A Cambridge Centre for Alternative Finance assessment has placed Ethereum near the efficient end of major proof-of-stake networks, despite its comparatively large absolute electricity demand. Researchers estimate that the blockchain consumes 7.87 gigawatt-hours annually, equivalent to continuous power draw of about 0.90 megawatts. More strikingly, Ethereum uses roughly 33 kilowatt-hours for every $1 million of market value, the second-lowest intensity in the comparison, behind BNB Chain. Ethereum now presents an environmental contradiction: a globally significant network operating with modest energy intensity. The finding reframes sustainability debates that still treat all public blockchains as similarly power hungry systems in today’s market.
Cambridge measures Ethereum from the wall socket
Cambridge built its estimate from physical infrastructure rather than theoretical validator assumptions. Researchers measured wall-plug electricity use across 20 combinations of Ethereum execution and consensus clients, then mapped those readings onto approximately 8,522 discoverable full nodes. A typical residential configuration drew a median 18 watts, while workstation-class equipment required about 152 watts. Factoring in the network’s hosting profile produced a weighted average of roughly 105 watts per node. The study separates the machines processing network data from nearly 894,000 validators securing Ethereum economically, avoiding the misleading assumption that every validator represents a separate computer drawing additional electricity overall.
The comparison becomes less intuitive when total consumption and economic scale are separated. Ethereum used more electricity than most proof-of-stake networks assessed, while Solana registered the highest annual demand at approximately 13.48 gigawatt-hours. Yet Solana’s market-value-adjusted intensity reached roughly 283 kilowatt-hours per $1 million, around 8.5 times Ethereum’s level. The networks examined consumed about 38 gigawatt-hours collectively. Scale makes Ethereum look power intensive in absolute terms but comparatively lean relative to the value it supports. That distinction matters because raw electricity totals can penalize larger networks without showing how much economic activity or capital each system represents at current valuations.
Low emissions leave concentration questions unresolved
Ethereum’s remaining climate impact depends increasingly on geography rather than consensus design. Cambridge estimated annual emissions of about 2.37 kilotonnes of carbon dioxide equivalent, approximately 99.98% below the final pre-Merge level. Sustainable sources provided 56.4% of the network’s electricity, including 39.4% from renewables and 17% from nuclear generation. Natural gas remained the largest individual source at 27.7%. The carbon footprint now follows the grids hosting Ethereum’s nodes more than the protocol itself, meaning emissions could decline as national power systems serving concentrated node hubs decarbonize, even if the blockchain’s electricity consumption remains broadly unchanged in absolute terms over coming years.
The findings also expose concentration risks concealed by low energy figures. The United States hosts about 31% of full nodes, followed by Germany at 16%, Finland at 8% and France at 6%. Roughly 64% operate in cloud or enterprise facilities, while three providers, Hetzner, Amazon Web Services and OVH, host around 40% of nodes. Meanwhile, about 79% of execution-layer nodes use Geth or Nethermind. Ethereum’s sustainability achievement does not automatically resolve its infrastructure dependencies. The network may consume little power relative to its market value, yet geographic, hosting and software concentration remain important resilience considerations for policymakers and participants alike.