What is bitcoin mining, and why does it use energy?

Quick answer: miners are computers that keep Bitcoin's ledger in order, and the electricity they burn is what makes cheating costly.

Miners run a never-ending contest: they try an enormous number of guesses until one finds the winning combination, adds a block of transactions to the ledger and collects a reward in bitcoin. It's called proof-of-work, and it works like a lottery where cleverness doesn't help: you can only keep trying, and every try costs electricity.

That is the point. According to the original Bitcoin whitepaper, anyone who wanted to alter an old block would have to redo the work of that block and of every block after it, and then outrun the honest miners. The same text names what is spent: CPU time and electricity. So energy isn't a side effect. It's the part that protects the ledger. And because the reward is paid in bitcoin, the more it is worth, the more machines are worth switching on.

How much energy does bitcoin mining consume?

Quick answer: a lot. Cambridge's estimate is about 138 TWh a year, around 0.5% of global electricity.

A terawatt-hour (TWh) is a billion kilowatt-hours. The Cambridge Digital Mining Industry Report (April 2025) estimates 138 TWh a year, up 17% on the year before, or 0.54% of global electricity. A preliminary preview from the same Cambridge centre, presented at a forum in Dallas and reported by trade press on 25 July 2026, points to about 190 TWh a year in December 2025; the full report is due later. Digiconomist shows 204.44 TWh (read on 14 October 2026; the page shows no date) using a different method, which starts from miner revenue and an assumed 5 cents per kWh. Nobody meters every machine, so these are all estimates.

Comparison, for scale only: the IEA estimates that global data centres used around 415 TWh in 2024, about 1.5% of global electricity. Cambridge's 138 TWh is about a third of that, the 190 TWh preview about 46% (our arithmetic). A comparison tells you size, not whether the use is worthwhile: pick a different yardstick and the headline changes.

Is bitcoin mining powered by renewable energy?

Quick answer: partly, and the share depends on who is counting.

In Cambridge's survey (49 firms, covering 48% of the network's computing power) 52.4% of the electricity is "sustainable" (42.6% renewables and 9.8% nuclear), natural gas is the biggest single source at 38.2% and coal is 8.9%. These are 2024 figures reported by the companies themselves. Cambridge puts emissions at 39.8 million tonnes of CO2 equivalent a year (about 0.08% of the global total), or 69.6 million with a different, IP-address-based model; Digiconomist says 114.03 million. In the 2026 preview the low-carbon share rises to 59.4% and hydropower overtakes gas as the top source, but those numbers are preliminary.

What does "excess energy" mean in bitcoin mining?

Quick answer: power that, at that moment and in that place, nobody is buying or nobody can deliver to buyers. A miner can be installed almost anywhere and, say its supporters, switched off quickly, which is why it can act as a customer of last resort. There are three documented types.

What is curtailment, the wind and solar that gets switched off?

When wind and solar produce more than the grid can use or carry, the operator asks plants to produce less. That is curtailment. The US EIA reports that California curtailed 3.4 million MWh (3.4 TWh) of wind and solar in 2024, 29% more than in 2023 and 93% of it solar. Had every kilowatt-hour gone to miners it would have been about 2.5% of Cambridge's 138 TWh (our arithmetic): real, but small.

In Great Britain the problem is wires, not oversupply. NESO, the system operator, explains that wind in the north can't always reach demand in the south, so it pays wind farms to switch off and pays other plants to switch on. Its August 2026 report puts wind curtailment at 413 GWh, 9.4% of the wind that could have been generated (13.3% in July 2026, 15.5% in August 2025). NESO doesn't mention mining; to use that power a machine would have to sit on the constrained side of the grid.

Texas is the documented case where miners did switch off. According to the MIT paper discussed below, during Winter Storm Elliott (December 2022) the large flexible loads in ERCOT's curtailment programme, almost all of them miners, cut about 1.4 GW. The authors add that the climate benefit of this kind of service is hard to measure.

What about hydropower?

Cheap hydro attracts miners, but surplus isn't guaranteed. In June 2018 Hydro-Québec paused processing service requests from crypto miners because demand exceeded capacity (that was 2018; check current rules). In Italy, a startup called Alps Blockchain puts miners inside hydroelectric plants and says up to 30% of a plant's output can go to mining, per Italian outlet greenme.it: company figures, not independently verified.

What about flared gas?

Oil wells also bring up gas; where there are no pipes or plants to sell it, it gets burned in a flare. The World Bank's 2026 tracker says 167 billion cubic metres were flared in 2025, worth an estimated $54 billion: the third annual rise in a row and the highest since 2019 (the page doesn't mention bitcoin). The mining idea is to run a generator at the well. Crusoe Energy, a US company quoted by Argus Media in 2021 for sites in North Dakota and Montana, says this cuts CO2-equivalent emissions by up to 63% compared with flaring: a company figure. Critics, as summarised by MIT, say it leaves the underlying fossil-fuel production untouched and may prolong it. The same paper cites a Science study estimating that US flares, counting unlit flares and incomplete combustion, destroy methane at 91.1% overall, not the 98% usually assumed.

What do independent studies and critics say?

Quick answer: that consumption is huge, that some benefits are real but hard to measure, and that reassuring figures need checking.

The closest thing to a neutral reading is MIT CEEPR working paper 2023-11, "Climate Impacts of Bitcoin Mining in the U.S." (June 2023). What it actually says:

• It analysed 13 listed miners operating in the US, about a quarter of the network at the end of 2022. Their electricity has a carbon intensity of 397 g CO2/kWh, almost the same as the US grid average (387), and their emissions (7.2 million tonnes a year) exceed Vermont's. The calculation uses average grid factors, not specific contracts.
• The Bitcoin Mining Council, an industry body, says 58.9% of mining energy is sustainable; the authors call that survey-based and unverifiable.
• Possible benefits (flexible load for the grid, flared gas, orphaned wells, renewables) are hard to quantify, and there is no complete record of miners who built new renewable capacity. Heat reuse looks limited to pilot projects.
• The answer is transparency and disclosure, to avoid greenwashing without assuming every kilowatt-hour is bad.

Limits: it is a working paper, so CEEPR says it has no selection or peer review; it covers the US and end-2022 data; and three of the authors say they co-founded a company that sells sustainability metrics for digital assets.

How do I spot a biased number?

Quick answer: ask who is speaking, how they measured, and exactly what they are talking about.

Four questions for any figure: who is giving it and what do they gain? How was it measured (survey, model, direct metering)? What period and region? And is it about renewable energy, excess energy or zero-emission energy? Those are three different things: "59% renewable" doesn't say where or when the power is used, or whether it would have been used anyway.

So, does bitcoin mining waste energy?

Quick answer: it depends on what you call waste and which power you mean.

Three honest answers. First: it is a huge, deliberate use of electricity. If you think a money system outside any government's control isn't worth that power, you'll call it waste; if you think it is, you'll call it the price of security. That's a value judgment, not a data point. Second: surplus energy exists and miners sometimes use it (Texas during grid stress, some flared gas, some hydro), but independent work says the climate benefits are hard to measure, and US-listed miners sat at the grid average. "It only uses wasted energy" and "it's boiling the planet" both oversimplify. Third: surplus isn't a fixed stock. Grids change, as Quebec's 2018 pause shows. Whenever you see a number, ask: from whom, and measured how?

Quick answer: about as much as the electricity it draws: an 800 W miner behaves like an 800 W space heater.

According to MARA, a mining company, nearly all the electricity used by mining servers is converted directly into thermal energy. So 800 W of mining is about 2,730 BTU per hour (our conversion), or 19.2 kWh of heat a day. Moving that heat into a water pipe takes a purpose-built system: a job for licensed professionals, and we give no wiring or plumbing instructions, DIY or otherwise.

When does a bitcoin heater beat a normal heater?

Quick answer: it beats a resistance heater; it doesn't beat a heat pump unless power is very cheap.

A heat pump moves heat instead of making it. The IEA says heat pumps on the market are three to five times more energy efficient than natural gas boilers (2022), and the US Department of Energy says heat pump water heaters can be two to three times more efficient than electric resistance ones (2020).

A worked example with the assumptions in the open: Canaan's Avalon Mini 3 (37.5 TH/s, 800 W, specs from a Hashrate Index blog that doesn't say whether it measured them), hashprice $39.87 (Hashrate Index, 28 September 2026), running 24/7, no pool fees. Gross revenue: about $1.50 a day. The US average residential electricity price was 18.31 cents per kWh in July 2026 (EIA, preliminary).

• Against a resistance heater: the same heat, plus about $1.50 a day in bitcoin.
• Against a heat pump with a COP of 3 (three units of heat per unit of electricity, our assumption): for 19.2 kWh of heat the pump uses 6.4 kWh, the miner 19.2. The extra 12.8 kWh costs more than the $1.50 of revenue whenever power costs more than about $0.117 per kWh. At the US average the balance is about -$0.85 a day.

These are examples, not forecasts. In summer, when you don't need heat, it is plain mining: see Is Bitcoin mining worth it in 2026?.

Why is a miner's heat "lukewarm"?

Quick answer: an air-cooled miner blows warm air; district heating needs much hotter water.

MARA says conventional air-cooled data centres produce heat at roughly 25-35 °C, while district heating networks generally need about 80 °C or more. According to the Hashrate Index data above, the exhaust air of an Avalon Mini 3 runs from about 35 to 55 °C depending on the mode. To go higher you need liquid cooling or heat pumps: MARA reports 50-78 °C from its miners and Canaan about 80 °C.

Who is really doing it? Cases with sources

Quick answer: real projects exist, mostly told by the companies. A 2023 MIT working paper said heat reuse looked limited to pilot projects (WP 2023-11).

Can I buy a bitcoin heater for my home?

Quick answer: yes, but the figures are the companies'.

Heatbit's official page presents the Bitair: 25 W, 1.2 TH/s, 25 to 44 dB, also sold as an air purifier, at $179. It says earnings are not guaranteed and that a solo miner's odds of finding a block are 1 in 17,200 a year. Canaan introduced the Avalon Mini 3 as a miner that warms rooms, at $899 on pre-order, according to its statement (The Block, January 2025). For "bitcoin water boilers" with a built-in miner we found no verifiable official spec sheets, so we quote no figures.

What about noise and safety?

An Antminer S21 XP makes 75 dB according to Kryptex; the Avalon Mini 3 about 33 to 55 dB by mode according to the Hashrate Index data; Heatbit claims 25-44 dB. Connecting a miner to electrical or plumbing systems is a job for a licensed professional.

Quick answer: five numbers.

• Hashprice: what one petahash per second of computing power earns per day, in dollars: $39.87 on 28 September 2026, with bitcoin at $83,479 and difficulty at 132.76 T (Hashrate Index).
• Efficiency: joules per terahash (J/TH), lower is better (Cambridge). Our example: an Antminer S21 XP, 270 TH/s, 3,645 W, 13.5 J/TH, 75 dB, about $6,000 (Kryptex, which doesn't name the seller).
• Price of power: the number that decides.
• Difficulty: it adjusts itself to the computing power on the network (whitepaper): more machines, less revenue for each.
• The halving: at block 1,050,000 the reward falls from 3.125 to 1.5625 bitcoin, estimated around 13 April 2028 (CoinWarz, date varies). With price and difficulty unchanged, revenue per TH/s roughly halves: transaction fees are only 0.70% of block rewards (Hashrate Index).

How much does a miner earn? A worked example

Quick answer: at US household prices it loses money; it earns only on cheap power, and the halving changes the maths.

Assumptions: S21 XP running 24/7, hashprice $39.87, pool fees and taxes ignored. Gross revenue: 270 TH/s × $39.87 per PH/s = about $10.77 a day. Consumption: 87.5 kWh a day. Break-even power price: $0.123 per kWh, falling to about $0.062 after the halving. For reference, the US average residential price was 18.31 cents per kWh in July 2026 (EIA, preliminary).

At Blockware's $0.07 all-in hosting rate the gross margin is about $4.64 a day, roughly 3.5 years to recover $6,000. With a 15% fee like Sazmining's, and power at the same rate (our assumption), about $3.03 a day and 5.4 years. After the halving, with everything else equal, both lose money. These are examples, not forecasts. If the heat replaces electric resistance heating the picture changes: see Bitcoin heater: does a mining space heater really work?.

Mine at home or use hosting?

Quick answer: at home you deal with noise, heat and your electrical service; with hosting the machine lives elsewhere and you rely on a provider.

At home, 75 dB is loud and 3,645 W is a lot of appliances at once. With hosting, a provider houses, powers and monitors the machine. Two examples, with what their own pages say (not recommendations):

• Sazmining (page read on 14 October 2026): you buy the machine and own it; 15% of the bitcoin mined; electricity at a fixed per-kWh rate with no markup (monthly fee non-refundable); sites named in its FAQ: Texas and Paraguay; Luxor (0.7% fee) and OCEAN pools. If the service ends within a year the machine passes to Sazmining; one-year manufacturer warranty; 5% discount if you pay in bitcoin or stablecoins; no refunds on machines.
• Blockware: $0.070 per kWh "all-in"; sites in Texas, North Dakota, Missouri and Kentucky; the customer owns the machines (info page).

How do bitcoin mining scams work?

Quick answer: guaranteed returns, cloud mining and fake shops.

Serious providers don't guarantee profits: Sazmining says so on its own page. The HashFlare case shows what cloud mining fraud looks like: according to the US Department of Justice, its equipment mined with less than 1% of the computing power it claimed, and customers had paid over $575 million; the founders were sentenced in 2025 (Decrypt). On fake shops, the FTC says to search the seller's name with "complaint" or "scam", never to buy from sellers who insist on gift cards, wire transfers, payment apps or cryptocurrency, and that paying by credit card protects you best. Some legitimate sellers accept bitcoin, so then it matters even more to know who they are.

What should I ask a miner seller?

1. Who are you? Company name, address, tax ID, how long have you sold miners: can I verify them?
2. New or used? Serial number, manufacturer's invoice, warranty: who honours it, and where?
3. Power draw (W) and efficiency (J/TH): where did you measure them, and do they match the manufacturer's sheet?
4. Final price with tax, duties, shipping, insurance and payment fees, in a written quote?
5. How do I pay and what protection do I get (card, bank transfer, bitcoin)? Refunds and cancellation?
6. Who owns the machine and the bitcoin? Does the payout go to my wallet? Which pool?
7. If the hosting ends or you go out of business, what happens to the machine, at what cost and how fast?
8. Power rate and fees: what do they cover (kWh, rack, network, repairs) and when can they change?

If any answer includes a guaranteed return, stop.

What are the hidden costs: duties, tax, shipping?

Duties depend on the machine's customs classification and origin. The Block reported total US duties, from 7 August 2025, of 21.6% on machines from Indonesia, Malaysia and Thailand and 57.6% on China-made ones (The Block); we couldn't confirm today's rates, so ask for the landed cost in writing. In the UK, GOV.UK says VAT applies to the whole package value (goods, postage, packaging, insurance) plus any duty, paid to the delivery company before delivery or on collection above £135.

Is bitcoin mining legal, and what about tax?

Just pointers. These official pages treat mining as a taxable activity, not a banned one; local rules on power, noise and zoning vary. The IRS says income from digital assets is taxable and lists mining among the receipts to report. HMRC says whether mining is a trade depends on activity, organisation, risk and commerciality; if not, the sterling value at receipt is miscellaneous income. The CRA says that in most cases mining is a business, and ASIC miners can qualify for capital cost allowance. Talk to an accountant before you buy.