Restaking

Restaking is the practice of reusing already-staked ETH (or its liquid staking tokens) to secure additional protocols beyond Ethereum itself, earning extra yield in exchange for taking on extra slashing risk. EigenLayer pioneered it; the model now spans multiple competing security layers.

What restaking actually is

Restaking lets you take ETH that's already staked — securing Ethereum's consensus and earning the base staking yield — and pledge that same capital a second time to secure other systems. Those systems are usually middleware that can't bootstrap their own validator set: oracles, data-availability layers, bridges, rollup sequencers, and increasingly AI/compute verification networks. The pitch is capital efficiency. One pool of stake rents out its economic security to many protocols at once, and stakers collect a stack of yields instead of one.

The term was popularized by EigenLayer, built by Sreeram Kannan's team. In EigenLayer's model, the protocols buying security are called AVSs — Actively Validated Services (the platform now markets the broader stack as EigenCloud). You deposit ETH or a liquid staking token, delegate to an operator, and that operator opts into running software for whichever AVSs they choose.

How it works under the hood

There are two flavors. Native restaking points your Ethereum validator's withdrawal credentials at the restaking contract, so misbehavior on an AVS can be penalized against your actual stake. Liquid restaking is the version most people touch: you deposit a liquid staking token (like stETH) into a protocol such as EtherFi, Renzo, or Kelp, and receive a liquid restaking token (LRT) — weETH, ezETH, rsETH — that stays tradeable and composable across DeFi while the underlying stake does the securing.

The critical mechanism is slashing. The whole value proposition rests on the threat that an operator who corrupts an AVS loses real money. On EigenLayer, slashing only went live in April 2025 — meaning for the protocol's entire explosive growth phase, the security being sold was, in my view, more theoretical than enforced. That matters when you're reasoning about what the yield is actually paying for.

Why it matters

Restaking is one of the more genuinely novel primitives DeFi has produced. Spinning up a new proof-of-stake network historically meant cold-starting a token and convincing validators to hold it — expensive and slow. Restaking turns Ethereum's roughly $100B+ of staked capital into a security marketplace any protocol can rent from. For builders, that collapses a brutal bootstrapping problem. The numbers followed: EigenLayer's TVL peaked near $28B during the pre-slashing points-farming frenzy, then fell hard toward $7B as farmers rotated out and risk got repriced, before recovering into the $15–19B range through early 2026 (Fensory). LRTs alone account for several billion, led by EtherFi's weETH.

Key risks and tradeoffs

The risks compound, literally. Restaking layers AVS slashing risk on top of base staking risk. LRTs add a third layer: wrapper-contract risk and depeg risk — Renzo's ezETH briefly depegged in April 2024 when a redemption mismatch hit thin liquidity, triggering cascading liquidations. The systemic worry the research community keeps flagging is correlated slashing: if many operators secure the same overlapping set of AVSs, a single bad event could slash a large fraction of restaked capital at once, with knock-on liquidations through every protocol holding the affected LRT as collateral. The data suggests this is manageable with conservative curation, but it's untested at scale under real stress.

There's also a quieter governance risk. Slashing decisions, dispute resolution, and which failures count as slashable are still maturing. You're trusting not just code but an evolving set of human and protocol judgments about when your money gets burned.

Current state (2026)

Restaking is no longer one protocol. EigenLayer still dominates, but Symbiotic (permissionless, configurable collateral and slashing) and Karak (multi-asset/"universal" restaking) have turned it into a category with real architectural disagreement about how flexible shared security should be. The frontier is moving toward vertical AVS specialization, multi-chain slashing, and extending the model to Bitcoin and AI compute verification.

What I'll be watching is the first genuinely large slashing event. Restaking's entire risk model is a promise that hasn't been collected on at scale — and how that day is handled will tell us whether shared security is infrastructure or just leverage with extra steps.

This is technical explanation, not financial advice. Restaking yields carry layered, sometimes correlated, loss risk; assess any protocol's audits and slashing terms yourself.

Frequently asked

What's the difference between staking and restaking?

Staking secures Ethereum's base layer and earns one yield. Restaking reuses that same staked capital to also secure additional protocols (AVSs), earning extra rewards in exchange for additional slashing exposure. You're not staking new money — you're putting existing stake to work a second time.

What is a liquid restaking token (LRT)?

An LRT is a tradeable receipt token — like EtherFi's weETH or Renzo's ezETH — that you get for depositing into a liquid restaking protocol. It represents your restaked position but stays liquid and usable across DeFi, so you keep composability while the underlying stake secures AVSs. The convenience adds wrapper and depeg risk.

Is restaking safe?

It's higher-risk than plain staking. You stack AVS slashing risk and (for LRTs) depeg and contract risk on top of normal staking risk, and correlated slashing across overlapping operators is a real systemic concern. Slashing on EigenLayer only went live in April 2025, so the model is still relatively unproven under stress.

Sources