What Is Restaking? How EigenLayer, LRTs, and Shared Security Actually Work in 2026

Restaking is the fastest-growing primitive in DeFi since liquid staking. It went from zero to nearly $20 billion in total value locked in under three years, and it changed how Ethereum security actually works. If you have ETH staked, restaking is the difference between earning one yield and earning three. If you build on Ethereum, it is the difference between bootstrapping your own validator set and renting trust on day one.

This is what restaking actually is, how the protocols work, what the yields look like in 2026, and where the money can disappear.

Primary data sources used in this article: EigenLayer documentation, DefiLlama restaking dashboard, and Dune Analytics.

The Problem Restaking Solves

Start with regular staking. When you stake 32 ETH on Ethereum, your capital does one job. It validates the Ethereum network and earns roughly 3% to 4% a year. That is it. The ETH is locked, securing one chain, generating one yield.

The problem this creates for new protocols is harder. Every oracle network, every bridge, every data availability layer, every cross-chain messaging system needs its own validator set with its own economic security. Bootstrapping that set from scratch is slow and expensive. A new oracle protocol launching today has to convince validators to lock up tokens, build slashing infrastructure, run nodes, and trust an untested system. The result is years of fragility before the network is actually secure.

Restaking flips the model. Your staked ETH stays where it is, securing Ethereum, but it also gets opted in to secure additional services. Those services pay you an extra yield for the additional security you are providing. The new protocol gets instant access to billions of dollars of economic backing. You get more yield on the same capital. Ethereum security becomes a shared resource that any application can rent.

That is the entire thesis in one paragraph. Everything else is execution.

How Restaking Actually Works

There are three ways to participate. Each has different requirements, different yields, and different risk profiles.

Native restaking is for users who run their own validator. You point your existing 32 ETH validator at a restaking protocol, download the operator software, and opt in to secure additional services. This gives you the most control and the highest yield per ETH, but it requires technical setup and active management. You also take on the full slashing risk directly.

LST restaking is for users who have liquid staking tokens like Lido’s stETH or Rocket Pool’s rETH. You deposit those tokens into a restaking protocol and they get put to work securing additional networks. No node operation required. You keep the LST tokens, they keep earning the base staking yield, and they earn the additional restaking yield on top.

Liquid Restaking Tokens (LRTs) are the easiest path. You deposit ETH or an LST into a protocol like Ether.fi or Renzo. They handle all the restaking complexity behind the scenes and issue you a new token, eETH or ezETH, that represents your restaked position. The LRT keeps accruing all three yield layers, and you can use it elsewhere in DeFi as collateral or in liquidity pools. This is the version most retail users actually interact with.

The services that get secured by all this restaked capital are called Actively Validated Services, or AVS. These can be oracles, bridges, data availability layers, sequencers for Layer 2 networks, decentralized AI inference, or any off-chain system that needs economic guarantees. Each AVS sets its own slashing conditions and pays its own rewards in its own token or in ETH.

The Three Main Protocols

The restaking market is concentrated. Three protocols hold essentially all the TVL, and the gap between them is enormous.

EigenLayer is the pioneer and the dominant protocol. As of early 2026, it holds approximately $19 billion in total value locked across 4.6 million ETH, with roughly 1,900 active operators. That works out to about 94% market share. The protocol was founded by Sreeram Kannan and Eigen Labs and went live on Ethereum mainnet in 2023. In 2026 it has expanded its scope beyond simple restaking into what the team calls a “verifiable cloud,” with services like EigenDA (data availability), EigenCompute (verifiable off-chain compute), and EigenVerify (dispute resolution). The native token is EIGEN. The ELIP-12 governance proposal introduced in Q1 2026 channels 20% of subsidized AVS rewards and 100% of EigenCloud fees into EIGEN buybacks.

Symbiotic is the modular challenger. It holds approximately $897 million in TVL across roughly 256,000 ETH, for a 5.5% market share. It launched on mainnet in January 2025, hit $200 million in deposits within 24 hours, and was the first restaking protocol to ship with full slashing functionality from day one. The Symbiotic architecture is permissionless. Any ERC-20 can be used as restakeable collateral, not just ETH and LSTs. Networks can customize their own vaults, operators, and slashing conditions. It is backed by Cyber Fund, founded by Lido co-founders Konstantin Lomashuk and Vasiliy Shapovalov, and by Paradigm.

Karak is the multi-asset protocol. It manages roughly $102 million in TVL across 29,000 ETH, about 0.6% of the market. Karak’s distinguishing feature is asset diversity. Beyond ETH and LSTs, it supports LP tokens, stablecoins, and wrapped Bitcoin (wBTC) as restaking collateral. Karak also runs its own Layer 2 called K2 as a sandbox for testing risk management and what it calls Distributed Security Services (DSS), Karak’s equivalent of AVS.

The strategic difference between the three matters. EigenLayer is curated and institutional. Symbiotic is permissionless and modular. Karak is multi-asset. Most new AVS still launch on EigenLayer because the security pool is twenty times deeper. Symbiotic and Karak compete for DeFi-native applications that want more flexibility or non-ETH collateral.

Outside Ethereum, the same primitive exists on Solana through Jito, which has built restaking on top of its dominant liquid staking position. Jito holds over 45% of Solana’s liquid staking market and offers a “triple yield” stack: base SOL staking, MEV tips from its Block Engine, and restaking rewards from Node Consensus Networks (NCNs).

Liquid Restaking Tokens: The Retail Layer

LRTs are how most users actually touch restaking. The protocols sit on top of EigenLayer and handle the complexity. You deposit, you get a token, the token earns yield in the background.

The market is more spread out than the underlying restaking layer. According to DefiLlama, the major LRT protocols in 2026:

Ether.fi is the largest. It issues eETH and holds the deepest TVL of any LRT protocol, in the multi-billion dollar range. Ether.fi’s distinguishing feature is non-custodial key management. Users keep control of their withdrawal keys even while their ETH is being restaked, which is unusual for the category. eETH has the deepest DeFi integration and the highest liquidity of any LRT.

Renzo issues ezETH and is the most aggressive on multi-chain expansion. The protocol has pushed ezETH deployment to Arbitrum, Blast, Linea, Mode, and other Layer 2 networks earlier and harder than competitors. The total ezETH supply is spread across multiple chains, which makes it the LRT of choice for users who want to use restaked exposure on L2s.

Kelp DAO issues rsETH and was the protocol that broke the points farming meta. Kelp Miles stacked on top of EigenLayer points pulled significant TVL during the early restaking boom. Kelp also expanded to Layer 2 access earlier than most. Notably, Kelp suffered a $300 million exploit in April 2026 that triggered roughly $5.4 billion in withdrawals across the restaking sector. The protocol survived but the incident reset the risk conversation around the entire category.

Puffer Finance issues pufETH and focuses on a different angle: anti-slashing technology and reducing the operator concentration risk that comes with most LRT protocols.

The basic LRT yield stack in 2026 looks like this: 3-4% from base Ethereum staking, 1-2% from EigenLayer AVS rewards, and a variable amount of points or token rewards on top. Total APY typically lands in the 4% to 7% range when paid in real yield, higher when speculative token incentives are included.

Recursive Restaking: Where the Higher Yields Come From

The 10% to 20% APY figures you see advertised come from a specific strategy called recursive restaking, or the yield loop.

The mechanics: deposit ETH into Ether.fi and receive eETH. Deposit the eETH as collateral on a lending protocol like Aave or Morpho. Borrow regular ETH against it at a lower rate than the eETH yield. Use the borrowed ETH to buy more eETH. Repeat.

Each loop multiplies your effective exposure to the restaking yield. A typical three-loop setup can take a 5% base yield to 12-15% effective APY. With aggressive loops, the number goes higher.

The catch is liquidation risk. Recursive positions are highly leveraged. If eETH temporarily depegs from ETH, which has happened, the entire position can be liquidated in minutes. This is not a passive yield strategy. It is a managed leveraged trade dressed up as staking.

The Risks That Actually Matter

Restaking stacks three categories of risk that do not stack in regular staking.

Slashing risk. Your ETH can be penalized on the base Ethereum layer for normal validator misbehavior. It can also be slashed on any AVS your operator opts in to. If you delegate to an operator that runs ten AVS and one of them has a bug or your operator misbehaves on it, you can lose principal. Each AVS sets its own slashing conditions, which means your downside scales with the number of services your operator secures.

Smart contract risk. A restaking position can touch six or seven smart contract systems in sequence. Ethereum staking contracts, EigenLayer core contracts, individual AVS contracts, an LRT issuer’s contracts, the LRT itself, a lending protocol if you are looping, an oracle for liquidation pricing. Each layer is a potential failure point. The Kelp DAO hack in April 2026 was a real-world demonstration of this. A bug in one part of the stack drained $300 million and triggered withdrawals across protocols that had no direct exposure to the exploited code.

Liquidity and exit risk. Unstaking from EigenLayer is not instant. Depending on which path you took in (native, LST, or LRT), exits can take days to weeks. During market stress, LRTs can temporarily depeg from underlying ETH because the secondary market gets thin and the primary redemption queue is long. If you are using LRTs as collateral and the peg slips, liquidations follow.

Centralization risk. EigenLayer holds 94% of the market. A meaningful exploit at EigenLayer would not just be a protocol failure. It would be a systemic event for everything secured by AVS, including data availability layers used by major rollups.

Bottom Line

Restaking is real infrastructure now, not a meme. It changed how new protocols launch and how staked capital earns. EigenLayer dominates the category and EigenLayer’s expansion into verifiable compute is the actual second act of the protocol, beyond just yield. LRTs are how most users participate, and Ether.fi, Renzo, Kelp DAO, and Puffer are the main names worth knowing. Base yields sit at 4-6%. Looped yields can hit 15-20% with real liquidation risk attached. The Kelp DAO exploit established that the smart contract risk is not theoretical, and concentration at the EigenLayer layer is the systemic question to watch through the rest of 2026.

This article is for informational purposes only and does not constitute financial advice.