for any machine: NZ$55 divided by 24 times its J/TH figure.
Divide that revenue by the 0.743 threshold and you have the breakeven power price for each class of hardware.
| Fleet efficiency | Example class | Revenue per kWh (NZ$) | Breakeven delivered power price vs COP 3.5 (NZ c/kWh) |
|---|---|---|---|
| 13.5 J/TH | latest generation air | 0.171 | 23.1 |
| 15.0 J/TH | current top tier | 0.154 | 20.8 |
| 17.5 J/TH | S21 class | 0.132 | 17.8 |
| 21.5 J/TH | S19 XP class | 0.108 | 14.5 |
| 29.5 J/TH | S19j Pro class | 0.078 | 10.6 |
Read the right hand column as: below this delivered power price, the miner beats the heat pump on running cost. Above it, it doesn't.
The whole thing hinges on hashprice, not on the heating. At last October's hashprice of about US$49 per PH/s per day, a 15 J/TH machine clears the threshold at any delivered power price below roughly 32 NZ c/kWh, which puts it in play even on a retail tariff. Stress it the other way to US$20 per PH/s per day and the threshold collapses to about 13 NZ c/kWh. Hashprice has fallen 37% since October and is down again this month as large operators strip hashrate and redeploy sites into AI and cloud contracts. Any deployment sized on today's hashprice is sized on the most favourable variable in the model. Run yours at a 40% haircut and see if it still stands up.
What the table leaves out
The comparison above is running cost only. Three other factors move real projects, and two of them tend to move against us.
Machine life. A heat pump is a 15 year asset. A miner is a 3 to 5 year economic asset, because efficiency improvements retire it long before it physically fails. Amortised over its life, an ASIC adds roughly 2 to 3 NZ c/kWh, and the immersion tank, exchanger, pump and controls add more. The offsetting factor is that ASIC capex per kW of heat is currently collapsing because of the bear market, which is exactly why this is a better year to buy hardware than to sell it. Get a quote for both paths. Do not assume either one is capex-free.
Electrical capacity. This is the one that quietly kills sites. To deliver 30 kW of heat, a heat pump draws about 8.6 kW. Miners draw about 33 kW. On a 63 A three phase supply, that is the difference between fitting comfortably and paying for a supply upgrade. If the switchboard and the incomer cannot take it, the running cost comparison never gets a chance to matter.
The temperature ceiling is a hardware choice, not a physical limit. Whatsminer's M63S+ and M63S++ are specified to 70°C outlet water, and MicroBT points to the M65S for anyone wanting 80°C. That clears the sanitising temperature the Building Code requires for stored potable hot water, so a hydro loop can do the whole job rather than just the bottom of the lift.
Where the miner genuinely wins
Against a heat pump, on a good site, the case is tight. Against these, it is not tight at all.
- Resistive electric elements. COP 1.0. The threshold factor becomes 0.10 and the miner wins on day one with any positive mining revenue. A surprising volume of New Zealand commercial hot water is still resistive.
- LPG and diesel. Bottled LPG at around NZ$2.50/kg delivers useful heat near 21 NZ c/kWh after boiler losses. A miner on 25 NZ c/kWh commercial power delivers heat at roughly 10 NZ c/kWh net of mining revenue. That is not a marginal win.
- Capex constrained retrofits where the existing plant works, a heat pump upgrade cannot be funded this year, and there is spare electrical headroom sitting idle.
- Year round, steady heat loads. Pools, spas, accommodation hot water, wash-down, greenhouse baseload. Seasonal space heating is the weakest application because the miner has nowhere useful to send heat for half the year.
- Sites that want a controllable, sheddable load. A miner can be curtailed in seconds. With hydro storage at 113% of average and wholesale prices at their lowest winter level in over a decade, a spot exposed site with a dispatchable load is in an unusually good position right now.
- Operators who want the Bitcoin. If accumulating BTC is a goal in its own right, the heat is what makes the power bill defensible. That is a legitimate position, but it should be stated openly rather than smuggled into an engineering case.
Where it loses
Be equally clear about this. Do not proceed if the site is on a standard retail tariff, if the only load is seasonal space heating, if the electrical supply is already tight, if the requirement is process heat above what a hydro loop can deliver, or if heat availability is critical and there is no backup source. In those cases the heat pump is the correct answer and we will tell you so.
The answer is usually both
The framing of miner versus heat pump is mostly a marketing artefact. On a real site the sensible configuration is often a miner carrying a steady baseload circuit, with a heat pump or existing plant sized to cover the periods when mining is curtailed or the hardware is down. With hydro hardware the split is no longer forced by temperature, so it comes down to redundancy: you want a heat source that still works when hashprice does not.
Check your own numbers
Everything above uses: hashprice US$32.21 per PH/s per day, NZD/USD 0.58, heat recovery efficiency 90%, heat pump COP 3.5. Change any one of those and the thresholds move. The condition is the one at the top of this article, and it takes about five minutes in a spreadsheet. Or start with our mining economics calculator.
If you want that run against your actual heat load, tariff and switchboard capacity, get in touch. We will give you the number even when the number says don't do it.
Related
- Digital Heating — recovery methods and applications
- Hosting — if the site arithmetic doesn't work but you still want the hashrate
Sources
- Hashprice and difficulty: Hashrate Index weekly roundup, 13 July 2026
- Hashrate redeployment to AI and cloud: Bitcoin.com, 15th difficulty adjustment of 2026
- Commercial heat pump COP: EECA, hot water heat pumps in commercial buildings
- Whatsminer hydro outlet temperature: M63S+ / M63S++ specification, MicroBT support on 80°C operation
- NZ hydro storage and wholesale prices: Newswire, winter 2026 power bills, Newswire, lowest winter wholesale prices in a decade
Discuss your project
Run the numbers on your actual site.
Share your heat load, delivered power price, and available electrical capacity. ASIC.NZ will give you an honest assessment—even when a heat pump is the better answer.