# Restaking after the hype: what is left of EigenLayer?

> Restaking secures $10 billion but barely generates fees. ether.fi abandons the business that made it famous and EigenLayer markets itself as EigenCloud.
> By José Maldonado · DeFi · 2026-10-19

Restaking promised that the same ETH staked on Ethereum would secure other services and generate a second income. In September 2026 that second income arrived all but empty. The category secured some $10.02 billion and produced $99,977 in fees over a week, according to [DefiLlama's tally](https://defillama.com/protocols/restaking), against $27.35 million for liquid staking with $51.87 billion locked.

That same month, the largest liquid restaking protocol announced it was leaving the business that made it known. ether.fi will thus cut this quarter its last structural link to EigenLayer, keeps less than 1% of its assets in restaking and its CEO, Mike Silagadze, explained it bluntly: no relevant yield and a perceived risk. Its weETH token, the most widely accepted as collateral in DeFi, stopped reusing ETH in August.

The episode forces a separation of two matters that the 2024 hype conflated. One is shared security infrastructure, which keeps operating and serves real networks. The other is the extra income promised to depositors, sustained by points and airdrops and never covered by demand for security. With the October 2026 data in front of us, that premium is measured in hundredths of a point and the business that sold it fits into a handful of protocols.

## A boom measured in deposits, not revenue

The trigger of the cycle was EigenLayer. At its peak it hovered around $19.7 billion and liquid restaking tokens grew more than 1,000% in the first six weeks of 2024. The mechanics were addictive: the base staking yield was kept, fees from validated services, the AVSs, were added, and it was rounded off with points redeemable for future tokens.

![EigenLayer keeps falling at its economic level as a protocol](https://content.observatorioblockchain.com/wp-content/uploads/2026/10/eigenlayer-now.png)
*EigenLayer keeps falling at its economic level as a protocol*

Deposits came in at a pace that none of those three sources could justify on its own. The market paid in advance for an income nobody had contracted yet, and that mismatch took two years to take its toll.

The rental logic was sound: a new service that does not want to build its own validator network rents Ethereum's and pays fees for it. Those fees were what should have reached the restaker, but points and emissions filled that gap while the market grew.

## The bill for a category with billions inside

The September figures are eloquent: $10.02 billion secured and $99,977 in fees over seven days, a ridiculous number for the capital retained. Liquid staking was then generating $27.35 million with $51.87 billion of capital, some 53 times more per dollar secured.

On October 16, the DefiLlama API returned $9.93 billion across 14 restaking protocols and $1.28 billion across 29 liquid restaking ones, with $258,133 and $1,107,925 weekly; liquid staking generated $28.43 million, more than ten times the sum of both.

The protocol-by-protocol detail, consulted on October 16, confirms the concentration: almost all the capital is in two names and fees do not follow size, with Kelp as the only partial exception thanks to its strategies.

Protocol

TVL ($ million)

7-day fees ($)

EigenCloud

6.285

233.467

Babylon Protocol

3.345

n/a

Kelp

990

536.728

Renzo

113

47.579

Mellow

24

429.855

The five largest liquid restaking tokens still standing added $953,350 in gross profit in the second quarter of 2026, against $2.18 million three quarters earlier. Puffer, which raised $23 million, booked $21,590 in the quarter and Swell stayed at $22,370.

In Kelp's books, rewards in EIGEN appear as $460,600 of revenue and as much again in cost: the token comes in and goes out to depositors without leaving any margin for the protocol.

## The premium that stayed in hundredths of a point

On September 27, 2026, the weETH token paid 2.29% a year while stETH, with no restaking at all, paid 2.19%: a single tenth of a point of difference. Between 95% and 100% of the measured yield of a liquid restaking token came from ordinary Ethereum staking.

The October 16 data keeps the pattern: stETH at 2.25% and weETH at 2.22%, with rsETH and ezETH moving according to the operator and the moment, with no stable premium to claim.

Any extra compensation arrived in tokens of the secured service itself or in EIGEN, whose value moves with the market and not with the service's revenue. A young AVS pays with its own token because it has little income in ETH or stablecoins, so the restaker takes on a new risk and receives an illiquid asset.

When that token falls, the income disappears from the spreadsheet. The second layer was never a solid cash flow, but a speculative position disguised as yield for the depositor.

## ether.fi stops being a restaking company

With that arithmetic, the exit of [ether.fi](https://www.ether.fi/) stops being a surprise. The firm removed restaking from weETH in August 2026 and anyone who wants it must opt for a separate token built on Symbiotic, the rival platform.

The business that sustains the company is another one: a card to spend without selling crypto, a lending market and treasury products. The card went from 17% of monthly revenue in January to 46% in July, and Silagadze argues that this line has already replaced what was lost in restaking and in the price of ETH.

His estimate is that neobanking moves some $300 billion a year, about three hundred times what DeFi generates, and that the company will raise its revenue pace by 38% this year while staking and restaking fall 70%.

It has not published the basis for that 38%, and DefiLlama shows gross profit down 47%, from $18.71 million in the third quarter of 2025 to $9.99 million in the second of 2026. Both things can be true at once: a future pace is not a closed quarter and revenue is not a margin.

The awkward fact is that restaking was still profitable when it decided to leave: in the second quarter, the card left $3.14 million and restaking on EigenLayer, $2.87 million, ahead of basic staking and the rest of the lines. It was not a ruinous business, but a small one with a risk the company preferred to stop carrying.

## The risk stopped being theoretical

Two changes closed the cycle. Points programs ran dry over the course of 2025 and slashing was activated on mainnet in April of that year: since then, an operator that goes down or signs contradictory messages can lose part of the ETH deposited. The promise required a premium to cover that risk, and that premium did not appear in the books.

In April 2026, an attacker drained 116,500 rsETH, some $292 million, from the bridge Kelp operated with LayerZero. The [incident report](https://layerzero.network/blog/layerzero-labs-kelpdao-incident-report) describes a single-verifier setup: the attacker poisoned two internal RPC nodes after tricking a developer in March and denied service to the external provider, forcing a valid attestation for a false message.

Attribution points to UNC4899, the North Korean group known as TraderTraitor. The theft did not happen in EigenLayer or in the token contract, but in the wrapper that any restaker accepted without looking too closely.

To that counterparty risk is added correlation risk, the hardest to explain. The same ETH can be exposed at once to the rules of several services, so that a failure in a small network punishes capital that sustains other infrastructure. It multiplies exposure rather than diversifying it, and that only pays off if the extra income is real.

## EigenLayer no longer sells restaking

The protocol itself changed its shop window. Eigen Labs launched EigenCloud in June 2025, with a $70 million purchase of EIGEN by a16z crypto, and today it sells verifiable computing instead of yield.

These are applications that prove that work done off-chain was executed properly, with restaking collateral as the bottom layer rather than as the product, and that is the difference from the 2024 shop window. Its holdings hover around $5.1 billion, far from the $22.06 billion of August 2025.

The second largest on the list is no longer an Ethereum derivative: Babylon accumulates $3.345 billion in restaking on Bitcoin, and Symbiotic, which activated slashing in January 2025, kept $480 million in September after a peak of $2.7 billion in December 2024.

What still works has a name of its own: EigenDA, the data availability service for rollups, operates on mainnet at 100 MB/s and remains the largest AVS by value secured, and Symbiotic has integrated more than 50 networks that pay for specific functions without selling themselves as a gateway to superior yield.

## What remains when the hype passes

Mature restaking looks less like a source of income and more like concentrated infrastructure: EigenCloud gathers almost two thirds of the capital and the five largest protocols corner around 98%.

The metric that will matter from now on will not be TVL, but revenue per unit of capital. It will take time to know whether the reinvention is enough to sustain a sector that was built on an income that never arrived.
