Three years ago, Ethereum did something almost no major infrastructure network has ever done: it cut its own power draw by over 99% without shutting down for a single block. No downtime, no forked chain, no do-over. Just a quiet Thursday night in September 2022 when the network's engine got swapped out mid-flight.
That event is known as the Merge, and it remains the single biggest case study anyone has for what "sustainable blockchain" could actually mean in practice — not as a marketing slide, but as measured electricity draw. So let's look at what the numbers actually say now, several years out, rather than what the press releases said the week it happened.
What the Merge actually changed
Before September 15, 2022, Ethereum ran on proof-of-work (PoW) — the same mechanism Bitcoin still uses. Miners raced to solve cryptographic puzzles using specialized hardware, and that race is what burned the electricity. More competing miners meant more total hashing power, which meant more total energy, regardless of how many transactions the network actually processed.
Proof-of-stake (PoS) replaced that entirely. Instead of miners competing on computation, validators are chosen to propose and confirm blocks based on how much ETH they've staked, with the right to validate rotating through the validator set. There's no puzzle-solving race — a validator node can run on hardware not much more powerful than a home server, because winning the right to validate isn't a function of raw compute.
The Cambridge Centre for Alternative Finance (CCAF), which has tracked this most rigorously, estimated that continuous power demand fell from about 2.4 gigawatts before the transition to 0.90 megawatts afterward — a reduction of more than 99.9%. To put 2.4 GW in perspective, researchers have repeatedly compared pre-Merge Ethereum's draw to the electricity demand of a nation-state the size of Iceland.
The current numbers (not the 2022 headlines)
A lot of "Ethereum went green" content still circulates screenshots from the week of the Merge itself. What's more useful is what CCAF found when it re-audited the network years later, using real hosting and hardware data instead of theoretical estimates.
The updated 2026 report puts Ethereum's annual electricity consumption at approximately 7.87 gigawatt-hours, drawn from an audit of 8,522 physical nodes, finding a network-weighted average of 105 watts per node. That's not a rough model — that's counted hardware.
The climate math follows from there: CCAF put the network's annual carbon footprint at roughly 2,370 tonnes of CO₂-equivalent, about 99.98% lower than the last level before the Merge. And on the energy mix itself, the researchers found that renewable energy supplied 39.4% of the network's power and nuclear energy supplied 17% — meaning well over half of Ethereum's already-tiny footprint comes from low-carbon sources in the first place.
How Ethereum stacks up against other proof-of-stake chains
Here's where it gets genuinely interesting, and where a lot of coverage stops too early. Post-Merge Ethereum isn't just cleaner than its old self — CCAF also placed it against other major PoS networks and found it's not automatically the frontrunner just because it's PoS.
By raw annual electricity use, Ethereum ranked second in the sample after Solana, which consumes about 13.48 GWh a year according to Crypto Carbon Ratings Institute data. That's a useful reminder that "proof-of-stake" isn't a single efficiency tier — validator count, node redundancy, and client software all matter.
Where Ethereum does pull ahead is when researchers adjust for the size of the network's economic footprint. Ethereum ranks second-lowest in market-value-adjusted energy intensity among the major proof-of-stake networks Cambridge studied globally, meaning it uses relatively little energy per dollar of value it secures — even though its absolute energy use isn't the lowest on the list.
The researchers were also careful to flag what these numbers don't tell you. The energy figures don't reflect network throughput and don't indicate energy use per individual transaction — the team explicitly discarded "energy per transaction" as a metric, since it can be manipulated simply by batching more activity onto the same base draw. If you see a headline claiming an exact watt-hour cost "per Ethereum transaction," treat it skeptically — that's not how the network's power draw actually works anymore.
Why proof-of-stake changes the underlying economics, not just the electricity bill
Alexander Neumüller, CCAF's research lead on digital assets, put it in a way that's easy to miss the significance of: "Under Proof-of-Stake, electricity is no longer the price of security."
That's the structural shift people miss when they only look at the kilowatt-hours. Under PoW, security and energy consumption were the same variable — you could not have one without burning more of the other. Attacking the network meant out-competing everyone else's hash power, so security scaled directly with global energy expenditure. Under PoS, security instead scales with capital at stake and the economic penalties (slashing) for misbehaving validators. You can add more security — more validators, more stake — without the network's total power draw scaling up in the same way. That decoupling is arguably a bigger deal than the headline percentage, because it means Ethereum's energy footprint doesn't have to keep growing as the network grows.
Where this leaves the "crypto is bad for the climate" conversation
It doesn't settle it — and it shouldn't, because Ethereum's transition doesn't say anything about Bitcoin, which still runs on PoW and still consumes energy at nation-state scale. Lumping "crypto" together as one energy story was never accurate, and the Merge is the clearest proof of that: two networks, same broad category, now sitting several orders of magnitude apart on power draw.
It also doesn't mean every claim wearing a "green blockchain" label deserves the benefit of the doubt. As Wes Ibarra, who covers the intersection of climate tech and crypto markets, puts it: "Most 'green crypto' claims fall apart the moment you ask for the methodology. My job is to ask anyway — and to give credit on the rare occasions the math actually holds up." The Ethereum Merge is one of those rare occasions — it's backed by repeated, independent, hardware-level audits rather than a single self-reported figure. That's a meaningfully higher bar than most sustainability claims in this industry clear, and it's worth being precise about why, instead of just repeating "99% less energy" as a slogan.
You can read more of Wes's reporting on where blockchain sustainability claims hold up — and where they don't — on his author page.
What's actually worth watching next
The Merge answered the electricity question. It didn't answer everything else people conflate with "blockchain and the environment" — e-waste from decommissioned mining hardware, the actual additionality of tokenized carbon credits, or whether DePIN sensor networks paying people in tokens produce data anyone can trust. Those are separate, ongoing stories, and lumping them in with the energy number is exactly the kind of shortcut worth resisting. For now, on the specific question of watts and grams of CO₂, Ethereum's proof-of-stake transition holds up to scrutiny — which, in this industry, is rarer than it should be.