What is the fundamental difference between Restaking and ordinary Ethereum Staking?
Ordinary staking locks up ETH to help validate Ethereum own blocks in exchange for roughly 3-4% base annual yield, and the only slashing risk you carry comes from Ethereum own consensus-layer rules. Restaking takes that same locked ETH and additionally delegates it to an operator who uses it to help secure other external services (AVSs), essentially, the same collateral takes on a second job.
The fundamental difference is in the nature of the risk: ordinary staking only requires understanding Ethereum own validator rules, while restaking requires understanding which AVSs your delegated operator has chosen to serve, what each AVS individual slashing conditions are, and whether those conditions could trigger simultaneously. Yield goes up as a result, but so does the scope of due diligence you actually need to do.
If I just deposit ETH into an LRT protocol like Kelp or Ether.fi, do I need to manage operators and AVSs myself?
No, that is precisely the point of an LRT protocol. It handles the technical decisions around operator selection and AVS allocation for you; you simply deposit ETH and receive a tradeable receipt Token (like rsETH), with a day-to-day experience similar to ordinary liquid Staking.
But this convenience does not make the risk disappear, it shifts and stacks it in a different place. You now additionally carry the risk of the LRT issuer own smart contracts, plus the risk of whatever Cross-Chain Bridge the issuer deploys to let the token be used across multiple chains. This is exactly where the Kelp DAO incident detonated, the problem was not a user picking the wrong AVS, it was a configuration flaw in the cross-chain bridge infrastructure the protocol itself had chosen. In other words, using an LRT protocol simplifies the operation, but it does not simplify the due diligence you should be doing, it just shifts the target of that evaluation from the AVS list to whether this LRT issuer contract and bridge architecture is actually sound.
How much risk does EigenLayer Unique Stake mechanism actually reduce?
Unique Stake addresses the concentration problem. Under earlier designs, delegated stake tended to be indiscriminately exposed across every AVS it touched, which meant a slashing event on one AVS could theoretically ripple across every other AVS that same stake was involved with, the correlated slashing risk described above. Unique Stake narrows that blast radius by letting stake be allocated to specific operator sets individually, rather than blanket-exposed everywhere.
What is worth understanding is that this does not reduce the number of Smart Contract layers your capital passes through. Your funds still travel through the Ethereum Staking layer, the EigenLayer core layer, and individual AVS layers, Unique Stake only narrows how far a slashing event can spread within that structure. It has nothing to do with an infrastructure configuration issue at the bridge layer, like the one behind the Kelp DAO incident, that is a risk sitting in an entirely different layer of the stack.
If I am already using an LRT protocol, what should I check after the Kelp incident?
The first step is confirming whether the LRT you hold depends on a Cross-Chain Bridge mechanism, if your position stays entirely on Ethereum mainnet and never bridges to another chain, an attack vector like Kelp does not apply to you. If your Token is genuinely deployed across multiple chains, it is worth understanding which cross-chain message verification architecture the protocol uses, a single verifier Node, or a multi-signature or multi-node verification setup, information that typically appears in a protocol security documentation or audit reports.
The second step is revisiting whether the APY your chosen LRT protocol advertises sits meaningfully above the current reasonable market range (roughly 4-7%). A yield well above that range usually means the protocol is plugged into more, or riskier, AVSs to generate the extra reward, not that it is simply more operationally efficient. Finally, it is worth remembering an easily overlooked detail from the Kelp incident: even if the protocol holding your funds was never hacked, you can still get caught up in the fallout, market freezes, liquidity crunches, if a lending market you use accepted the affected token as collateral. Risk assessment can not stop at the protocol you interact with directly; it has to extend to how the asset you are holding gets used elsewhere.
You stake ETH on Ethereum and earn roughly 3-4% annually, the base reward for helping secure consensus. Restaking asks a simple but consequential question: can that same locked ETH also help secure other systems at the same time, in exchange for extra yield? The answer is yes, but the price is that your capital now passes through several distinct layers of smart contracts in sequence, and a failure in any one layer can drag down protocols that have nothing to do with you. The April 2026 Kelp DAO incident was the first time this stacked-risk structure showed its cost at scale.
EigenLayer is the dominant restaking protocol by market share, with Total Value Locked approaching $20 billion across roughly 4.6 million ETH in early 2026, about 94% of the restaking market. The mechanics break down into a few roles: users delegate already-staked native ETH or a liquid Staking Token (like stETH) to an operator; the operator uses that delegated capital to help secure Actively Validated Services (AVSs), external systems like oracles, cross-chain bridges, data availability layers, or other off-chain services that need economic security guarantees; in exchange, the operator and the delegators behind them earn extra rewards paid by the AVS, typically in the AVS own token or in ETH. The business logic is straightforward: a new protocol that wants to borrow Ethereum large, economically proven validator set does not need to bootstrap its own from scratch, it effectively rents existing security instead.
The core constraint on staked ETH is slashing, a portion of stake gets permanently confiscated if a validator misbehaves. Restaking layers this mechanism on top of itself: the same ETH is now exposed simultaneously to Ethereum own consensus-layer slashing rules and to whatever slashing conditions each individual AVS your operator serves has defined. If one AVS rules are poorly designed, or an operator misbehaves on that AVS, a slashing event can theoretically ripple across multiple services at once, what the industry calls correlated slashing risk. EigenLayer 2026 Unique Stake mechanism lets stake be allocated to specific operator sets rather than exposed indiscriminately across every AVS, which reduces concentration of that risk, but it addresses how concentrated the exposure is, not how many Smart Contract layers the capital passes through.
Most users do not run validators themselves; instead they deposit ETH through a liquid restaking protocol like Kelp, Ether.fi, or Renzo and receive a tradeable receipt token in return, Kelp issues rsETH, for instance, that represents a claim on the underlying restaked position while remaining liquid enough to use elsewhere in DeFi. This convenience layer quietly stacks on a fresh set of smart contract risks: the LRT issuer own contracts, and, if the token needs to move to other chains, a Cross-Chain Bridge contracts on top of that. Trace the full path and capital can pass through Ethereum staking contracts, EigenLayer core contracts, individual AVS contracts, the LRT issuer contracts, and bridge contracts, five or six distinct smart contract systems stacked in sequence, each one an independent point of failure.
On April 18, 2026, an attacker drained roughly $292 million from Kelp DAO cross-chain bridge, 116,500 rsETH, about 18% of the token Circulating Supply, making it the largest DeFi exploit of the year. Critically, the root cause sat neither in EigenLayer core contracts nor in any flaw in Kelp staking logic. It was in the LayerZero cross-chain messaging layer Kelp used to bridge rsETH to more than 20 other chains, which relied on a 1-of-1 verifier Node configuration: an attacker only needed to convince a single verification node that a forged cross-chain message was legitimate to trigger the bridge into releasing funds. The attacker compromised infrastructure nodes and launched a denial-of-service attack against others to pull this off; Kelp emergency pause mechanism did not activate until 46 minutes after the funds were drained, though it did Block two subsequent, larger repeat attempts. The attack was later attributed to North Korea Lazarus Group. The stolen rsETH was then used as collateral to drain real assets from lending platforms like Aave, forcing Aave, SparkLend, and Fluid to freeze their rsETH markets to contain Bad Debt, and roughly $13 billion in TVL exited related protocols within two days, protocols whose own contracts were never touched, caught up purely through their connection to the affected asset.
If you are considering putting ETH into restaking for extra yield, the first thing worth understanding is exactly how many contract layers your capital passes through, not just the advertised APY. The typical 2026 restaking yield stack runs roughly 3-4% base staking plus 1-2% AVS rewards plus variable token incentives, landing total APY somewhere around 4-7%; an LRT protocol advertising yield well above that range usually means it is plugged into more, and riskier, AVSs, not that it found free money. The real lesson from Kelp is not the tired warning that bridges are dangerous. It is that the vulnerability sat entirely outside the restaking mechanism itself and outside the smart contract logic auditors typically scrutinize most closely, in an easily overlooked piece of infrastructure configuration, a single-verifier bridge setup. That means reviewing a protocol smart contract code alone is not enough; every layer you are delegating through, operator, AVS, LRT issuer, bridge, deserves its own individual risk assessment, rather than trusting the entire stack as one indivisible black box.