Ethereum Merge Dropped Energy Use 99.9%, Cambridge Reports

The Ethereum Merge energy story just got a hard number behind it. New research from the Cambridge Centre for Alternative Finance puts Ethereum's annual electricity consumption at 7.87 gigawatt-hours and continuous power demand at 0.90 megawatts, down from roughly 2.4 gigawatts before the September 2022 transition off proof-of-work. That's a reduction exceeding 99.9%. Cambridge built those figures from actual hardware measurements, not industry estimates, and annual carbon emissions came in at approximately 2.37 kilotonnes of CO2 equivalent.

How Cambridge Measured What Others Estimated

Most crypto energy studies pick a single assumed wattage and multiply it across all known nodes. Cambridge didn't take that shortcut. Researchers from the Cambridge Centre for Alternative Finance tested 20 combinations of Ethereum's main execution and consensus clients across two distinct hardware setups: a lightweight residential system that drew a median 18 watts, and a professional workstation pulling around 152 watts. The resulting network-weighted average landed at roughly 105 watts per node.

The team identified 8,522 discoverable full nodes across the network. About 36% operated on residential hardware; the remaining 64% ran in cloud or enterprise data centers. Geographically, the United States hosted 31% of those nodes, followed by Germany at 16%, Finland at 8%, and France at 6%. Those four countries together accounted for nearly 62% of the full-node network in the study.

That methodology distinction carries real weight. When you test actual client software on actual hardware configurations rather than plugging one flat estimate into a spreadsheet, you get a figure that reflects how the network actually operates. The 7.87 GWh result Cambridge produced isn't a guess or a midpoint in a range. It's derived from controlled testing across multiple client and hardware combinations.

What Does Ethereum's Energy Use Look Like Compared to Other PoS Chains?

Here's the part that deserves more scrutiny than the headline number suggests. In raw electricity terms, Ethereum doesn't claim the lowest spot. Solana topped Cambridge's comparison at about 13.48 GWh per year, with Ethereum second at 7.87 GWh. Networks like NEAR, Tron, and TON fell in the 3.6 to 5.1 GWh range. Cardano and BNB Chain both stayed below 1 GWh. The top-tier PoS networks Cambridge studied consumed 38 GWh combined.

Divide by market value, though, and the ranking flips. Ethereum consumed about 33 kilowatt-hours per $1 million of market cap, the second-lowest rate measured. BNB Chain was the only network that came in lower. Solana registered roughly 283 kWh per $1 million of market value, about 8.5 times Ethereum's rate. Cambridge was clear that this market-value framing is what places Ethereum near the bottom of energy intensity among major proof-of-stake chains.

Cambridge was also deliberate about what it would not calculate. A per-transaction metric was off the table because roughly 92% of Ethereum ecosystem activity now settles on layer-2 scaling networks. Applying mainnet-only data to a system where most transactions never touch mainnet would produce a misleading result. So Cambridge skipped it.

The report does not claim Ethereum uses the least electricity of any proof-of-stake network. That's worth stating plainly, because the framing matters. It uses more than most peers in absolute terms. The efficiency case only holds up after you divide by market cap, and Cambridge labels that distinction clearly rather than letting the 99.9% figure carry more weight than it should.

The Grid Mix That Shapes Ethereum's Carbon Footprint

The Ethereum Merge on September 15, 2022 was the inflection point for everything in this report. Before it, competitive proof-of-work mining set the energy floor. The network's continuous power demand sat at roughly 2.4 gigawatts. After the Merge, validators replaced miners, and that demand dropped to 0.90 megawatts. The security model changed. So did the electricity bill.

Where the electricity comes from now shapes most of what remains. Renewable energy supplies 39.4% of Ethereum's power, and nuclear contributes another 17%, giving a combined low-carbon share of 56.4%. Fossil fuels account for the other 43.6%, with natural gas the single largest source at 27.7%.

That breakdown is partly a node geography problem. Node operators in countries with cleaner grids lower the average carbon intensity. Operators in regions with heavier fossil fuel dependence push it back up. The 2.37 kilotonnes of CO2 equivalent Cambridge calculated isn't a fixed ceiling. It's a snapshot that moves with the grid composition at each node location.

Under Proof-of-Stake, electricity is no longer the price of security.

What Cambridge's Forward Guidance Actually Says

Cambridge kept its projections deliberately open. Lighter stateless verification could lower hardware requirements for node operators, reducing the per-node energy draw. But broader participation in running nodes could offset those gains entirely. The report treats future electricity demand as an open question, not a trend line.

That restraint is what makes this study worth taking seriously. Energy reports in the crypto space often tilt toward a preferred conclusion by building in efficiency assumptions that haven't materialized yet. Cambridge didn't do that. The 2.37 kilotonnes figure is a data point, not a destination. Where Ethereum's emissions land next depends on which direction the grid moves, not just on what the protocol does.