At 11:47 PM UTC on April 24, 2024, a liquidation bot on Morpho Blue executed a forced sale of ezETH collateral at roughly $700 per token. The nominal price of ETH was near $3,700. Within 40 minutes, over $23 million worth of leveraged restaking positions on Morpho had been liquidated, and another cluster of forced sales was unfolding on Gearbox. No validator had been slashed. No protocol had been hacked. The underlying ETH had gone nowhere. What collapsed was a market microstructure: thin DEX liquidity, a gated withdrawal queue, and borrowed leverage all converged at the moment the Renzo team announced a disappointing airdrop.
That episode is the sharpest available illustration of what liquid restaking tokens (LRTs) actually are — and why they carry a fundamentally different risk profile from the liquid staking tokens (LSTs) that preceded them.
LRT vs LST: The Critical Distinction
An LST such as stETH or rETH wraps a plain Ethereum staking position. One validator set, one slashing condition (Ethereum's beacon chain), one direct relationship between the token and deposited ETH. When you hold stETH, you are exposed to exactly one slashing surface: an Ethereum validator double-signing or going offline.
An LRT wraps a restaked position. The underlying ETH is simultaneously delegated to EigenLayer operators, each of which opts into one or more Actively Validated Services (AVSs). Each AVS defines its own slashing conditions, independent of the beacon chain. When you hold an LRT, you are implicitly accepting the union of all slashing surfaces across every operator and every AVS in the protocol's basket. That is not an additive stack — it is multiplicative.
There is a second, equally important distinction: redemption. An LST holder can generally redeem at close to par via on-chain redemption contracts or deep DEX liquidity. An LRT holder cannot. Before your ETH comes back, it must exit the EigenLayer strategy contracts, clear the EigenLayer 7-day unstaking delay, and in some cases wait for beacon chain validator exits on top of that. Total withdrawal windows range from roughly 2 days (for small amounts that hit a buffer pool) to 15 days for a full redemption through Renzo. During market stress, that window is precisely when everyone wants liquidity.
The liquid staking tokens and the slashing risk at the validator layer article in our Ethereum series covers the beacon-chain slashing model in detail; restaking adds layers of slashing risk on top of that foundation.
Key distinction: An LST gives you staked ETH with one slashing surface. An LRT gives you restaked ETH with as many slashing surfaces as the operators in the basket collectively service.
For DeFi composability, the rebasing versus non-rebasing axis also matters. Rebasing tokens (like eETH) automatically adjust your wallet balance when rewards accrue — useful for seeing income, awkward as collateral. Non-rebasing reward-bearing tokens (weETH, ezETH, rsETH) keep the token count constant and instead let the exchange rate rise against ETH, making them far easier to integrate with Aave, Pendle, and Morpho.
The Major LRTs: A Field Comparison
The LRT market is dominated by a handful of protocols, each with a distinct design philosophy.
| Protocol | Token | Type | TVL (Jan 2026) | Withdrawal Window | Governance Token |
|---|---|---|---|---|---|
| ether.fi | eETH / weETH | Rebasing (eETH) / Non-rebasing (weETH) | ~$7.8B | ~1-7 days | ETHFI |
| Renzo | ezETH | Non-rebasing (reward-bearing) | Declined post-depeg | Up to 15 days | REZ |
| KelpDAO | rsETH | Non-rebasing (reward-bearing) | $1B+ | 7-10 days (buffer: ~2 days) | KEP |
| Puffer Finance | pufETH | Non-rebasing | ~$247M | ~7-10 days | PUFFER |
| Swell | rswETH / swETH | Non-rebasing | ~$300M combined | ~7-10 days | SWELL |
ether.fi is the clear market leader at approximately $7.8 billion TVL as of January 2026, managing over 4.5 million ETH across more than 142,000 validators. The eETH / weETH split addresses a practical DeFi problem: eETH's rebasing nature makes it awkward as collateral, so wrapping it into weETH freezes the share count and lets the yield appear as a rising exchange rate instead. At peak, over 70% of all weETH supply was deployed in Pendle yield markets; Curve pools carried $800 million or more in liquidity, achieving under 0.05% slippage on a $10 million swap.
Renzo (ezETH) was a major protocol by TVL, peaking above $3 billion in April 2024. ezETH is reward-bearing: both Ethereum PoS staking rewards and EigenLayer AVS restaking rewards auto-compound into the exchange rate, so holders do not need to manually claim anything. The protocol routes deposits into EigenLayer strategy contracts across multiple operators, diversifying — but also accumulating — slashing exposure. As of January 2026, TVL had declined substantially following the April 2024 de-peg event, which is examined in detail below.
KelpDAO (rsETH) reached over $1 billion in TVL. rsETH accepts deposits in ETH or whitelisted LSTs (stETH, ETHx), aggregates them into EigenLayer strategy contracts, and issues rsETH as a single receipt token. Operator delegation is managed by Kelp governance rather than individual users. A buffer pool was introduced to allow small-amount withdrawals in approximately 2 days rather than the full 7-10 day EigenLayer delay. rsETH was widely used as collateral on Aave and Morpho. Part 4 of this series covers a larger incident affecting KelpDAO in April 2026 — that event is beyond the scope of this piece.
Puffer Finance (pufETH) sits at roughly $247 million TVL and emphasises a "native liquid restaking" model with anti-slashing validator keys, attempting to reduce the beacon-chain slashing surface before adding EigenLayer exposure on top. Swell (rswETH / swETH) holds approximately $300 million in combined TVL across its liquid restaking and L2 products, including Swellchain, which launched in December 2024.
The combined LRT category carried an estimated $18-24 billion in TVL during 2025-2026, with ether.fi alone representing roughly 40-50% of the LRT subset.
Token Mechanics Deep Dive
Understanding exactly how each token accrues value — and where that value can diverge from ETH — is essential before entering any leveraged position.
eETH is a rebasing token. When the EtherFiOracle publishes a reward report, total pooled ETH increases without any increase in the share count, causing every holder's eETH balance to auto-increment. This is intuitive for a wallet display but creates accounting problems when used as lending collateral: most money-market protocols prefer that the unit count of a collateral token stays stable.
weETH wraps eETH into a non-rebasing format. Wrapping freezes the share count; yield is reflected instead in a rising weETH/ETH exchange rate. The wrapper is the version that lives on Aave v3 as collateral, trades on Curve with deep liquidity, and serves as the principal token in Pendle yield strategies.
ezETH is non-rebasing and reward-bearing from inception. Value accrues via two compounding streams — Ethereum PoS staking rewards and EigenLayer AVS restaking rewards — both folded into the exchange rate with no manual claim required. This clean design made ezETH attractive for use as collateral, but it also concentrated risk: a sudden forced unwinding of the exchange rate on secondary markets cannot be arbitraged closed when the redemption path is gated behind a 15-day queue.
rsETH follows the same reward-bearing pattern. Deposits of ETH or LSTs enter EigenLayer strategy contracts; rsETH represents an aggregated, governance-curated restaked claim. The exchange rate rises as rewards accumulate. Kelp's buffer pool mechanism was a direct response to the liquidity mismatch problem — acknowledging that a 7-10 day exit window is incompatible with DeFi collateral use at scale.
For on-chain lending health factors and liquidation mechanics, the key variable is whether the collateral token's price can be reliably marked by a liquidation bot. When an LRT de-pegs on a thin DEX, the oracle reads a price that diverges from fundamental value — and borrows get liquidated at the DEX price, not the redemption value.
The Points Era and the Airdrop Flywheel (2024)
The explosive TVL growth of LRTs in 2024 was almost entirely incentive-driven rather than yield-driven. EigenLayer had not yet launched a live AVS or distributed a single dollar of real restaking revenue. Instead, it distributed points — off-chain accounting units that users accumulated by depositing into EigenLayer-aware protocols. LRT protocols layered their own points programmes on top, meaning a user who held weETH or ezETH was simultaneously accumulating EigenLayer points and protocol-specific points, each redeemable for a future token airdrop.
EigenLayer TVL grew from roughly $1 billion in January 2024 to over $15 billion by mid-2024. The driver was not yield; it was speculation on airdrop value. This created a structurally fragile capital base: much of the deposited ETH was in leveraged loops — users borrowing against their LRT to buy more LRT, amplifying points accrual. When EigenLayer mainnet launched in April 2024 and airdrop announcements began, the incentive to hold reversed sharply.
By 2026, the dynamics had matured. EigenLayer's slashing mechanism went live on April 17, 2025. AVS revenues began appearing on-chain. LRT yields started reflecting real restaking economics rather than points speculation. But the 2024 cycle left a durable lesson: capital attracted by airdrop incentives is concentrated in leveraged positions and exits rapidly the moment the incentive inverts.
Part 2 of this series covers how the operator and AVS layer works in detail.
The April 2024 ezETH De-Peg: A Case Study in LRT Fragility
On April 24, 2024, Renzo published the parameters for its REZ token airdrop. Season 1 allocated 5% of total REZ supply to points holders — a figure the community widely considered far below expectations.
Within minutes, large holders began selling ezETH on Uniswap. Because ezETH had no instant redemption mechanism, sellers had no alternative to the DEX. Uniswap liquidity for ezETH was thin relative to the size of the positions seeking exit. The price printed at approximately $688-$700, against a nominal ETH price of around $3,700 — a discount of roughly 80% from par. That discount was not a reflection of any underlying loss; it was a pure market microstructure failure on illiquid secondary markets.
The de-peg's damage was amplified by leverage loops. A common strategy had been to deposit ezETH on Gearbox or Morpho Blue, borrow ETH against it, purchase more ezETH with the proceeds, and repeat. When the oracle-read price of ezETH fell, health factors on leveraged positions breached liquidation thresholds. Liquidation bots began force-selling ezETH collateral into the same thin market, pushing the price lower, triggering further health-factor breaches, and cascading into a liquidation spiral.
Total liquidations across the event reached approximately $56-65 million: roughly $23 million on Morpho Blue across 146 users, and approximately 10,650 ezETH force-sold on Gearbox. ezETH recovered toward par over the following days as panic subsided and it became clear that no EigenLayer slashing had occurred and no protocol was insolvent. The damage was entirely a function of microstructure.
The oracle price feeds that determine when collateral positions are liquidated are a key vulnerability here: when a liquidation oracle reads a DEX spot price during a thin-liquidity panic, it marks collateral at a discount that does not reflect the real redemption value of the underlying — but the liquidation executes anyway.
This event also illustrates the broader pattern explored in composability and how a single depeg can propagate across interconnected DeFi protocols: ezETH's price distress in Uniswap pools immediately propagated to Morpho and Gearbox health factors, with no circuit breaker to slow the cascade.
The primary lesson is a precise one: the ezETH de-peg was caused not by slashing, not by protocol insolvency, and not by any failure of EigenLayer's infrastructure. It was caused by the intersection of thin secondary-market liquidity, a withdrawal queue that prevented arbitrage, and leveraged positions that amplified forced selling.
Risks to Understand Before You Participate
Holding an LRT means accepting a layered risk stack that is qualitatively different from holding an LST. The categories below are not exhaustive, but they cover the most material exposures.
Slashing risk (operator layer): If an EigenLayer operator to which your LRT protocol has delegated stake is slashed for misbehaviour on any AVS, that loss is socialised across all LRT holders in proportion to their share. EigenLayer's slashing mechanism went live on April 17, 2025. The more AVSs a given operator services, the more independent slashing conditions it is exposed to simultaneously.
De-peg and secondary-market risk: Because LRTs cannot be instantly redeemed for ETH at par, any mass exit creates DEX sell pressure against illiquid pools. The ezETH episode is the canonical example. Leverage loops transform a peg wobble into a liquidation cascade.
Withdrawal queue risk: Redemption windows range from approximately 2 days (Kelp buffer pool for small amounts) to 15 days (full Renzo unstaking). During market stress, the periods when holders most urgently want liquidity are precisely the periods when the withdrawal queue is longest.
Smart contract and bridge risk: The LRT system spans multiple contract layers — deposit contracts, EigenLayer strategy contracts, operator delegation modules, and for multi-chain protocols, cross-chain bridges. Each layer is an independent attack surface. Part 4 of this series, restaking, slashing, and how LRT bridge-layer risk materialised in April 2026, examines what happens when that bridge layer fails.
Concentration and governance risk: LRT protocols make operator-selection and AVS-onboarding decisions on behalf of all holders via on-chain governance. Individual token holders rarely scrutinise each new AVS being added to the basket, meaning a single governance vote can silently expand the slashing surface for the entire TVL base.
Points hangover: Positions entered specifically to accumulate airdrop points may be over-leveraged and under-researched. The incentive that attracted capital disappears at the moment of the airdrop, which is often precisely when mass exits occur and thin liquidity is most exposed.
Key Takeaways
- LRTs and LSTs are not interchangeable: an LRT carries the union of all slashing surfaces of every operator and AVS in the basket, plus a withdrawal queue that prevents instant arbitrage redemption — a fundamentally different risk profile.
- Token structure matters for DeFi integration: non-rebasing reward-bearing tokens (weETH, ezETH, rsETH) are better collateral than rebasing tokens, but "better" does not mean "safe" — the ezETH de-peg showed that a rising exchange rate cannot be redeemed instantly when the market panics.
- The April 2024 ezETH episode was a microstructure failure, not a protocol failure: no slashing occurred; the damage came from thin DEX liquidity, a gated redemption queue, and leveraged loops that cascaded into $56-65 million in liquidations.
- Leverage amplifies every LRT risk: an unleveraged LRT holder weathered the ezETH de-peg; a leveraged one was liquidated. Before using any LRT as collateral in a loop strategy, model the health factor at a 20-30% discount from par.
- Ether.fi's scale does not equal safety: market leadership gives weETH deeper liquidity and more DeFi integrations, but it also means ether.fi's operator basket carries the largest absolute restaking exposure in the category. Evaluate operators and AVSs, not just TVL rankings.
