What is restaking? EigenLayer, liquid restaking tokens, slashing and the 2026 reality explained

Who this is for: Treasurers, fund analysts, founders and serious ETH or BTC holders who want to understand what restaking sells, who pays for it, how slashing and LRT wrappers add risk on top of plain staking, and how to compare a staked, liquid-staked or liquid-restaked position like for like.

Proof-of-stake chains pay validators to lock capital and behave. Restaking asks a simple question: if 43.8 million ETH is already locked and earning about 2.6 percent, why not rent that collateral to other systems that need economic security, such as data availability layers, oracles and bridges, and collect a second stream of fees? That idea, launched by EigenLayer in 2023, pulled in more than $20bn within a year and spawned wrapper tokens, points programmes and copycats on Bitcoin and Solana.

It matters now because the experiment has run long enough to read the results. Slashing did not go live until April 2025, the incentive tokens have been distributed, the largest LRT issuer has walked away from EigenLayer, and the biggest loss came from a bridge configuration rather than a validator fault. Meanwhile Bitcoin staking through Babylon has become the second-largest pool of restaked capital.

This guide explains the mechanism (operators, AVSs, operator sets, slashing), the main venues, what LRTs are and why they broke, how yields compare with plain staking, and what the 2026 numbers say. It builds on CCS’s guides to staking and Ethereum Layer 2s, the most common customers for restaked security.

Restaking by the numbers

What restaking is and what it is not

Staking on Ethereum means depositing 32 ETH per validator, attesting to blocks, and accepting that the deposit can be slashed if the validator signs conflicting messages. The reward is new ETH issuance plus tips and MEV: about 2.56 percent a year as of October 2026, according to Staking Rewards, with 43.84 million ETH (35.9 percent of supply) staked.

Restaking adds a second contract on top. The staker agrees that the same ETH (or an LST such as stETH) may also be slashed if a second set of rules is broken, in exchange for fees from whichever services rely on that promise. The ETH is not moved or lent; what is reused is the right to confiscate it. This is why critics call restaking rehypothecation: one unit of collateral underwrites more than one liability.

Three things restaking is not. It is not a deposit account; rewards depend on which services exist and what they pay. It is not insurance against validator failure; it adds slashing conditions rather than removing any. And it is not the same as a liquid restaking token, a wrapper issued by a third party that restakes on the holder’s behalf and carries its own contract, bridge and liquidity risks, as the April 2026 Kelp DAO exploit showed at a cost of roughly $292m.

Why anyone wants to buy security

A new oracle, bridge or data availability layer has a cold-start problem: nobody trusts it until value stands behind its operators, and launching its own token creates a small, volatile security budget. Renting a slice of Ethereum’s $119bn of staked ETH (Staking Rewards, October 2026) is faster and, in theory, harder to attack. EigenLayer calls these customers Actively Validated Services; Jito on Solana calls them Node Consensus Networks (NCNs). The buyer pays operators in its own token, ETH or stablecoins, and operators pass most of it to the restakers who delegated to them.

How EigenLayer works: operators, AVSs and operator sets

EigenLayer, built by Eigen Labs and launched on mainnet in June 2023 with deposit caps, has three roles. Restakers supply collateral, natively by pointing a validator’s withdrawal credentials at an EigenPod contract, or by depositing LSTs, EIGEN or other ERC-20s, and delegate the whole balance to one operator at a time. Operators run the software for each AVS. AVSs define the tasks, rewards and slashing conditions.

The marketplace opened on 9 April 2024, when caps were lifted and the first AVSs registered. Within two weeks TVL passed $15bn, with Google Cloud, Coinbase and HashKey among the operators, according to The Block on 23 April 2024. By April 2025 Cointelegraph counted more than 30 AVSs live on mainnet.

Operator sets and unique stake

The original design let any AVS slash any delegated stake, so a restaker could not size their exposure. The April 2025 slashing release (ELIPs 002 to 004) replaced that with operator sets. Each AVS creates sets with their own rules. An operator joins a set and allocates a specific amount of “unique stake” to it; only that amount can be slashed by that AVS, and the same stake cannot sit in two sets at once. Operators publish an allocation delay so restakers who dislike a new allocation can undelegate first.

1. A restaker deposits stETH into EigenLayer and delegates to an operator.
2. The operator allocates, say, 40 percent of that stETH to EigenDA’s operator set and 20 percent to a bridge AVS, leaving 40 percent unslashable by any AVS.
3. Each AVS pays rewards to its operator set, daily since Rewards v2 (ELIP-001, Q1 2025). Operators take a fee they set per AVS, replacing the earlier fixed 10 percent default.
4. If the operator misbehaves on the bridge AVS, that AVS can slash up to the 20 percent allocation; the rest is untouched.
5. Slashed shares sit for a 7-day resolution delay (ELIP-016) before being burned or, in a redistributable set, sent to the AVS’s pre-set recipient.

Rewards, EIGEN and EigenCloud

Restakers earn three things: the underlying ETH staking yield, AVS rewards in whatever asset the AVS chooses, and EIGEN programmatic incentives. EIGEN launched as a non-transferable “stakedrop” in 2024 and became tradable later that year; CoinGecko records an all-time high of $5.65 on 16 December 2024 and a price of about $0.26 with a $248m market capitalisation on 6 October 2026, about 95 percent below the peak.

In mid-2025 Eigen Labs repositioned the stack as EigenCloud, a “verifiable cloud” combining EigenDA (data availability), EigenVerify (dispute resolution) and EigenCompute (off-chain execution). The 17 June 2025 launch came with a $70m EIGEN purchase by a16z crypto and was followed on 8 July 2025 by a 25 percent staff cut. The message: restaked ETH is the collateral layer for a cloud business, not the product itself.

How slashing works and when it arrived

For its first 22 months EigenLayer had no slashing: deposits grew past $20bn by June 2024 with no way for a restaker to lose principal to an AVS. Slashing reached testnet on 19 December 2024 and mainnet on 17 April 2025, which Eigen Labs described as the protocol becoming feature-complete. The rollout was opt-in: operators had to allocate stake to slashable sets, and restakers could undelegate during the allocation delay.

Three parameters define the exposure. The withdrawal delay is 100,800 blocks, about 14 days, during which queued withdrawals remain slashable; the deallocation delay is also about 14 days; and the slash resolution delay is 50,400 blocks, about 7 days. An AVS does not need an on-chain proof: the documentation states that slashing “does not have to be objectively attributable”, so the AVS’s own governance or committee can decide, which makes that governance a direct input to the restaker’s risk.

Redistribution: slashing that pays someone

On 22 July 2025 EigenLayer activated redistribution. In a standard operator set, slashed assets are destroyed. In a redistributable set, flagged at creation and never convertible later, slashed assets go to an immutable recipient address chosen by the AVS. The use cases are insurance-like: Cap, a stablecoin protocol, uses it so operators who fail to deliver yield compensate holders. Native ETH and EIGEN cannot be redistributed; only LSTs and other ERC-20s qualify. EigenLayer’s own documentation warns that redistribution turns a slash from a burn into a transfer, so a compromised AVS governance key becomes a route to steal collateral. Public reporting of individual mainnet slashes remains thin; every large loss in restaking to date has come from wrappers and bridges, not AVS penalties.

Liquid restaking tokens: ether.fi, Renzo and Kelp

Native restaking requires a validator or an LST deposit locked behind a 14-day exit queue. LRT issuers removed both frictions: deposit ETH or an LST, receive a liquid receipt (weETH from ether.fi, ezETH from Renzo, rsETH from Kelp DAO, pufETH from Puffer), and let the issuer choose operators, allocate to AVSs and manage withdrawals. The receipt trades on DEXs and can be posted as collateral on Aave and Morpho, the source of both the appeal and the systemic risk.

The Block counted close to $8bn across LRTs on 1 April 2024, led by ether.fi above $3.2bn and Renzo at $2bn. DWF Labs put the sector at 3.34 million ETH, about $11.3bn, on 26 July 2024, more than triple the February 2024 level. Most of that capital was chasing EigenLayer points and issuer points, each convertible into an expected airdrop. ETHFI set its all-time high of $8.53 on 27 March 2024 and REZ peaked at $0.28 on 30 April 2024; by 6 October 2026 they traded at about $0.75 and $0.004 respectively, per CoinGecko and DefiLlama.

Why the largest issuer left

In August 2026 ether.fi removed all EigenLayer exposure from weETH, turning it into a plain liquid staking token, and created weETHs on Symbiotic for users who still wanted restaking. Less than 1 percent of assets remained with EigenLayer, with EigenPod credentials due to be removed in Q4 2026. Chief executive Mike Silagadze told The Defiant on 6 August 2026 it was the “end of an era” and said restaking no longer offered meaningful yield while carrying perceived risk. The numbers agree: the top five LRTs together earned about $953,000 of gross profit in Q2 2026 against $2.18m in Q3 2025, and in the week to 8 September 2026 restaking generated about $100,000 of fees on $10bn of TVL while liquid staking generated $27.4m on $51.9bn, per data cited by TechFlow and Blockhead.

Beyond Ethereum: Symbiotic, Babylon, Jito and Solayer

Symbiotic: from restaking to collateral markets

Symbiotic, backed by Paradigm with about $34.8m raised, took the opposite design: a minimal immutable core, permissionless vaults accepting any ERC-20, and networks that bring their own slashing logic plus optional “resolvers” who can veto a slash. It launched on mainnet in January 2025 with slashing live from day one. In a 1 July 2026 CCS interview, co-founder and CEO Misha Putiatin explained the Core V2 pivot to “collateral markets” for credit, insurance and real-world-asset liquidity, with about $300m deployed at the time. DefiLlama now lists Symbiotic under Collateral Markets, with $487m of TVL, more than 80 vaults, 74,000 stakers and 111 operators as of 6 October 2026.

Babylon: Bitcoin staking without a bridge

Babylon lets BTC holders lock coins in a self-custodial timelock script on Bitcoin itself and delegate “finality provider” rights to a proof-of-stake chain, without bridging or wrapping. Slashing uses extractable one-time signatures (EOTS): a provider who signs two conflicting blocks leaks its private key, which lets anyone burn the staked BTC. Deposits opened in capped phases, with Cap-1 at 1,000 BTC in August 2024 and about 23,000 BTC ($1.4bn) by 9 October 2024, led by Lombard and Solv, per Blockworks. The Babylon Genesis chain and BABY token launched in April 2025. As of 6 October 2026 DefiLlama puts Babylon at $3.49bn, 32 percent of all restaking TVL, with BABY at a $67m market capitalisation. CCS interviewed Babylon Labs’ Clayton Menzel in May 2025.

Solana: Jito and Solayer

Solana restaking has stayed small. Jito Restaking tokenises deposits into Vault Receipt Tokens and lets Node Consensus Networks such as Switchboard and Squads rent them across 10 vaults; Solayer issues sSOL. DefiLlama shows Jito Restaking at $20.5m and Solayer at $12.4m on 6 October 2026, each a fraction of a percent of the category. Solana’s higher native staking yield leaves less room for a restaking premium to matter; see the CCS guide to Solana for the base economics.

Yields: restaking versus plain staking in 2026

The baseline is Ethereum’s own reward rate, 2.56 percent as of 6 October 2026, which every restaker also earns. In 2024 the second layer paid in points, a claim on future tokens rather than cash flow. In 2026, measured in fees actually paid by AVSs, it pays little: the week-to-8-September-2026 figure cited by TechFlow (about $100,000 on $10bn of TVL) annualises to roughly 0.05 percent, and EigenLayer’s 30-day fees on DefiLlama were $141,000 on 6 October 2026 with no protocol revenue retained. EIGEN incentives add a token-denominated layer, but EIGEN is down about 95 percent from its December 2024 high.

Marketing figures should be read carefully. Dashboards still quote “up to” rates of 4 to 12 percent, which bundle base staking, token incentives, points at an assumed price and sometimes leverage loops. Symbiotic’s DefiLlama page shows an average 22.65 percent APY across 12 pools in October 2026, but those are collateral-market vaults paying in project tokens for credit and insurance underwriting, not ETH restaking yield. The gap between headline and realised fees is the main reason ether.fi left.

Systemic risk: rehypothecation, slashing and depegs

Restaking stacks liabilities on one asset. A unit of stETH inside an LRT inside a lending market carries at least five layers of risk: the validator, Lido’s contracts, EigenLayer’s contracts and the operator’s AVS allocations (slashable again, with 14-day exits), the LRT issuer’s contracts and bridges, and the lender’s liquidation logic. A problem anywhere prices into the most liquid layer, the LRT, first.

Correlated slashing. If one operator runs many AVSs on shared infrastructure and suffers a key compromise or bug, several AVSs can slash the same delegated pool at once. Operator sets cap each AVS at its allocation, but they do not stop one operator failure from triggering slashes across every set it joined.

Redistribution as an attack surface. Because redistributable sets send slashed funds to a chosen address, an AVS whose governance multisig is compromised can slash and collect. EigenLayer’s documentation lists compromised AVS governance and operator-AVS collusion as the main vectors.

LRT depegs and the Kelp case. LRTs trade at a discount whenever redemptions slow or confidence drops, and lenders that accepted them near par become the shock absorber. On 18 April 2026 an attacker forged a LayerZero message to Kelp’s rsETH bridge, which accepted a single verifier (a “1-of-1 DVN”), and released 116,500 rsETH from mainnet escrow, about 18 percent of supply and worth roughly $292m, with no backing. The attacker posted the tokens on Aave v3 and borrowed about $196m of WETH before Kelp paused contracts 46 minutes later. Aave froze rsETH and its TVL fell about $6.6bn in 24 hours; bad-debt estimates ranged from $124m to $230m against an Umbrella safety module and treasury backstop. On 21 April 2026 Arbitrum’s Security Council voted to seize 30,766 ETH (about $71m) of the proceeds mid-bridge, itself a contested intervention. Preliminary attribution pointed to North Korea’s Lazarus Group. CCS covered the Aave fallout and the Arbitrum seizure. The lesson: restaking’s largest realised loss had nothing to do with validators or AVSs. It came from the wrapper and its bridge.

Institutional uptake and the 2026 state

Institutions touched restaking as operators (Google Cloud, Coinbase and HashKey registered on EigenLayer in April 2024), as investors (a16z crypto led a $100m Series B in February 2024 and bought $70m of EIGEN in June 2025) and as customers (Sony’s Soneium L2 adopted an EigenLayer-secured fast-finality layer in April 2025, cutting finality to under 10 seconds). What has not happened at scale is institutional capital restaking its own ETH; the custody, exit-delay and slashing-governance questions are hard to clear with a risk committee, and regulated providers have mostly stayed at the plain-staking layer covered in the CCS custody guide.

As of October 2026 the sector is smaller and more specialised. DefiLlama’s restaking category holds $10.9bn across 14 protocols, two-thirds EigenLayer and most of the rest Babylon; LRTs are a $1.45bn category dominated by Kelp. EigenLayer’s TVL has rebuilt from a 2026 low (Blockhead reported about $5.1bn in late September 2026; DefiLlama showed $7.1bn on 6 October, up 8.9 percent on the month), and on 30 September 2026 the protocol confirmed operators can post Bitcoin-linked collateral alongside staked ETH, per CCS reporting. EIGEN has a 36.82 million token unlock due on 1 November 2026.

How we got here: a timeline

Worked example: 100 ETH, three ways to stake it

How to evaluate a restaking position: a checklist

• What is the realised uplift, in ETH, over the last 90 days? Ignore “up to” figures. Ask for AVS rewards and token incentives actually distributed, net of fees, divided by TVL. In 2026 the sector number was far below 1 percent.
• Which operator sets hold my stake, and can any redistribute? Redistributable sets convert a slash into a payment to a third party. Know the recipient and who controls the AVS’s slashing decision.
• How is a slash decided? On-chain proof is best. A committee or vote means the AVS’s key management is your risk. EigenLayer permits non-attributable slashing by design.
• How long does exit take, and what happens in the queue? EigenLayer withdrawals take about 14 days and stay slashable; LRT redemptions add the issuer’s queue. In a crisis the only exit is the secondary market at a discount.
• If it is an LRT, how is the bridge secured? Kelp’s $292m loss came from a 1-of-1 verifier. Ask for the verifier set, the pause authority and whether Layer 2 balances are backed by mainnet escrow.
• Who else holds this token as collateral? An LRT widely used on Aave or Morpho will be discounted in a stress event even if your position is fine, as April 2026 showed.

Risks and open questions

The structural question is whether demand for rented security can pay for the capital that arrived to supply it. Rollups, the obvious customers, buy data availability at prices measured in thousands of dollars a month, not hundreds of millions, and oracles and bridges have been slow to pay restakers more than token incentives. Until an AVS category generates fees comparable to Ethereum’s own issuance, restaking yields will stay a rounding error on top of staking yields.

The second question is governance of slashing. Non-attributable slashing and redistribution make restaking useful for subjective services, but they move trust from code to committees, and a captured AVS governance process is now a direct route to delegated collateral. The third is contagion through lending markets: Aave absorbed roughly $196m of bad debt from a wrapper it did not control, and whether any LRT regains collateral status at scale is an open question for 2027. Finally, the tax and securities treatment of AVS rewards remains unaddressed by the IRS and the SEC, and custodians have mostly declined to offer restaking until it is.

What to watch next

Glossary

Restaking

Pledging already-staked assets, or tokens representing them, as slashable collateral for additional services in exchange for extra rewards.

AVS (Actively Validated Service)

EigenLayer’s term for a service that rents restaked security: a data availability layer, oracle, bridge or fast-finality layer.

Operator

A node runner that accepts delegated restaked assets and runs the software for one or more AVSs, taking a fee from rewards.

Operator set

A group created by an AVS that defines tasks, rewards and slashing rules; operators allocate a set amount of stake to each one they join.

Redistribution

A slashing mode, live on EigenLayer since July 2025, in which slashed assets go to a pre-set recipient rather than being burned.

LST (liquid staking token)

A transferable token representing staked ETH and its rewards, such as stETH or rETH.

LRT (liquid restaking token)

A transferable token issued by a third party that restakes deposits on the holder’s behalf, such as weETH, rsETH or ezETH.

EigenPod

A per-user contract that receives a validator’s withdrawal credentials so its balance can be natively restaked.

EOTS

Extractable one-time signatures, Babylon’s scheme under which a double-signing finality provider leaks the key needed to slash its BTC.

DVN

Decentralised verifier network, the parties that must attest to a LayerZero cross-chain message; Kelp’s bridge required only one.

Why it matters

Restaking was the clearest test so far of whether crypto-economic security can be sold as a commodity. The supply side showed up in extraordinary size: more than $20bn of ETH within a year, then billions of dollars of BTC through Babylon. The demand side has not. Three years in, fees paid by services renting that security are a small fraction of a percent of the collateral supplied, the largest wrapper issuer has left, and the largest loss came from a bridge rather than from slashing.

That does not make the idea dead. Babylon has turned restaking into the first credible way to earn on native Bitcoin without a custodian, EigenCloud has made its collateral pool the backing for a verifiable computing business, and Symbiotic has found the same machinery works for insurance and credit. What the 2024 to 2026 cycle settled is the order of operations: security has to be demanded before it is worth supplying, and a wrapper’s own contracts are a bigger risk to holders than the validators underneath. Anyone comparing a staked, liquid-staked or liquid-restaked position in 2026 should start from that finding rather than from the headline APY.

Sources

1. EigenCloud docs: Slashing concept, accessed October 6, 2026
2. EigenCloud docs: Redistribution, accessed October 6, 2026
3. Eigen Labs: Redistribution is live on mainnet, July 22, 2025
4. DefiLlama: Restaking protocols by TVL, October 6, 2026
5. DefiLlama: Liquid restaking protocols by TVL, October 6, 2026
6. DefiLlama: EigenCloud, October 6, 2026
7. Staking Rewards: Ethereum staking reward rate, October 6, 2026
8. CoinGecko: EigenCloud (EIGEN), October 6, 2026
9. The Defiant: ether.fi removes restaking from weETH, nearing a full EigenLayer exit, August 6, 2026
10. Blockhead: Ether.fi to fully exit EigenLayer restaking as sector profits collapse, September 29, 2026
11. TechFlow: As the restaking craze subsides, why is ether.fi bidding farewell to EigenLayer?, September 2026
12. The Block: Kelp DAO’s rsETH bridge apparently exploited for roughly $292 million, April 18, 2026
13. DeFi Prime: The KelpDAO rsETH exploit: $292M minted from a 1-of-1 bridge, April 2026
14. Blockworks: EigenCloud launches aiming to bring verifiability to everything, June 17, 2025
15. The Block: a16z crypto invests $70 million in direct EigenLayer token deal, June 17, 2025
16. Cointelegraph: EigenLayer to begin slashing restakers in April, April 3, 2025
17. Kiln: EigenLayer unveils Rewards v2 and slashing for 2025, January 15, 2025
18. ForkLog: Eigen Labs reduces workforce by 25% to focus on EigenCloud, July 8, 2025
19. The Block: EigenLayer’s TVL crosses $15 billion, April 23, 2024
20. The Block: Liquid restaking platforms jump to near $8 billion in TVL, April 1, 2024
21. DWF Labs: Liquid restaking and tokens (LRT), overview of top protocols, July 26, 2024
22. Blockworks: Bitcoin-staking protocol Babylon attracts $1.4B in deposits, October 9, 2024
23. Crypto Coin Show: EigenLayer adds Bitcoin collateral to its restaking protocol alongside staked ETH, September 30, 2026

Disclosure: This guide is for education only and is not investment, legal or tax advice.