Restaking in 2026: The Premium Melted at a 4% Risk-Free Rate

Restaking was the loudest idea in decentralized finance for two years running. The pitch was simple and seductive: you have already locked ETH to help secure Ethereum, so why let that capital do one job when it could do several? Re-pledge it to other networks, the argument went, and collect extra yield on top of your base staking reward. At the peak, tens of billions of dollars chased that promise.

By October 2026 the promise has quietly deflated, and the cause is not a dramatic exploit or a regulatory crackdown. It is arithmetic. On 16 September the Federal Reserve raised interest rates for the first time since 2023, pushing the target range to 3.75 to 4.00 percent. A three-month US Treasury bill now yields about 4.00 percent, and on-chain money-market tokens pay nearly as much with no risk of being slashed. Against that backdrop the one or two extra percentage points restaking offered, most of it paid in volatile tokens rather than cash, stopped being worth the added risk. This is a full explainer of how restaking works, where the yield really came from, and why the market, led by the sector’s own flagship, has been walking away.

What restaking actually is

Start with plain staking. To help secure Ethereum you lock 32 ETH, run a validator, and earn a reward for proposing and attesting to blocks. That reward is the base layer of every yield discussed here. Restaking, a term coined by EigenLayer, lets you take that same staked ETH (or a liquid token that represents it) and re-pledge it to secure additional services that are not Ethereum itself. Those services are called Actively Validated Services, or AVS: oracle networks, cross-chain bridges, data-availability layers, rollup sequencers, and co-processors that need their own economic security but cannot easily bootstrap a validator set of their own.

The idea arrived with enormous momentum. Restaking was a defining DeFi narrative of 2024 and 2025, and at the peak EigenLayer’s value locked reached roughly 22 billion dollars in August 2025 before the long slide, with the broader liquid restaking sector having crested even earlier. Understanding why that capital is now leaving means understanding what it was promised, and what it was actually paid.

In exchange for renting out your security you earn extra rewards and accept extra slashing conditions: if the service you back misbehaves, your capital can be penalized on top of Ethereum’s own rules. The elegant version of the story is that Ethereum’s enormous pool of staked ETH is idle collateral, and restaking turns it into a marketplace where new networks buy security by the slice. The less elegant version was flagged early by Ethereum’s co-founder.

In a May 2023 essay titled Don’t overload Ethereum’s consensus, Vitalik Buterin warned that “any expansion of the ‘duties’ of Ethereum’s consensus increases the costs, complexities and risks of running a validator,” and cautioned against designs that could pressure the community into bailing out a failed service. Sreeram Kannan, the founder of EigenLayer, pushed back in a September 2023 CoinDesk interview, arguing that “anything that restaking can do, already liquid staking can do, so I view restaking as a lesser risk than liquid staking.” Three years on, both men have a claim to being right: the risk was real, and the demand that was supposed to justify it never fully arrived.

The yield stack: base, liquid, restaked

To see why the premium mattered, line up the layers of Ethereum yield from safest to riskiest. Each layer adds return, but it also adds something new that can go wrong. The base reward is the block reward a validator earns, the same economic engine that, in proof-of-work, pays a miner instead of a validator. Liquid staking wraps that staked position in a tradeable token such as stETH or weETH so the capital stays usable in DeFi. Restaking sits on top, adding the AVS premium and the slashing exposure that comes with it. A liquid restaking token, or LRT, then wraps the restaked position so it, too, can be reused as collateral.

As of early October the base staking reward sits near 2.63 percent, with roughly 43.7 million ETH staked across about 871,000 validators, just under 36 percent of supply. That base number is the foundation every higher layer builds on, and it is denominated in ETH, not dollars, a distinction that becomes decisive later.

AVS, or who is supposed to pay for rented security

Everything in restaking rests on one assumption: that AVS will pay real money to rent Ethereum’s security. On the supply side, the capital showed up in force. EigenLayer crossed five million ETH in deposits in September, spread across roughly eighteen active security networks, and a cohort of well-funded AVS launched: EigenDA for data availability, Lagrange for zero-knowledge proofs, Omni for interoperability, and others.

The demand side never matched it. Paying customers stayed scarce, and the fees they generated were tiny next to the capital waiting to be rented. One blunt comparison from late September captured the gap: the restaking category held about 10 billion dollars of capital but produced only 99,977 dollars in fees in a single week, while plain liquid staking’s 51.87 billion dollars generated 27.35 million dollars, meaning ordinary staking earned roughly 53 times more per dollar secured. Usage told the same story: daily active addresses touching the main restaking contracts fell below 100 across August and September, down from tens of thousands at the spring peak. The marketplace for rented security had plenty of guards for hire and almost no one booking them.

The imbalance has a structural cause, not just a cyclical one. For an AVS, rented security is a cost line, not a product, so each one buys only as much as it truly needs, and many discovered they needed far less than restaking’s boosters assumed. A data-availability layer or an oracle can often bootstrap adequate security from its own token or a smaller trusted set, and only a handful of services have the transaction volume to justify paying real fees for Ethereum-grade security. Supply was mobilized like a commodity; demand behaved like a trickle of bespoke contracts.

The number that broke the pitch

Restaking’s yield was never judged in a vacuum. Every investor compares it to the next-best safe return, the opportunity cost of capital, and in 2026 that comparison moved sharply against it. On 16 September the Federal Reserve, under Chair Kevin Warsh, raised its policy rate by a quarter point to a range of 3.75 to 4.00 percent in a unanimous vote, the first hike since 2023. “The plain fact is that inflation is too high and has been for too long,” Warsh said, adding that the move would “support a timelier return to the committee’s 2 percent goal.” Markets put the odds of a further hike at the 28 October meeting near a coin flip, with most forecasters expecting at least one more before year-end.

That matters because it reset the hurdle every yield must clear. When cash earns almost nothing, a thin ETH-denominated premium on top of staking looks attractive. When a Treasury bill pays 4 percent with no credit risk and no slashing, an extra point or two of restaking yield, carrying real tail risk and often paid in a falling token, no longer clears the bar. The same macro turn that put Bitcoin back in the headlines this October quietly raised the cost of capital under every crypto yield product, and restaking, which always lived or died on a slim spread, was the most exposed of all.

On-chain, risk-free, and composable

A defender of restaking could once argue that Treasury bills are beside the point: they are off-chain, awkward for a crypto user to hold, and useless as DeFi collateral. That argument has collapsed, because the risk-free rate is now on-chain too. Tokenized US Treasury products hold about 14.77 billion dollars and pay a roughly 3.6 percent average yield across more than 100 products, with BlackRock’s BUIDL, Ondo’s USDY and OUSG, Circle’s USYC and Franklin Templeton’s BENJI leading the field. The Sky Savings Rate pays about 3.60 percent on its sUSDS token, and accredited wrappers such as Ondo’s OUSG have paid close to 4.8 percent. Every one of these tokens is liquid and composable, the very properties that made liquid restaking tokens attractive, except they carry no slashing risk at all.

Put the instruments side by side and the problem becomes visual. The safest options now sit at the top of the yield range, and the riskiest option, restaking, sits near the bottom once you strip out token incentives.

Composability was the one structural edge restaking tokens held over plain staking: you could earn the yield and still post the token as collateral elsewhere, putting the same capital to work twice. Tokenized Treasuries now offer that same trick. An investor can hold BUIDL or OUSG, earn a dollar yield anchored to the front of the Treasury curve, and borrow against it in the same DeFi venues that accept a liquid restaking token. Once the safe instrument is also the composable one, the restaking token’s remaining advantage is mostly the extra risk it carries.

Read down that table and the logic of the whole year falls out of it. Why accept stacked slashing exposure for a yield that, measured in honest fees, barely registers, when a tokenized Treasury pays more, in dollars, with none of the downside?

Where the yield actually came from: points, not fees

If AVS fees were negligible, where did all that advertised restaking yield come from? Mostly from points, the off-chain loyalty tallies protocols handed out in anticipation of a future token airdrop. Depositors farmed EigenLayer points, then liquid restaking tokens layered their own points on top, and the headline annual rates circulating in 2024 and 2025 were in large part a bet on the value of tokens that had not yet traded. A point is, in effect, a loan against a token’s future price, which is why this is the same pattern that makes most new token listings lose money: when the token finally prints and falls, the yield it implied evaporates.

The mechanism was reflexive, which is why it unwound so fast. Points attracted deposits, deposits lifted expectations for the eventual token, a richer expected token value justified a higher implied yield, and that higher yield pulled in still more deposits. Every step depended on the token being worth more later. When the tokens finally traded and drifted lower, the same loop ran in reverse: the implied yield fell, deposits looked for the exit, and there was no stream of real fees underneath to arrest the slide.

Fall it did. EIGEN, EigenLayer’s token, trades near 0.26 dollars, down about 95 percent from its December 2024 peak of 5.65 dollars, with a market cap around 256 million dollars after a scheduled monthly unlock on 1 October lifted circulating supply past 970 million. The project’s ELIP-012 overhaul routes a share of AVS fees and all EigenCloud fees into buying back EIGEN, but with protocol revenue this thin the buyback is far too small to offset the unlocks. The token-funded model shows up starkly in the profit numbers: the five largest remaining liquid restaking tokens, Renzo, Kelp, Swell, Puffer and Bedrock, together earned gross profit of just 953,350 dollars in the second quarter, down from 2.18 million dollars three quarters earlier.

The market’s verdict: ether.fi walks away

The clearest signal that the premium had melted came not from a critic but from the company that embodied restaking. ether.fi was the largest liquid restaking token issuer, a business built on re-pledging staked ETH to EigenLayer. In August it stripped restaking out of weETH, its flagship token, reverting it to a plain liquid staking token, and moved restaking into a separate, opt-in token called weETHs, built on Symbiotic rather than EigenLayer. By late September the firm confirmed it would sever its last structural ties to EigenLayer this quarter, with less than 1 percent of its assets still restaked.

Chief executive Mike Silagadze was blunt about why. “There were no meaningful yield opportunities in restaking and there was some perceived risk from stakers, so we decided it made sense to exit,” he told CoinDesk. Weeks earlier, as the unwinding began, he had struck a more wistful note, telling The Defiant it was “end of an era” and that he still believed “restaking will come back in one form or another, I think it was just a bit too early.”

Tellingly, ether.fi has not suffered for walking away. Its ETHFI token trades near 0.73 dollars with protocol value locked around 5.18 billion dollars, holding up far better than the restaking tokens because the company pivoted into consumer finance: debit cards, portfolio loans and payments. Silagadze says neobank revenue has “fully replaced the revenue lost from restaking and lower ETH price,” with card fees climbing from 17 percent of monthly revenue in January to 46 percent by July. The brand that once personified restaking now earns its keep selling banking, and the market has rewarded the switch.

Slashing: the risk you are no longer paid for

Restaking’s defenders always conceded it carried more risk; the premium was supposed to compensate for it. EigenLayer turned on slashing on mainnet in April 2025, which means restaked capital can now be penalized if an AVS or its operators break the rules. Stack several AVS on the same collateral and you stack their slashing conditions too, and because operators and collateral overlap heavily across services, those risks are correlated rather than independent. One bad event can ripple.

Correlation is the part that is easy to underprice. Because the same large operators run many services and the same collateral backs several of them, a single bug in a widely used AVS, or one operator’s misconfiguration, can trigger penalties across a swath of restakers at once rather than in isolation. That is close to the scenario Buterin worried about in 2023: a failure large enough that the wider community faces pressure to intervene. A premium that barely exists cannot pay for a tail that correlated.

The problem in 2026 is not that the risk is unusually high; it is that the compensation disappeared. When the AVS premium is mostly token emissions and the real fee yield rounds to zero, a restaker is accepting correlated tail risk in return for a coupon paid in a token that has fallen 95 percent. That is the opposite of how risk pricing is supposed to work. Rational capital demands more yield for more risk, and restaking in late 2026 offered less yield for more risk, a combination with only one natural outcome.

Rehypothecation and the Kelp lesson

Liquid restaking tokens made the risk worse in a subtle way. Because an LRT is itself a tradeable token, it gets reused as collateral across other protocols, borrowed against, looped and rehypothecated, so a single underlying deposit can support several layers of leverage. That is efficient when everything works and dangerous when one link breaks.

The sector learned this in April 2026, when an attacker exploited a cross-chain bridge to mint about 116,500 rsETH out of thin air, roughly 292 million dollars of Kelp DAO’s liquid restaking token, then deposited it as collateral on Aave to borrow real assets, leaving close to 196 million dollars of bad debt and knocking billions off Aave’s value locked over a single weekend. The funds were eventually recovered through a coordinated effort, and the fault lay in the bridge configuration rather than the base protocols, but the lesson stuck: an LRT yield came bundled with the security of every bridge and contract it touched. That is exactly the kind of cross-chain weak point that has drained more value from crypto than almost any single smart-contract bug. Paying for that hidden risk made sense when the yield was rich; it makes none when the yield is gone.

The platforms, and who survived the repricing

Restaking is not one protocol but several competing designs, and the repricing has sorted them in an ironic way. EigenLayer, the pioneer, accepts ETH and liquid staking tokens, issues the EIGEN token, and still holds the most capital, around 5.1 billion dollars after falling from a 22 billion dollar peak in August 2025. Symbiotic takes almost any ERC-20 as collateral, has deliberately never launched a public token, and in July 2026 pivoted with its Core V2 release toward broader collateral markets, routing idle vault capital into lending venues like Aave and Morpho; it is where ether.fi’s opt-in restaking now lives. Karak is the smaller third player. Babylon takes a different base asset entirely, letting Bitcoin holders lock coins through Bitcoin’s own timelock scripts to secure proof-of-stake chains, with no wrapping or bridging.

The pattern in that table is the punchline of the year. The projects that avoided a speculative token, Symbiotic with no token at all and Babylon anchored to Bitcoin rather than its own emissions, have held together better than the token-rich pioneer. Babylon’s BABY token is down about 92 percent from its high and EIGEN about 95 percent, but Babylon’s Bitcoin staking base and Symbiotic’s collateral-markets pivot each gave them somewhere to go that did not depend on paying yield in a falling coin.

Babylon is the most interesting outlier because it never touched the Ethereum yield stack at all. It lets Bitcoin holders commit coins through Bitcoin’s native timelock scripts, so the asset never leaves Bitcoin and is never wrapped, and that locked BTC can then secure proof-of-stake chains that opt in. A planned integration would let this native Bitcoin serve as collateral in an Aave V4 vault, though that work remains in testing. It is a useful reminder that restaking the idea is not the same as restaking the Ethereum token economy that soured around it.

EigenCloud’s bet: manufacture the demand

If paying tenants will not come to rent security, one answer is to build the tenants yourself. That is the logic behind EigenLayer’s rebrand to EigenCloud and its push into verifiable cloud services: EigenDA for data availability, plus EigenCompute and EigenVerify for off-chain computation and dispute resolution, much of it aimed at artificial-intelligence workloads that need their outputs proven rather than trusted. Andreessen Horowitz backed the pivot with a 70 million dollar token purchase, and Kannan has framed the goal as making virtually anything verifiable on-chain, closing the gap between what developers want to build and what blockchains currently allow.

The strategic read is candid: restaking could not find enough external customers, so EigenCloud is trying to become its own biggest customer, generating real fees from compute and AI rather than waiting for AVS demand that never scaled. It is a plausible second act, and it rhymes with a broader 2026 theme of crypto networks chasing verifiable AI and decentralized compute as a source of genuine revenue. Whether it generates enough fees to matter to EIGEN holders is still an open question; for now the buyback it funds is a trickle against a flood of unlocks.

What US regulators say about staking yield

Regulation did not kill restaking, but it shapes how the yield can be offered in the United States. In May 2025 the Securities and Exchange Commission’s Division of Corporation Finance issued a statement on protocol staking describing it as an administrative or ministerial activity rather than a securities transaction, and in August 2025 it extended similar comfort to certain liquid staking activities, provided the provider stays administrative and does not guarantee or set returns.

The catch sits in the fine print. The protocol-staking carve-out explicitly excludes arrangements whose tokens carry “intrinsic economic properties or rights, such as generating a passive yield,” which is a fair description of a restaking or liquid restaking token sold on its promised rate. Those are staff statements, not rules, and a future commission can reverse them, a live concern while the agency operates with a thin roster of commissioners and its longer-term crypto framework stays unsettled in Congress. For now, pure protocol participation sits on firmer ground than a yield-bearing restaking product marketed to the public.

Would restaking come back?

Silagadze’s “a bit too early” is worth taking seriously, because the thesis was never nonsense; the timing and the token incentives were. Three things would change the math. The first is the rate environment: if the Fed reverses course and cash yields fall back toward zero, a thin ETH premium becomes attractive again, though the 28 October meeting points the other way for now. The second is real AVS demand: a single application that genuinely needs rented security and pays real fees for it would do more for the sector than any points campaign. The third is EigenCloud’s compute bet actually generating revenue, which would give the model a cash flow that does not depend on token emissions.

There is also a quieter path that needs none of those: time. Token unlocks eventually finish, the weakest projects close, and the capital that remains is the capital that genuinely wants the risk. A smaller restaking sector funded by real AVS fees, priced honestly against the risk-free rate, could be a healthier business than the points-fueled version that peaked in 2024. Smaller, in this case, might mean truer.

Until one of those lands, the honest framing is that restaking has been repriced, not disproven. The infrastructure works, slashing is live, and billions in capital still sit ready. What vanished was the premium, and with it the reason for most holders to take the risk.

The bottom line for holders

If you are weighing a restaking or liquid restaking position in late 2026, three questions cut through the marketing. First, is the yield real fees or token emissions? A rate quoted in points is a bet on an unlaunched or falling token, not income. Second, is it denominated in ETH or in dollars? A 4 percent ETH yield is not the same as a 4 percent dollar yield if you measure your wealth in dollars and ETH is volatile. Third, does the all-in yield clear the roughly 4 percent risk-free rate after you price the slashing, smart-contract and bridge risk you are taking on?

For most holders in this environment, the answer to that last question is no, which is precisely the conclusion the market reached on its own. Restaking turned staked ETH into a security marketplace with real technology behind it. It just launched into a world that, a few rate hikes later, could buy a safer yield with one click. The premium did not get stolen or banned. It melted.

Frequently Asked Questions

What is restaking in simple terms?

Restaking lets you take ETH you have already staked to secure Ethereum and re-pledge it to secure additional networks, called Actively Validated Services, such as oracles, bridges and data-availability layers. In return you earn extra rewards and accept extra penalties if those services misbehave. EigenLayer introduced the idea, and Symbiotic, Karak and Babylon offer their own versions.

Is restaking still worth it in 2026?

For most holders, the risk-adjusted answer is no. With US Treasury bills paying about 4 percent and on-chain money-market tokens paying close to it with no slashing risk, restaking’s extra yield of roughly one to two points, much of it in volatile tokens, no longer compensates for the added risk. The sector’s largest issuer, ether.fi, is exiting restaking for exactly this reason.

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

Staking locks ETH to secure Ethereum and earns a base reward near 2.6 percent. Liquid staking gives you a tradeable token, such as stETH or weETH, that represents the staked position so your capital stays usable. Restaking goes one step further, re-pledging that staked ETH to secure other networks for extra yield and extra slashing risk.

Why did EIGEN and other restaking tokens fall so much?

Much of restaking’s advertised yield came from points that anticipated token airdrops rather than from real fees. When the tokens launched and the networks they secured generated little revenue, the implied yield collapsed and the tokens fell with it. EIGEN is down about 95 percent from its 2024 peak, and the five largest liquid restaking tokens earned under one million dollars of combined gross profit in a recent quarter.

What are the main risks of restaking?

The main risks are stacked slashing, where backing several services multiplies the ways your capital can be penalized; smart-contract and bridge failures, as in the April 2026 Kelp exploit that created about 292 million dollars of bad debt; and token risk, since much of the yield is paid in assets that can fall sharply. Rehypothecation, reusing the same token as collateral across many protocols, can amplify all of these.

By Yuki Tanaka, HOGE Wire staff writer covering DeFi and on-chain markets.