ether.fi was the best-known name in liquid restaking during 2024. On 29 October 2026 its own documentation page says that its eETH and weETH tokens do not bundle additional restaking rewards or EigenLayer AVS slashing exposure. The same page reports that less than 1% of the protocol’s assets were still restaked as of August 2026, that the remainder was due to be removed by the end of the third quarter, and that holders who want restaking exposure must opt into a separate product such as weETHs, which the same documentation files under the label “Super Symbiotic” rather than EigenLayer. The flagship of the category has, in effect, stepped out of it.
That is a strange place to start a lesson on restaking, and a useful one. Restaking is the idea that ether already staked to secure Ethereum can be pledged a second time to secure other services, for extra rewards and extra risk. EigenLayer turned the idea into contracts, slashing on it went live on 17 April 2025, and the money that flowed in peaked in August 2025 before falling by roughly three quarters in dollar terms. The machinery comes first, then the two real failures of the layer built on top of it (Renzo in April 2024, Kelp in April 2026), and the figures end where the opening began: with most of the money gone.
One reminder first. The previous lesson, liquid staking tokens such as stETH and rETH, covered how Lido and Rocket blockchain that runs exactly as written when it is called, holding and moving funds according to its code, without an intermediary able to change the outcome.">smart contract holding reserves of two or more tokens, supplied by users, that a decentralized exchange uses to execute trades automatically at a price set by a formula.">Pool stake ether for you and hand back a token, why stETH traded at a discount in June 2022 and why withdrawals changed that. The token mechanics themselves (a balance that rebases versus a token whose exchange rate climbs) are in our guide to interest-bearing tokens. This lesson assumes both and builds one floor higher.
Restaking: the same staked ether pledged a second time
On Ethereum, a validator locks at least 32 ETH. If it signs two conflicting blocks or otherwise breaks the protocol’s rules, part of that stake is destroyed. That threat of loss, called slashing, is what makes the network expensive to attack. The stake is collateral for one job: honest validation of Ethereum.
Restaking asks a simple question: why only one job? A new network, such as a data availability layer, an oracle, a bridge or a network of RPC nodes, also needs participants who will lose money if they cheat. Launching its own token and persuading people to stake it is slow and costly, and a young token is cheap to buy up and attack. If ether stakers agree to put their existing stake at risk for that network too, the network borrows security that already exists. The stakers earn a second stream of rewards. In exchange they accept a second set of rules under which their stake can be cut.
Two routes lead in. With native restaking, a solo staker points the withdrawal credentials of an Ethereum validator to a contract called an EigenPod, so the beacon-chain ETH itself counts as restaked. With token restaking, a holder deposits a staking means depositing a proof of stake coin with a protocol that stakes it for you and issues a transferable token (such as Lido stETH or rETH) representing the deposit plus rewards, which you can use or sell while the coins stay staked.">liquid staking token such as stETH into an EigenLayer strategy contract. Either way the ether keeps earning ordinary staking rewards; restaking adds a layer on top of it and never replaces it.
How EigenLayer works: restakers, operators and AVSs
EigenLayer, built by Eigen Labs, has three kinds of participants. Restakers deposit ETH or a liquid staking token and delegate it. Operators receive that delegated stake and run the off-chain software that the services need; the restaker keeps ownership of the funds but takes on the operator’s behaviour as a risk. AVSs are the services themselves. Eigen Labs first called them Actively Validated Services and, in a post of 19 February 2025, renamed the acronym Autonomous Verifiable Services.
EigenDA, Eigen Labs’ own data availability service for rollups, is the best-known AVS. When slashing went live, Eigen Labs’ announcement of 17 April 2025 counted 190 AVSs in development, 40 of them on mainnet, and named LayerZero and Infura’s Decentralized Infrastructure Network (DIN) as launch partners for slashing. Those counts are from April 2025; we found no more recent official total.
A restaker therefore makes two choices, often without noticing the second. The first is to deposit. The second is to delegate to an operator, and that operator then decides which AVSs to serve. The restaker’s exposure is the sum of the operator’s decisions.
Shared security: what an AVS actually buys
What an AVS buys is a promise backed by money: a set of operators whose delegated stake can be cut if they break that AVS’s rules. The security is economic. It is measured by how much stake an attacker would have to put at risk, or acquire, to corrupt the service, and by whether the slashing rule can actually detect the fault on-chain. A data availability service can prove that an operator signed for data it did not store; a service whose faults are subjective is much harder to protect with slashing.
In 2023 and 2024 the pitch was that Ethereum stakers would rent out this security as a market: AVSs pay rewards, restakers and operators collect them. In practice, much of the yield on restaked ETH in the early years came from token incentives and points programmes rather than from fees paid by AVSs. That matters for the rest of this lesson, because when the incentives were cut, the money moved.
EIGEN, incentives and the EigenCloud name
EIGEN is the protocol’s token. Since 17 June 2025 Eigen Labs has marketed the whole stack as EigenCloud, a developer platform it describes as “Built atop EigenLayer and powered by the EIGEN token”. EigenDA and newer services such as EigenCompute sit under that name, and DefiLlama now lists the protocol as EigenCloud. The contracts, the restaking mechanics and the documentation sections are still called EigenLayer, which is the name used in this lesson.
The incentives changed in two steps. On 18 December 2025 Eigen Labs presented ELIP-12, a proposal for an Incentives Committee that would move EIGEN rewards away from idle stake and towards slashable and redistributable stake; on stake subsidised by EIGEN incentives, the committee would propose a 20% fee on AVS rewards. The committee went live with its first update, posted by the Eigen Foundation on 12 March 2026. From 19 March, programmatic incentives went only to stake supporting productive AVSs, ETH incentives were cut by 50% and EIGEN incentives by 75%, and all ETH incentives were at first directed to EigenDA. Undelegated stake, or stake with an operator serving no such AVS, no longer receives any of them.
Slashing on EigenLayer since 17 April 2025
For almost two years EigenLayer accepted deposits without any way to slash them. That ended on 17 April 2025, when Eigen Labs announced that slashing was live on mainnet. It came with two building blocks that decide who can lose what.
The first is the operator set. An AVS defines one or more groups of operators, each with its own tasks and its own slashing conditions. Slashing is opt-in: in Eigen Labs’ guide for stakers, “Stakers will not automatically be subject to slashing when the feature is released.” Stake becomes slashable only when an operator registers for a slashable operator set and allocates stake to it.
The second is unique stake allocation. An operator earmarks a portion of its delegated stake, called a magnitude, for a given operator set. Only that set’s AVS can slash that portion. If an operator gives 30% of its stake to one AVS and 20% to another, a fault on the first can cut at most the first 30%; the second AVS has no claim on it. This is the design answer to the fear that one failure would be multiplied across every service the same ether secures.
Delays protect restakers from surprises. EigenLayer’s contract documentation on GitHub sets the allocation configuration delay at 126,000 blocks on mainnet, about 17.5 days, which the stakers’ guide describes as the time an operator’s first allocation to an operator set takes. Deallocations take 100,800 blocks, about 14 days, and the allocation stays slashable during that window. Withdrawals by restakers follow the same rhythm: the DelegationManager documentation gives a minimum withdrawal delay of 100,800 blocks (14 days) on mainnet, and shares waiting in the queue remain slashable until the delay ends. For native restakers, leaving Ethereum’s validator queue adds its own wait on top.
Since 22 July 2025, AVSs can also choose redistribution: slashed funds are sent to a chosen address instead of being burned, which lets a lending or insurance service compensate its users from the slashed stake. It is a feature of specific operator sets, flagged as redistributing, and it changes the incentive picture: whoever receives slashed funds has a reason to see slashing happen.
Has anyone actually been slashed? We found no report of an operator being slashed on EigenLayer mainnet as of 29 October 2026, in Eigen Labs’ posts or in press coverage we could open. That is absence of evidence, not proof, and it cuts both ways: the operators may simply have behaved, or the slashing rules have not yet been tested by a real dispute at scale.
Liquid restaking tokens: what ether.fi, Renzo and Kelp issue
Restaking on EigenLayer directly means choosing an operator, waiting 14 days to withdraw and holding a position that no other app understands. Liquid restaking protocols wrap all of that. You deposit ETH or an LST; the protocol stakes and restakes it, chooses operators and AVSs, and gives you a receipt token: ether.fi’s eETH (rebasing) and weETH (non-rebasing), Renzo’s ezETH, Kelp’s rsETH and Puffer’s token were among the main ones. The receipt token’s exchange rate against ETH is meant to rise with rewards, exactly like rETH or wstETH, and the token can be traded, bridged or posted as collateral.
That convenience adds two layers that EigenLayer itself does not have. One is the wrapper’s own contracts: minting, redemption, bridging and pricing. The other is the market price of the receipt token. A liquid restaking token can only be redeemed through the issuer’s queue, which inherits EigenLayer’s 14-day withdrawal delay plus any validator exit time. Anyone who wants out faster has to sell on a DEX, and the price there is whatever buyers will pay that day. Points programmes and airdrops in 2024 also pushed many holders into these tokens for reasons unrelated to security, which made the exits crowded when the airdrops arrived.
ether.fi took restaking out of weETH in 2026
ether.fi’s current eETH documentation describes a wind-down: the protocol previously restaked a large part of its assets through EigenLayer, less than 1% remained restaked as of August 2026, the rest was planned for removal by the end of the third quarter and EigenPod withdrawal credentials by the fourth quarter. It says users who want more restaking exposure must opt into a separate product. Redemption timing “depends on available protocol liquidity, Ethereum validator withdrawals, and network conditions”, according to the same page. Older ether.fi pages still describe eETH as natively restaked, so the current product page is the one to read.
DefiLlama still lists ether.fi Stake, now in its Liquid Staking category, at 4.43 billion dollars on 29 October 2026, against a peak of 12.43 billion on 14 August 2025. Most of that is plain staked ether. Adding it to a “restaking total” would now be wrong.
Renzo ezETH, 24 April 2024: a receipt token is only a market price
The first visible crack in the LRT layer came on 24 April 2024. A day after Binance Labs announced the distribution plan for Renzo’s REZ token, ezETH fell 18.3% to about 2,642 dollars within minutes, The Block reported, and CoinDesk found it as low as 750 dollars on Uniswap around 02:45 UTC. By early afternoon New York time it traded at 3,077 dollars against ether at 3,179 dollars: largely recovered, still below the asset it was supposed to track.
Nothing was hacked and no operator was slashed. Holders who wanted to leave, many of them disappointed by the airdrop, could not redeem ezETH for ETH quickly, so they sold on DEXs into pools too shallow to absorb them. Positions that used ezETH as collateral were liquidated at the depressed price: Hitesh Malviya, founder of the research firm DYOR, told CoinDesk the slippage “caused massive liquidation” on Gearbox and Morpho. Totals published afterwards differ from one outlet to the next, and we found no primary tally of the amount liquidated. The lesson is the one stETH taught in 2022, one floor up: a peg without a fast redemption path is a market price, and a lending market’s oracle will treat it as one.
Kelp rsETH, 18 April 2026: the bridge failed, not the restaking
The second case was much larger. On 18 April 2026, rsETH, Kelp’s liquid restaking token, was released without backing from the Ethereum side of its LayerZero bridge. Aave’s governance forum post of 24 April 2026 (ARFC rsETH incident funding update) says “a vulnerability enabled an unauthorised release of rsETH from the Ethereum LayerZero adapter”. Citing a LlamaRisk report, it counts 152,577 rsETH extracted from the LayerZero lockbox, an original shortfall of about 163,183 ETH at 1.0696 rsETH per ETH. On the day of the attack, The Block, reading the transaction’s on-chain data, reported 116,500 rsETH released, about 292 million dollars and roughly 18% of circulating supply. The forum post does not explain the gap between the two counts. The attacker posted rsETH as collateral on Aave V3 and borrowed WETH against it, which is how a wrapper failure became a lending-market problem; that side of the story is told in our lesson on crypto lending and borrowing.
The same ARFC lists the recovery effort as of late April. Kelp recovered and froze 40,373 rsETH, the Arbitrum Security Council froze 30,766 ETH that the attacker still held on Arbitrum, and liquidating the attacker’s positions was expected to bring back more; together that left a residual gap of about 75,081 ETH (the post also gives 75,228 in a table). To close it, a coalition called DeFi United gathered donations, Mantle committed a credit facility of up to 30,000 ETH, and the post asked the Aave DAO to contribute 25,000 ETH from its treasury. LayerZero and Kelp publicly disagreed over whose configuration allowed the release; that dispute is unresolved, and as of the sources we could open on 29 October 2026, a final accounting of the loss and of rsETH’s backing had not been published. DefiLlama still counted 0.91 billion dollars of ETH locked in Kelp on 29 October 2026, against 1.59 billion on 17 April.
For a restaking lesson the point is precise. EigenLayer’s slashing was not involved; the exploited piece was the cross-chain release path of a receipt token, the kind of risk covered in our guide to bridges and cross-chain liquidity. But that receipt token was accepted as collateral on the largest lending venues, so a bridge failure in one wrapper turned into a liquidity event across DeFi.
Is restaking safe? The risks stacked layer by layer
Restaking is not risk-free, and the extra yield, when there is any, is payment for added risk. The useful exercise is to name each layer a restaked position passes through and what can go wrong there. The table pairs each layer with a real event where one exists; where none exists, it says so.
| Layer | What can go wrong | Real case or status (29 October 2026) |
|---|---|---|
| Ethereum staking | Validator penalties and slashing; exit queue delays | Ethereum’s own rules, covered in the previous lesson |
| Liquid staking token used as deposit | Discount to ETH, issuer contract risk | stETH discount in June 2022 |
| EigenLayer contracts | Bug in delegation, allocation or withdrawal code | We found no exploit of EigenLayer’s core contracts |
| Operator and AVS | Operator fault, or a faulty slashing rule in the AVS code | Opt-in slashing live since 17 April 2025; we found no report of a mainnet slashing |
| Liquid restaking wrapper | Mint, redemption or bridge bug; poor operator choices | Kelp rsETH bridge release, 18 April 2026 |
| Receipt token market | Depeg when holders rush for the exit | Renzo ezETH, 24 April 2024 |
| Lending market collateral | Liquidations at a depressed price, bad debt | ezETH liquidations (2024); rsETH on Aave (2026) |
Read top to bottom, each row adds a failure mode to the ones above it. A holder of an LRT used as collateral carries all seven. A solo staker who restakes natively with a carefully chosen operator carries three or four. The same word, restaking, covers both.
Exit time is the other risk to count. Leaving EigenLayer takes at least 14 days after the withdrawal is queued, during which the shares stay slashable, plus the validator exit for native stake. An LRT holder in a hurry depends on DEX liquidity instead, which is exactly what failed in April 2024.
“Cascading slashing”: the feared scenario and what really cascaded
The fear most often attached to restaking is cascading slashing: one AVS slashes an operator, the same ether was securing ten other services, those lose their security at once, more faults follow, and losses spread through every LRT and every lending market that holds them. Before 2025 this was a reasonable worry, because the design for slashing was not public.
Two facts narrow it now. Slashing is opt-in per operator set, and unique stake allocation means a given slice of stake can be slashed by one AVS only, so one fault cannot be charged twice against the same slice. A slashed operator still loses stake, and every restaker delegated to it shares that loss, but the multiplication across services that the scenario assumes is what the allocation model was built to prevent. And, as of 29 October 2026, we found no report of a mainnet slashing event at all, cascading or not.
The cascades that did happen ran through the layers above EigenLayer: a receipt token losing its market price (Renzo), a wrapper’s bridge releasing unbacked tokens (Kelp), and lending markets that had accepted those tokens as good collateral. That pattern of stacked dependencies is the subject of our lesson on DeFi composability, and restaking is one of its clearest examples. None of the losses described in this lesson started with a slashing event.
Symbiotic, Babylon, Jito and other restaking designs
Symbiotic went live on Ethereum mainnet on 28 January 2025. Its co-founder Misha Putiatin told The Block it was “the first fully permissionless restaking protocol to go live in a production environment”. It accepts any ERC-20 token as collateral, so a network can be secured by stablecoins or by its own token as well as by ether. It had raised a 5.8 million dollar seed round led by Paradigm and cyber.Fund in June 2024. Its DefiLlama description now speaks of “collateral markets” rather than restaking, and DefiLlama files it under that category, with 317 million dollars on 29 October 2026 against a peak of 2.70 billion on 7 December 2024.
Babylon applies the idea to bitcoin, which has no smart contracts and no native staking. In phase 1, launched in August 2024, bitcoins are locked in a self-custodial staking script on the Bitcoin chain with a maximum time-lock of 64,000 blocks and an unbonding period of 1,008 blocks (about a week); “there is no slashing of staked bitcoin during Phase-1”, and the coins are not transferred to the finality providers. The initial cap was 1,000 BTC. DefiLlama counts 3.25 billion dollars locked in Babylon on 29 October 2026, all of it bitcoin. More on the project on our Babylon project page.
Jito Restaking on Solana uses vaults that hold SPL tokens such as JitoSOL, issue vault receipt tokens and delegate to operators serving node consensus networks. Its documentation says the slashing functionality is still being developed and that the slashing program is not currently live. DefiLlama shows 17 million dollars in it on 29 October 2026, down from about 400 million on 1 July 2025, so Solana restaking is small and, per its own docs, without live slashing.
Among the 2024 names, Puffer’s liquid restaking token is still listed by DefiLlama with about 26 million dollars on 29 October 2026, and Renzo describes itself as a platform for staking, restaking and capital deployment, with 105 million dollars, down from 4.05 billion on 28 May 2024. Karak, another name often cited in 2024, appears nowhere in the sources we could open for 2026, so we cannot state its current status.
Restaking by the numbers on 29 October 2026
The figures below come from DefiLlama’s API, retrieved on 29 October 2026 around 09:50 UTC. They are in dollars, so they move with the price of ether and bitcoin as well as with deposits and withdrawals. DefiLlama’s categories also matter: ether.fi is now filed under liquid staking, Symbiotic under collateral markets. How these totals are built, and what they hide, is explained in our guide to measuring DeFi TVL.
| Protocol (DefiLlama category) | TVL on 29 Oct 2026 | Peak (date) |
|---|---|---|
| EigenCloud / EigenLayer (Restaking) | 5.97 billion dollars | 22.06 billion (14 Aug 2025) |
| ether.fi Stake (Liquid Staking) | 4.43 billion dollars | 12.43 billion (14 Aug 2025) |
| Babylon Protocol (Restaking, bitcoin) | 3.25 billion dollars | 7.14 billion (7 Oct 2025) |
| Kelp (Liquid Restaking) | 0.91 billion dollars | 2.09 billion (18 Sep 2025) |
| Symbiotic (Collateral Markets) | 0.32 billion dollars | 2.70 billion (7 Dec 2024) |
| Renzo (Liquid Restaking) | 0.11 billion dollars | 4.05 billion (28 May 2024) |
| Jito Restaking (Restaking) | 0.02 billion dollars | 0.40 billion (1 Jul 2025) |
EigenLayer’s figure is down about 73% from its August 2025 peak. Its history on DefiLlama shows 15.63 billion dollars on the day of the Renzo depeg in April 2024, 7.07 billion on the day slashing went live in April 2025, the 22.06 billion peak in August 2025, 9.92 billion a year after slashing and 5.97 billion now. Of today’s total, DefiLlama lists about 5.34 billion under WETH and about 0.50 billion under stETH, the rest in other liquid staking tokens.
The timing tells a story the protocol’s own posts also tell. In a post of 27 February 2026, Eigen Labs’ founder Sreeram Kannan framed the year around AI agents and wrote that “Markets, however, shifted from valuing future infrastructure to valuing immediate revenue.” The cut in incentives followed in March; by August ether.fi reported less than 1% of its assets still restaked. Restaked ETH that once chased points and rewards has largely gone back to plain staking.
Looking at restaking without risking money
This lesson does not suggest a deposit. If you want to see the machinery, most of it can be read without signing anything:
- Open the protocol’s page on DefiLlama and compare the TVL chart with the dates above: the slashing launch, the August 2025 peak and the March 2026 incentive cut.
- Read the GitHub repository Layr-Labs/eigenlayer-contracts: its README lists the current deployments, including the Sepolia and Hoodi testnets, and the docs folder gives every delay in blocks.
- Before holding any liquid restaking token, find the issuer’s current page on slashing risk and redemption, as ether.fi’s eETH page does, and check which operators and AVSs it uses, if any.
- On a lending market, look up whether the token is accepted as collateral, at what loan-to-value and with what oracle. That is where Renzo’s and Kelp’s problems spread.
Developers can test on a testnet: the repository’s documentation gives a minimum withdrawal delay of 50 blocks (about 10 minutes) on testnets. Its deployment table lists native restaking as supported on Hoodi but not on Sepolia, where EigenPod functions are paused, so token restaking is the only route on Sepolia. Test tokens have no value. On mainnet, every deposit, delegation, withdrawal queue and completion is a separate transaction with its own gas cost, and the money is locked for at least 14 days on the way out. Anyone who still tries it with a small amount should count both before starting.
Rules and taxes for US and UK readers
On 5 August 2025 the SEC’s Division of Corporation Finance published a staff statement on certain liquid staking activities, taking the view that the activities it covers do not involve the offer and sale of securities. It is a staff view, not a Commission rule, and it does not stretch to restaking: a footnote says “This statement does not address ‘restaking,'”. It also excludes providers that decide whether or how much to stake, guarantee rewards, or enable receipt tokens to generate additional returns. Liquid restaking tokens therefore sit outside the only US staff guidance on staking receipts.
For US taxes, IRS Revenue Ruling 2023-14 holds that a cash-method taxpayer who stakes a proof-of-stake token includes the fair market value of rewards in gross income when they gain dominion and control over them. It does not address restaking rewards, points, airdrops or the swap of ETH into a receipt token, which may be a taxable disposal. Our guide to crypto taxes in the United States covers the basics; for restaking, the treatment is unsettled and worth taking to a tax adviser.
In the United Kingdom, we found no HMRC guidance written for restaking: how its rules on staking returns and on disposals apply to depositing ETH into an LRT, or to restaking rewards, has not been spelled out. The general rules are in our guide to crypto taxes in the UK.
Where this lesson sits in the DeFi path
Restaking closes module 2 of the DeFi course. If the receipt tokens are still unclear, go back to the lesson on stETH, rETH and liquid staking; the protocol itself has its own page, EigenLayer on Coindaro, with its current figures. A later lesson opens the next module with decentralised exchanges and automated market makers, the venues where receipt tokens such as ezETH find, or fail to find, a buyer.
