cryptonist
Fibre optic strands gathered into a bundle, their ends lit

DeFi

What liquid staking is, and what the token claims

Liquid staking swaps running a validator for holding a token. The token is a claim on ether that someone else has staked, plus the rewards it earns, less the pool’s fee, and it can be sold while the ether underneath stays locked. Two designs carry the rewards differently, one by changing the balance and one by changing the rate. The claim can fall short in four places: slashing passed through to holders, a market price below the redemption value, the pool’s contracts and governance, and a concentration of validators that is a risk to Ethereum as well as to the people holding the token.

Staking without a pool

Ethereum’s proof of stake pays validators to check and propose blocks, and destroys part of their stake when they break its rules. ethereum.org’s proof-of-stake page describes the way in: a validator deposits ether into the deposit contract, runs an execution client, a consensus client and a validator client, and joins an activation queue that limits how fast new validators arrive. The home staking page gives the size: at least 32 ETH to activate a validator, which can hold up to 2048 ETH if it uses compounding withdrawal credentials. How the network itself works is covered in what Ethereum is.

The rules that make stake worth trusting are the ones that make it risky. Proposing two blocks for one slot, or signing contradictory attestations, is slashable. The penalty depends on how many validators are slashed around the same time: under 0.1% of the stake for one validator slashed alone, up to all of it in a mass slashing. Leaving takes time too. Withdrawals have been possible since the Shanghai/Capella upgrade of 12 April 2023, according to ethereum.org’s withdrawals page, but a validator that exits waits in a queue whose length depends on demand. Who pays staking rewards, and why issuance moves value from holders who do not stake to those who do, is set out in where crypto yield comes from.

A solo staker therefore needs 32 ETH, a machine kept online and the patience for two queues. Pooling removes the first two requirements. ethereum.org’s page on pooled staking is direct about the price: the protocol does not support pooling natively, so every pool was built separately, and every pool puts an intermediary between the staker and Ethereum.

What the token is a claim on

A liquid staking protocol takes deposits of any size, stakes them through node operators and mints a token to the depositor’s wallet. ethereum.org describes that token as an ERC-20 claim on the staked ether and the rewards it earns. The ether stays on the consensus layer; the token can be held, transferred, sold, or deposited somewhere else as collateral.

The holder is not a staker in the protocol’s eyes. Ethereum pays rewards to validators and does not know the token exists, so what the holder owns is a claim on a set of contracts, on the operators who run the validators, and on whoever governs both. In normal conditions the difference does not show. It shows in exactly the conditions the rest of this piece is about.

Two protocols show how the parts are arranged.

At Lido, stETH is minted one for one against the ether deposited, according to Lido’s integration guide. The validators are run by node operators grouped into staking modules, and the staking router documentation says a module is registered only through the Lido DAO’s voting process. The DAO’s LDO holders also set fees and assign node operators and oracles, Lido’s DAO page says. The fee is a share of staking rewards divided between a module’s operators and the DAO treasury. The staking router documentation uses 5% to each as its example, and requires the two together to be equal across modules.

At Rocket Pool, a node operator bonds 4 ETH of its own per validator, and the other 28 ETH comes from depositors, who receive rETH, according to Rocket Pool’s staking overview. The operator’s ether sits at risk beside the depositors’. Deposits made directly through the protocol pay a 0.05% fee, its staking guide says.

There are two ways out. A holder can redeem through the protocol, which depends on unstaked ether being available or on validators exiting through the consensus-layer queue, or can sell the token to someone else for whatever the market pays. The first returns the redemption value, eventually; the second is quick and returns the market price. One change since the Pectra upgrade of May 2025 bears on the first route. Under EIP-7002, the holder of a validator’s withdrawal address can trigger its exit from the execution layer, and ethereum.org notes that staking protocols can use this to exit validators without relying on their node operators to cooperate.

Rebasing and value-accruing tokens

Both designs pass on the same rewards. They differ in where the rewards show up.

A rebasing token changes its holders’ balances. stETH is the usual example. Lido’s integration guide says balances are normally recalculated once a day, when Lido’s oracle reports the consensus-layer balance, and that a report takes effect once five of nine oracle daemons agree on it. Every address holding stETH sees its balance change at the same moment, with no transfer. Underneath, the contract counts shares: each holder owns a share of the ether the protocol controls, and the report changes how much ether a share is worth. A rebase can go down. If Lido’s validators are slashed or penalised, balances fall, though the guide says no daily rebase had been negative at the time of writing.

A value-accruing token keeps the balance and changes the rate. rETH works this way. The number of tokens stays fixed and each one becomes redeemable for more ether over time; Rocket Pool’s documentation says the rate is updated about every 24 hours from the beacon-chain rewards its operators earn. Its worked example has a holder deposit 10 ETH for 10 rETH when the two were equal; later, with 128 ETH staked and validator balances grown to 160 ETH, the 10 rETH return 12.5 ETH. Those are the documentation’s illustrative figures, not a rate anyone should expect.

One design can be turned into the other. wstETH wraps stETH: its balance does not rebase, and its price in stETH changes instead. The difference matters to any contract that holds the token, because accounting written on the assumption that balances change only when tokens move breaks when they change on their own. Lido’s guide records that Aave V2 needed a custom implementation to pass stETH rebases on to its depositors, with rounding effects of a wei or two, while Aave V3 and V4 take wstETH; it recommends wstETH unless an application means to support rebasing. ethereum.org adds that the two designs are treated differently for tax in some jurisdictions, which is a matter for the rules where a holder lives rather than for the token.

Where the claim can fall short

ethereum.org’s list of risks for these tokens begins with the risks of staking and adds the pool’s own. Four matter most.

Slashing is passed through. If a pool’s validators are slashed or penalised, the loss is typically shared across all holders under the protocol’s rules. For stETH that means a negative rebase. Lido’s withdrawal queue has a second route for the same loss: a request is normally paid at the rate that applied when it was placed, but after significant losses from slashing and penalties it is paid at a lower rate reported by the oracle, conditions the guide says had not arisen on mainnet. Because the protocol’s penalty grows with the number of validators slashed around the same time, a fault shared by many validators under one operator costs each of them more than an isolated fault would. Lido’s oracle also has a sanity check that rejects any report showing staked ether down by more than 5%, and the guide says that when such a check fails the report cannot be finalised and the rebase does not happen.

The market price can leave the redemption value. Because the token trades freely, ethereum.org says, its price can fall below the value of the ether behind it, particularly under stress, and if redemptions are slow, selling at a discount may be the only fast way out. Lido’s own guide tells integrators that stETH can at times be bought at a discount on decentralised exchanges such as Curve or Balancer. Rocket Pool’s documentation says redemption through the protocol works only while its pool holds enough ether, and otherwise points holders to a decentralised exchange, probably at a small premium. A discount with a working redemption route behind it is a different object from one without, the distinction what holds a stablecoin peg draws for depegs; here the route exists, and the question is how long it takes. Lido’s queue is first in, first out, takes requests of up to 1,000 stETH each, and a request cannot be cancelled once made. A holder who has posted the token as collateral has a further question, which price the lending market reads, and how DeFi lending works covers it.

The contracts, and whoever can change them. The ether sits in contracts that can contain bugs, and how crypto gets hacked traces what exploited contracts have cost across the recorded incidents. Fees, operator sets and the token’s own mechanics can be changed through governance and upgrades, and ethereum.org points out that a holder typically has no vote. The daily rebase also depends on the oracle: if its daemons do not reach quorum, or Ethereum is not finalising, Lido’s guide says no report is submitted and the rebase waits.

Concentration. ethereum.org warns that large amounts of stake under a few organisations create the conditions for censorship, value extraction and single points of failure. The thresholds are arithmetic, and ethereum.org’s page on attacks sets them out. A third of the stake can stop the chain finalising. 34%, with some control over when messages arrive, could make two conflicting chains final. More than half can dominate the fork choice and censor. Two-thirds can finalise a chain of its own choosing. ethereum.org’s pooled-staking page puts liquid staking protocols at around a third of all staked ether, without dating the figure. A note by Danny Ryan, which the same page lists as further reading, argues that a liquid staking protocol past those thresholds becomes a basis for a cartel. Whoever decides which node operators join and leave can make them censor or extract value together, and that holds whether the decision rests with a governance token or with an automated profitability test. He adds that the risk to Ethereum and the risk to the pooled capital are bound together, since the community’s defence against such a cartel would fall on the stake. When he wrote it, only a validator’s own signing key could exit it; EIP-7002 has since changed that.

The scale, from the snapshot

As of 22 November 2026, per DefiLlama, the protocols it files under Liquid Staking with at least $1M each held $39.1B between them, across 109 protocols, listed on the liquid staking category page. $28.9B of that was on Ethereum, where 24 of the protocols held at least $1M each. Lido held $16.4B, almost all of it on Ethereum: 41.9% of the category total and 56.9% of the category’s Ethereum figure (computed). Rocket Pool held $805.8M. These are shares of a DefiLlama category total, not of staked ether. The category runs across Solana, TON and other chains, and it includes exchange-branded tokens such as Binance staked ETH, at $6.5B, beside protocol-run pools.

Three cautions from what TVL measures apply. The figures are dollars at the snapshot’s prices, so deposits held in ether move with ether’s price: Lido’s figure was 29.0% below its level 30 days earlier and Rocket Pool’s 31.0%, and a change of that size can come from price alone, with nothing withdrawn. A token deposited in a lending market or a restaking protocol is counted again in that protocol’s figure. And DefiLlama’s data definitions say liquid staking is left out of chain totals by default. Its fee figure for a liquid staking protocol is the staking rewards the deposits generated, not what the protocol kept, as the piece on protocol fees and revenue explains; it recorded $52.7M for Lido over the previous 30 days.

What to check before holding one

  1. How redemption works, and how long it takes. A buffer of unstaked ether, a withdrawal queue, the consensus-layer exit queue, or only a market.
  2. Who chooses the node operators, and who can change the fee. A DAO vote, a bond the operator posts, or a company’s terms.
  3. How a slashing loss reaches the holder. A smaller balance, a lower rate, or a discounted withdrawal.
  4. Which design it is. A rebasing balance and a fixed one behave differently in a wallet and in any contract that holds them, and the wrapped version may be the one an application accepts.
  5. How much of all stake the pool holds. Past a third, its size is Ethereum’s problem as well as the holder’s.

Committing the same staked ether to secure other services adds a layer with slashing rules of its own, which what restaking is sets out.