A single Bitcoin transaction carried an average water footprint of about 16,000 litres in 2021 once the electricity and cooling behind it were counted — roughly 6.2 million times the water behind one credit-card swipe, and the figure rises and falls with the price of Bitcoin itself

In 2021, the average transaction recorded on the Bitcoin blockchain carried an estimated water footprint of roughly 16,000 litres, once the water used to generate the electricity behind the mining process and to cool the machines doing the work is added up. That figure, drawn from a peer-reviewed study published in the Cell Press journal Cell Reports Sustainability, works out to roughly 6.2 million times the water footprint of a single credit-card swipe, based on de Vries’ own estimate — discussed in a later interview with Greenpeace rather than the paper itself — that one card transaction carries a footprint of around 2.6 millilitres.

The research, titled “Bitcoin’s growing water footprint,” was published online in late November 2023 by Alex de Vries, a data scientist with a background in financial economics who tracks the industry’s resource use. De Vries built his estimate around two components. The larger of the two, accounting for roughly 80 to 90 percent of the total, is indirect water use: the water consumed at power plants to generate the electricity that miners buy, whether through evaporative cooling at thermoelectric plants, hydropower reservoir evaporation, or other generation methods. The paper calculates this by combining data on where Bitcoin’s computing power is located with regional water-intensity figures, expressed in litres of water per kilowatt-hour of electricity produced on that grid.

The remaining share, direct water use, covers cooling and humidification equipment on-site at mining facilities, measured using an industry metric called water usage effectiveness, or WUE. For large-scale US mining operations, the study estimated a weighted-average WUE of between 0.25 and 1.03 litres per kilowatt-hour, depending on climate and facility design. Location matters enormously to the total: the study estimated Kazakhstan accounted for 63 percent of Bitcoin’s 2021 water footprint while hosting only around 14 percent of its computing power. Average water intensity across Bitcoin’s power mix rose from 8.63 litres per kilowatt-hour in 2020 to 15.0 litres per kilowatt-hour in 2021, the paper found.

What the comparisons mean

News coverage of the study, including reporting by ScienceDaily and Fast Company, popularized the comparison that a single Bitcoin transaction’s water footprint is roughly equivalent to filling a backyard swimming pool. The 6.2-million-times-a-credit-card-swipe comparison, cited in de Vries’ own commentary and repeated across outlets covering the paper, follows from dividing the 16,000-plus-litre figure by that credit-card baseline, and it moves with Bitcoin’s price. When the market fell during the 2022 downturn, mining activity and its associated electricity and water use fell with it, only to climb back toward record levels as the price recovered in 2023. Because a higher Bitcoin price makes mining more profitable and draws more computing power onto the network, the water footprint tracks the price cycle instead of holding steady. De Vries projected the network’s total 2023 water footprint at roughly 2,237 gigaliters, an increase over 2021 despite the intervening price crash.

Extrapolated to the United States, which absorbed a large share of mining capacity after China’s ban, de Vries estimated domestic mining operations consume water at a scale comparable to roughly 300,000 US households, or a city on the order of Washington, DC, a comparison he discussed in an interview with Greenpeace.

Where the estimate is contested

The methodology has drawn pushback. Critics covered in a DL News investigation note that per-transaction water figures are largely independent of transaction count — a block can bundle thousands of individual payments, so dividing total resource use by the number of on-chain transactions can make the per-transaction number swing based on how the network happens to be used that day — a shift in usage patterns, not a genuine efficiency change.

Bitcoin advocate Daniel Batten has gone further, arguing in commentary covered by the same report that the swimming-pool framing overstates the case by counting indirect water use from power generation that would occur with or without Bitcoin mining, and that applying the same accounting method to other electricity-intensive industries — banking, AI, search engines, construction, or media — would produce similarly large-sounding numbers. Separately, a UN-affiliated water researcher, Kaveh Madani, flagged in the same report that “water footprint” and “water consumption” are technical terms with distinct meanings in hydrology, and that using them interchangeably in public communication risks overstating the case. The Register’s coverage of the dispute frames it as a live disagreement between the paper’s author and its critics — not a settled question.

Why it matters beyond the headline number

The debate sits inside a larger and less disputed trend: much of Bitcoin mining has moved into regions and grids with high water intensity, including coal-heavy and drought-prone areas, since China’s 2021 mining ban scattered the industry’s computing power across new jurisdictions. That shift, documented in mining-location data tracked separately by the University of Cambridge, means Bitcoin’s water footprint is now heavily influenced by decisions about where mining farms are sited and what power sources they draw on, in addition to how much Bitcoin is worth on any given day. For a network whose environmental accounting has focused almost entirely on carbon and electricity for over a decade, the 2023 paper marked one of the first attempts to put a number on the water side of the ledger as well.