What is restaking?
Plain staking lets your ETH secure exactly one thing: Ethereum. Restaking is the idea that the same staked ETH can also opt in to securing a second protocol — a new oracle network, a bridge, a data-availability layer — in exchange for additional rewards. If your validator misbehaves on that second protocol, the same ETH you posted to Ethereum gets slashed.
This is the core thesis of EigenLayer, which launched on mainnet in 2024 and grew to over $20B in TVL during the points-farming era. Symbiotic, Karak and Babylon (for BTC) followed with variations. Together they form a new market: shared crypto-economic security.
Restaking is "leverage" applied to security. The same dollar of stake backs multiple promises — and a default on any one of them can wipe the principal.
EigenLayer and AVSs
An AVS (Actively Validated Service) is any protocol that wants its own validator set without bootstrapping new economic security from scratch. It plugs into EigenLayer, defines its own slashing conditions, and pays rewards in its own token (or in ETH) to operators who opt in.
Restakers delegate their stake to operators. Operators choose which AVSs to support. When a restaker deposits an LST or natively-restaked ETH, their stake is exposed to:
- The base Ethereum slashing conditions (the validator's day job).
- Every slashing condition of every AVS the operator opts in to.
- Any operator-level misbehaviour penalties.
For AVSs, this is enormous: launching a new protocol no longer requires a year of token incentives to attract honest validators. For restakers, it means yields stack — but so do the ways to lose.
From restaking to LRT
Native restaking on EigenLayer is operationally heavy: you have to choose operators, manage AVS selection, track points, and queue withdrawals through a long timelock. Liquid restaking tokens abstract all of that away.
You deposit ETH (or stETH, wstETH, cbETH, etc.) into an LRT protocol. The protocol restakes for you, picks operators and AVSs on your behalf, and gives you a single token (eETH, ezETH, rsETH, pufETH…) that represents your share. The LRT is fungible, tradeable, and accepted by lending markets — just like an LST, but two layers deep.
The risk stack
Holding an LRT means you are simultaneously exposed to risks at five different layers. Failures stack — if any one breaks, your principal can move sharply.
For a user holding bridged rsETH on an L2, all five layers are live at the same time.
Major liquid restaking protocols
The LRT market consolidated quickly into a handful of protocols. Numbers are approximate as of April 2026.
| Protocol | Token | TVL | Notable feature | Status |
|---|---|---|---|---|
| ether.fi | eETH / weETH | $6.2B | Native restaking, non-custodial keys | Operating |
| Renzo | ezETH | $1.4B | Auto-routes across many AVSs | Operating |
| KelpDAO | rsETH | $1.0B* | Multi-asset restaking, multi-chain | Bridge exploit Apr 2026 |
| Puffer | pufETH | $580M | Anti-slashing pre-confirmations | Operating |
| Swell | rswETH | $340M | Tied to Swell L2 rollup | Operating |
| EigenPie / Magpie | m<LST> | $210M | Per-LST isolated vaults | Operating |
| Mellow | Various | $420M | Modular LRT vault factory (on Symbiotic) | Operating |
* KelpDAO TVL post-exploit, with rsETH bridge contracts paused on most chains.
Restaking-specific risks
Correlated AVS slashing
An operator running buggy software for one AVS can be slashed there and trigger downtime penalties on Ethereum at the same time. Losses correlate across what looked like diversified positions.
Opaque AVS exposure
Most LRTs don't tell you in real time which AVSs your stake is securing. Operators add and drop AVSs by governance vote — you find out after the fact.
Long withdrawal queues
EigenLayer enforces a 7–14 day withdrawal delay. During stress, the LRT trades at a discount in the secondary market while everyone waits for native exits to clear.
Compounded smart-contract risk
You're trusting the LST contracts, the LRT contracts, EigenLayer's contracts, every AVS's contracts, and (if bridged) a bridge's contracts. The probability of any bug is higher than for any single layer.
Bridge exploit risk
Most LRTs are bridged to L2s for DeFi composability. A bridge mint can create unbacked LRT supply — exactly what happened to KelpDAO in April 2026. See the case study →
Lending-market contagion
LRTs are widely accepted as collateral on Aave, Spark, Fluid, Morpho. A 15% LRT drawdown becomes 60%+ liquidation cascades for leveraged users, and bad debt for the lender.
Operator concentration
A small number of professional operators run most of the restaked stake. A single operator outage or compromise can affect multiple LRTs at once.
Governance attack surface
An LRT DAO that can change operator allow-lists, AVS selections or slashing logic is a target. Several LRTs have already had to pause governance after attempted hostile takeovers.
The "points" trap
Through 2024 and most of 2025, LRTs were marketed almost entirely on points: deposit now, accumulate points, redeem them later for an airdrop. Users layered LSTs into LRTs into looping leverage strategies on lending markets to multiply their points exposure.
That's still common today. Two things to keep in mind:
- Points are not yield. They are a marketing claim against a future token of indeterminate value. Many points programs eventually issued tokens worth a fraction of what users assumed.
- Looping is leverage. Borrowing ETH against your LRT and depositing again triples or quadruples your AVS exposure. A single AVS slashing event can wipe the entire stack.
Several of the most-leveraged LRT loop strategies in 2025 took 80–95% drawdowns when the underlying LRT depegged for 24–48 hours due to oracle issues, even though the LRT eventually recovered. Liquidations are permanent.
Evaluation checklist
- What's actually restaked? Native ETH, an LST, or an LRT-of-LRT? Each adds a risk layer.
- Which AVSs is the protocol opted into right now? Look for a public, real-time dashboard.
- What slashing conditions do those AVSs impose? Read the actual slashing math, not the marketing.
- What's the operator set? One curated provider, a permissionless market, or a multi-tier system?
- Where are the contracts deployed? If it's a bridged LRT on an L2, check who runs the bridge and what their verifier setup is — not just "audited by X".
- What's the redemption queue? Native withdrawal time, current backlog, and any per-day caps.
- Has the protocol been audited since the last upgrade? Many incidents originated in unaudited proxy upgrades.
- If you're tempted by points, what is the maximum total airdrop value as a percentage of TVL? It's almost always smaller than people assume.