This is a more detailed version of introductory article I wrote at ethereum.org
Ethereum's current energy expenditure with proof-of-work (PoW) is too high. Resolving energy expenditure concerns without sacrificing security and decentralization is a significant technical challenge and has been a focus of research and development for years. This article will explore why building Ethereum has had a substantial environmental impact, how it compares to other chains and how the upcoming network upgrade to proof-of-stake (PoS) will dramatically change this. This is a undeniably a very controversial issue that attracts a lot of vitriol on all sides of the debate. This article is an attempt to pick some signal out of the noise.
Transactions on the PoW Ethereum blockchain are validated by miners. Miners bundle together transactions into ordered blocks and add them to the blockchain. The new blocks get broadcast to all the other node operators who run the transactions independently and verify that they are valid. Any dishonesty shows up as an inconsistency between different nodes. Honest blocks are added to the blockchain and become an immutable part of the chain’s history.
This only works if there is a cost associated with mining and unpredictability about which specific node submits the next block. This is a design feature that ensures miners have “skin in the game” - they have expended a valuable asset (energy and capital to buy hardware) in order to participate in securing the network and therefore have something to lose by acting dishonestly. These conditions are met by requiring a miner to solve a computational puzzle faster than any other miner. A valid solution demonstrates that the miner has the necessary hardware and has expended energy - there is no known way to solve the puzzle other than brute force computation. This creates competition between miners - the fastest computer wins the race more often and is rewarded in ether. To successfully defraud the blockchain, a dishonest miner would have to consistently win the proof-of-work race, which is very unlikely and prohibitively expensive. A malicious miner would need more than half of the total computing power of the entire network. Then, they could force an alternative history into the canonical chain. This is unlikely because of the enormous investment in hardware and energy expenditure that would be required. Therefore, under PoW, energy keeps the network secure.
Proof-of-work is a robust way to secure the network and enforce honest changes to the blockchain, but it is problematic for several reasons. Since the right to mine a block comes from a brute force computation race, miners can increase their odds of success by investing in more powerful hardware. Whereas for Bitcoin this can include producing ASICs (application specific integrated circuits) that optimize the hardware specifically for PoW calculations, Ethereum’s PoW, although it is not technically ASIC-resistant it is still primarily mined using GPUs (Graphical Processing Units) rather than ASICs and FGPAs (Field-programmable gate arrays). This limits the hardware turnover and energy expenditure but has other consequences, such as limiting the supply of GPUs for other purposes (like gaming).
There are many nuanced debates around PoW energy consumption and its relationship to environmentalism. By definition there is no escaping the energy expenditure associated with PoW - it is precisely that energy expenditure that secures the network from spam and malicious actions. However, the implications for carbon emissions can be ameliorated to some degree by choosing to use energy from responsible, renewable sources. There are examples of energy for mining coming from excess that would otherwise go to waste, or being used to regulate energy outputs from powerplants. This is a sticky issue with strong counter-arguments that we will return to later in this article.
Ethereum has used proof-of-work since genesis, but a greener future for Ethereum is already being built in the form of a proof-of-stake (PoS) chain. Migrating off of proof-of-work and onto proof-of-stake has always been a fundamental goal of Ethereum. However, developing a proof-of-stake system that adheres to Ethereum's core principles of security and decentralization is not trivial. It has required a lot of research and breakthroughs in cryptography, cryptoeconomics, and mechanism design to get to a point where the transition is possible.
Under proof-of-stake, PoW puzzle-solving is not necessary. Miners are replaced by validators who perform the same function except that instead of expending their assets in the form of computational work, they stake ether as collateral against dishonest behavior. If the validator is lazy (offline when they are supposed to fulfill some validator duty) their staked ether can slowly leak away, while provably dishonest behavior results in the staked assets being "slashed". This strongly incentivizes active and honest participation in securing the network. This disincentive structure allows for network security with proof-of-stake while eliminating the need to expend energy on brute-force computations. Detailed explanations of the network security under proof-of-stake can be found here and here.
The hardware arms race that incentivizes PoW miners to consistently upgrade their hardware to more and more energy-hungry systems is removed under PoS. Validators can participate with normal home computers or even low power devices such as Raspberry Pi. This removes a significant barrier to entry and widens participation in securing the network. However, there is still a substantial cost to becoming an Ethereum validator - the minimum stake is 32 ether. When the PoS architecture was designed this was a relatively low amount but now the price of ether has increased dramatically - 32 ether is worth about $100,000 at the time of writing this article. Nevertheless there are already more than 291,000 validators, each having staked at least 32 ether, currently securing the Beacon Chain. There is thought to be, on average, 5.5 validators per node, meaning those 291000 validators are probably distributed across about 52,000 nodes.
It is important to note that part of the reason why Ethereum is now able to switch its security mechanism from PoW to PoS is because the ether token has economic value and liquidity. If the ether price was negligible then a malicious actor would have little to lose by defrauding the network (i.e. sufficient stake to take control of the network could be bought cheaply) and any penalties incurred by misbehavior would be inconsequential. It is only because ether is valuable and tradeable that validators have actual risk and real incentives. While it is easy to demonise PoW in favour of low-carbon PoS, arguably, the only reason ether has this property of being valuable and tradeable is because a real-world commodity (energy) has been burned to create the ether that currently exists. That value is now baked into the token making it a viable asset to stake. Ethereum’s higher carbon past has enabled its low carbon future.
Ethereum’s PoS chain is known as the Beacon Chain. It has been running since November 2020 has not yet been merged with Ethereum mainnet. However, it is running full PoS consensus so we can estimate Ethereum's post-merge energy usage. A recent estimate suggested that the merge to PoS could result in a 99.95% reduction in total energy use, with PoW being ~2000x more energy-efficient than PoW. The energy expenditure of Ethereum will be roughly equal to the cost of running a modest home computer for each node on the network.
In the article written for ethereum.org we originally included a figure showing per-transaction energy expenditure for Bitcoin, PoW Ethereum and PoS Ethereum originally posted on the Ethereum Foundation blog. However, I’m unsure how useful estimating per-transaction expenditure really is. This is because the energy required to mine a block is independent of the number (or type) of transactions within it. The energy required to mine a block on a PoW chain is determined by the block difficulty, which in turn is a tuning parameter that regulates the block times (e.g. if blocks are being produced too rapidly then the difficulty is increased and vice-versa). A per transaction unit of energy expenditure implies that fewer transactions would lead to smaller energy expenditure, which is not the case. For Ethereum the transaction throughput is not only that of Ethereum’s mainnet - it is also the sum of the transaction throughput of all of its “layer 2” rollups, which - to my knowledge - have not been included in calculations so far and would drastically reduce them. This is why tools that compare energy consumption per transaction across platforms are misleading. This is also why alarmism around the environmental impact of NFTs is often misleading - they often seem to be rooted in a fundamental misunderstanding of how and why blockchains expend energy.
More relevant, in my opinion is the overall energy consumption and carbon footprint of the network as a whole. From those values one can examine what that network offers to its users and to society at large and make a more holistic evaluation of whether that energy expenditure is justified or not. Per transaction measurements, on the other hand, imply the value of the network only comes from its role in transferring crypto between accounts and prohibit an honest cost-benefit analysis.
Whole network energy consumption and carbon footprints for Bitcoin and Ethereum are estimated at Digiconomist.net. At the time of writing this article Ethereum’s total energy consumption is ~112 TWh/yr, equivalent to that of the Netherlands, with a carbon emission equivalent to that of Singapore (53 MT/yr). For comparison, Bitcoin currently expends about 200 TWh/yr energy and emits about 100 MT/yr C, while also generating about 32,000 T of electrical waste from obsolete hardware per year. These are clearly very substantial figures. Switching off Ethereum’s PoW in favor of PoS will reduce this energy expenditure by more than 99.95%, implying that the total energy expenditure for securing Ethereum is expected to be closer to 0.01 TW/yr.
To put Ethereum's energy consumption in context, we can compare annualized estimates for other industries. If we take Ethereum to be a platform for securely holding digital assets as investments, perhaps we can compare to mining gold, which has been estimated to expend about 240 TWh/yr. As a digital payments platform we could perhaps compare to PayPal (estimated to consume about 0.26 TWh/yr). As an entertainment platform we could perhaps compare to the gaming industry which has been estimated to expend about 34 TW/yr. Estimates of energy consumption by Netflix range dramatically between about 0.45TWhr/yr (their own estimates reported in 2019) up to about 94 TWh/yr (as estimated by Shift Project) - there is some discussion about the assumptions underlying these estimates available on Carbon Brief. Alternatively, Ethereum could be compared to Youtube which has been estimated to expend about 244 TWh/yr, although these values depend a lot on the type of device videos are streamed on and the energy-efficiency of underlying infrastructure such as data centers. Estimates of Youtube's energy expenditure have been broken down by channel and individual videos. Those estimates imply that people consumed 45 times more energy watching Gangnam Style in 2019 than proof-of-stake Ethereum will use in a year.
PoW chains undeniably expend a huge amount of energy. Supporters of PoW argue that this energy expenditure is what gives the network security and is therefore necessary. There have also been several arguments in favour of PoW stimulating investment in renewable energy and incentivizing capture of “stranded” energy that would otherwise be wasted. At face value these arguments fail as a concession to environmentalists because the truly green option is not burning the energy at all. Energy spent mining crypto, even renewable energy, is energy that could be used for something else. The crux of the issue is that some people believe the energy expenditure required to secure a PoW chain to be worth it, others don’t.
The transition of Ethereum from PoW to PoS should help to mitigate the environmental concerns. Certainly, PoW is hard to justify now that evidence in favour of PoS being at least as secure and decentralized is constantly growing. The emissions resulting from the PoS chain will be as good as negligible - orders of magnitude below most other industries Ethereum could be compared to. The interesting thing is that Ethereum is not one thing - it is a means for decentralising anything in a secure, trustless, provable way. Pricing in DAOs, NFTs, DeFi, DeSci, ReFi, public goods funding, metaverse, web3 and a whole host of other applications, all at a minimal energy cost seems like a clear win for PoS Ethereum. With a blockchain as a backend, circular, less wasteful economies can be established using, for example NFTs and global markets to connect buyers and sellers, or preventing double-spending of carbon offsets.
There is also a substantial and very active ReFi scene building on Ethereum that will be truly unlocked at the moment of the merge. ReFi is the term given to applications that use DeFi building blocks to create applications with provable positive externalities that improve the environment. Ultimately, the goal of a ReFi application layer should first be to repay Ethereum’s carbon debt from the PoW years and then, as ReFi apps scale, to make Ethereum not only Carbon negative, but overwhelmingly environmentally positive.
The transition to PoS solves the ongoing emissions problem for Ethereum. It has often been stated that The Merge reduced the world’s energy consumption by 0.2% because it switched off mining, but I find this argument slightly facile because it is 0.2% that Ethereum was adding in the first place. It’s more that Ethereum stopped adding 0.2% to the world’s energy consumption. While there is real cause to celebrate the environmental impact of The Merge we also shouldn’t use it to greenwash. The Merge did not address the substantial carbon debt that Ethereum has accrued in its first 8 years. It will take years of running PoS bring the total expended energy down to an acceptable average. This is where I am hopeful the emerging ReFi community will help. There is a very interesting nascent scene building tools for manipulating carbon markets as well as using DeFi principles to incentivize land regeneration, carbon capture through regenerative agriculture and other related projects. If this ecosystem continues to gather pace and lock away more carbon, it could help Ethereum to quickly become net Carbon negative. These projects may also be able to help other apps on Ethereum to offset their historic emissions by connecting them to legitimate carbon offsetting schemes. It would also be interesting to explore the possibility of an offsetting scheme where major organizations in the Ethereum ecosystem repaid the existing Carbon debt by gradually buying tokenized carbon credits. A quick back-of-the envelope calculation suggests around $30,000,000 would make Ethereum whole with respect to its carbon expenditure at a price of $1.67 per Tonne.
Ethereum has been an aspirant PoS chain since its inception, and has long promised to turn off PoW in favor of PoS. However, it has proven to be a major technical challenge. This is because the Ethereum developer community refuses to compromise on security, scaleability and decentralization. Even with the Beacon Chain up and running the precise roadmap for Ethereum has changed several times. Initially, the roadmap was divided into phases, with scaling by sharding the blockchain happening before the merge to PoS. However, the rapid development of layer 2 “rollups” allowed Ethereum to scale sufficiently that the merge was brought forwards in time, before sharding. New software clients had to be developed from the ground up to deal with the new security and consensus mechanisms. The community had to buy in to the idea by staking their ether in a deposit contract. Then, the new client implementations had to be extremely rigorously tested. This process has taken several years, but now there are four merge-ready clients and an expectation that PoW will be switched off for good within the next 3-4 months.
While Ethereum's energy consumption has historically been substantial, there has been a major investment of developer time and intellect into transitioning from energy-hungry to energy-efficient block validation. To quote Bankless, the best way to reduce the energy consumed by proof-of-work is simply to "turn it off", which is the approach Ethereum has committed to take.