Restaking Explained: How Staked ETH Secures Everything Else

The month restaking stopped being the default

On August 6, 2026, ether.fi, the largest liquid restaking protocol in crypto, did something that would have sounded absurd a year earlier: it switched restaking off by default. Its flagship token, weETH, had routed user deposits into restaking since launch. From that date weETH reverted to a plain liquid staking token, and restaking was carved out into a separate, opt-in product called weETHs that delegates to a rival network rather than to the category’s pioneer. As the change landed, weETHs held roughly $17.7 million and paid about 3.50%, a rounding error next to ether.fi’s multi-billion-dollar book.

That one decision is the clearest sign yet of how the mood has turned. Restaking spent 2024 and early 2025 as the most hyped idea in Ethereum, a mechanism that promised to let a single unit of staked ETH earn several streams of yield at once. By the middle of 2026 the biggest issuer of restaked assets decided the risk was not worth bundling into a product most people use as collateral. This article explains what restaking is, how it works, where its yield and its danger actually come from, and why 2026 became the year the market started taking it apart.

Staking, then liquid staking, then restaking

Start with the base layer. Ethereum runs on proof of stake, which means validators post collateral, 32 ETH each, and earn rewards for proposing and attesting to blocks honestly. Misbehave and a portion of that collateral is destroyed, a penalty called slashing. As of early August 2026 roughly 41.4 million ETH, about 34% of all ETH, was staked across hundreds of thousands of validators, and the base consensus reward had fallen to around 2.6% a year, a three-year low driven by the sheer volume of ETH competing for a fixed reward pool.

Two problems come with raw staking: you need 32 ETH to run your own validator, and staked ETH is locked, so it cannot be used elsewhere. Liquid staking solved both. Deposit any amount with a protocol like Lido or Rocket Pool and you receive a liquid token, stETH or rETH, that accrues staking rewards while staying tradable and usable across DeFi. If you want the mechanics of that layer, our guide to Lido, Rocket Pool and Frax walks through how each one works.

Restaking is the next step out on the risk curve. Instead of letting your staked ETH secure only Ethereum, you re-pledge it, or the liquid token that represents it, to secure additional systems as well. In return you earn extra rewards from those systems, and you accept extra ways to be slashed. The idea was coined by EigenLayer, and for two years it was treated as close to free money. It is not free, and 2026 is the year the bill came due.

What restaking actually does: renting out Ethereum’s trust

To see why anyone bothers, look at what a new piece of crypto infrastructure has to build before it can launch. An oracle that reports prices, a bridge that moves assets between chains, a data-availability layer, a rollup’s sequencer, a zero-knowledge prover: each needs a set of operators putting real value at stake so that lying or going offline costs them money. Traditionally every one of those services had to bootstrap its own token and its own staking economy, an expensive and slow process that also produces weak security in the early days, when the token is cheap and easy to attack.

Restaking offers a shortcut. These services, called Actively Validated Services, or AVS, can rent security from the pool of ETH that is already staked and already worth hundreds of billions of dollars. Operators opt in to run the extra software, restakers back them with their ETH, and in exchange the AVS pays fees or tokens. Sreeram Kannan, the founder of Eigen Labs, put the pitch to CoinDesk in 2023 this way: “Anything that restaking can do, already liquid staking can do, so I view restaking as a lesser risk than liquid staking.” That framing looks generous in hindsight, but it captured the ambition: turn Ethereum’s idle security into a marketplace anyone can buy from.

Not everyone cheered. Ethereum co-founder Vitalik Buterin published a widely-read warning, Don’t overload Ethereum’s consensus, arguing that “Any expansion of the ‘duties’ of Ethereum’s consensus increases the costs, complexities and risks of running a validator.” His fear was that a large restaking system could grow important enough to demand a bailout, dragging Ethereum’s social layer into disputes that have nothing to do with securing the base chain. Three years later that concern reads less like theory and more like a preview.

The restaking map: five networks, five bets

The category is no longer one protocol. Five networks now define it, and they have drifted apart in strategy as the hype cooled. EigenLayer, since rebranded as EigenCloud, remains the anchor. It pioneered restaking on Ethereum, turned slashing on for real in April 2025, and has expanded into a wider pitch it calls a verifiable cloud, adding data availability, off-chain compute and dispute resolution to plain restaking.

Symbiotic took a more permissionless design, accepting almost any ERC-20 as collateral and letting each network set its own slashing rules; it is now the network ether.fi chose for weETHs. Karak, the third mover, has all but left the category, rebranding as OpenGDP and repositioning around tokenized real-world assets. Babylon brought the model to Bitcoin. And SSV Network approaches shared security from the validator-infrastructure side with its Based Applications design. The table below sketches the field.

Liquid restaking tokens and the rehypothecation problem

Just as liquid staking wrapped staked ETH in a tradable token, liquid restaking wrapped restaked ETH in one too. These liquid restaking tokens, or LRTs, include ether.fi’s eETH and weETH, Renzo’s ezETH, Kelp’s rsETH and Puffer’s pufETH. Each represents a deposit that is simultaneously securing Ethereum, securing one or more AVS, and sitting in your wallet as a token you can lend, borrow against or pool.

That is the feature and the flaw. The same token can be posted as collateral on a lending market, which mints a claim against it, which can be redeposited, which can be borrowed against again. The industry’s polite word for this is rehypothecation; the plain-English version is that several different lenders can end up believing they have first claim on the same underlying ETH. When you hold an LRT you do not hold ETH, you hold a claim on a claim, and the distinction matters enormously in a crisis. It is the same lesson our feature on crypto custody draws from the exchange failures: control of a token is not the same as ownership of the asset beneath it.

LRTs also carry peg risk. Their market price should track the value of the ETH plus rewards they represent, but redemption is not instant, so in a panic the token can trade below its supposed value. Renzo’s ezETH briefly depegged in April 2024 when a wave of holders tried to exit at once. A token that is money-good on a calm Tuesday can gap lower the moment everyone reaches for the door together, and leveraged loops built on top of it turn that gap into forced liquidations.

Where the restaking yield actually comes from

The number that sold restaking was a headline APY, often quoted in the high single digits or low teens. To judge whether it is real, you have to break it into its parts, because the sources differ wildly in how durable they are.

Stacked together, those lines produced eye-catching quotes during the points era. Strip out the emissions and the speculative points, though, and what remains is the roughly 2.6% base reward plus a slim margin of genuine AVS fees. Worse for the yield story, the base is under pressure from Ethereum’s own roadmap. A proposal filed in early August 2026, EIP-8361, would introduce a tapered issuance burn that destroys a growing share of validator rewards as the staking ratio climbs, which at today’s levels would push the consensus yield from about 2.6% toward 1.2% over roughly 18 months. If the base shrinks, restaking’s extra basis points have to do even more of the work to justify the extra risk.

Slashing and smart-contract risk: the price of the extra yield

Every extra reward stream comes with an extra way to lose money. On plain Ethereum, slashing punishes a narrow set of provable faults. Restaking layers new slashing conditions on top, one set per AVS an operator secures. EigenLayer made those penalties live on mainnet in April 2025, which turned restaking from a rewards-only program into one with real downside. If an operator you back misconfigures a node, signs a conflicting message or fails an AVS’s liveness rule, the ETH standing behind them can be cut.

The subtler danger is correlation. A single operator often secures many AVS at once, and many restakers back the same handful of large operators, so one bad bug can trigger slashing across a wide swath of the network at the same time. That is the too-big-to-fail scenario Buterin warned about: a loss large enough that the community faces pressure to intervene rather than let it stand.

Then there is code. A restaked deposit can pass through a liquid staking contract, a restaking contract, an AVS contract and an LRT wrapper, each an independent piece of software that can be exploited. The attack surface is additive, and in practice it is the wrappers and bridges, not Ethereum’s consensus, that have failed. Our running catalog of what actually breaks crypto in 2026 keeps returning to the same theme: the losses come from the glue between systems, not the systems themselves.

The Kelp weekend: when restaking risk turned systemic

The clearest illustration arrived over a single weekend in April 2026. An attacker exploited Kelp DAO’s cross-chain bridge to mint about 116,500 rsETH out of thin air, roughly $292 million of freshly conjured liquid restaking tokens. The tokens were not stolen from a vault; they were newly created, which meant they looked legitimate to any system that trusted rsETH’s reported supply.

The attacker then deposited that rsETH as collateral on Aave and borrowed real assets against it, leaving Aave with roughly $196 million of bad debt once the fake collateral was recognized. Aave’s total value locked fell by about $6 billion over the weekend as users pulled funds, and the AAVE token dropped sharply. Aave founder Stani Kulechov moved fast, telling users on X, as quoted by CoinDesk, that “rsETH has been frozen on Aave V3 and V4, the asset does not have any borrowing power as a measure due to KelpDAO bridge exploit that happened outside of Aave.” He personally pledged 5,000 ETH toward the recovery.

The point is not that Aave or Kelp were uniquely careless. It is that a liquid restaking token was accepted as blue-chip collateral without pricing in the risk of the bridge that minted it, and a failure two protocols away became Aave’s problem in hours. Restaking’s promise is shared security; its shadow is shared fragility. Anyone weighing LRTs as collateral should read our risk map for on-chain credit, because the collateral you accept is only as safe as the weakest system that can mint it. A recovery coalition of Lido, ether.fi, Consensys and others eventually made depositors whole, but the episode reset how the market thinks about restaked collateral.

The TVL round trip: from mania to reality

Follow the money and the story tells itself. EigenLayer’s total value locked peaked near $19 billion in early 2026 during the scramble for points and airdrops. By the middle of August 2026 EigenCloud held around $5 billion, still the largest in the category but a fraction of the peak, and the entire restaking sector tracked by DefiLlama sat near $7.9 billion across roughly a dozen protocols. Deposits left as the points dried up and the promised AVS fee economy failed to materialize at the scale the valuations implied.

Token prices tell the same story more bluntly. EIGEN, the EigenCloud token, traded near $0.17 in mid-August 2026 for a market capitalization around $149 million, roughly 97% below its December 2024 peak. SSV, the largest distributed-validator token, changed hands near $2.08 for a market value around $31 million, and even ETH itself traded around $1,883, some 62% under its August 2025 high. The scoreboard below captures the drawdown.

None of this means restaking is dead. EigenCloud’s TVL even ticked up slightly over the trailing month. But the gap between a token worth $149 million and a protocol securing $5 billion, or between BABY’s small market value and the more than $5 billion in Bitcoin it helps secure, is the market’s way of saying it has not decided what these networks are worth, or whether the tokens capture any of the value the protocols create.

Bitcoin gets in on it: Babylon and native-BTC restaking

Restaking is not only an Ethereum story. Babylon brought the idea to Bitcoin, and it did so without the wrapping or bridging that has caused so many losses. Using Bitcoin’s own timelock scripting, a holder can lock BTC on the Bitcoin base chain itself, keeping self-custody, and that locked BTC then backs proof-of-stake networks that opt in for security. If the staker breaks the rules, the script allows their BTC to be slashed; if they behave, they earn rewards from the chains they secure.

The scale is real. Babylon’s own dashboard reported about 56,853 BTC staked, worth roughly $5.64 billion, making it the largest Bitcoin staking system by that measure, though the figure has sat unchanged for weeks and reads more like a periodically-updated headline than a live counter. Its BABY token, launched with the network in April 2025, trades at a small fraction of the value it secures, the same token-versus-protocol gap visible across the sector.

The most watched next step is lending. Aave Labs and Babylon have been building an integration to let native Bitcoin serve as collateral in Aave V4 through trustless Bitcoin vaults, with BTC locked in a Taproot output on Bitcoin rather than a wrapped IOU on Ethereum. It reached public testnet during 2026 but, as of this writing, has not shipped to mainnet; governance still has to finalize risk parameters, oracle design and liquidation mechanics before real money moves. It is a reminder that the interesting part of restaking is increasingly what plugs into it, not the restaking itself.

Why the market is unbundling restaking

Return to ether.fi’s decision, because it is the thesis of 2026 in miniature. For two years the winning product design bundled everything together: deposit ETH, receive one token, and let it earn staking rewards and restaking rewards and points all at once. That bundle was easy to market and easy to grow. It was also, it turned out, hard to use as collateral, because a lender pricing weETH had to price Ethereum staking risk, EigenLayer slashing risk, individual AVS risk and LRT peg risk all in one number.

By splitting weETH back into clean staking and separating restaking into opt-in weETHs, ether.fi is admitting that most holders wanted the base asset, not the extra risk, and that serious collateral needs a simple risk profile. The direction of travel across the field is the same. Symbiotic, which now hosts ether.fi’s restaking, keeps its restaking layer explicitly separate from any staking token. Karak walked away from restaking entirely by becoming OpenGDP. The market is not abolishing restaking; it is unbundling it, turning a default into a choice. We trace that market-structure shift in depth in our companion piece on why the market is unbundling restaking.

There is an irony here worth stating plainly. Kannan argued in 2023 that restaking was a lesser risk than liquid staking. The revealed preference of 2026, led by the largest restaker, is the opposite: keep liquid staking clean and quarantine restaking for those who explicitly want it. The idea survives, but as an add-on rather than the main course.

What the SEC says, and where restaking sits

The US regulatory backdrop shifted in restaking’s favor during 2025, though not without limits. In May 2025 the Securities and Exchange Commission’s Division of Corporation Finance issued a staff statement that ordinary protocol staking, whether solo, delegated or custodial, is not by itself a securities transaction. Commissioner Hester Peirce underlined the point in a companion note titled Providing Security is not a Security. In August 2025 the SEC extended that comfort to liquid staking and staking receipt tokens such as stETH, and in March 2026 a joint SEC and CFTC interpretive release began drawing clearer lines around which digital assets count as commodities.

Restaking sits at the edge of that comfort zone rather than squarely inside it. The safe-harbor logic rests on the idea that stakers are paid for providing a technical service, not for the efforts of a promoter. Restaking complicates that in two ways: an AVS pays rewards that look more like a return on a managed enterprise, and LRT issuers actively choose operators and strategies on a holder’s behalf. The SEC’s own guardrails already exclude arrangements where a custodian guarantees a return or exercises discretion over how funds are deployed, and a liquid restaking token that routes your ETH across a shifting set of AVS looks a lot like discretion. For now the question is unsettled, which is its own kind of risk for anyone building a product on top of restaked yield.

Is restaking worth it in 2026?

For most holders, the honest answer is that plain staking or clean liquid staking now captures the large majority of the reward with a fraction of the complexity. Restaking still makes sense for a narrower audience: sophisticated users who understand slashing, operators who can evaluate which AVS are worth securing, and funds that treat the extra basis points as a portfolio position rather than a set-and-forget yield. If you do choose to restake, a short checklist keeps you on the right side of the risks.

• Separate the yield: know how much of the quoted APY is real staking reward versus token emissions and points that can evaporate.
• Read the slashing conditions of every AVS your operator secures, not just the headline network.
• Prefer restaking you opt into deliberately, like ether.fi’s weETHs model, over products that restake by default.
• Treat LRTs as claims, not as ETH, and assume redemption can be slow or discounted in a panic.
• If you post an LRT as collateral, size the position for a depeg, not for a calm market.

Restaking was sold as a way to make the same ETH work several jobs at once. That is still possible, and the underlying engineering, shared security rented from a deep pool of stake, is genuinely useful for the oracles, bridges and provers that need it. What changed in 2026 is that the market stopped treating it as a free lunch and started pricing it as what it always was: an extra return for an extra, and sometimes correlated, set of risks.

Frequently Asked Questions

What is restaking in simple terms?

Restaking lets you take ETH that is already staked to secure Ethereum and re-pledge it to secure additional services, such as oracles or bridges, at the same time. You earn extra rewards from those services and take on extra ways to be penalized. EigenLayer, now called EigenCloud, introduced the idea, and it is the higher-risk step beyond ordinary staking and liquid staking.

Is restaking safe?

Restaking adds risk on top of staking. You face extra slashing conditions from every service your operator secures, smart-contract risk across several layers of software, and, if you hold a liquid restaking token, the chance it trades below the value it represents. The April 2026 Kelp DAO exploit, which minted about $292 million of fake rsETH and left Aave with roughly $196 million of bad debt, showed how one failure can spread. It is not inherently unsafe, but it is meaningfully riskier than plain staking.

How much can you earn from restaking?

The durable part of the yield is Ethereum’s base staking reward, around 2.6% a year in mid-2026, plus a thin margin of real fees paid by the services buying security. Much of the eye-catching double-digit APY quoted during 2024 and 2025 came from token emissions and points, which are dilutive or speculative and have largely faded. A pending Ethereum proposal, EIP-8361, could push the base reward toward 1.2% over the coming years, squeezing yields further.

What is the difference between staking, liquid staking and restaking?

Staking locks 32 ETH to run a validator and secure Ethereum. Liquid staking lets you deposit any amount and receive a tradable token, such as stETH or rETH, that keeps earning while staying usable. Restaking goes one step further by re-pledging that staked ETH to secure other systems for additional rewards and additional risk. In 2026 the trend is to keep these layers separate rather than bundling restaking into the staking token by default.

Why is ether.fi removing restaking from weETH?

On August 6, 2026, ether.fi made weETH a plain liquid staking token and moved restaking into a separate, opt-in token called weETHs built on Symbiotic. The reasoning is that most holders use weETH as collateral and want a simple, predictable risk profile, while restaking bundles in slashing and peg risks that make the asset harder to lend against. It reflects the broader 2026 shift toward unbundling restaking and making it a deliberate choice rather than a default.

By Nathan Reed, DeFi correspondent at HOGE Wire.