Restaking Explained: Renting Ethereum’s Trust in 2026

Restaking is the idea that the money securing Ethereum can work a second job. You stake ETH to help run the network and earn a modest reward; then you pledge that same staked position again to vouch for other services, oracles, bridges, data-availability layers, co-processors, and you collect an extra fee for the favor. The pitch is seductive: instead of every new protocol bootstrapping its own token and its own validator set, it can simply rent trust from the tens of billions of dollars already staked on Ethereum.

Coined by EigenLayer in 2023, restaking went from whiteboard concept to one of the largest capital pools in decentralized finance in under two years, then spent most of 2026 unwinding. This guide walks through what restaking is, how the stack fits together from a single staked coin up to a liquid restaking token, who the players are (EigenLayer, Symbiotic, Karak, and Bitcoin-native Babylon), what can go wrong, and why a product that once held more than $22 billion now struggles to earn six figures in a week. It also takes on the question most explainers dodge: in late 2026, is restaking actually worth it to you?

What Restaking Actually Means

Start with plain staking. To validate Ethereum you lock 32 ETH (or a fraction of that through a pool), run software that proposes and attests to blocks, and in return you earn freshly issued ETH plus a share of transaction fees and MEV. The network enforces honesty with slashing: break the rules and the protocol destroys part of your stake. That bonded capital is the whole point. It is what makes an attack expensive.

Restaking reuses that bond. Through a restaking protocol, you agree that your staked ETH (or a token representing it) can also be slashed if you misbehave while helping to secure some other service. In EigenLayer’s vocabulary those services are Actively Validated Services, or AVS. Each AVS writes its own rules and its own penalties, and each pays operators for the coverage. You are, in effect, renting out Ethereum’s trust and getting paid rent, while accepting new ways to lose your principal.

The appeal to builders is obvious. Standing up a credible validator set is slow and expensive, and a brand-new token makes a poor security budget because it can collapse the moment it is attacked. Pooled, or shared, security lets a project borrow Ethereum’s hard-won credibility on day one. The appeal to stakers is extra yield. The catch, which took the market most of two years to internalize, is that extra yield and extra slashing are the same coin seen from two sides.

Not everyone was enthusiastic. Ethereum co-founder Vitalik Buterin warned against the concept early. In his May 2023 essay Don’t overload Ethereum’s consensus, he cautioned that “any expansion of the ‘duties’ of Ethereum’s consensus increases the costs, complexities and risks of running a validator,” and worried that stacking extra obligations onto the base layer could drag the whole chain into bailing out a service that grew too large to fail.

The Staking Stack: From ETH to a Liquid Restaking Token

Restaking sits at the top of a ladder, and each rung adds a token, a yield source, and a risk. Understanding the ladder is the fastest way to understand what you actually hold.

The first rung is native stake: ETH bonded to a validator. The second is the liquid staking token, or LST, such as Lido’s stETH or Rocket Pool’s rETH, which hands you a tradable receipt for staked ETH so your capital is not frozen. The third rung is the restaked position, where that staked ETH (or the LST) is committed to an AVS. The fourth is the liquid restaking token, or LRT, such as ether.fi’s eETH or Renzo’s ezETH, a receipt for the restaked deposit that you can then lend, borrow against, or loop inside DeFi.

By the top of the ladder, a single unit of ETH can be backing a validator, a liquid staking receipt, a restaking commitment, a liquid restaking receipt, and a leveraged DeFi position all at once. That is rehypothecation, the same asset pledged behind several promises, and it is the source of both restaking’s capital efficiency and its correlated-failure risk. When one layer breaks, the layers above it can unwind together.

How EigenLayer Works: Operators, AVS, and Rented Security

EigenLayer is the protocol that coined restaking and still defines the category. The mechanics have three roles. Stakers deposit ETH or an LST and delegate it to an operator. Operators run the actual software for the services they choose to secure. AVS are those services; each one registers its slashing conditions and its reward schedule. A staker’s capital can be slashed only for the AVS its operator has opted into, which is meant to keep risk a matter of choice rather than a blanket exposure.

Slashing went live on EigenLayer mainnet on an opt-in basis in April 2025, the moment restaking stopped being a pure points game and started carrying real penalties. The EIGEN token adds a second mechanism for faults that cannot be proven on-chain, so-called intersubjective or unattributable faults, where token holders can socially agree that a service misbehaved and slash accordingly.

Getting out is its own topic. Native restakers hold their ETH in a contract called an EigenPod, and unwinding a position means queuing a withdrawal, waiting out a delay designed to give AVS time to settle any pending penalties, and only then reclaiming the stake. A draft governance proposal known as ELIP-018, nicknamed RETIRE, was floated in 2026 specifically to let pod owners switch off restaking and exit cleanly, a tacit admission that the original design made leaving harder than joining. For anyone weighing an LRT, the redemption path matters as much as the yield.

In 2025 the project rebranded from EigenLayer to EigenCloud, signaling a push beyond pure restaking toward verifiable cloud services, a pivot covered later in this guide. The core restaking machine still runs: thousands of operators and a security pool measured in the billions. But the roster of live services it secures is, tellingly, short. Fewer than two dozen AVS are actually in production, and only a handful pay fees worth mentioning.

What an AVS Buys, and Why So Few Are Buying

An AVS is buying one thing: the credible threat that if its operators cheat, they lose money. That threat lets a service inherit Ethereum-grade economic security without issuing its own staking token. The canonical customers are infrastructure, not consumer apps.

• EigenDA, EigenLayer’s own data-availability layer and its first live AVS.
• Lagrange and Brevis, zero-knowledge coprocessors that prove facts about chain state.
• eOracle, a restaking-secured oracle network.
• Omni and AltLayer’s MACH services, which sell fast cross-rollup messaging and finality.
• Witness Chain, which verifies physical infrastructure and proof-of-location claims.

Here is the structural problem that defined 2026. Restaking is a two-sided market, and only one side showed up. The supply of capital willing to be rented ballooned into the billions, pulled in by points and token incentives. The demand side, services willing to pay real fees for security, barely materialized. Most of the yield restakers earned was not fees from paying customers; it was emissions, new tokens printed to subsidize deposits. When the incentives thinned, there was little underneath. A marketplace with enormous supply and almost no paying demand does not clear at a high price. It clears near zero.

Why did so much capital pile in anyway? Points. Before EIGEN existed, EigenLayer awarded restakers points toward a future airdrop, and LRT issuers layered their own points on top, so depositors chased a reward that was really a bet on two unlaunched tokens at once. That reflexive loop, deposit to farm points, points imply a token, token price pulls in more deposits, inflated the numbers on the way up and unwound just as fast on the way down. Once EIGEN listed and then slid, the engine ran in reverse, which is a large part of why the supply side evaporated in 2026.

Liquid Restaking Tokens and the Leverage They Hide

If EigenLayer is the engine, liquid restaking tokens were the turbocharger. An LRT issuer, ether.fi (eETH and weETH), Renzo (ezETH), Kelp (rsETH), or Puffer (pufETH), takes your deposit, restakes it across AVS, and hands you a liquid token you can use elsewhere. ether.fi grew into the largest issuer by far, at times holding the majority of the liquid restaking segment.

LRTs are where rehypothecation gets dense. The same ETH underpins the validator, the staking receipt, the restaking commitment, and the LRT, and holders then deposit the LRT as collateral to borrow more, sometimes looping the trade several times. In a calm market this is simply leverage. In a stressed one it is a chain of dominoes.

The clearest warning came in April 2026, when an attacker exploited Kelp DAO’s cross-chain bridge to mint roughly 116,500 rsETH out of thin air, a sum worth around $292 million, then posted it as collateral on Aave to borrow real assets, leaving tens of millions in bad debt before a coordinated rescue restored the backing, as CoinDesk reported. The lesson was not that restaking math failed; it was that a liquid restaking token is only as safe as the weakest bridge in its plumbing, and lending markets had priced none of that. It is the same pattern that recurs across the industry, where cross-chain infrastructure, not the core protocol, is the honeypot, a dynamic explored in our look at what 830 bridge bugs reveal about cross-chain security.

The Competition: Symbiotic, Karak, and the No-Token Model

EigenLayer was never the only answer. Symbiotic, backed by a venture round led by Pantera Capital in 2025, took a more permissive design: any ERC-20 can serve as collateral, slashing is active from day one, and disputed faults are settled by a configurable resolver. Crucially, Symbiotic launched without a public token, and in mid-2026 it steered its Core V2 toward composable collateral markets. Co-founder Misha Putiatin framed the shift in a July 2026 release: “Collateral markets are not a new financial idea. What is new is making them composable, enforceable, and capital-efficient onchain.”

Karak, built by Andalusia Labs, was the third major venue; it has since drifted toward a broader execution-layer identity. The pattern worth noting is the irony at the center of the sector: the protocols without a tradable token held up better through the downturn than the token-rich leader. EigenLayer’s EIGEN trades more than 96% below its all-time high, while the no-token venues kept quietly securing whatever demand existed. The token, it turned out, was the speculative wrapper, not the product.

Babylon and Bitcoin Restaking

Restaking is not an Ethereum-only story. Babylon brings the same idea to Bitcoin, and it does so without wrapping or bridging the coins. Using Bitcoin’s own timelock scripting, a holder can lock BTC in a self-custodied script and pledge it to secure proof-of-stake chains that opt in, called Bitcoin Supercharged Networks. Misbehave and the locked BTC can be slashed; behave and you earn a reward for lending Bitcoin’s enormous security budget to younger chains.

Babylon’s genesis mainnet and its BABY token both launched in April 2025. By late October 2026 the protocol held roughly $3.4 billion in value, according to DefiLlama, making it the second-largest restaking protocol by that measure and the dominant force in Bitcoin DeFi. Because the figure is denominated in BTC, it has held up better than Ethereum restaking’s dollar TVL, even as the BABY token sank more than 91% from its peak. A planned integration that would let native Bitcoin back loans on Aave V4 remains on testnet, not mainnet, a reminder that the ambitious plumbing is still being built.

Babylon is the useful counter-example. Its TVL held not because Bitcoin restaking solved the demand problem, but because locking idle BTC for any yield is attractive to a holder base that otherwise earns nothing, and because the market it serves is distinct from Ethereum’s crowded AVS race.

SSV and the Based-Applications Alternative

A different response to the shared-security question comes from SSV Network, the largest provider of distributed validator technology on Ethereum, which splits a validator key across several non-trusting operators so no single machine holds it. In early 2025 SSV unveiled SSV 2.0 and its concept of based applications, or bApps, which founder Alon Muroch said would “profoundly change the restaking market.”

The design flips restaking’s risk model. Validators opt in with a separate participation key, and their 32 ETH principal is never slashable; only optional delegated capital can be penalized, and each bApp sets its own risk tolerance. Where EigenLayer-style restaking is additive, piling new slashing conditions onto the same bond, SSV’s model is deliberately subtractive about what it puts at risk. It is less a version of restaking than an alternative answer to the same problem: how to let a new service borrow credible security without minting a fragile token of its own.

Slashing: The Risk You Are Actually Paid to Take

Yield in restaking is compensation for slashing risk, so it pays to know exactly what can be slashed. There are three layers. First, Ethereum’s base-layer penalties for double-signing or prolonged downtime, which exist with or without restaking. Second, AVS-defined slashing, where a service penalizes operators for failing its specific duties. Third, EigenLayer’s intersubjective slashing for faults that cannot be proven automatically, which leans on token holders to adjudicate.

Layered on top is smart-contract risk. Restaking adds new, complex contracts and brand-new AVS code, all of it holding or governing large sums, all of it a fresh target. This is why audits and formal verification matter more here than almost anywhere else in DeFi; the firms that build the tooling for this work, profiled in our piece on Trail of Bits and crypto’s audit toolchain, are effectively the first line of defense for a restaker’s principal.

The subtler danger is correlation. A single popular operator may secure many AVS at once, and a single widely used AVS may be trusted by many operators. A bug or a coordinated fault in the wrong place could slash a large slice of the pool simultaneously, which is the systemic concern Buterin flagged in 2023. The irony of 2026 is that catastrophic slashing has been rare; the risk that was so carefully priced never really fired. What fired instead was the demand side failing to show up, which hollowed out the reward long before the risk ever did.

It is fair to ask whether any of this risk is theoretical. Base-layer slashing is real and happens routinely to misconfigured validators, usually in small amounts. AVS-level slashing, by contrast, has been used sparingly since it went live, partly because so few services impose demanding conditions and partly because operators are cautious. That is cold comfort, though. The absence of a large slashing event so far says more about thin usage than about the design being safe; a busier, higher-stakes AVS economy would test the penalties far harder than anything seen to date.

The Numbers Behind the 2026 Reckoning

The single most damning statistic of the year is a fee comparison. According to reporting by Blockhead drawing on DefiLlama data, the entire restaking category secured around $10 billion in assets yet generated roughly $99,977 in fees in a single week, while plain liquid staking, with about $51.87 billion locked, produced some $27.35 million in the same period. Measured per dollar of security provided, ordinary staking earned roughly 53 times more. The product that promised a richer yield delivered a poorer one.

The capital voted accordingly. EigenLayer’s total value locked sits near $5 billion, down from a peak above $22 billion in August 2025, per DefiLlama. More starkly, the amount of native ETH actually restaked collapsed from a peak around 3.5 million to roughly 643,000 by the end of September 2026, according to an on-chain investigation by Bitquery, which also found the shrunken remainder had reconcentrated onto two operators, Kraken at about 43% and P2P.org at about 18%. A system sold on permissionless decentralization ended the year leaning on two names.

Put that fee gap in plain terms. If restaking secures roughly $10 billion and collects about $100,000 a week, the services it protects are paying on the order of a hundredth of a percent a year for that protection. No token buyback, no points program, and no amount of narrative can manufacture a durable business out of a fee that thin. It is the clearest possible evidence that security, in the abstract, is something almost everyone wants and almost no one will pay much for.

The tokens tell the same story. As of October 26, 2026, EIGEN changed hands around $0.197, some 96.5% below its December 2024 peak; ether.fi’s ETHFI traded near $0.61 (down roughly 93% from its high) and Babylon’s BABY near $0.0137 (down about 92%), all per CoinGecko, while ETH itself sat near $2,300. EIGEN also fell about 16% over the trailing week, underperforming ETH’s roughly 5% pullback, as token-unlock pressure met a soft risk-asset tape. EigenLayer’s own ELIP-012 reforms route a share of AVS fees and all EigenCloud revenue into buying back EIGEN, but with gross fees this low the buyback is a trickle against monthly token unlocks.

The ether.fi Exit and the Market’s Verdict

No single event captured restaking’s repricing like the retreat of its biggest champion. Over 2026, ether.fi, the largest liquid restaking issuer, stripped restaking out of its flagship weETH token, reverting it to a plain liquid staking token, and isolated restaking in a separate, opt-in token called weETHs built on Symbiotic rather than EigenLayer. By the fourth quarter the firm was severing its last structural tie to EigenLayer, with less than 1% of its assets still restaked, as reported by The Defiant.

Founder and chief executive Mike Silagadze did not hide the symbolism. “End of an era. Sad,” he wrote, adding that he still believed restaking would return in one form or another and had simply arrived too early. In blunter comments to CoinDesk, relayed by Blockhead, he explained the exit in pure business terms: “There were no meaningful yield opportunities in restaking and there was some perceived risk from stakers, so we decided it made sense to exit.”

Where did ether.fi go? Into banking. The company leaned into a neobank strategy built around crypto debit cards and on-chain lending, and told reporters the new revenue had effectively replaced what restaking lost. That is the quiet tell of the year: the brand that once embodied restaking now earns its keep selling payments and credit, and its ETHFI token has held up on the strength of that pivot rather than anything to do with AVS. The market’s verdict was not delivered in a manifesto; it was delivered in a treasury reallocation.

EigenCloud: Restaking’s Pivot to Verifiable AI

Faced with a security product few wanted to pay for, EigenLayer’s answer was to change what it sells. The EigenCloud rebrand repositions restaked security as the trust backbone for verifiable cloud services: EigenDA for data availability, EigenVerify for dispute resolution, and EigenCompute for running workloads whose results can be checked rather than trusted. The thesis is that artificial-intelligence agents transacting on-chain will need a way to prove their computations were honest, and that cryptoeconomic security can underwrite that proof. The strategy drew a $70 million token purchase from a16z in 2025, and founder Sreeram Kannan has pitched it as bringing crypto-grade verifiability to ordinary cloud workloads.

It is a credible pivot and a revealing one. Verifiable compute is a real and growing market, as the throughput numbers in our scoreboard on decentralized inference make plain. But it also means the flagship restaking protocol is increasingly manufacturing its own demand, becoming the anchor tenant for the very security it sells, because outside tenants proved so scarce. Whether that is shrewd vertical integration or a tacit admission that the open AVS marketplace did not work is a matter of interpretation.

It is worth recalling that Kannan himself never oversold restaking’s safety. Back in a 2023 CoinDesk interview he argued that “anything that restaking can do, already liquid staking can do, so I view restaking as a lesser risk than liquid staking,” a careful framing that reads, three years on, as more prescient than promotional.

Regulation: Where the SEC Draws the Line

For a US holder, the regulatory picture turned friendlier in 2025, though with a catch that matters specifically for restaking. In a May 2025 statement, the SEC’s Division of Corporation Finance said that certain protocol staking activities are administrative or ministerial and do not amount to securities transactions. A follow-up August 2025 statement extended similar comfort to liquid staking receipt tokens, provided the issuer stays a passive agent and does not decide whether, when, or how much to stake, or promise a set return.

The catch is in the fine print. The protocol-staking carve-out explicitly set aside arrangements built around assets with intrinsic economic properties such as generating passive yield. A restaking token or an LRT engineered to pay active, service-derived yield plausibly falls outside that safe harbor, and staff statements are guidance, not rules; they can be withdrawn. The line between a passive receipt and a yield product is exactly where restaking lives, which keeps it in a grayer zone than plain staking.

Two practical notes follow. Custodial restaking-as-a-service, where an intermediary makes the choices for you, looks more like a regulated activity than solo protocol interaction, a distinction that runs through the broader debate over who the gatekeepers are, covered in our guide to DeFi compliance in 2026. And rewards are generally taxable when received, which turns a stream of restaking payouts into a stream of taxable events well before you sell, a trap we unpack in our explainer on crypto staking taxes.

Should You Restake? A Practical Framework

Strip away the narrative and restaking is a yield-for-risk trade, so judge it the way you would any yield. The base case in late 2026 is unflattering. Native ETH staking pays somewhere around 2.6% to 3%; restaking historically added perhaps one to two percentage points on top, and most of that premium arrived as tokens, not fees. Against that, the US three-month Treasury bill yields about 4.1%, per Trading Economics, and even on-chain tokenized Treasuries pay roughly 3.6% with no slashing at all, across nearly $15 billion of product tracked by RWA.xyz. When the risk-free rate clears 4% and composable, zero-slashing yield exists on-chain, nobody needs to be paid a thin premium to accept slashing, bridge, and smart-contract risk.

If you are still weighing it, four questions cut through most of the noise:

• Is the yield fees or emissions? Fees from paying customers are durable; token emissions are a countdown.
• Is the token rehypothecated? Trace how many promises your one unit of ETH is backing before you add leverage on top.
• Can you actually exit? Check the withdrawal queue and the LRT’s redemption path before you need them, not after.
• Do you understand the exact slashing conditions? If you cannot name what would cost you money, you are not being paid enough to take the risk.

The macro backdrop reinforces the caution. Heading into the Federal Reserve’s October 28 meeting, markets had swung to expecting a hold, with any rate hike largely deferred to December, per prediction-market and futures pricing compiled by DefiRate; risk assets were soft, ETH traded near $2,300, and EIGEN underperformed the broader market on the week. None of that is restaking-specific, but it is the environment in which any extra-yield pitch now has to compete.

The honest conclusion is not that restaking is dead. The technology works, the slashing fired cleanly when it was tested, and shared security remains a genuinely useful primitive for the handful of services that need it. What died in 2026 was the belief that renting Ethereum’s trust would be a lucrative business for the people supplying the capital. For most holders right now, plain staking or a tokenized Treasury captures more yield for less risk, and the burden of proof sits squarely on any restaking product claiming otherwise.

Frequently Asked Questions

What is restaking in simple terms?

Restaking lets you reuse staked ETH (or a token that represents it) to help secure extra services beyond Ethereum itself, such as oracles, bridges, and data-availability layers, in exchange for additional rewards. The trade-off is that your stake can be slashed for misbehavior on those extra services, so you take on more risk for more yield.

Is restaking safe?

Restaking adds several risks on top of normal staking: service-specific slashing, smart-contract bugs in new restaking and AVS code, and, for liquid restaking tokens, bridge and rehypothecation risk, as the April 2026 Kelp DAO exploit showed. Catastrophic slashing has been rare in practice, but the extra yield exists precisely because the extra risk is real.

What is the difference between restaking and liquid staking?

Liquid staking gives you a tradable receipt for staked ETH and earns Ethereum’s base rewards; restaking goes a step further by pledging that stake to secure additional services for extra fees and extra slashing conditions. In 2026, liquid staking earned roughly 53 times more in fees per dollar secured, which is why much of the capital moved back toward plain staking.

Why did EigenLayer’s TVL and the EIGEN token fall so much in 2026?

Demand for restaking security never matched the supply of capital, so most rewards came from token emissions rather than real fees; when those thinned, deposits and prices fell. EigenLayer’s TVL dropped from above $22 billion in August 2025 to around $5 billion, and EIGEN traded more than 96% below its all-time high by late October 2026.

Is restaking still worth it in 2026?

For most holders, probably not on yield alone. With the US risk-free rate above 4% and on-chain tokenized Treasuries paying around 3.6% without slashing, restaking’s thin, emissions-heavy premium rarely compensates for its added risks, though the underlying technology remains useful for services that genuinely need shared security.

By Yuki Tanaka, DeFi and on-chain markets correspondent, HOGE Wire.