Restaking is one of the more genuinely novel ideas to come out of Ethereum’s post-Merge architecture. EigenLayer lets validators take their staked ETH - already locked up to secure Ethereum - and reuse it as cryptoeconomic collateral for other protocols, called Actively Validated Services (AVSs). Bridges, oracles, rollup sequencers: all of them can borrow Ethereum’s security without bootstrapping their own validator sets from scratch.
The pitch is efficient. Why fragment economic security across dozens of separate staking pools when you can concentrate it?
The problem is that concentrating security also concentrates failure.
What Slashing Actually Means at Scale
Slashing is the mechanism that makes staking honest. If a validator behaves maliciously - double-signing a block, for instance - a portion of their staked ETH is destroyed. The threat of loss is what keeps validators in line.
When a validator opts into multiple AVSs through EigenLayer, they’re accepting slashing conditions from each of those services simultaneously. Under normal circumstances, that’s fine. Under correlated stress - a bug in a shared AVS codebase, a coordinated attack targeting multiple services, or a widespread oracle failure - a single validator can be slashed by several protocols at once.
The restaked ETH backing those validators doesn’t multiply. It’s the same collateral being stretched across more potential loss events.
The Leverage Nobody Calls Leverage
This is, in a structural sense, leverage on security. The collateral base stays fixed while the obligations it’s meant to cover expand. EigenLayer’s documentation acknowledges this risk exists and places responsibility on operators to manage their AVS exposure carefully. In practice, that means the system’s safety depends heavily on individual validator discipline and good AVS design - neither of which is uniformly reliable.
There’s also a more subtle issue: correlated slashing can be amplified by the fact that many restakers delegate to the same large operators. If a major node operator runs into a slashing event across multiple AVSs simultaneously, the knock-on effect on ETH’s staking yield - and on the AVSs that relied on that operator’s collateral - could move markets before anyone has fully processed what happened.
This Isn’t an Argument Against Restaking
EigenLayer fills a real gap. Bootstrapping cryptoeconomic security is genuinely hard, and the protocol has attracted significant TVL since its mainnet launch.
But the framing around restaking has leaned heavily on capital efficiency and barely at all on the scenario where two or three AVSs face slashing conditions in the same week. That scenario isn’t theoretical - it’s a question of when the right combination of bugs and incentives lines up. The architecture tolerates it poorly, and that deserves more honest discussion than it’s currently getting.