EigenLayer crossed $15 billion in total value locked at its peak, and the pitch was straightforward: validators could take their staked ETH and pledge it again to secure additional networks - called Actively Validated Services - without withdrawing from Ethereum. Extra yield, same collateral. The problem is that framing obscures what restaking actually does to the risk profile of the underlying stake.
One Asset, Multiple Slashing Conditions
When ETH is restaked across several AVSs simultaneously, each service can impose its own slashing conditions. A validator who misbehaves - or whose operator software has a bug - on a given AVS can have their ETH slashed by that AVS’s rules, independent of anything happening on Ethereum mainnet. The restaker has effectively exposed a single pool of collateral to multiple, distinct fault domains.
This is not theoretical. Slashing on Ethereum mainnet requires a provably malicious act: double-signing a block, for instance. AVS slashing conditions can be considerably more opaque, governed by on-chain logic that varies by service. Some AVS contracts are upgradeable, meaning the slashing conditions themselves can change after a validator has already opted in.
The analogy to rehypothecation in traditional finance is imperfect but instructive. A prime broker who lends out a client’s securities has legal constraints on how many times those assets can be pledged. Restaking has no equivalent ceiling baked into the protocol by default.
The Operator Layer Adds Another Variable
Most retail restakers do not interact directly with AVSs. They delegate to operators - entities who manage the actual validation work across services. That introduces a principal-agent problem: the restaker carries the slashing risk while the operator makes the operational decisions. EigenLayer’s documentation acknowledges this delegation model, but the economic misalignment is structurally embedded. Operators can earn fees while restakers absorb tail risk.
Liquid restaking protocols like EtherFi and Renzo add a further abstraction layer, issuing receipt tokens that get deployed into DeFi. At that point, the original ETH is backing an LRT, which is backing positions in lending protocols, which may be collateral for other positions. The stack becomes difficult to unwind under stress.
This Isn’t an Argument Against Restaking
The underlying mechanic is genuinely useful. Bootstrapping economic security for new networks is a hard problem, and restaking offers a real solution. But the yield being offered is compensation for layered, compounding risk - not a return on otherwise idle capital. Treating it as the latter is where validators and liquidity providers tend to get surprised.