Restaking Risk Explained: Slashing, Cascades and Cover

Restaking turned a simple idea, reusing already staked ether to secure more than one system, into one of Ethereum’s largest and most contested experiments. By mid-2026 the category had cycled through a multi-billion dollar boom, a sharp drawdown, a real nine-figure exploit, and now a scramble to build the insurance and risk-management tools that were missing the first time around. This is not another walkthrough of what an AVS is. It is a look at the part of restaking that decides whether any of it is actually safe to use: how slashing risk spreads between protocols, why that risk behaves differently from ordinary staking risk, and what insurers, engineers, and institutions are doing about it in 2026.

What Restaking Actually Means in 2026

Staking Ethereum means locking ether to help run the base chain: validators propose and attest to blocks and earn a native yield that has generally sat in the high twos to low threes percent range through 2026. Roughly a third of ETH’s supply is staked this way. Restaking takes that same staked ether, or a liquid staking token representing it, and pledges it a second time to secure additional software: oracles, bridges, data availability layers, rollup sequencers, and other services EigenLayer calls Actively Validated Services, or AVSs. In exchange, restakers can earn extra yield on top of their base staking return. In exchange for that yield, they take on extra slashing conditions, meaning more ways their stake can be cut if something goes wrong.

EigenLayer popularized the model and has since folded in a broader EigenCloud brand covering EigenDA (a data availability network), EigenCompute, and EigenVerify, extending the same restaked security idea to verifiable off-chain computation and AI agent accountability. A parallel liquid restaking token (LRT) market grew up around the same idea: deposit ETH into ether.fi, Renzo, Kelp DAO, or Puffer, receive a liquid token such as eETH, ezETH, rsETH, or pufETH, and stay free to use that token elsewhere in DeFi while the underlying ETH is restaked. For a primer on the plain liquid staking layer this all sits on top of, see our comparison of Lido, Rocket Pool, and Frax.

EIGEN, the token behind EigenLayer’s coordination and governance layer, trades for roughly $0.24 with a market capitalization near $178 million as of late July 2026, according to CoinGecko, a small fraction of its December 2024 all time high above $5.60. That gap between a shrunken token price and a still large restaked asset base is itself a clue to where this piece is headed: the economics of restaking and the risk of restaking have not moved together.

Why Restaking Risk Is Not Just Staking Risk

A solo Ethereum validator answers to one rulebook. Go offline for too long and it loses a small, predictable amount through inactivity leaks. Sign two conflicting blocks, or otherwise break a narrow set of protocol rules, and it can be slashed for a capped, known percentage of its 32 ETH. It is a bounded, well understood risk that node operators have priced for years.

Restaking multiplies the number of rulebooks bearing on that same 32 ETH. Each AVS a validator opts into writes its own slashing conditions: an oracle network might slash for reporting a bad price, a bridge validator set might slash for signing an invalid message, a rollup sequencer AVS might slash for censoring transactions. None of that is hypothetical. Ethereum co-founder Vitalik Buterin flagged the structural version of this problem before EigenLayer’s mainnet even launched, arguing in a widely read 2023 essay that “any expansion of the ‘duties’ of Ethereum’s consensus increases the costs, complexities and risks of running a validator,” and that stacking too many financial obligations onto the same validator set threatens the fragile social consensus that lets the chain function at all.

EigenLayer’s own founder, Sreeram Kannan, has pushed back on the sharpest version of that critique without dismissing the underlying concern. In a 2023 interview he argued that “anything that restaking can do, already liquid staking can do,” and that he personally views “restaking as a lesser risk than liquid staking,” while separately warning builders against assuming any single AVS is too big to fail, since Ethereum’s community could simply choose not to bail one out. Three years on, the debate has moved from theory to something closer to an engineering and insurance problem, which is where the rest of this piece spends its time.

The Correlated Slashing Problem

Restaking practitioners have a specific term for the risk Buterin was gesturing at: correlated slashing. The idea is straightforward once stated. If one pool of staked ETH, or one operator running many validators, secures several AVSs at once, a single failure does not necessarily stay contained to the AVS that caused it.

Matt Leisinger, co-founder and chief product officer at Alluvial, the company behind the institutional liquid staking protocol Liquid Collective, has described the mechanism plainly: “restaked tokens are often exposed to multiple validator networks. If one network underperforms or violates protocol rules, slashing penalties can cascade across all restaked layers.” The practical risk is less about any single AVS behaving maliciously and more about shared infrastructure. If many AVSs run on the same client software, the same operator set, or the same oracle feed, a bug or outage in that shared layer can trigger slashing across services that otherwise have nothing to do with each other.

This is not just a talking point; it has drawn formal academic attention. A 2024 paper by Naveen Durvasula and Columbia computer scientist Tim Roughgarden, “Robust Restaking Networks,” works out the math of how much a restaking network can absorb before a small loss turns into a much larger one. Their headline result is a kind of safety margin: if the cost of attacking a restaking network always exceeds the attacker’s profit by a fixed cushion (their example uses 10 percent), then a sudden loss equal to 0.1 percent of total stake cannot cascade into a loss of more than roughly 1.1 percent. In plain terms, overcollateralization is not just a nice to have; it is the variable that determines whether a small local failure stays small or turns into a chain reaction across every AVS an operator has opted into. The paper’s broader point is that robustness in a restaking network is a property protocols can actually measure and expose to users, rather than a vague promise.

Case Study: How the Kelp DAO Exploit Became Aave’s Problem

The clearest real world illustration of cascading restaking risk so far did not come from an AVS slashing event at all. It came from a cross-chain bridge. On April 19, 2026, an attacker exploited a vulnerability in the cross-chain bridge underpinning Kelp DAO’s liquid restaking token, rsETH, minting roughly 116,500 rsETH, about 18 percent of the token’s then existing supply, without actually depositing the ETH that was supposed to back it, according to CoinDesk’s reporting, which valued the exploit at roughly $292 million. The attacker then deposited a large share of that manufactured rsETH as collateral on Aave V3 and borrowed against it, leaving Aave holding overvalued collateral once the manipulation was discovered. Aave’s own total value locked fell from about $26.4 billion to roughly $20 billion over the following weekend, and the bad debt left on Aave’s books was estimated at around $196 million.

The recovery effort is arguably as instructive as the exploit itself, because it shows how quickly restaking risk becomes everybody’s problem once an LRT is accepted as collateral elsewhere in DeFi. Aave founder Stani Kulechov rallied a coalition that reporting has called “DeFi United,” personally pledging ETH toward the shortfall and posting that “rsETH has been frozen on Aave V3 and V4 … due to KelpDAO bridge exploit that happened outside of Aave,” adding that “Aave is my life’s work and we’re working nonstop to find the best possible outcome for users.” By around mid-2026, Kelp reported rsETH fully backed again across mainnet and its layer-2 deployments, with its bridge migrated from LayerZero’s OFT standard to Chainlink’s CCIP and a requirement for multiple independent attestors before large transfers settle.

Our earlier anatomy of a bridge hack covers why cross-chain infrastructure keeps being crypto’s weakest link, and the Kelp incident fits that pattern precisely: the AVS and staking layers were never touched, the bridge connecting an LRT to the rest of DeFi was. The lesson for anyone holding or lending against an LRT is that the risk of restaking is not confined to the restaking protocol itself. It runs through every bridge, oracle, and lending market that treats a liquid restaking token as ordinary collateral.

How EigenLayer Redesigned Slashing to Contain the Blast Radius

EigenLayer’s own engineering response to the correlated slashing problem predates most of the incidents above. In a design the team calls the EigenLayer Security Model, stake is organized into Operator Sets, groups of operators that a staker delegates to and that an AVS assigns specific tasks. The key mechanism is what EigenLayer calls Unique Stake Allocation: an operator decides in advance what share of its delegated stake any single Operator Set is allowed to slash, and critically, “a unit of ETH can only be slashed by a single Operator Set at any given time.” That single sentence is a direct technical answer to Leisinger’s cascade concern. Rather than an operator’s entire balance being exposed to every AVS it joins simultaneously, each unit of restaked ETH has one, and only one, path to being slashed at any moment.

The stated goal, per EigenLayer’s own writeup, was to let AVSs onboard permissionlessly, without a common veto committee vetting every new service before it can touch restaked capital, while still keeping the blast radius of any one AVS’s failure limited to the stake actually allocated to it. Slashing itself went live on EigenLayer’s mainnet on April 17, 2025, giving AVSs the ability to actually enforce these conditions on chain rather than relying on reputational or off-chain penalties. As of mid-2026, EigenLayer counts roughly 190 AVSs in development and around 40 live on mainnet, spread across more than 2,000 registered operators, though the useful, fee generating subset of that AVS count remains much smaller than the headline figure suggests.

None of this eliminates restaking risk. It converts an unbounded, network wide risk into a series of bounded, per-AVS risks that an operator, and in theory a staker, can actually see and choose to accept or avoid. Whether ordinary users read those disclosures before restaking is a separate question, one that pushes the remaining risk back onto due diligence rather than protocol design, the subject of a later section.

The New Insurance Layer: ether.fi, Nexus Mutual and Liquid Collective

If 2025 was restaking’s growth phase, mid-2026 looks like its insurance phase. On July 17, 2026, ether.fi, the largest liquid restaking provider with more than $6 billion in assets across its Cash, Stake, and Liquid product lines, announced what it called crypto’s largest ever ETH slashing cover: a policy with Nexus Mutual protecting its validators against up to 15,000 ETH in slashing penalties, an amount that Decrypt reported exceeds all historical Ethereum slashing losses combined. ether.fi founder and chief executive Mike Silagadze framed it as a bet on trust: “we’ve always believed the safest protocols will ultimately win. That’s why we’ve invested heavily in audits, operational security, staking architecture, and now the largest insurance program in the industry.” Nexus Mutual founder Hugh Karp, whose organization has covered more than $7 billion in smart contract, slashing, and other digital asset risks since 2019, called it “a historic step,” adding that his team has “known the ether.fi team since before it was ether.fi, and they’ve been focused on risk from day one.”

ether.fi is not alone. Liquid Collective, the institutional liquid staking protocol built with Coinbase, Kraken, Figment, and Kiln via parent company Alluvial, runs its own slashing coverage program, also underwritten in part by Nexus Mutual. That program layers three lines of defense:

• Nexus Mutual coverage that scales dynamically with the protocol’s total value locked, up to $5 million per incident in its published terms
• A node operator commitment of up to $5 million per operator
• A slashing coverage treasury funded by a small percentage of network rewards

The design lets a slashing event adjust the underlying exchange rate of Liquid Collective’s LsETH token rather than punishing individual holders directly. Liquid Collective sits on the staking side of the spectrum rather than the AVS restaking side, but the same logic applies: as more institutional capital touches staked and restaked ETH, insurance stops being a marketing bullet point and starts being required infrastructure, much like custody and audits became baseline expectations in the years after FTX. Whether these policies are large enough relative to the capital they cover is the next question serious allocators are asking, and it is one insurers and protocols themselves are still working through in public.

Bitcoin Restaking’s Different Risk Model: Babylon

Not all restaking runs through Ethereum, and not all of it carries the same risk shape. Babylon lets Bitcoin holders lock BTC directly, using Bitcoin’s own timelock scripting rather than wrapping or bridging the coins onto another chain, and use that locked BTC to provide economic security to proof-of-stake networks that opt in as what Babylon calls Bitcoin Supercharged Networks. Because the BTC never leaves Bitcoin or gets represented by a bridged token, the specific bridge exploit risk that hit Kelp DAO does not apply in the same way. Babylon’s Genesis mainnet and its BABY token both launched on April 10, 2025, with an airdrop split across early stakers, staking rewards, and open source contributors.

The scale is real: Babylon’s own dashboard shows 56,853 BTC staked, worth roughly $5.64 billion, as of mid-2026, making it one of the largest Bitcoin staking systems by that measure. The BABY token itself has traded far below its April 2025 all time high of $0.1661, a reminder that even a restaking design built to avoid bridge risk is not immune to the token price collapse that has hit almost every other project in this category; EIGEN and SSV show the same pattern despite very different underlying architectures. Babylon’s risk still runs through the finality providers stakers delegate to and the specific slashing conditions of whichever BSN a delegation secures, so the due diligence questions are the same in spirit as anywhere else in restaking, just applied to a different base asset.

Institutional custodians have started building around this. Hong Kong based custodian Hex Trust rolled out a Babylon integration letting clients stake BTC directly from custody, without bridging, delegating to Hex Trust itself as what Babylon calls an Institutional Finality Provider. Hex Trust chief commercial officer Calvin Shen called it “an enormous opportunity for Hex Trust and our clients,” saying that “unlocking the value of Bitcoin through staking, and ultimately through DeFi, is a huge milestone for the industry.”

SSV and the Distributed Validator Approach to Risk Isolation

SSV Network approaches the same problem from a different angle: distributed validator technology, or DVT. Instead of one operator holding a validator’s signing key, SSV splits it into key shares distributed across four or more independent, non-trusting operators using Shamir secret sharing, with an IBFT consensus process and threshold BLS signatures coordinating them. The pitch is fault tolerance for the base validator itself: no single operator going offline or misbehaving can bring the validator down or leak its key.

SSV’s own restaking layer, unveiled in January 2025 under the name Based Applications, or bApps, builds an explicit risk boundary into the design. Validators opt in using participation keys rather than withdrawal keys, and the core 32 ETH backing a validator is never slashable through a bApp; only additional, voluntarily delegated capital is exposed. SSV describes this as a “Risk Expressive Model,” letting each based application set its own risk and reward terms rather than imposing one blanket slashing regime across the network. SSV founder and chief executive Alon Muroch has called the bApps design the company’s “biggest, most ambitious project,” one that “will profoundly change the restaking market.” The network also launched cSSV in April 2026, converting staked SSV into a liquid, ETH denominated reward instrument as part of a push to tie the SSV token’s value more directly to real network fees, an acknowledgment, in Muroch’s own words elsewhere, that “the token’s value is largely detached from ETH staking rewards” as originally designed. SSV itself trades around $2.07, down roughly 97 percent from its March 2024 high.

Whether isolating the slashable base, SSV’s approach, or isolating stake per Operator Set, EigenLayer’s approach, proves the more durable design is still an open question. Both are explicit attempts to solve the same correlated risk problem this piece has been describing, using different points of leverage.

Symbiotic and Karak’s OpenGDP Pivot: When a Rival Leaves the Category

Not every competitor is converging on the same answer, and one of them has stopped trying to answer the question at all. Symbiotic, EigenLayer’s most direct rival, has run for roughly two years and raised $34.8 million (a seed round co-led by Paradigm and Cyber Fund, followed by a $29 million Series A led by Pantera Capital) without ever launching a public token, according to FinanceFeeds, which quotes Symbiotic co-founder Misha Putiatin making the case that the category’s real opportunity has moved beyond restaking narrowly defined: “the RWA market has crossed $33 billion, but most of those assets still can’t be redeemed on demand.”

That framing is now a product, too. Symbiotic launched Liquid Lane in June 2026, extending its vault infrastructure to real world asset use cases like credit underwriting and stablecoin liquidity, and followed it on July 1, 2026 with Core V2, which The Block reported pivots the protocol’s positioning explicitly beyond restaking toward what Symbiotic calls collateral markets, claiming meaningfully better capital efficiency than standalone liquidity pools. Putiatin described the shift plainly: “Symbiotic Core changes that: committed capital now functions as a liquidity sleeve, settling redemptions without sitting idle.”

Karak has gone further still. The protocol, which raised a $48 million Series A led by Lightspeed Venture Partners in December 2023, rebranded entirely in July 2026, dropping the Karak name for OpenGDP, “the operating layer for real-world economic execution.” Its own site no longer mentions restaking, AVSs, or shared security anywhere in its main copy; the pitch is now tokenizing real economies, stablecoin settlement rails, and institutional coordination tools, close to a full exit from the category it started in. Our sector map of restaking’s 2026 landscape covers where each major player now stands in more detail. Read together, Symbiotic’s pivot and Karak’s rebrand say something about how the market is pricing pure restaking’s future: adjacent categories like tokenized real world assets look more durable to at least two of the field’s founding teams than doubling down on AVS security alone.

How Institutions Actually Vet a Restaking Position

For a retail user, restaking risk mostly comes down to which LRT to hold and how much yield is worth chasing. For an institution moving eight or nine figures, the process looks more like corporate credit underwriting than a yield search. Due diligence teams at custodians and treasuries generally work through the same handful of questions regardless of which protocol they are evaluating:

• What exactly can this AVS slash for, and is the penalty capped
• Who has the authority to trigger a slashing event, and how quickly
• How concentrated is the actual operator set behind the numbers shown on a dashboard
• Does exposure need to be split across multiple operators rather than concentrated with whichever one offers the best headline yield
• What insurance, if any, sits underneath the position, and how large is it relative to the capital at risk

Because slashing risk is non-isolated by default, the correlated slashing problem described earlier in this piece, the standard institutional response is to spread restaked exposure across multiple operators rather than concentrating it. Real capital has started moving on these terms rather than staying purely theoretical. Nasdaq listed SharpLink Gaming, an Ethereum treasury company chaired by Consensys founder Joseph Lubin, deployed $170 million of a planned $200 million ETH treasury program into EigenCloud AVS restaking alongside ether.fi, routed through Consensys’s own Linea network and held with Anchorage Digital Bank as qualified custodian, a structure detailed in CoinDesk’s coverage. Separately, Flow Traders, the Euronext Amsterdam listed market maker best known for ETF liquidity provision, went live in late 2025 as an operator on Cap, a private credit AVS built on EigenLayer, accessing restaking backed cryptoeconomic guarantees to support USDC liquidity rather than restaking for its own sake.

Neither example eliminates the underlying risks this piece has walked through. What they show is that the institutional playbook has shifted from asking whether to touch restaking at all to asking which specific operators, custodians, and insurance layers make a given position defensible on a risk committee memo.

The Regulatory Gap: What the SEC Has, and Has Not, Said

Regulatory clarity has lagged the product itself, which matters more for restaking than for plain staking. In a May 2025 statement, the SEC’s Division of Corporation Finance said that protocol staking, whether solo, delegated, or run through a custodian, generally does not involve a securities transaction under the Howey test, paired with a companion statement from Commissioner Hester Peirce titled “Providing Security is not a ‘Security.’” A follow up in August 2025 extended similar treatment to liquid staking and staking receipt tokens like stETH and eETH style wrappers, according to a subsequent SEC release.

Restaking and liquid restaking tokens sit outside both statements. Neither the May nor the August guidance addresses a token that has been staked, then restaked into a separate set of AVS specific slashing conditions, then wrapped again into an LRT that circulates as DeFi collateral, the exact structure that turned Kelp DAO’s bridge exploit into an Aave balance sheet problem. That gap is not a minor technicality: both existing statements are non-binding staff level guidance, not rules adopted by a Commission vote, meaning they carry no force of law and could be withdrawn by a future Commission with different priorities. Our broader look at how US crypto enforcement actually maps out in 2026 covers the wider policy backdrop this sits inside, including how much of the current hands off posture rests on personnel and priorities that could change. Until restaking and LRTs get their own explicit treatment, institutions building compliance programs around this asset class are, in effect, extrapolating from staking guidance written for a simpler product.

Restaking by the Numbers: A Protocol Comparison

The figures below are a snapshot, not a forecast. Total value locked in particular varies meaningfully by tracker and by the week across this sector, and should be read as directional rather than exact.

How the Industry Is Responding to Correlated Slashing

Put side by side, the mitigation layer for restaking risk breaks down into a few distinct approaches, some architectural and some financial:

Where Restaking Risk Goes From Here

Put the last year of restaking history together and a pattern emerges that is more interesting than any single price chart. The category’s engineering has genuinely improved: EigenLayer’s Unique Stake Allocation and SSV’s Risk Expressive Model both directly target the correlated slashing problem that worried Buterin in 2023, and real insurance capacity, ether.fi’s 15,000 ETH cover and Liquid Collective’s layered Nexus Mutual program, has shown up for the first time at anything like institutional scale. Real institutional capital has started using the rails, from SharpLink’s treasury deployment to Flow Traders’ operator role on Cap.

At the same time, none of that risk infrastructure is yet sized for the amounts restaking claims to secure. A 15,000 ETH policy and a $5 million per incident cap are meaningful firsts, not a ceiling on what a systemic AVS failure could cost across billions of dollars in restaked value. Regulatory guidance still stops at plain staking. And two of the category’s founding challengers, Symbiotic and Karak, have spent 2026 moving their core pitch toward tokenized real world assets rather than doubling down on AVS security, a tell about where the smart money inside those teams thinks the more durable business actually is.

Restaking is not going away. EigenCloud, Babylon, and SSV all show real usage well beyond their token prices. But the version of restaking gaining institutional traction in 2026 looks less like the points driven, yield chasing product of 2024, and more like a specific, insured, operator vetted instrument that risk committees can actually underwrite, a narrower and slower moving thing than the original pitch.

Frequently Asked Questions

What is restaking in crypto?

Restaking is the practice of reusing already staked ether, or a token representing it, to provide economic security to additional protocols beyond the Ethereum base chain, commonly oracles, bridges, data availability networks, and rollup sequencers, known collectively as Actively Validated Services, or AVSs. In exchange for extra yield, restakers accept extra slashing conditions specific to each service they opt into. EigenLayer popularized the model; Symbiotic, Karak (now OpenGDP), Babylon, and SSV Network have each built variations on it.

What is the difference between staking and restaking?

Staking secures only Ethereum’s base chain and carries one well defined slashing risk, downtime or equivocation. Restaking takes that same staked ETH, or a liquid staking token, and pledges it a second time to secure other services, each with its own separate slashing rules. Staking risk is bounded and standardized; restaking risk is layered, since a single unit of ETH can end up subject to several different rulebooks depending on how many AVSs an operator has opted into.

What is correlated slashing risk?

Correlated slashing describes a scenario where a single underlying failure, such as a software bug, an operator outage, or a compromised bridge, triggers slashing or losses across multiple AVSs or protocols at once, rather than staying contained to the service where the problem originated. It is a central concern in restaking because the same stake, or the same operator, commonly secures several services simultaneously. EigenLayer’s Unique Stake Allocation and SSV’s Risk Expressive Model are both engineering responses aimed specifically at limiting how far a single failure can spread.

Is restaking safe for institutions?

Restaking carries real, well documented risks, including the April 2026 Kelp DAO bridge exploit that produced roughly $196 million in bad debt on Aave, but institutions increasingly participate through a specific due diligence process: vetting individual operators, distributing exposure across several of them, checking each AVS’s slashing terms, and increasingly requiring insurance coverage such as ether.fi’s Nexus Mutual backed policy. Whether it counts as safe depends heavily on which protocol, which operator, and how much insurance and diversification sit underneath a given position, rather than being a single yes or no answer for the category as a whole.

Does the SEC regulate restaking and liquid restaking tokens?

Not explicitly. The SEC’s Division of Corporation Finance said in May 2025 that ordinary protocol staking generally is not a securities transaction, and extended similar treatment to liquid staking tokens in August 2025, but neither statement addresses restaking or liquid restaking tokens specifically. Both statements are also non-binding staff level guidance rather than formal rules, meaning restaking currently operates in a regulatory gray zone that a future Commission could fill in either direction.

By the HOGE Wire markets desk.