Restaking Explained: Where’s the Demand for Rented Security?

In 2023 and 2024, restaking was the loudest idea in Ethereum. It promised to take the tens of billions of dollars already locked up securing the network and rent that same security out to a second wave of applications, all for extra yield. Capital poured in. By early 2026, restaking protocols had gathered enough deposits to rank among the largest applications in all of crypto.

Then the story changed. The token most associated with the sector, EIGEN, now trades near $0.17 with a market value around $129 million, a fall of more than 96 percent from its December 2024 peak, according to CoinGecko. Sector deposits are well off their highs. And the question that hovered over restaking from the beginning is now impossible to ignore: there is plenty of security for rent, but who is actually renting it?

This is an explainer, but not the usual one. Instead of walking through the mechanics and stopping there, it treats restaking as what it really is, a two-sided market, and asks whether the demand side ever showed up. Along the way you will get the full picture: what restaking is, where the yield comes from, who the major players are, what can go wrong, and where United States regulators have landed. Prices are in US dollars, and the regulator in focus is the Securities and Exchange Commission.

What restaking is, and why 2026 is its reckoning

Restaking, in one line: it lets capital that is already securing one blockchain be pledged a second time to secure other things, in exchange for extra rewards. On Ethereum, that means taking ether you have staked (or a liquid token that represents it) and committing it to also back oracles, bridges, data-availability layers, or rollup infrastructure. If those services misbehave, your stake can be slashed by their rules as well as by Ethereum’s.

The idea, coined by EigenLayer, was seductive. Ethereum spends an enormous amount of capital, tens of millions of ether, to secure a single chain. Restaking proposed to recycle that security budget so that a new protocol would not have to bootstrap its own validator set and its own token from scratch. Instead it could rent trust from Ethereum’s existing stakers, the way a startup rents cloud servers rather than building a data center.

For a while the market believed. Deposits climbed into the tens of billions of dollars by most trackers; liquid restaking tokens became some of the fastest-growing assets in DeFi; every new middleware project described itself as an actively validated service. The category briefly looked like Ethereum’s next foundational layer.

2026 is where the enthusiasm meets the ledger. EIGEN, the token most associated with restaking, changes hands near $0.17, a drop of more than 96 percent from its December 2024 high per CoinGecko. Sector deposits are far below their peak. The mechanics work; slashing is live; the plumbing is real. What is missing is proof that a large, paying market for rented security actually exists. That gap, between an abundant supply of security and uncertain demand for it, is the real story of restaking in 2026, and the thread this explainer follows.

From staking to restaking: what actually gets reused

Start with plain staking. To help run Ethereum you lock 32 ether into a validator, or you use a staking pool that does it for you. In return the network pays issuance, and your validator collects transaction tips and MEV. If your validator goes offline or signs conflicting messages, a portion of that stake is slashed. This is the base layer of crypto-economic security: honest behavior is rewarded, provable misbehavior is punished, and the punishment is denominated in a valuable asset.

Restaking asks a simple question. That staked ether is sitting there, bonded and slashable, doing one job. Could it do a second? Could the same bond that keeps a validator honest on Ethereum also keep an oracle honest, or a bridge, or a cross-chain messaging layer? If so, the new service borrows Ethereum-grade security instead of minting a fresh token and begging validators to hold it.

The services that buy this borrowed security are called Actively Validated Services, or AVSs. An AVS is any system that needs a decentralized set of operators to do a job correctly and can define a rule for when they have done it wrong. Oracles, data-availability layers, keeper networks, sequencers, zero-knowledge proof committees, and interoperability hubs all fit the shape.

• Data availability: services such as EigenDA that store rollup transaction data cheaply and prove it was published.
• Oracles and price feeds: networks that report off-chain data on-chain and can be slashed for reporting falsely.
• Bridges and interoperability: cross-chain messaging layers that move value and must not forge messages.
• Coprocessors and proving: zero-knowledge or compute committees that run heavy work off-chain and attest to the result.

The chain of custody runs like this: a restaker delegates stake to an operator; the operator opts in to one or more AVSs and runs their software; the AVS pays rewards; and if the operator breaks an AVS’s rules, the restaker’s stake is exposed to that AVS’s slashing. In effect, restaking turns Ethereum’s security budget into something that can be sold, much the way Bitcoin mining pools sell hashpower to whoever needs a block built. The difference is that restaking is selling a promise of honesty, backed by the threat of a fine.

Two markets in one: the supply and demand of trust

Most restaking explainers describe one side of the trade: how you deposit, what token you get, how the yield accrues. But a security marketplace only functions if both sides show up. On one side sits the supply of security: stakers and operators offering their bonded capital for rent. On the other sits demand: services willing to pay real money for that capital’s guarantees.

The supply side arrived instantly. Restaking offered yield on top of yield, and capital never needs to be asked twice. The demand side is harder. An AVS only pays for pooled security when renting it is cheaper and better than the alternatives, which include issuing its own token, running a permissioned committee, or simply trusting a multisig. Plenty of services looked at that trade and decided the alternatives were good enough.

That imbalance is why the sector’s numbers can look so strange: billions of dollars locked, and yet, by the reading of protocol dashboards such as DefiLlama, very little fee revenue flowing back from the services that security is supposed to secure. Supply is not the constraint. Demand is.

Several designs now compete to close that gap, each with a different bet about what the demand side actually wants. Before we tour them, it helps to see how they line up.

The math of rented security: cost of corruption vs profit from corruption

To see why demand matters, it helps to understand how rented security is priced. EigenLayer’s own framing borrows two terms: the cost of corruption and the profit from corruption. The cost of corruption is how much an attacker would lose, in slashed stake, to break a service. The profit from corruption is how much they would gain by breaking it, say by feeding a lending market a false price and draining it. A service is economically secure only while the cost of corruption comfortably exceeds the profit from corruption.

Restaking’s pitch is that pooling stake raises the cost of corruption cheaply. Instead of a small new protocol scraping together, say, a few tens of millions of dollars of its own thin token to secure itself, it can tap a shared pool worth billions, so an attacker faces the sum of all that staked value rather than one illiquid coin. That is a genuine improvement, and it is why serious researchers took the idea seriously.

But the same pooling creates the sector’s signature danger. If one unit of stake backs many services at once, a single failure can be slashed across all of them, and losses can cascade. Academic work has tried to bound this: a 2024 paper by Naveen Durvasula and Tim Roughgarden, Robust Restaking Networks, shows that with enough overcollateralization the cascade stays contained, but only when services do not over-share the same stake.

EigenLayer’s answer is to isolate risk with what it calls operator sets and unique stake allocation, so that a given slice of stake can be slashed by only one service at a time rather than by all of them simultaneously. The economics, in other words, can be made to work. The open question is whether enough services want to pay for the guarantee in the first place.

The supply side: LRTs, operators, and an eager pile of capital

On the supply side, the innovation that made restaking explode was the liquid restaking token, or LRT. Locking stake into an AVS is illiquid and technical, so issuers began handing depositors a freely transferable token that represents the restaked position. Deposit ether, receive an LRT, and keep something tradeable that still earns restaking rewards.

ether.fi’s eETH became the largest of these by a wide margin, managing several billion dollars in deposits. Renzo’s ezETH, Puffer’s pufETH, and Kelp’s rsETH round out the field. Because these tokens are liquid, holders rarely stop at holding: they lend them, borrow against them, and pool them into automated market makers to squeeze out more yield, a practice known as rehypothecation.

This is where supply became almost too eager. Each layer of reuse, restake the ether, tokenize the position, lend the token, borrow against it, buy more, stacks another claim on the same underlying bond. It amplifies returns in calm markets and amplifies losses when something breaks, because a single depeg or exploit now ripples through every venue that accepted the token.

Operators are the other half of supply. They are the professional node runners who actually execute AVS software on behalf of delegators. A couple of thousand operators run on EigenLayer, but the work concentrates: running many AVSs well demands serious infrastructure, so a handful of large operators end up carrying most of the load, which quietly recentralizes the very security restaking set out to distribute. The table below shows how the three layers stack up.

The demand side: who buys pooled security, and what do they pay?

Now the hard side. Two years into the experiment, the uncomfortable truth is that most restaking yield has not come from services paying for security. It has come from token emissions, protocols printing their own tokens to reward depositors while the market for actual security demand catches up.

The clearest evidence sits on the fee line. Even as restaked deposits ran into the billions, protocol dashboards such as DefiLlama have shown restaking revenue running far behind the capital committed, at times close to zero. In plain terms: a great deal of stake is standing guard over services that are paying very little, if anything, for the watch. Yield that looks like a return on security is often just a subsidy paid in a freshly minted token.

Some real demand does exist. EigenDA, EigenLayer’s own data-availability service, sells blob space to rollups and competes with alternatives like Celestia and Ethereum’s native blobs. A growing set of AVSs secure oracles, bridges, coprocessors, and, increasingly, AI systems that need a way to prove their outputs were computed honestly rather than fabricated, an emerging demand source as autonomous AI agents move on-chain. But measured against the mountain of capital on the supply side, paying demand is still a trickle.

This is the reckoning behind the token charts. A restaking token can only hold its value over the long run if the network underneath it earns fees, not just prints incentives. Markets have started pricing that distinction aggressively, which is why tokens across the sector trade far below their launch valuations even as the technology matures. The security marketplace has abundant sellers; it is short on buyers.

EigenLayer becomes EigenCloud: a bet on verifiable AI

The sector’s original protocol has responded to the demand problem by widening its ambitions. EigenLayer rebranded to EigenCloud, repositioning from a pure restaking layer into what it calls a verifiable cloud: alongside EigenDA for data availability, it now pushes EigenCompute for off-chain computation and EigenVerify as a dispute layer, all backed by restaked security. The bet is that the demand restaking always lacked will come from developers who want cloud-style services with cryptographic guarantees, especially for AI.

To fix the emissions-versus-fees problem directly, EigenLayer’s governance passed ELIP-012 in December 2025, establishing a new emissions controller and an incentives committee, documented in the foundation’s public ELIP repository. The redesign steers token rewards toward productive stake, capital actively securing live, fee-generating services, rather than idle deposits, and routes a share of AVS rewards plus all net cloud fees into buying back EIGEN. It is an explicit attempt to make the token track real revenue instead of hype.

Whether it works depends entirely on that revenue materializing. Slashing has been live on EigenLayer’s mainnet since April 2025, the security model has been battle-tested, and venture backers put fresh money behind the EigenCloud pivot in 2025. But the market’s verdict, an EIGEN price down more than 96 percent from its peak per CoinGecko, reflects impatience: investors want to see fees, not roadmaps. The turn toward verifiable AI is the company’s answer to where those fees might finally come from.

Babylon: Bitcoin learns to earn without leaving Bitcoin

Ethereum is not the only chain with a security budget to rent. Babylon brought the same idea to Bitcoin, and did it without the wrapping and bridging that make Bitcoin DeFi so hazardous. Using Bitcoin’s own timelock scripting, a holder can lock BTC on the Bitcoin base chain itself and have that locked coin help secure proof-of-stake networks that opt in. No custodian holds the coins; no synthetic wrapped token crosses a bridge.

Babylon’s genesis mainnet and its BABY token both launched in April 2025. By the reading on its own dashboard, the system has attracted on the order of tens of thousands of BTC worth several billion dollars, which would make it the largest Bitcoin staking system by that measure (that figure updates slowly, so treat it as a recent snapshot rather than a live counter). BABY itself trades near $0.013, far below its April 2025 high per CoinGecko, echoing the same emissions-versus-fees skepticism that dogs EIGEN.

Babylon faces the identical demand question in a different accent: the supply of stakeable Bitcoin is effectively unlimited relative to the number of proof-of-stake chains willing to pay Bitcoin holders for security. Its longer-term thesis leans on integrations that would let native Bitcoin serve as collateral in lending, but as of mid-2026 the marquee example, a native-BTC vault for a major lending protocol, remained in testing rather than live on mainnet.

Symbiotic, Karak and OpenGDP: restaking without a token

Not every challenger believes a token is the answer, and one has decided restaking itself may not be. Symbiotic built a restaking layer that accepts almost any ERC-20 as collateral, not just staked ether, with slashing enabled from day one and dispute resolution left to each network’s chosen resolver. It has raised around $35 million and, notably, still has no public token, running instead on a points program while it builds. Its 2026 pivot, Core V2, reframes the product around collateral markets: committed capital that can back insurance, credit, and tokenized-asset liquidity rather than only middleware security.

Symbiotic co-founder Misha Putiatin has pitched that shift toward real-world assets bluntly, telling FinanceFeeds that “the RWA market has crossed $33 billion, but most of those assets still can’t be redeemed on demand,” the liquidity gap his product aims to fill. It is a telling move: a restaking protocol looking for demand outside restaking.

Karak went further still. The protocol rebranded to OpenGDP and repositioned around tokenizing real-world economic activity, dropping the restaking-and-shared-security language from its site almost entirely. Like Symbiotic, it never launched a liquid, widely traded token. The lesson these two draw is pointed: if a token trades down more than 90 percent while a tokenless rival keeps operating, maybe the token was never the product. Their retreat from the category is itself a data point about how thin demand for pure rented security turned out to be.

SSV and the based-applications alternative to restaking

A different school of thought argues that Ethereum should extend its security to new applications without the extra slashing layers restaking piles on. SSV Network, the largest distributed-validator-technology provider on Ethereum, splits a single validator key across several non-trusting operators so that no one machine holds it, which hardens staking against downtime and single-operator failure.

Its 2.0 design introduces based applications, or bApps, where developers can build services secured by Ethereum validators who opt in with participation keys. Crucially, a validator’s core 32-ether principal is never put at slashing risk; only optional, separately delegated capital can be slashed. The pitch is that this is a safer way to share security, an infinite-sum arrangement rather than the zero-sum gamble its founder Alon Muroch ascribes to conventional restaking, because it does not force validators to bet their principal on every new service. SSV has also moved its fee model to ether, so token holders earn staking-denominated rewards, an attempt, like EigenLayer’s, to tie token value to real network activity.

The market has been no kinder here. SSV trades near $2.10, down about 97 percent from its 2024 high per CoinGecko, with a market value around $31 million despite the network securing billions of dollars in staked ether. Once again, adoption of the underlying technology and the price of its token have moved in opposite directions, the through-line of the entire sector.

When the plumbing breaks: the Kelp and Aave contagion

If you want a single event that captures restaking’s hidden risks, look at April 2026. Attackers exploited a cross-chain bridge belonging to Kelp DAO, an LRT issuer, and minted roughly 116,500 rsETH out of thin air, about $292 million at the time, according to CoinDesk. They did not stop there. Because rsETH was accepted as collateral in DeFi lending markets, the attackers deposited the fraudulent tokens on Aave and borrowed against them, leaving the lender with around $196 million of bad debt and triggering a multi-billion-dollar drop in Aave’s total deposits over a single weekend.

The incident is the perfect illustration of rehypothecation risk. An LRT is only as sound as the bridge and accounting that back it, but once it circulates through lending markets and other venues, its failure is no longer contained; it becomes everyone’s problem. Aave founder Stani Kulechov, coordinating the response, said 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,” per CoinDesk, and he pledged personally to the recovery.

There was, at least, a functional ending. A coalition including Lido, ether.fi, Consensys, and Kulechov himself backstopped the shortfall, the fraudulent tokens were burned, and by mid-2026 Kelp reported rsETH fully backed again, with its bridge migrated to a more conservative design. The lesson stuck: the danger in restaking is rarely the base protocol; it is everything built on top of it that treats a restaked token as if it were plain cash.

Slashing, correlated risk, and the price of a single bug

Set the exotic contagion aside and the everyday risk of restaking is slashing itself. Every AVS a unit of stake backs adds another rule that can cost you money. Most losses will not come from an operator turning malicious; they will come from bugs, a faulty AVS contract, a misconfigured operator, an oracle that reports garbage, any of which can trigger penalties that flow straight back to delegators who never touched the code.

The nightmare scenario is correlated slashing: one shared dependency, a common client, a popular operator, a single library, failing in a way that gets punished across many services at once. This is the cascade the Robust Restaking Networks paper models, and it is why serious operators cap how much stake they re-pledge and why EigenLayer isolates slashing to one service at a time. Restaking multiplies both your yield and your simultaneous ways to lose.

An insurance market has grown up to price this. In 2026 the liquid restaking issuer ether.fi arranged what Decrypt called crypto’s largest ETH slashing cover, a Nexus Mutual policy protecting up to 15,000 ether against slashing losses. ether.fi founder Mike Silagadze and Nexus Mutual founder Hugh Karp both framed it as the kind of guardrail restaking needs to become institution-ready. The existence of the product cuts two ways: it shows the risk is real enough to insure, and it shows the sector maturing toward the safeguards traditional finance takes for granted.

Restaking and the SEC: is rented yield a security?

For United States readers, the regulatory picture has actually brightened. The Securities and Exchange Commission spent years treating staking-for-others as a likely securities offering, a stance capped by a settlement with Kraken over its staking-as-a-service product in 2023. The posture softened sharply in 2025. The SEC’s Division of Corporation Finance issued a staff statement that protocol staking, whether solo, delegated, or custodial, is not in itself a securities transaction, accompanied by Commissioner Hester Peirce’s memorably titled note Providing Security is not a Security. A follow-up in August 2025 extended that comfort to liquid staking and staking receipt tokens.

For restaking, this is helpful but not a clean bill of health. Plain staking rewards look increasingly safe from securities treatment; a packaged product that promises a fixed return, exercises discretion over the depositor’s assets, or markets pooled yield from third-party services sits on murkier ground. The more a restaking product resembles a managed yield fund rather than a passive protocol interaction, the more it risks tripping the securities line, and slashing complicates the picture further by putting principal at risk in ways a simple deposit does not.

The same agency that decides how staking rewards are treated also shapes how crypto ETFs get approved, and the two threads are converging: staking-enabled exchange-traded products have begun to launch, pulling regulated capital toward on-chain yield. If that capital ever flows from plain staking into restaking, it will be because the demand-and-revenue question at the heart of this article finally has a convincing answer.

Does the security marketplace clear? The 2026 verdict

So where does that leave restaking in 2026? The technology is no longer speculative. Slashing works, the risk models are published, insurance exists, regulators have drawn clearer lines, and multiple credible teams keep shipping. As a feat of mechanism design, restaking is real.

As a market, it has not yet cleared. Supply, staked ether, Bitcoin, and any-asset collateral, is effectively limitless and arrived years ago. Demand, services willing to pay meaningful fees for rented security rather than take it for free with a token subsidy, remains thin. That is the plain reading of prices down more than 90 percent across EIGEN, SSV, and BABY while the underlying deposits stay in the billions. Vitalik Buterin warned about exactly this tension early, writing in 2023 that “any expansion of the ‘duties’ of Ethereum’s consensus increases the costs, complexities and risks of running a validator,” per his blog; the sector spent two years testing how far to push those duties.

Even restaking’s architect concedes the modesty of the win. In a 2023 interview, Eigen Labs founder Sreeram Kannan told CoinDesk that “anything that restaking can do, already liquid staking can do, so I view restaking as a lesser risk than liquid staking,” framing restaking as an incremental improvement rather than a revolution.

The bull case is that verifiable cloud services and AI, which genuinely need cheap, credibly neutral security and are only now arriving, become the paying customers restaking always lacked. The bear case is that most applications keep choosing their own tokens or a trusted committee, and pooled security remains an elegant answer to a question few services were desperate to ask. The 2026 verdict is not failure; it is a not-yet. The security is built and waiting. Whether the demand ever shows up in size is the only question that still matters.

Frequently Asked Questions

What is restaking in simple terms?

Restaking lets capital that already secures one blockchain be pledged a second time to secure other services, in exchange for extra rewards. On Ethereum, you take staked ether (or a liquid token representing it) and commit it to also back things like oracles, bridges, or data-availability layers. The trade-off is extra reward for extra slashing risk if those services misbehave.

What are the risks of restaking?

The main risk is slashing: every service your stake backs adds another rule that can cost you money, and losses often come from bugs rather than attacks. The most dangerous version is correlated slashing, where one shared failure is punished across many services at once. Liquid restaking tokens add rehypothecation risk, as the April 2026 Kelp and Aave incident showed, when a bridge exploit spread into lending markets.

How much can you earn from restaking?

Historically most restaking yield has come from token emissions rather than fees paid by the services being secured, so headline rates can overstate durable income. Base Ethereum staking pays a few percent a year, and restaking layers additional rewards on top, but much of that has been subsidy. As token incentives taper, real yield depends on whether services actually pay for security, which remains the sector’s open question.

What is the difference between staking and restaking?

Plain staking bonds ether to secure Ethereum itself and is slashable only by Ethereum’s rules. Restaking reuses that same bonded stake to secure additional services, so it earns extra rewards but is also exposed to each of those services’ slashing conditions. Liquid restaking tokens go one step further by making the restaked position tradeable.

Is restaking legal in the United States?

Interacting with a decentralized restaking protocol is not banned, and in 2025 the SEC’s staff said protocol staking is not in itself a securities transaction, later extending that view to liquid staking tokens. However, packaged products that promise fixed returns or manage users’ assets can still raise securities questions, so legal treatment depends on how a specific product is structured rather than on restaking as a concept.

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