What Is Restaking? EigenLayer, AVS, and the $15 Billion Reuse of Ethereum's Security

Ethereum's proof-of-stake works because validators put 32 ETH at stake — collateral that gets slashed if they misbehave. That same 32 ETH does exactly one job: securing Ethereum block production.

Restaking is the idea that the same collateral can do more than one job — securing Ethereum and a data availability layer, and an oracle, and a cross-chain messaging protocol — without anyone ever moving the underlying ETH. EigenLayer was the first protocol to ship this primitive at scale. As of early 2026 it commands roughly $15 billion in restaked TVL and ~94% market share of the restaking category, with 40 active services running on mainnet, 190 in development, and 2,000+ registered operators.

Slashing — the missing piece that made restaking real instead of theoretical — went live on EigenLayer mainnet on April 17, 2025. That single deployment transformed restaking from "promise" to "feature-complete."

This guide explains how restaking works, what an AVS actually is, why liquid restaking tokens exist, and where the model can break.

Why Restaking Exists at All

To understand restaking, first understand the problem it solves: the cost of bootstrapping a new decentralized network's security.

Suppose you want to launch a new oracle network, or a cross-chain bridge, or a decentralized sequencer for a Layer 2. You need a set of independent operators, you need them to put real economic stake at risk, and you need that stake to be large enough that attacking your network costs more than the attack profits.

Building that from scratch means launching your own token, distributing it widely enough to be credible, and convincing validators to stake meaningful capital in your token before your network has any users. This is the cold-start problem and it is why most novel infrastructure protocols fail before they ship.

Restaking solves this by letting new networks rent Ethereum's already-staked capital instead of bootstrapping their own. Validators who have already locked 32 ETH (or held a liquid staking token like stETH) can opt in to back additional services. Those services pay rewards directly. The validator earns Ethereum staking yield plus AVS rewards on the same collateral. The new service gets billions of dollars of cryptoeconomic security at launch instead of zero.

The mechanism is shared security. The infrastructure is EigenLayer. The services renting that security are called AVSs.

AVS — The "Active Validated Service" Layer

An AVS (Actively Validated Service) is any decentralized network that needs cryptoeconomic security and chooses to source it from EigenLayer rather than running its own validator set.

The AVS pays restaked operators in fees or its own token. Operators run the AVS's required software (a node, a verifier, an oracle data fetcher, etc.) and commit their restaked ETH as collateral. If they break the AVS's rules — go offline, sign conflicting messages, fail an attestation — the AVS can submit on-chain proof, and the operator's stake gets slashed.

As of the slashing launch in April 2025, two flagship AVSs anchored the live ecosystem:

• LayerZero uses restaked operators as Decentralized Verifier Networks (DVNs) for cross-chain message verification. LayerZero CEO Bryan Pellegrino noted that slashing brings "stronger enforcement guarantees to the AVS ecosystem" and enables "cryptoeconomic verification of messages."
• Infura's Decentralized Infrastructure Network (DIN) uses restaked operators to serve RPC requests with verifiable Service Level Agreements — a long-standing weak point in the Ethereum stack since most apps still depend on a handful of centralized RPC providers.

Beyond those two flagships, the active mainnet AVS roster includes:

• EigenDA — EigenLayer's own data availability layer, the single largest consumer of restaked security and direct competitor to Celestia.
• EigenCompute — verifiable off-chain execution, mainnet alpha launched January 2026.
• Hyperlane, Omni, Espresso, Witness Chain, AltLayer — interop, sequencers, watchtowers, and rollup services.

The roadmap of 190 AVSs in development covers oracles, MEV protection, watchtowers for optimistic rollups, decentralized AI inference verification, and zk-proof generation marketplaces.

How Slashing Actually Works (And Why It's the Whole Point)

Without slashing, restaking is theater. An operator can promise to follow an AVS's rules, but if the protocol can't actually burn their stake when they break those rules, the "security" is just a polite request.

Slashing on EigenLayer works through two primitives that the protocol introduced specifically to handle restaking's cross-AVS complexity:

Unique Stake Allocation — A mechanism that lets operators earmark portions of their delegated stake for specific AVSs, ensuring only one AVS can slash any given unit of stake. Without this, an operator securing five AVSs would face compounding risk: a fault on AVS A could trigger slashing that also wipes out collateral committed to AVSs B through E. Unique Stake Allocation breaks that compounding by partitioning the stake. Operators can still serve many AVSs — they just can't double-pledge the same unit of collateral as if it were unencumbered.

Operator Sets — A protocol primitive that lets each AVS organize operators into distinct groups with different responsibilities and different slashing conditions. Think of an AVS as a hospital and Operator Sets as specialized teams — surgeons with strict slashing for botched operations, intake nurses with much milder penalties for paperwork errors. Different work, different consequences, all governed under one protocol contract.

The slashing logic itself runs on Ethereum. AVSs deploy a contract that defines slashable conditions in code; when an operator violates those conditions, anyone (typically a watchtower run by another operator) can submit a proof that triggers the burn. This is the "verifiable cloud" thesis — services that can mathematically prove their own correctness or get penalized on-chain when they can't.

Liquid Restaking Tokens (LRTs) — The Layer Most Users Actually Touch

Most users will never operate a node or directly interact with EigenLayer contracts. They'll get restaking exposure through Liquid Restaking Tokens (LRTs) — tradable ERC-20s that represent a deposit into a restaking strategy run by a third-party protocol.

An LRT protocol takes user ETH (or stETH from Lido), deposits it into EigenLayer, allocates it across selected AVSs, and issues a liquid token in return. Users can hold the LRT for yield, sell it at any time on a DEX, or use it as collateral elsewhere in DeFi.

The major LRT protocols by TVL in 2026:

Source: protocol disclosures and DefiLlama, early 2026.

The LRT category exploded in 2024 because it took restaking — which was operationally heavy — and packaged it into a single token a normal DeFi user could hold. It also introduced an additional layer of risk that the operator-run model doesn't have: the LRT issuer is a smart-contract counterparty, and the user has no direct claim on the underlying restaked ETH except through the LRT contract.

That extra layer matters. In April 2026, Kelp DAO lost $292 million when an attacker exploited a LayerZero bridge integration — not a flaw in EigenLayer itself, but a flaw in the layer above it. The lesson generalizes: each layer of abstraction in the restaking stack adds yield, and adds attack surface.

The Yield Stack — Where Returns Actually Come From

A restaked ETH position can earn from up to four distinct sources, layered:

1. Native Ethereum staking yield — currently ~3% APR, paid in ETH from issuance plus tx fees minus MEV burn.
2. AVS rewards — paid by each AVS the operator opts into, typically a mix of the AVS's native token plus some ETH/stables. Yields here are highly variable; early-stage AVSs often pay 5-10% in token incentives, mature ones pay closer to fee revenue (1-3%).
3. EIGEN emissions — EigenLayer's governance token, distributed to participants based on activity. The ELIP-12 governance proposal launching in Q1 2026 establishes an Incentives Committee to direct EIGEN emissions specifically toward fee-generating AVSs (vs. spreading them broadly).
4. LRT-level rewards — LRT protocols often layer additional points or token incentives on top to attract deposits.

The headline numbers ("up to 15% APR on restaking!") almost always assume all four layers stacked, all layers paying in volatile tokens at peak prices. A more realistic conservative restaker — native staking + 2-3 high-quality AVSs, no points farming, no LRT layer — is looking at something like 5-7% blended APR in ETH and stablecoin terms, with the AVS portion subject to genuine slashing risk.

The Risks Nobody Quite Wants to Price

Restaking introduces risks that don't exist in plain ETH staking. The honest ones are:

1. Stacked slashing exposure. Even with Unique Stake Allocation segmenting collateral, an operator running multiple AVSs faces correlated risk: a single buggy node software release, a single ops mistake, a single cloud-provider outage can trigger faults across multiple AVSs simultaneously. Slashing that ought to be 0.1% per fault can compound into a much larger loss when correlated faults hit at once.

2. AVS quality is uneven. Not every AVS is LayerZero or EigenDA. Some are speculative early-stage infra with poorly defined slashing conditions, ambiguous fault attribution, or governance that can change slashing parameters after operators have committed stake. Picking AVSs is now an active risk decision, not a passive one.

3. Centralization at the top of the stack. EigenLayer has ~94% market share of restaking. The top three LRTs hold the majority of liquid restaking TVL. The top handful of operators secure most of the active AVSs. A protocol-level bug or governance failure at any of these chokepoints is a systemic Ethereum-adjacent risk, not just a local one.

4. Liquidity-driven peg risk on LRTs. LRTs trade on DEXes against ETH. If a large holder needs to exit faster than the EigenLayer withdrawal queue allows (currently a multi-day cooldown), they sell on DEX into a thin LRT/ETH pool and the LRT depegs. The deeper the unwind, the more depeg, and other holders of the same LRT see paper losses they didn't choose. This has happened in smaller magnitudes already.

5. Smart-contract risk at the LRT layer. Documented in the Kelp DAO LayerZero bridge incident — your restaking yield is only as safe as the weakest contract between your wallet and the underlying restaked ETH.

Restaking's Place in the Ethereum Stack, in 2026

Restaking has settled into a clear role in the Ethereum stack: it is the shared-security layer that lets infrastructure projects ship without needing to bootstrap their own validator economy. EigenLayer is the dominant primitive; LRTs are the consumer wrapper; AVSs are the demand side; slashing is the enforcement mechanism that finally ties the loop closed.

The category will not stay still. Symbiotic, Karak, and other restaking protocols are competing on permissionless AVS launches and broader collateral support (BTC, stables, project tokens beyond ETH/LST). The LRT layer will likely consolidate as the highest-yield-points-farming era ends and users start asking which protocols have the strongest risk frameworks rather than the highest token-weighted APR. The AVS roadmap will get pruned — the 190-in-development number will shrink as projects discover that running a real node operator marketplace is harder than launching a token.

Three checkpoints to watch through the rest of 2026:

1. EigenCompute mainnet graduation — moves the protocol from "secure other people's services" to "host other people's compute," directly competing with AWS/GCP for verifiable workloads.
2. First public slashing event of size. Small slashings have happened. A multi-million-dollar slashing on a major AVS will be the first real stress test of how operators, LRTs, and DEX liquidity all respond simultaneously.
3. ELIP-12 implementation — concentrating EIGEN emissions on fee-generating AVSs separates real economic demand from incentive farming. Watch which AVSs survive the cut.

The shorthand: restaking turned Ethereum's $80B+ staked ETH into rentable security collateral. That's a genuinely new primitive, and it's still early enough that the structure of the market is being decided in real time.

Sources

• Slashing Goes Live on Mainnet — EigenCloud Blog (April 17, 2025)
• EigenLayer AVS Mainnet Launch — EigenCloud Blog
• Intro to Slashing on EigenLayer: AVS Edition — EigenCloud Blog
• EigenLayer Mainnet Slashing — Blockdaemon institutional staker guide
• EigenLayer Tokenomics — Tokenomics.com
• EigenLayer's biggest risk may be centralization — Blockworks
• Sreeram Kannan profile — X
• BlockAI News internal: Kelp DAO Loses $292M in 2026's Biggest DeFi Hack

Stay close to BlockAI News.

• 📧 Subscribe → daily Web3 × AI brief
• 💬 Join the community on Telegram
• 𝕏 Follow @BlockAI_News on X

The first nine-figure slashing event won't be telegraphed in advance. We'll have the operator post-mortem, the LRT depeg chart, and the AVS-by-AVS exposure breakdown the day it lands.