ether.fi announced that it will lift the final structural binding with EigenLayer and withdraw from the re-staking track. Data shows that re-staking returns are much lower than ordinary staking, and hacking incidents have exposed additional risks. The project turned to new banking services such as encrypted consumer cards and loans to achieve income diversification.
Author: Oliver Knight, coindesk
Compiled by: Chopper, Foresight News
At the end of the quarter, ether.fi will remove the final structural binding between its staking tokens and the re-staking protocol EigenLayer. EigenPod withdrawal certificates will be removed before the end of the year, and by then less than 1% of assets will still be participating in re-pledge.
When ether.fi is launched in 2024, the ETH deposited by users will automatically be re-pledged on EigenLayer. In August this year, the project stripped the re-staking function from weETH. weETH is ether.fi’s most liquid pledged token and is widely used as collateral in DeFi. After being stripped, it becomes an ordinary liquid pledged token. If users still want to participate in re-staking, they must separately choose another token deployed on the competing platform Symbiotic.
Mike Silagadze, CEO of ether.fi, said in an interview with CoinDesk that this decision was based on risk considerations: "There are no considerable profit opportunities in the re-staking track, and staking users also perceive risks, so we chose to withdraw."
The initial vision of re-pledge
Staking is locking ETH to ensure the security of the Ethereum network and earn income on it. The idea behind re-staking is that the same locked ETH can take on a second job: EigenLayer will rent out this network security capability to other services such as oracles and data availability layers, and these services will pay for it. Depositors can get double returns on the same funds. Liquidity re-pledged tokens are built on this mechanism. Depositors will receive tradable certificates without having to lock up and wait. The tokens can be sold or used as collateral elsewhere. weETH is the largest token among them.
EigenLayer's locked assets peaked at $19.7 billion, and the market value of liquid re-pledged tokens increased more than 10 times in the first six weeks of 2024. However, the fee paid by the demander for purchasing security services has never been able to cover the basic pledge income and additional premium, and the second layer of income promised by re-pledge has never been realized.
On September 8, DefiLlama data showed that the total locked-up volume of the re-pledge sector was US$10.02 billion, and only US$99,977 in handling fees were generated in the past week; the liquidity pledge sector locked up US$51.87 billion, and the weekly handling fees were as high as US$27.35 million. Converted, for the same size of assets, the income generated by ordinary pledging is approximately 53 times that of re-pledge.
There is almost no profit from the re-pledge agreement
Two things have further reduced the motivation to participate in re-pledge: the points incentive program to subsidize deposits will gradually end in 2025; in April 2025, the slashing mechanism will be officially launched. Penalty means that when a validator goes offline, signs conflicting messages and other violations, the protocol will confiscate part of their pledged ETH as a penalty. This means that re-pledge has gone from being only a theoretical risk in the past to actually causing losses, but there is no additional income to compensate for the risk.
Leaving aside ether.fi, the other projects on the track are all relatively small. The remaining five leading liquid re-pledge protocols, Renzo, Kelp, Swell, Puffer Finance, and Bedrock, had a total gross profit of only US$953,300 in the second quarter of 2026. Three quarters ago, the combined gross profit of these five companies was still US$2.18 million. Puffer, which raised $23 million in financing, had a gross profit of only $21,590 in the quarter; Swell’s was $22,370.
DefiLlama data also reveals that the truly profitable part of these businesses has never been the re-staking itself. Taking Kelp as an example, the EIGEN token reward book records a gross income of US$460,600, and the cost of revenue is also US$460,600 - the rewards are fully distributed to depositors, and no penny is retained by the agreement. The same goes for Puffer and Swell. All profits from these protocols come from the ordinary staking fees at the bottom of the re-pledge.
Hacking incident
On April 18, 2026, an attacker exploited the KelpDAO cross-chain bridge vulnerability to mint 116,500 rsETH out of thin air in 46 minutes, worth approximately US$293 million, without any ETH asset endorsement behind it. The attacker deposits counterfeit tokens into Aave as collateral and lends out real ETH. About $6 billion was removed from Aave in the days that followed, with potential bad debt ranging from $123 million to $230 million. In May, Aave announced a comprehensive reform of collateral and currency listing standards, and simultaneously reviewed the project’s network security, technical architecture, and price fluctuation risks.
Silagadze does not view the incident as a failure of the leverage mechanism. He said: "The root cause of Kelp's theft is a security flaw in the cross-chain bridge, which has nothing to do with leverage. Ether.fi's market parameter settings in Aave are extremely conservative, and we attach great importance to security."
Objectively speaking, EigenLayer itself did not malfunction, no forfeiture occurred, and the core re-staking mechanism did not collapse. The weak point was the cross-chain bridge. But this is precisely the reason why this hacking incident severely damaged the liquid re-staking track: the loss of funds occurred in the packaging token (that is, the tradable certificate) on the upper layer of the re-staking, rather than the underlying protocol of the re-staking. By April, holding such tokens is equivalent to bearing an additional layer of risk of smart contract being attacked, but without corresponding additional benefits. The income from wrapping tokens is no longer enough to cover its own risks.
Where did the money go?
The capital withdrawn from the re-hypothecation track did not leave the crypto lending market, but simply moved from ETH assets to USD assets.
The mainstream gameplay in 2024 is: pledge ETH, re-pledge, encapsulate into liquid re-pledge tokens, mortgage loans, and add positions, thereby superimposing exposure to a single asset. By 2026, similar treasury operations are moved to select vaults. Selected vaults are a type of lending pool where an external institution (treasury manager), rather than the lending agreement itself, determines which assets the pool accepts and what terms it sets, and the manager divides the funds accordingly. Morpho is the largest platform for this type of business, with a locked position of approximately US$5.8 billion.
Similar to the re-pledge logic, depositors only get a certificate, the risk is set by a third party, and the certificate can be accepted as collateral by another platform.
Select vaults have also experienced Kelp-like crashes. On November 4, 2025, Stream Finance disclosed losses of approximately $93 million and suspended withdrawals. Its yield-based U.S. dollar stablecoin, xUSD, is designed to be anchored to $1, and its price plummeted 77% in a single day. The treasury manager builds a vault in Morpho, and depositors deposit real stablecoins in exchange for xUSD, and borrow stablecoins to continue to increase their holdings of xUSD, artificially pushing up the token price and deviating from actual asset support. The market has been valuing xUSD at a fixed $1, rather than the true market price, so when the price plummeted, the automatic liquidation that was supposed to be triggered did not take effect. According to subsequent research statistics, the debt exposure related to the lending platform is approximately US$285 million. Another USD token (65% backed by Stream’s loans) plummeted around 98% before being liquidated.
ether.fi’s withdrawal from re-staking also raises a common problem faced by all liquid re-staking protocols: when the business it was started on is no longer profitable, what should we do next? The answer given by ether.fi is: no longer just a staking service provider.
ether.fi’s second growth curve
ether.fi has now launched an encrypted consumption card, which allows users to swipe their cards for consumption without selling encrypted assets. At the same time, it has launched a lending market on Optimism, the second-layer network of Ethereum, and has built a series of vaults. In August this year, ether.fi further expanded its new banking business and launched tokenized stocks, metal assets, and portfolio asset loans built on Aave.
Silagadze believes that the annual revenue of the new banking track is approximately US$300 billion, which is 300 times the size of the DeFi market. The proportion of consumer card business revenue increased from 17% in January to 46% in July. He said: "New banking revenue has fully covered the revenue gap caused by the shrinkage of re-hypothecation and the decline in ETH prices. We expect the overall annualized revenue to increase by approximately 38% this year, while the revenue from staking and re-hypothecation has declined by 70%. The diversification of revenue has been a great success."
DefiLlama's data for the same period showed a different trend: ether.fi gross profit fell from US$18.71 million in the third quarter of 2025 to US$9.99 million in the second quarter of 2026, a decrease of 47%. The two sets of data do not conflict: revenue is not equal to gross profit, and annualized expectations are not equal to past quarterly financial reports, but ether.fi did not disclose the calculation caliber for the 38% growth rate.
ether.fi gross profit of each business in the second quarter of 2026
In the second quarter, credit card fees generated $3.14 million in gross profit. EigenLayer’s re-pledge business contributed $2.87 million, which is higher than the ETH basic staking business and higher than the sum of treasury, lending and management fees. According to DefiLlama statistics, re-staking is still the second most profitable business of ether.fi when it decides to exit the track.
Silagadze disputes DefiLlama's statistical caliber. The cashback rewards were included in DefiLlama's statistics as both $5.83 million in revenue and $5.83 million in costs, and did not contribute to profits. He explained: "This cash rebate was subsidized by third-party partners in the early days. Now the rules have changed, and current revenue statistics no longer include such subsidies. However, DefiLlama's data still uses the old statistical method in the most recent quarter." He also said that migrating the cash library to Aave replaced a customized debt management system, reducing risks. "We have notified users multiple times, and users freely choose to accept this change."
Technology itself does not fail
This is not to say that rehypothecation technology itself has failed. EigenLayer's data availability service EigenDA mainnet throughput can reach 100MB/s, and the guaranteed asset scale remains the first in the industry. Symbiotic, which migrates ether.fi's re-staking business, has integrated with more than 50 networks.
The core conflict is not whether the re-pledge technology can run smoothly, but whether it can generate enough revenue to support the business model. For a protocol that regards re-staking as its only core product, the answer is no.
EigenLayer no longer regards re-pledge as its main product. The brand has been renamed as EigenCloud. It now focuses on verifiable calculations to help applications prove that off-chain calculation results are authentic and trustworthy; re-pledge collateral has become the underlying infrastructure and is no longer a core product for external sales. Its locked position has fallen to $5.1 billion from a high of $220.6 billion in August 2025.
The industry has yet to come to a conclusion. The lock-up scale has shrunk by about 75%, but the guaranteed service volume has not decreased. Is this considered a failure on the track, or is the previous scale four times larger than the actual demand? Silagadze did not wait for the dust to settle, and led ether.fi to withdraw from the staking track in advance.