ether.fi was 71% of EigenLayer's native ETH. By July it was gone.
In the summer of 2025, 3.5 million ETH sat restaked on EigenLayer. By September 2026 only a fifth of it was left. We followed the exit on-chain, day by day, to see who left and what became of the ETH.
- 71%
- 5,869
- 745,594
- 23,653
- 43%
- 30 days
01 · One night in JulyOne address, 72 transactions, under two hours
Late on July 7, one Ethereum address began sending transactions that pulled stake off EigenLayer. It sent 72 of them in under two hours. When it stopped, three firms that run validators had lost three quarters of a million ETH of stake between them.
By the next afternoon the job was done. All of that stake had belonged to one customer, ether.fi, and almost none of it was left.
ether.fi is a staking project. You give it ETH, it stakes that for you, and you get a token back, eETH, that you can trade or use in DeFi. EigenLayer, or Eigen for short, is where ether.fi sent the stake to earn a second yield, a practice called restaking. Nobody sent more. At the end of March 2026, ether.fi accounted for 71% of the native ETH on Eigen, meaning ETH staked straight into validators.
So the short answer to what happened to EigenLayer in 2026 is that one customer left. The exit took less than three months from the first cut to the last, and every step of it is on-chain.
02 · The basicsRestaking in one minute
Staking is the simple part. You lock ETH in a validator to help run Ethereum, and it earns a yield. Our piece on the MetaMask Staking incident shows what that job looks like when it goes wrong.
Restaking gives the same ETH a second job. On EigenLayer, which now calls itself EigenCloud, the stake is pledged to secure other services, such as oracles and data layers. Eigen calls them AVSs. They are meant to pay for that security, which would give the staker a second yield on top of the first. The pledge has a cost, though. If the firm running those services breaks the rules, part of the stake can be slashed.
The firms that do this are specialists. We call them node firms, meaning companies that run the validators and the extra services for a staker. The staker picks one and delegates its stake to it.
A liquid restaking token, or LRT, wraps all of this in one token. ether.fi has eETH and its wrapped form, weETH, while Renzo has ezETH and Kelp has rsETH. People trade them and post them as collateral to borrow.
03 · The riseHow ether.fi became most of EigenLayer
Eigen let stakers pick node firms from April 2024, and ETH poured in. Within three months about 2.5 million native ETH was staked with them, just under half of it with firms working for ether.fi. Those firms are easy to spot, because each lists itself under ether.fi's name plus its own, such as "ether.fi-8 - DSRV". In March 2026 there were 11 of them holding stake.
The total peaked in July 2025 at just under 3.5 million ETH. After that everyone else slowly shrank while ether.fi held steady, so its share kept growing. By the following March it stood at 71%.
Points did much of the early work, since both projects handed them out to early users ahead of their token launches. Those programs wound down through 2025, and slashing went live in April of that year, CoinDesk reported. From then on restaking carried a real risk, and the open question was whether the second yield paid for it.
04 · The exitFour rounds of cuts in 78 days
The first cut came in April 2026, days after a hack at another restaking project. On April 18, an attacker used Kelp's bridge to mint rsETH with nothing behind it, then borrowed real ETH against it on Aave. Eigen itself was untouched, but it was a bad week to hold an LRT, and billions left Aave.
Three days later, on April 21, ether.fi's node firms lost about a tenth of their stake, in a long run of small transactions spread across the day. Two more big cuts followed that week.
Then it went quiet, and came back in rounds. There were two cuts in mid-May, then six in eight days in June, each between about 80,000 and 100,000 ETH. On June 23 the stake went the other way, and about a third of a million ETH was added back. We do not know why.
Two weeks later came the big one. On July 7 and 8, more than 1.1 million ETH of stake was pulled in under a day, which left ether.fi's firms with a few thousand.
| Day, 2026 | Change, ETH | Firms | Left, ETH |
|---|---|---|---|
| Apr 21 | -224,832 | 2 | 2,035,041 |
| Apr 23 | -181,997 | 6 | 1,838,772 |
| Apr 24 | -86,473 | 2 | 1,752,298 |
| May 13 | -88,169 | 1 | 1,620,826 |
| May 16 | -100,500 | 1 | 1,520,326 |
| Jun 5 | -91,733 | 3 | 1,381,465 |
| Jun 8 | -94,379 | 2 | 1,287,087 |
| Jun 9 | -97,441 | 2 | 1,189,645 |
| Jun 10 | -79,945 | 1 | 1,109,700 |
| Jun 11 | -100,826 | 1 | 1,008,873 |
| Jun 12 | -93,560 | 2 | 915,313 |
| Jun 23 | +330,170 | 7 | 1,248,226 |
| Jul 7 | -548,990 | 7 | 589,318 |
| Jul 8 | -583,449 | 3 | 5,869 |
Who did the pulling? The stake sat in 341 small contracts, and each one points to ether.fi's main pool, the contract that issues eETH. The transactions went through ether.fi's own node manager, a contract that pool names. We ranked the 400 stakers that pulled the most out of Eigen this year, and four fifths of what they took came from those contracts. Every one of its node firms ended up close to zero.
| Node firm | Mar 30 | Jul 12 | Sep 30 |
|---|---|---|---|
| Node.Monster | 424,783 | 1,605 | 2,249 |
| Chainnodes | 407,220 | 873 | 2,333 |
| DSRV (8) | 270,105 | 360 | 1,346 |
| Finoa | 244,641 | 993 | 1,339 |
| P2P.org | 217,830 | 487 | 2,804 |
| Pier Two | 189,196 | 491 | 534 |
| Validation Cloud | 144,147 | 137 | 192 |
| Cosmostation | 142,195 | 606 | 1,055 |
| DSRV (3) | 109,366 | 195 | 9,374 |
| Allnodes | 80,370 | 240 | 484 |
| Nethermind | 30,020 | 282 | 692 |
The news came later. On August 7, CoinDesk reported that ether.fi had removed restaking from weETH that week. In late September it reported that under 1% of the project's assets were still restaked as of August. On-chain, the stake had been about that low since July 8.
05 · What movedThe pledge left and the validators stayed
It would be easy to read all that as 2.25 million ETH walking out of EigenLayer. The chain shows something less dramatic. When stake leaves Eigen, the ETH itself does not have to go anywhere, because a staker can take back the pledge and leave the validators running.
ether.fi's contracts show just that. Their restaked balance fell 99% between the end of March and the end of September, while the count of live validators in their Eigen pods fell by less than half. A pod is the contract a validator pays out to, so those validators are still wired to Eigen. That last link is due to go by the end of the year, per ether.fi's docs as CoinDesk reported them.
Why the validator count fell at all, we cannot say. Some may have exited to pay out holders, and some may have been merged into bigger validators, which Ethereum has allowed since the Pectra upgrade.
eETH holders did move, on their own clock. eETH supply, which tracks the ETH held with ether.fi, fell by more than a third in April and May, a net drop of over a million ETH. Then it turned. Four months later it had won back about 40% of that drop, while the pledge stayed near zero.
So the two came apart. In March, the stake on Eigen was equal to about three quarters of eETH supply, and at the end of June it was still close to two thirds. A month later it was 1%.
06 · The stETH sideThe stETH went first
Native stake was only part of what ether.fi had on Eigen. Eigen also runs a pool for Lido's stETH, and at the end of January one of its contracts owned two thirds of it.
It began pulling out in February, two months before the Kelp hack. By the end of April it owned none of the pool, and the pool itself has shrunk by four fifths since January.
The stETH did not go far. At the end of the quarter the same contract held more than half a million stETH in its own balance, outside Eigen.
07 · Who is leftEigenLayer without ether.fi
Six months on, the native stake on Eigen is a fifth of the March figure. Node firms named Kraken hold 43% of it. Add P2P.org and you are past 60%.
Kraken's own stake took a round trip, falling by almost nine tenths between March and mid-July and then coming nearly all the way back by the end of September. We have not traced why.
The other LRTs went their own ways. Renzo's node firms lost almost four fifths of their stake over those six months, while Kelp's barely moved. The field thinned out too. In March, 46 node firms held 1,000 ETH or more. Six months later, 27 did.
08 · WhyWhat happened to EigenLayer
Mike Silagadze, ether.fi's CEO, gave the reason to CoinDesk in late September. There were "no meaningful yield opportunities in restaking", he said, and stakers saw some risk in it. So ether.fi left.
The fee numbers in the same report back him on yield. Citing DefiLlama, CoinDesk said the whole restaking sector earned about $100,000 in fees in one week of early September. Plain liquid staking earned $27 million.
The project has not dropped restaking for good. CoinDesk reports that holders who want it can opt in to a separate token, weETHs, built on Symbiotic, a rival to Eigen. The weETHs token we found on Ethereum had a supply of about 8,000 at the end of the quarter.
On motive the chain is silent, but the order of events is clear. The stETH began leaving two months before the Kelp hack, and the native ETH three days after it. Whether Kelp sped up a plan already under way is a question for ether.fi.
So what happened to EigenLayer? Its biggest customer found the second yield too small for the risk, by its CEO's account, and took back 2.25 million ETH of stake in 78 days. If you hold eETH or weETH, your token now earns the base staking yield, and about 1% of the ETH behind it is still pledged to Eigen. If you count on Eigen to secure a service, the native stake behind it is a fifth of what it was in March, and most of it sits with two names.
09 · How we measuredWhat we counted, and what we left out
We read Ethereum's record of EigenLayer's own contracts. Each time stake is added to a node firm or taken from it, the contract logs an event. We added those up day by day from April 2024 to the end of September 2026. The running totals match what the contract itself reports on the four dates we checked.
A node firm counts as ether.fi's if the name it lists says so. We checked that against the contracts doing the staking, which all point to its pool. We did not measure validator balances or rewards, and on Symbiotic we looked only at the token's supply. Days are UTC.
10 · The recordThe contracts and transactions behind the numbers
| What | Address or transaction |
|---|---|
| EigenLayer delegation | 0x3905…f37a |
| EigenLayer pod manager | 0x91e6…a338 |
| EigenLayer stETH pool | 0x93c4…564d |
| ether.fi pool | 0x3088…f216 |
| ether.fi node manager | 0x8b71…916f |
| eETH | 0x35fa…8ac2 |
| weETHs | 0x917c…9d88 |
| ether.fi stETH holder | 0x1b7a…6fff |
| ether.fi-10, Node.Monster | 0x6794…e051 |
| ether.fi-7, Chainnodes | 0x8e7e…a56f |
| Sender, July 7 and 8 | 0x7835…6d2a |
| First cut, April 21 | 0x2401…d7e7 |
| July 7, 23:00 UTC | 0x5f7d…492f |
| July 8, 00:17 UTC | 0x8a9b…ca54 |
| July 8, 14:24 UTC | 0xe886…da19 |
FAQ
What happened to EigenLayer in 2026?
Its largest staker left. At the end of March 2026, node firms working for ether.fi held 71% of the native ETH restaked on EigenLayer. By July 8 almost none of that stake was left. Six months on from March, the total across all stakers was a fifth of what it had been.
Did ether.fi leave EigenLayer?
Almost entirely. Its native stake fell by more than 99% between April 21 and July 8, 2026, and its stETH had left EigenLayer's stETH pool by the end of April. At the end of September, over 20,000 of its validators were still wired to EigenLayer pods, though the stake behind them is no longer pledged.
Is weETH still a restaking token?
CoinDesk reported on August 7, 2026 that ether.fi had removed restaking from weETH and moved it to a separate opt-in token, weETHs. On-chain, the stake its node firms hold on EigenLayer has been about 1% of eETH supply since July.
Why did ether.fi stop restaking?
CEO Mike Silagadze told CoinDesk in late September 2026 that restaking offered no yield worth having and that stakers saw some risk in it. The on-chain record shows when the stake left. It cannot show why.
How much ETH is restaked on EigenLayer now?
About 643,000 native ETH was staked with EigenLayer's node firms at the end of September 2026. That is less than a fifth of the peak the summer before. EigenLayer's stETH pool has shrunk by a similar share.
Who holds the most stake on EigenLayer now?
Node firms named Kraken hold 43% of the native stake on EigenLayer, and P2P.org holds 18%. The five largest node firms hold more than two thirds between them. These are end-of-September figures.
What is native restaking?
Native restaking means pledging ETH that is staked directly in Ethereum validators, as opposed to pledging a staking token such as stETH. The validator is set to pay out to an EigenLayer contract called a pod.
Ask about any staker or node firm in plain English
Every figure above came from data anyone can query. The Bitquery MCP server puts it behind an AI assistant, so you can ask how much stake a node firm holds on EigenLayer today, which wallets pulled out this week, or what a contract did on a given day, without writing the query yourself.
All figures run to September 30, 2026. Stake on EigenLayer changes daily, so later counts will differ.
Native ETH here means ETH delegated to a node firm on EigenLayer, summed from the delegation contract's own events. ETH held in pods without a node firm is not counted, and neither are staking tokens other than the stETH pool covered in section 06.
A node firm is counted as ether.fi's when its published name contains ether.fi. As a second check, the 341 contracts that held the stake each name ether.fi's pool. A node firm working for ether.fi under another name would be missed.
Pod and validator counts were read from the contracts at two blocks, March 31 and September 30, 2026. The pods' restaked balance on March 31 (2.25 million ETH) is read from the pod contracts, so it differs slightly from the 2.26 million delegated to node firms, which is summed from events. We did not measure validator balances, so we cannot say how much ETH sits in the validators still live in those pods.
Statements attributed to ether.fi or its chief executive are as reported by CoinDesk on August 7 and September 28, 2026. We did not speak to ether.fi, EigenLayer, Kraken or any other party named here.
The article places the Kelp hack and the first cut three days apart. It does not claim that one caused the other. One ether.fi node firm lost about 80,000 ETH of stake on February 26, 2026; the combined stake was then flat until April 21, which is where the article starts counting.
This article is provided for informational and educational purposes only and reflects analysis of publicly available on-chain data as of the dates indicated. It does not constitute legal, financial, compliance, or investment advice, and nothing in it is a recommendation to buy, sell, hold or stake any token or asset.
The findings describe stake, contracts and transactions observed in Ethereum data between April 9, 2024 and September 30, 2026. Attributions come from names that node firms published on-chain and from contract state; they may be incomplete or incorrect and may be revised as more data becomes available.
References to ether.fi, EigenLayer, EigenCloud, Kraken, P2P.org, Renzo, Kelp, Lido, Symbiotic or any other named party describe what the record shows about contracts and addresses linked to them. They are not statements about any company's solvency, security practices, disclosures or conduct, and nothing here asserts that any party acted unlawfully or improperly.
Nothing herein should be relied upon as a definitive determination of fact. Readers should conduct their own independent verification before taking any action. The authors and publisher accept no liability for any loss or damage arising from reliance on this material. All trademarks and company names are the property of their respective owners.
Reported by Gaurav Agarwal for Bitquery Research, with AI tools; every figure was checked against the raw data.