Bitcoin's energy use: the honest numbers come as a range, not a headline
The University of Cambridge's best guess puts Bitcoin's electricity use at about 189 terawatt-hours in 2025, with a wide band of uncertainty around it. A Cambridge survey of miners found just over half their power came from renewables and nuclear, and most of the rest from gas and coal.
In the crypto village, the right to write the next page of the notebook doesn't go to the oldest or the richest. It goes to whoever solves a pointless puzzle first. Solving it takes lamp oil, lots of it, burned in every house that wants a chance.
That's the point. Burning oil makes it expensive for Ben to rewrite old pages, because he would have to out-burn everyone else. The cost is the lock.
So how much oil does the village burn? The honest answer isn't one number. It's a best guess with a floor and a ceiling, and the people who measure it say so themselves.
What it is
Bitcoin uses proof of work: miners run special computers that make trillions of guesses per second, and the winner adds the next block. Those computers run on electricity.
Nobody can meter the whole network, because miners are private companies and individuals spread across the world. So researchers estimate. The best-known estimate comes from the Cambridge Centre for Alternative Finance, which runs the Cambridge Bitcoin Electricity Consumption Index (CBECI).
How it works: how Cambridge builds the range
Cambridge starts from what is public: how much computing work the network does each day, and how efficient the available mining machines are. Then it makes three assumptions, set out in its methodology (CBECI methodology):
- The floor. Every miner uses only the most efficient machines. That gives the "theoretical minimum".
- The ceiling. Every miner uses the least efficient machines that still make a profit. That gives the "theoretical maximum".
- The best guess. Miners use "a combination of profitable hardware". This is the number usually quoted.
Now the figures, read from Cambridge's public data on 18 November 2026, at its default assumption of 5 US cents per kilowatt-hour:
- Whole years. The best-guess total was 158.72 terawatt-hours (TWh) for 2024 and 188.56 TWh for 2025 (CBECI yearly data).
- One day, with the band. The last daily reading in the data set, for 17 May 2026, put the annualised best guess at 165.38 TWh, with a floor of 81.94 TWh and a ceiling of 312.56 TWh (CBECI daily data).
Look at that band. The ceiling is almost four times the floor. Anyone who quotes a single number without the range is selling you a certainty that the researchers don't claim.
A second Cambridge study came at it from the other side: it asked the miners. Its April 2025 industry report estimated mining's annual electricity use at about 138 TWh, about 0.54% of global electricity consumption (Cambridge, 2025). Two methods from the same university, two different numbers. That's what honest measuring looks like.
Compare it with something named
A terawatt-hour means little on its own. So compare like with like: same year, named source.
The International Energy Agency estimates that data centres used about 415 TWh in 2024, around 1.5% of the world's electricity (IEA, 2025). Cambridge's best guess for Bitcoin in 2024 was 158.72 TWh. Divide: Bitcoin used roughly 38% as much electricity as all the world's data centres put together.
That comparison cuts both ways. Bitcoin is one network. Data centres run search, streaming, banking, email and AI for billions of people. Whether Bitcoin's share is worth it is a question of what you think it does. The numbers can't answer that for you.
Where the power comes from
The same Cambridge industry report surveyed 49 mining companies, together covering nearly 48% of the network's computing power. Their electricity mix:
- renewables 42.6%
- natural gas 38.2%
- nuclear 9.8%
- coal 8.9%
So about 52.4% came from renewables and nuclear combined, which the report calls sustainable. Now the caveat, in the report's own words: the respondents were "U.S.-centric", which "likely affects precise estimates of global activity". Miners in some countries, Russia for one, are under-represented (Cambridge, 2025).
Why it matters
- Critics have a point. Close to 200 terawatt-hours a year, with close to half from fossil fuels in the survey, is real demand on real power grids.
- Defenders have a point too. The energy is what makes rewriting history expensive, and a majority of surveyed miners' power came from low-carbon sources.
- Both sides like to drop the range. Critics quote the ceiling, fans quote the floor. Cambridge publishes all three.
Where people get it wrong
- "Bitcoin uses X TWh, full stop." It's an estimate. Ask for the range, the date and the electricity price assumed.
- "Each transaction uses as much power as a house for a week." Mining power doesn't rise with each transaction. Miners compete for the block reward whether a block holds ten transactions or thousands.
- "Mining runs entirely on renewables." Cambridge's survey found about 47% fossil fuels, and the survey leans towards US miners.
- "Mining runs entirely on coal." The same survey found coal at 8.9%.
- "All crypto uses this much." Coins on proof of stake don't mine at all. Our article on Ethereum explains its switch in 2022.
Ask for three numbers
Next time someone quotes Bitcoin's energy use at you, angry or proud, ask for three things: the source, the date and the range. If they have all three, you can have a real conversation.
If they have one big number and a strong opinion, they're not measuring the lamp oil. They're burning it.
Frequently asked questions
How much electricity does Bitcoin use?
The Cambridge Bitcoin Electricity Consumption Index puts its best guess at 188.56 terawatt-hours for 2025, at its default electricity price assumption. It's an estimate inside a wide range, because nobody can meter every miner.
Why is Bitcoin's energy use an estimate?
Miners are private operators spread across many countries, and no one sees all their electricity bills. Cambridge estimates from the network's computing power and the efficiency of mining machines, giving a floor, a ceiling and a best guess.
What energy sources do Bitcoin miners use?
A Cambridge survey published in April 2025, covering 49 firms with nearly 48% of the network's computing power, found 42.6% renewables, 38.2% natural gas, 9.8% nuclear and 8.9% coal. Cambridge notes the sample leans towards US miners.
Does Bitcoin use more electricity than data centres?
No. The IEA estimates data centres used about 415 terawatt-hours in 2024. Cambridge's best guess for Bitcoin that year was 158.72 terawatt-hours, roughly 38% of that figure.
Does each Bitcoin transaction use a fixed amount of energy?
No. Mining power depends on the reward miners compete for, not on how many transactions are in a block. Dividing total energy by the number of transactions gives a figure that sounds precise but doesn't describe how the system works.
Sources
- Cambridge Centre for Alternative Finance, Cambridge Bitcoin Electricity Consumption Index (CBECI)
- Cambridge Centre for Alternative Finance, CBECI methodology
- Cambridge Centre for Alternative Finance, CBECI yearly consumption data at the default electricity cost (read on 18 November 2026)
- Cambridge Centre for Alternative Finance, CBECI daily data at the default electricity cost (read on 18 November 2026)
- Cambridge Centre for Alternative Finance, Cambridge Digital Mining Industry Report (April 2025)
- International Energy Agency, Energy and AI: executive summary (April 2025)

