Explainers
What Ethereum is, and what its on-chain data shows
Ethereum is a general-purpose blockchain: a shared ledger that runs programs, called smart contracts, which anyone can deploy and anyone can call, paying in the chain’s own token, ether. Since September 2022 it has been secured by validators who lock up ether rather than by miners, and a transaction’s fee is split between a base fee the protocol destroys and a tip for the validator. The headline figure on this site’s data page for it measures deposits in its applications, and it leaves out the ether staked to secure the chain.
What the chain is for
A token, a lending market or an exchange on Ethereum is a contract: code with balances attached, stored on the chain, run identically by every node that checks a block. Using one means sending a transaction that calls its code, and the block that includes the transaction records the change of state. That model, accounts with balances and contracts with code, is what other networks copy when they call themselves EVM-compatible, after the Ethereum Virtual Machine that executes the code. It is also why an Ethereum address is the same string on every such network, a convenience with one expensive failure mode.
The network went live on 30 July 2015, when its first release, Frontier, began producing blocks, according to ethereum.org’s record of past upgrades. No one person owns or controls the protocol, says ethereum.org’s governance page, which describes how changes are made instead. Anyone in the community can write an Ethereum Improvement Proposal. Proposals that change the protocol are discussed on the developers’ AllCoreDevs calls, and a change reaches the network only when the teams that maintain the client software ship it in an upgrade and node operators choose to run it. Those teams are separate: the governance page lists go-ethereum, Nethermind, Besu, Erigon and Reth for the execution layer, and Prysm, Lighthouse, Nimbus, Teku, Lodestar and Grandine for the consensus layer.
The Ethereum Foundation’s own page on ethereum.org calls it a non-profit that funds protocol development and the wider ecosystem, and says it does not control or lead Ethereum and is not the only organisation funding its development.
How consensus works
Ethereum began with proof of work. The Merge, executed on 15 September 2022, joined the original chain to the Beacon Chain, a proof-of-stake chain that had run alongside it since 1 December 2020, and switched mining off. Ethereum.org puts the resulting fall in the network’s energy use at about 99.95%.
Under proof of stake, a validator deposits 32 ETH into a deposit contract and runs three pieces of software: an execution client, a consensus client and a validator client. Time is cut into slots of 12 seconds and epochs of 32 slots. In each slot one validator, chosen at random, proposes a block, and a committee of others attests to it, which is a signed vote that the block is valid.
Finality comes in steps. The first block of each epoch is a checkpoint, and validators vote for pairs of checkpoints. When a pair attracts votes representing at least two-thirds of all staked ether, the later one is justified and the earlier one finalised. Ethereum.org’s description is that a finalised block cannot be reverted without slashing a majority of stakers, which is what makes an attack on it uneconomic.
Slashing is the penalty for the two behaviours the protocol treats as dishonest: proposing two blocks for one slot, or signing contradictory attestations. The amount depends on how many validators are slashed at around the same time. Ethereum.org gives the range as less than 0.1% of stake for one validator slashed alone and up to all of it in a mass slashing, so a validator slashed alone loses far less than one slashed alongside many others.
The Pectra upgrade of 7 May 2025 changed one number in this design. Through EIP-7251 it raised the maximum effective balance of a single validator to 2,048 ETH, which ethereum.org says improves capital efficiency for stakers.
What ether does, and how fees are set
Every operation a transaction performs has a cost in gas, and gas is paid in ether. The London upgrade of 5 August 2021 brought in EIP-1559, which divides the fee in two. The base fee per unit of gas is set by the protocol from how full the previous block was: it rises when blocks run above their target size and falls when they run below, by at most 12.5% from one block to the next, according to ethereum.org’s gas documentation. The base fee is burned, removed from supply. The priority fee, a tip the sender chooses, goes to the validator. What gas is and how its price is set covers the mechanics in full.
The burn was a deliberate design choice. EIP-1559’s authors wrote that it ensures only ether can pay for transactions and that it counterbalances the inflation from newly issued ether, which now goes to validators. Whether the supply grows or shrinks over a period depends on which of the two is larger, and the burn side depends on how much the network is used.
Ether’s second job is security. It is the stake a validator puts at risk, and holders without 32 ETH or the wish to run software can pool their ether with a staking service. Some of those services issue a token representing the staked ether, which can be used elsewhere in DeFi; liquid staking explains what that token is a claim on. Liquid staking is also the largest category among the fifteen protocols listed on Ethereum’s data page.
The rollup roadmap
Ethereum.org’s scaling page says the chain is scaled using layer 2s, also called rollups, which batch transactions together off the main chain and send the output to Ethereum, so that anyone can check the result against data posted there.
That data was the expensive part: ethereum.org says storing it permanently on Ethereum accounted for over 90% of what rollup users paid. The Dencun upgrade of 13 March 2024 added EIP-4844, known as proto-danksharding, which gives rollups a cheaper, temporary place for that data, called blobs, deleted from Ethereum once it is no longer needed. The fuller version, danksharding, is a later stage that needs new ways of checking the data is available. The most recent upgrade on ethereum.org’s list is Fusaka, activated on 3 December 2025.
The roadmap shapes how Ethereum appears in the data. A rollup is a chain of its own, with its own page here, and layer 2s explained and what Base is take up how they work. What stays on Ethereum is the bridge contract that holds assets for each of them, and the canonical bridges of three other networks appear among Ethereum’s fifteen largest protocols in the snapshot: Arbitrum Bridge at $2.8B, Base Bridge at $2.6B and Polygon Bridge at $2.1B.
What has gone wrong
The DAO, 2016. This crisis began in a contract on the chain, and the chain itself was changed to answer it. Ethereum.org records that an insecure contract, The DAO, was drained of more than 3.6 million ETH, and that a hard fork at block 1,920,000 on 20 July 2016 moved the funds to a new contract with one function, withdraw, from which holders could reclaim 1 ETH for every 100 DAO tokens. Not every participant accepted the fork, and the chain that continued without it became Ethereum Classic. DefiLlama’s hacks record lists the incident on 17 June 2016, classified as reentrancy, at $60.0M, the amount its record gives.
Denial of service, 2016. In September and October 2016 the network came under denial-of-service attacks that exploited operations priced too low in gas for the work they caused. Ethereum.org’s upgrade record calls the Tangerine Whistle fork of 18 October 2016 the first response: its EIP-150 raised the gas costs of the operations that could be used in such attacks.
Finality delays, May 2023. On 11 May 2023, from about 20:19 UTC, block production dropped sharply and finalisation was delayed by four epochs; the next day the same thing happened for nine epochs, long enough for an inactivity penalty to apply. In both cases the chain recovered without intervention. The post-mortem written by the Prysm client’s developers and published by Offchain Labs counts about 47 and about 149 missed blocks, and traces the cause to consensus clients, Prysm among them, that could not efficiently process valid attestations pointing at an old checkpoint and so spent their resources recomputing old states.
The episode fits the risk ethereum.org’s client-diversity page describes: a bug in a consensus client run by more than a third of nodes could stop the chain finalising, so the network is safer when no single client is that large.
DefiLlama’s hacks record, as committed to this site, lists 564 incidents tagged Ethereum between 2016 and 2026, and files 415 of them against DeFi protocols and 59 against centralised exchanges; the Ethereum hacks page breaks them down by year, and an incident that touched several chains is counted in full on each. How crypto gets hacked reads the whole record by class of failure.
What the on-chain data shows
As of 22 November 2026, per DefiLlama, Ethereum had $35.7B of total value locked, down 23.0% over 30 days, which ranks it first of the 121 chains this site tracks (its data page). DefiLlama lists 1,830 protocols deployed on it, and $128.0B of stablecoin supply circulates there.
The total value locked figure is the value of assets deposited in DeFi protocols on the chain, at the snapshot date’s prices. The chain figure leaves out liquid staking, bridges and deposits DefiLlama flags as double-counted, and ether staked directly with validators is not counted at all: DefiLlama’s FAQ says it counts no chain’s staked assets. What TVL measures sets out those rules. The practical result on Ethereum is that the headline excludes the ether doing the chain’s most basic job, securing it.
The page’s list of the fifteen largest protocols shows the excluded categories, because each protocol there is shown at its own figure on the chain:
| Protocol | Category | TVL on Ethereum |
|---|---|---|
| Lido | Liquid Staking | $16.4B |
| Aave V3 | Lending | $13.0B |
| SSV Network | Staking Pool | $9.0B |
| LayerZero V2 | Bridge | $6.5B |
| Binance staked ETH | Liquid Staking | $6.2B |
The fifteen add up to $84.5B, more than twice the chain figure (computed). Three liquid-staking entries account for $25.8B of that and five bridge entries for $17.8B, the second figure combining the two DefiLlama labels the page uses, Bridge and Canonical Bridge (both computed). The two totals are not meant to reconcile, and the list is not a breakdown of the headline.
The 30-day fall needs the same care. TVL is quantity times price, so it falls when deposits leave and when the prices of the deposited tokens fall, and the snapshot carries only the dollar figure, so it cannot say which happened. The seven-day change in the same snapshot is a fall of 3.1%.
Stablecoin supply on the chain, $128.0B, is 3.6 times its total value locked (computed). Even if every dollar of that TVL were stablecoins, which it is not, at least $92.3B of the supply would sit outside the deposits the figure counts (computed). Where it sits instead, and what it is used for, the snapshot does not say: it records balances, not uses. This site’s stablecoin table, built from a separate DefiLlama feed, places $129.8B of supply on Ethereum, of which Tether is $66.0B and USD Coin $41.3B, 50.9% and 31.8% of Ethereum’s total in that table (computed). Where stablecoin supply lives reads the table chain by chain.
What the page cannot tell you
How much ether is staked. Native staking is outside the TVL rules, so neither the headline nor the list measures it, and liquid-staking entries cover only the part of staking that issues a token.
What happens on rollups. Activity that moved to layer 2s is counted on those chains’ pages, so a smaller Ethereum figure can reflect where the roadmap sent applications as much as any loss of use.
Who uses it. A dollar figure says how much is deposited, not by how many people, and fees, which say more about use, are on the protocol pages, under methods the piece on protocol fees and revenue sets out.