Restaking: A Reference

Why restaking matters in 2026

Restaking (restaking is the practice of using already-staked ETH or LSTs as cryptoeconomic security for additional services beyond Ethereum consensus) remains Ethereum's most consequential new primitive since liquid staking — but 2026 changed the risk calculus: EigenLayer's canonical restaked TVL sits at ~$6.5B after a post-slashing-launch repricing, EIGEN trades at ~$0.25 (95% off ATH), slashing is live and contractually enforceable since April 2025, and the April 2026 Kelp DAO $292M bridge exploit showed that LRT-layer risk is real and separable from the core protocol. Last verified: 2026-05-27.

Restaking means renting your stake's slashing capacity to other protocols, which can take a share of it if you misbehave on their service. The yield is a fee for absorbing extra risk, not a bonus for showing up. Vitalik Buterin's 2023 essay "Don't overload Ethereum's consensus" explains why pooled restaking risk is a systemic concern, not just an individual one.

EigenCloud (formerly EigenLayer, rebranded June 2025) holds ~$6.5B in canonical restaked TVL per DefiLlama, controlling roughly 58% of an ~$11.3B total restaking market. Gross figures cited elsewhere run higher because they include restaked LSTs double-counted against both the LST protocol and EigenLayer; the ~$6.5B canonical number reflects unique ETH-denominated collateral actually held in EigenLayer's contracts. The EIGEN token trades at ~$0.25, approximately 95% below its $5.65 all-time high. Five LRT protocols — Ether.fi, Puffer, Renzo, Kelp, and Swell — control the vast majority of the restaking market, per DefiLlama. Realised cash yield is 3.5–5.0%; most headline APYs are speculative points and token incentives.

How restaking actually works

You stake ETH, opt your validator (or LST) into EigenLayer, delegate to an operator running an AVS, and earn extra yield in exchange for additional slashing exposure stacked on top of base ETH consensus risk. The slashing risk is cumulative across every AVS you opt into — and as of April 17, 2025, slashing is live on mainnet with contractually enforceable penalties. Last verified: 2026-05-27.

1. You stake ETH to a validator (solo, Lido, Rocket Pool, etc.). This earns base consensus yield (~3.0–3.5%) and exposes you to standard Ethereum slashing.
2. You opt into EigenLayer — your validator (or your LST) is now restakable. The validator's withdrawal credentials get pointed at an EigenLayer "EigenPod," giving EigenLayer's contracts the right to slash if AVS rules are broken.
3. An AVS (Actively Validated Service) requests cryptoeconomic security from EigenLayer, defining its own slashing conditions and reward terms.
4. You delegate to an operator who runs the AVS's required software. The operator is the actor who can be slashed if they misbehave on the AVS.
5. You earn extra yield from the AVS — paid in fees, tokens, or points.
6. You take on extra slashing exposure — if the operator misbehaves on the AVS, you lose part of your stake. This risk is contractually live as of April 2025; zero slashing events have been triggered through May 2026, but the mechanism is no longer notional.

The key insight: each AVS adds its own slashing conditions on top of base ETH consensus slashing. Restakers earn yield from multiple sources, but their slashing scope is the sum of all opted-in conditions. Many users in 2024–2025 treated restaking as a free-lunch yield boost while slashing was inactive; that excuse is gone.

The four ways to participate

Native restaking (run a validator, opt into EigenLayer), LST restaking (deposit stETH/rETH/cbETH), LRT (deposit to Ether.fi/Renzo/Kelp/Puffer/Swell), or fixed-yield via Pendle PT-eETH. Each strips out a different combination of operational complexity, smart-contract risk, and yield ceiling. Last verified: 2026-05-27.

1. Native restaking

Run your own validator, opt into EigenLayer directly. Highest yield, highest operational complexity. Requires 32 ETH, a node setup, and ongoing operational discipline (uptime monitoring, key management, AVS software upkeep).

2. LST restaking

Restake your stETH / rETH / cbETH directly into EigenLayer. No validator ops needed. Slightly lower yield than native because you pay LST issuer commission (typically 10%) on top of standard restaking economics. Adds LST issuer risk (Lido governance, Rocket Pool node-operator slashing pool) on top of EigenLayer risk.

3. Liquid restaking token (LRT)

Deposit ETH or LST to Ether.fi / Renzo / Puffer / Kelp / Swell. They handle EigenLayer plus operator selection and AVS allocation. You receive a transferable LRT (eETH, ezETH, pufETH, rsETH, rswETH) usable in DeFi. This is what the vast majority of restaking TVL chose — and also the layer where the largest restaking-adjacent exploit to date occurred (Kelp DAO, $292M, April 2026, via a LayerZero bridge vulnerability, not EigenLayer itself).

4. Restaking through a derivative

Pendle (Pendle is a yield-tokenization protocol that splits yield-bearing assets into Principal Tokens with fixed yield and Yield Tokens with floating yield) PT-eETH / PT-weETH / PT-pufETH lock fixed yield without ongoing restaking exposure. The PT holder gets a fixed rate to a maturity date; the YT holder takes on the speculative side (variable yield plus airdrop points). Best for users who want predictable returns and don't want to track AVS-level risk.

What is an AVS?

An AVS (an Actively Validated Service is any protocol that pays restaked ETH holders for cryptoeconomic security and defines its own slashing conditions) is a protocol that pays restakers for cryptoeconomic security: EigenDA (data availability), Lagrange (ZK coprocessors), AltLayer (RaaS), Hyperlane (interop), eOracle (oracles). Each AVS sets its own slashing conditions and reward terms. EigenCloud has 60+ AVSs live across 1,900+ active operators. Last verified: 2026-05-27.

An AVS is any protocol that uses restaked ETH for cryptoeconomic security:

Each AVS sets its own slashing conditions (e.g., "if you sign a conflicting attestation in our protocol, lose X% of your delegation") and reward distribution. Slashing went live across EigenLayer on April 17, 2025; the first wave of AVSs with enforceable slashing includes EigenDA, Lagrange, and Hyperlane. Reading the AVS docs is non-optional if you care about your worst-case downside: a restaker delegating across five AVSs takes on the union of five distinct slashing rule sets.

The five major LRTs

An LRT (a Liquid Restaking Token is a tokenised receipt for restaked ETH that remains usable as DeFi collateral) tokenises a restaked position. Ether.fi leads with ~$2.8B+ TVL; the April 2026 Kelp DAO $292M exploit — a LayerZero bridge failure, not an EigenLayer contract bug — confirmed that LRT-layer risk is structurally distinct from base restaking risk. Last verified: 2026-05-27.

Kelp DAO status as of late May 2026: The April 18 exploit drained 116,500 rsETH ($292M) via a compromised LayerZero 1-of-1 DVN bridge, leaving ~$190M bad debt on Aave. Recovery was nearing completion by late May 2026 (~116,000 rsETH returned). Kelp migrated rsETH bridge infrastructure from LayerZero to Chainlink CCIP, and Aave/Kelp restored rsETH operations. LayerZero later acknowledged it "made a mistake." The exploit has been attributed to North Korea's Lazarus Group.

Restaking vs LSTs vs solo staking: the comparison most explainers skip

Solo staking maximizes alignment and minimizes counterparty risk but requires 32 ETH and ops work. LSTs (Lido, Rocket Pool) trade a small commission for full DeFi composability. LRTs trade additional smart-contract and AVS slashing risk for an extra 0.5–1.5% realised yield. The risk-adjusted answer for most holders is still a plain LST. Last verified: 2026-05-27.

Risk layers

Six risks stack: ETH consensus slashing, AVS slashing (live since April 2025), operator misbehavior, LRT smart-contract risk, bridge/infrastructure risk (Kelp DAO $292M, April 2026), and concentration if one operator or AVS becomes systemically important. Zero slashing events have fired through May 2026, but the mechanism is contractually enforceable. Last verified: 2026-05-27.

Restaking adds risk on top of holding plain ETH; each layer is independent and additive.

• ETH consensus slashing. Same as plain staking — penalties for double-signing, surround voting, or extended downtime. Well-understood, small in absolute terms, and operationally mitigated by professional validator setups.
• AVS slashing. Each AVS adds its own slashing conditions. Slashing went live on EigenLayer mainnet on April 17, 2025. Zero events have fired through May 2026, but the first wave of AVSs (EigenDA, Lagrange, Hyperlane) now have enforceable slashing live in production. The risk is real, not notional.
• Operator risk. Operators can be slashed for misbehaving on AVSs. If you delegate to a bad operator, your stake takes the hit alongside theirs. Operator quality varies; professional operators (P2P, Stakefish, Figment) have maintained zero-slash track records.
• LRT smart-contract risk. The LRT contract itself can be exploited. The Kelp DAO $292M exploit (April 18, 2026) was not an EigenLayer contract bug — it was a bridge infrastructure failure: LayerZero's 1-of-1 DVN configuration was compromised by attackers (attributed to Lazarus Group) who poisoned two of LayerZero's RPC nodes, allowing them to forge cross-chain messages and drain 116,500 rsETH. The root cause is a configuration risk (single DVN, no second verifier) that 40% of LayerZero-integrated protocols share. Kelp migrated to Chainlink CCIP (which requires 16 independent validators per message) by May 2026.
• Depeg risk. LRTs can depeg from ETH during forced redemptions, large unwinds, or stress events. Renzo's ezETH briefly depegged ~10% in April 2024 during the points-farming unwind. Pendle PT holders and leveraged loopers got liquidated en masse. The Kelp exploit also caused rsETH to suspend operations briefly, highlighting bridge-driven depeg risk.
• Concentration risk. If one AVS or one operator becomes systemically important and fails, multiple LRTs get hit simultaneously. The Kelp exploit left ~$190M bad debt on Aave — a concrete example of how LRT stress propagates across DeFi lending, matching Vitalik's 2023 warning about overloading Ethereum consensus-adjacent security.

Yield reality

Realised cash yield from restaking is 3.5–5.0% — base ETH staking (3.0–3.5%) plus AVS rewards (0.5–1.5%). Headline 8–15% APYs depend on points and token incentives that may never realise. EigenCloud runs a -$12.7M annual earnings deficit (spending ~$57M in incentives vs ~$14M in fees), meaning AVS yield is currently cross-subsidised. Size positions on the cash number, not the headline. Last verified: 2026-05-27.

What is the real cash yield from restaking?

Realised cash yield in 2026: roughly 3.5–5.0%. Headline 8–15% APYs include token plus points speculation. The 2024 EigenLayer airdrop disappointed many farmers; the 2024 LRT points-into-token conversions (Renzo, Ether.fi, Kelp) delivered but at lower implied valuations than the points-market had priced. The pattern: realised airdrop value is usually 30–60% of what the points-market projected at peak. Size positions on the cash number, not the headline. The fact that EigenCloud is running a significant incentive deficit — spending ~$57M/yr in incentives against ~$14M/yr in fees — means current AVS yields are subsidised; they may compress as subsidy winds down.

Common misconceptions

"Restaking is free yield" — no, it is compensation for extra slashing now live in production. "LRTs are like LSTs" — the Kelp $292M bridge exploit and the ezETH ~10% depeg in 2024 said otherwise. "AVSs don't actually slash" — true through March 2025, false thereafter. "Pendle PT means no risk" — PT is fixed-yield, not zero-risk, and the underlying LRT can still depeg or be exploited. Last verified: 2026-05-27.

• "Restaking is free extra yield." It is compensation for absorbing additional slashing capacity. Cash yield is 0.5–1.5% above base ETH staking — well below headline 8–15%. Slashing is now live; the free-lunch framing is gone.
• "LRTs behave like LSTs." The Kelp DAO $292M exploit and the ezETH ~10% depeg in April 2024 are the two clearest counterexamples. LRTs add a contract layer, an operator layer, AVS-level slashing exposure, and bridge infrastructure risk that LSTs don't carry. The peg can break independently of the underlying ETH, for reasons (bridge compromise, operator failure, mass liquidation) unrelated to Ethereum consensus.
• "AVSs don't actually slash." Through early 2025, mostly true. Slashing went live on EigenLayer mainnet on April 17, 2025. As of May 2026, zero events have fired — but the conditions are enforceable and real.
• "Pendle PT is risk-free." PT-eETH is fixed-yield to maturity, but the maturity payout is in the underlying LRT. If the LRT depegs (as ezETH did in April 2024) or the underlying protocol is compromised, PT holders take the hit too.
• "Diversifying across LRTs eliminates the risk." It diversifies LRT-issuer-specific risk, but the underlying restaked ETH still passes through EigenLayer's contracts and the same overlapping operator set. AVS-level and EigenLayer-level risk remain correlated across LRTs. The Kelp exploit also demonstrated that bridge infrastructure risk is entirely independent of both EigenLayer and LRT-issuer risk.

Practical recommendations

First-time stakers: plain stETH or rETH. DeFi users: weETH on Aave — conservatively sized. Yield maxers: cap LRT exposure under 20% across 2+ LRTs. Fixed-yield seekers: Pendle PT-eETH. Don't size on speculative airdrop points and don't loop LRTs. Last verified: 2026-05-27.

• First-time stakers: Use plain stETH (Lido) or rETH (Rocket Pool). Skip restaking entirely until you understand the slashing scope of each AVS and the bridge/contract risk in any LRT you consider.
• DeFi users: Use weETH (wrapped Ether.fi) — the most-accepted LRT collateral on Aave and Morpho. Keep LTV conservative; an LRT depeg of 10% with 75% LTV is an instant liquidation, as April 2024 ezETH holders discovered.
• Yield maxers: Cap LRT exposure at under 20% of ETH allocation. Diversify across 2+ LRT issuers to mitigate single-protocol risk (but note EigenLayer-level and bridge-level risk remain correlated).
• Fixed-yield seekers: Pendle PT-eETH or PT-weETH locks 5–6% for 3–12 months without active restaking management. Still exposed to LRT depeg and EigenLayer contract risk.
• Don't loop LRTs. Looping (deposit LRT, borrow ETH, swap to LRT, redeposit) amplified both gains and losses during the April 2024 ezETH depeg, with many vaults liquidating cascadingly. The leverage looks free until it is not.
• Don't farm points. If your investment thesis depends on a future airdrop, you are speculating, not investing. The realised airdrop is usually 30–60% of the peak points-market price.

Looking ahead

A few specific signals worth tracking through 2027:

• EigenCloud's unit economics. The protocol is running ~$57M/yr in incentive spend against ~$14M/yr in AVS fee revenue as of mid-2026. Restaking as a sustainable primitive requires AVS fees to grow materially or incentives to compress. Watch whether the gap closes.
• AVS slashing events in production. Zero events have fired through May 2026. The first live slashing event will be a major market test — how much does a legitimate slash move LRT prices? The answer will reprice the entire restaking risk premium.
• Bridge architecture after Kelp. The Kelp exploit has accelerated a broader audit of bridge DVN configurations across LayerZero-integrated protocols. 40% of protocols on LayerZero use the same 1-of-1 configuration that was exploited. Watch for industry-wide migration to multi-DVN or alternative bridge standards (Chainlink CCIP, Wormhole).
• Symbiotic and Karak gaining share. EigenCloud holds ~$6.5B canonical TVL; Symbiotic has grown to $1B+ TVL with a permissionless, multi-asset approach. The TVL split and whether multi-asset restaking opens new AVS revenue categories are worth tracking.
• Native restaking on L2s. The first restaking primitives that secure L2 services (sequencers, fast-finality bridges) using L1 restaked ETH are shipping. If sequencer decentralization on Arbitrum or Optimism uses restaking, the addressable AVS revenue grows materially.

Verdict

Plain LSTs are the better risk-adjusted choice for ~80% of ETH holders. The 20% who actively want LRT exposure should cap allocation at 20% of ETH, diversify across 2+ LRT issuers, ignore speculative point/token rewards in sizing, and read each AVS's slashing conditions and bridge security model before deploying. Last verified: 2026-05-27.

For 80% of ETH holders, plain LSTs are the better choice — base yield is already attractive at 3.0–3.5%, and the extra 0.5–1.5% from restaking comes with disproportionate complexity, smart-contract risk, bridge infrastructure risk, and slashing exposure most users will not model correctly.

For the 20% who actively want LRT exposure: cap at 20% of ETH allocation, diversify across 2+ LRTs, and do not size on speculative point/token rewards. Read each AVS's slashing conditions before opting in. The April 2026 Kelp $292M bridge exploit — now fully attributed to a LayerZero DVN infrastructure failure and Lazarus Group, not EigenLayer itself — and the April 2024 ezETH depeg are both within the historical track record of this primitive. The risk is specific and learnable; ignoring it is how the losses happened.

---

Related: Best ETH Restaking Protocols 2026 · Best Liquid Staking Tokens 2026 · How to Stake ETH