Ethereum's continuous power demand now sits at approximately 0.90 megawatts, according to a new study from the Cambridge Centre for Alternative Finance. Annual electricity consumption is estimated at 7.87 gigawatt-hours, with carbon emissions of roughly 2.37 kilotonnes of carbon dioxide equivalent per year. For context, the British Museum uses about 16.18 gigawatt-hours annually, meaning Ethereum now consumes less electricity than a single mid-sized public institution.
The Cambridge Centre for Alternative Finance noted that Ethereum's "electricity consumption and carbon emissions have fallen by more than 99% since it migrated to Proof-of-Stake, transforming its environmental footprint to a level well below that of small public institutions."
The drop traces directly to the Merge on September 15, 2022, when Ethereum replaced proof-of-work mining with proof-of-stake validation. Cambridge estimated that continuous power demand fell from approximately 2.4 gigawatts before the transition to 0.90 megawatts afterward, a reduction the study calculated at more than 99.9%. Alexander Neumüller, research lead at Cambridge's digital assets energy program, summarized the structural shift:
"Under Proof-of-Stake, electricity is no longer the price of security."
How Cambridge measured the network
Rather than applying a single assumed figure to every node, Cambridge tested 20 combinations of Ethereum's main execution and consensus software clients across two hardware setups. A lighter residential system drew a median of 18 watts. A workstation used for professional deployments drew approximately 152 watts. From those measurements, the study calculated a network-weighted average of roughly 105 watts per node.
Researchers identified 8,522 discoverable full nodes. Around 64% operated in cloud or enterprise data centers, while 36% ran on residential hardware. The United States hosted 31% of the nodes, Germany hosted 16%, Finland 8%, and France 6%. Those four countries together accounted for nearly 62% of the full-node network Cambridge measured.
Cambridge also declined to produce a per-transaction energy estimate. The reason was that approximately 92% of Ethereum ecosystem transactions now settle on scaling networks, which would make a mainnet-only calculation incomplete and potentially misleading.
How Ethereum compares to other proof-of-stake networks
Ethereum's absolute electricity consumption remained higher than most proof-of-stake networks in the study. Solana ranked highest at about 13.48 gigawatt-hours annually, placing it above Ethereum on both total consumption and market-value-adjusted intensity. The full set of networks in Cambridge's top-tier proof-of-stake comparison used 38 gigawatt-hours combined.
The more significant ranking appears when Cambridge adjusts electricity use for market value. Ethereum consumed roughly 33 kilowatt-hours per $1 million of market value. That placed it second-lowest among the networks assessed, behind only BNB Chain. Solana recorded approximately 283 kilowatt-hours per $1 million, roughly 8.5 times Ethereum's rate. NEAR, Tron, and TON ranged from 3.6 to 5.1 gigawatt-hours annually, while Cardano and BNB Chain stayed below 1 gigawatt-hour.
What grid sources now determine about emissions
With proof-of-work mining removed from the picture, Ethereum's remaining carbon footprint depends almost entirely on the electricity grids that serve its nodes. Cambridge estimated that renewable energy supplied 39.4% of the network's power and nuclear energy supplied 17%, putting low-carbon sources at 56.4% of the total electricity mix. Fossil fuels supplied the remaining 43.6%, with natural gas representing the largest single fossil source at 27.7%.
The geographic distribution of nodes drives that mix. Nodes concentrated in the United States and Germany pull from grids with different carbon intensities than those in Finland and France, where nuclear and hydropower have larger shares. Cambridge said local grid carbon intensity now determines Ethereum's carbon output more than the proof-of-stake protocol itself.
The report treats future demand as uncertain. Cambridge noted that lighter stateless verification could reduce hardware requirements, but wider node participation could offset those efficiency gains. The study does not project that total electricity use will fall further on the assumption that technical improvements will dominate network growth.
Alex Shepel
Disclaimer: All materials on this site are for informational purposes only. None of the material should be interpreted as investment advice. Please note that, despite the nature of much of the material created and hosted on this website, HODL FM operates as a media and informational platform, not a provider of financial advisory services. The opinions of authors and other contributors are their own and should not be taken as financial advice. If you require advice, HODL FM strongly recommends contacting a qualified industry professional.
