ether.fi (ETHFI)

Overview

ether.fi is a non-custodial liquid restaking protocol on Ethereum. A user deposits ETH and receives eETH (or its wrapped, DeFi-composable form weETH), a liquid token that represents a position which is staked on the beacon chain and simultaneously restaked through EigenLayer. Unlike custodial staking services, node-operation keys stay with the staker rather than the protocol, which ether.fi frames as its core "non-custodial" design (see the protocol documentation).

By 2026 ether.fi had grown into the largest liquid restaking protocol and one of the largest liquid-staking protocols overall, second only to Lido by total value locked – live figures are tracked on DeFiLlama. Governance and the protocol treasury are directed by holders of the ETHFI token through the ether.fi DAO.

From staking to restaking

ether.fi sits one layer beyond classic liquid staking. Protocols like Lido (stETH) and Rocket Pool (rETH) issue a liquid receipt for ETH that is only validating Ethereum. ether.fi's eETH is additionally restaked into EigenLayer, where the same ETH secures actively validated services (AVSs) in exchange for extra rewards – and takes on the extra slashing surface that restaking introduces.

This makes ether.fi the reference example of the liquid restaking token (LRT) archetype: a governance-bearing DAO wrapped around a staking position whose risk and reward are a function of a separate restaking market it does not itself control. Its EigenLayer dependency is why the two projects are tightly coupled in this directory – ether.fi is consistently among EigenLayer's largest single sources of restaked ETH.

ETHFI: the governance token

ETHFI is the ether.fi DAO's governance and coordination token, announced in March 2024. Key parameters from ether.fi's own announcement:

• Fixed supply: 1,000,000,000 ETHFI, with "no further issuance."
• Initial circulating supply: 115,200,000 ETHFI (11.52% of max supply) at launch.
• Airdrops: Season 1 distributed 6% of supply, claimable from 18 March 2024 over a 90-day window; Season 2 distributed a further 5% (snapshot 15 March 2024), with unclaimed Season 1 tokens rolling into Season 2. Whale wallets were subject to a 3-month vest.
• Full distribution is scheduled to complete by the end of 2030.

The announcement scopes what ETHFI actually decides: the Grants Program, "protocol longevity, vision, and key economic parameters, such as protocol fees," greenlighting of node operators and contributor permissions, and treasury-diversification activities. Holders can also stake ETHFI, receiving sETHFI. The staking documentation describes a vault that restakes the deposit on Karak for loyalty and Karak points rather than paying a share of protocol revenue, and it is explicit that staking does not cost the holder their vote: staked ETHFI votes through the etherfi-dao.eth Snapshot space – 8,643 followers and 15 proposals when the space was read off Snapshot's public API on 7 October 2026 – though staking re-delegates away from any delegation previously made through Agora.

How governance works

ether.fi's process is documented in the official governance resources:

• Discussion: proposals are posted on the Discourse forum "with context and discussion before and during the voting window."
• Voting: votes are cast on Snapshot (etherfi-dao.eth) with a 4-day window. A vote can be opened at the same time a proposal is submitted.
• Quorum & threshold: "1M ETHFI is required for quorum in order for a vote to be valid," and a proposal "passes if it receives at least 1M ETHFI in approval votes."
• Weight & delegation: voting power is proportional to ETHFI (and sETHFI) held, counted across Ethereum and Arbitrum both as a raw balance and as on-chain ERC20Votes weight. Delegation happens by calling delegate() on the token; the space registers no Snapshot delegation portal, and the Foundation's own portal at vote.ether.fi/delegates is no longer publicly readable (HTTP 401, checked 13 August 2026).
• Who may propose: not holders. The Snapshot space sets onlyMembers, so only three allowlisted addresses can open a vote (see below).

The Foundation-and-multisig model

ether.fi separates deciding from executing through the ether.fi Foundation. A Proposer "submits proposals for the protocol changes and governance decisions, including treasury spend, initiating community votes for token holder approval." A Multi-Sig Committee then "implements these decisions, handles emergency actions, and ensures proposals align with the Foundation's objectives."

This is a distinct archetype from the fully on-chain Governor-style DAOs elsewhere in this directory: the binding tally is an off-chain Snapshot vote, and a trusted multisig carries out the result. The Foundation states that "further governance mechanisms will be rolled out in phases," aiming toward "full decentralization with an ecosystem of active contributors" – a candid progressive-decentralization posture rather than a claim of already-trustless execution. That gap between an off-chain vote and a multisig that executes it is exactly the failure mode this wiki catalogues in how DAOs fail.

Governance in practice

Beyond parameter changes, the ether.fi treasury and its product surface (Stake, Liquid, and the Cash card) give holders real economic levers. Treasury-diversification and token-value questions – including proposals to deploy treasury toward ETHFI buybacks – have been brought to the DAO's Snapshot, exercising the "key economic parameters" and "treasury diversification" mandate the token was given. Because these are Foundation-executed multisig actions rather than trustless on-chain execution, they are a live test of the phased-decentralization promise above.

What the record shows

Everything the Foundation documents about this DAO is checkable at source. Snapshot's public GraphQL hub returns the etherfi-dao.eth space's live configuration and its full proposal history, and the chains return the token state underneath it. The readings below were taken on 13 August 2026 – Ethereum block 25,742,725, Arbitrum block 493,948,614.

The published numbers hold

The space's voting.period is 345,600 seconds, which is exactly the four days the Foundation documents, and voting.quorum is 1,000,000, matching the “1M ETHFI” threshold quoted above. Both are settings on the voting venue rather than protocol code, so the space's admins could change either without a vote – but as configured they are what the DAO says they are. The space has 8,632 followers and 14 proposals.

Proposal rights are an allowlist, not a threshold

The space sets filters.onlyMembers: true, so a proposal can only be opened by an address on its authorized list, and that list holds three entries: two admins and one member. The Foundation's Proposer role is therefore not a convention the community observes, it is a permission at the voting venue, and no ETHFI balance confers it. A proposal threshold denominated in tokens can always be met by a large enough holder, and a refundable deposit only prices the attempt; an allowlist is the one gate a balance cannot pass. It is the strict end of the space the wiki surveys under proposal throttles and rate limits, and it is what makes the progressive-decentralization framing above load-bearing rather than decorative.

Fifteen proposals, and an electorate that emptied

The complete history, read from the hub:

The weight cast per proposal is roughly flat across two years – 1.1M to 5.1M ETHFI throughout – while the number of addresses casting it fell from 3,556 to 55 in April 2026, a decline of 98.5%. The August 2026 vote drew 80, which is 0.93% of the space's 8,643 followers. This is the shape the wiki describes under voter apathy, but with a detail worth isolating: the collapse is in participants, not in participation weight.

A weight quorum cannot detect the difference

On the April 2026 proposal, 55 addresses cast 3,163,434 ETHFI between them. The largest single voter, 0x9E27…0bc5, cast 1,862,679 – 58.9% of everything cast, and on its own 1.86× the 1,000,000 quorum. The two largest together carried 79.5%.

Because quorum here is a test of weight and not of headcount, an electorate can empty out entirely without the threshold ever registering it: one holder of that size clears quorum alone, and the vote is valid. That is the structural point behind quorum and threshold design – a weight quorum measures whether enough tokens showed up, never whether enough people did.

The August 2026 vote and its largest voter

The fifteenth proposal, opened on 30 August 2026 by one of the space's two admin addresses, asked holders to confirm three things: that ether.fi Labs pays the Foundation a share of its monthly revenue from card interchange, swaps and staking (about never whether enough people did.

.33m in the month before), which the Foundation spends on ETHFI every week through automated TWAP purchases rather than at its discretion; that every user reward on ether.fi is paid in ETHFI bought on the market; and that the Foundation may draw up to 20m ETHFI it already holds as a buffer when purchases fall short. It closed on 3 September with 1,141,999 ETHFI for and none against, cast by 80 addresses, 1.14 times the quorum.

The holder who carried 58.9% of the April vote did not vote. The largest voter this time cast 341,591 ETHFI, and without it the remaining 79 addresses would have cast 800,408, short of the 1,000,000 quorum with no vote against. A weight quorum lets one holder validate a vote alone, and it lets one holder's absence sink a vote nobody opposed. As with April, the proposal has no thread on the forum: its /latest.json, read on 7 October 2026, still lists the October 2025 buy-back as the newest proposal topic.

The forum stopped before the proposals did

The documented process says proposals are posted on the Discourse forum “with context and discussion before and during the voting window.” The forum's newest topic, read from its own /latest.json on 13 August 2026, is the buy-back proposal of 30 October 2025. The April 2026 Snapshot vote has no thread there at all: a Discourse search for rsETH, the subject of that proposal, returns zero topics. The DAO's most recent decision was taken at the venue but not at the forum the process names, which is the durability problem the wiki tracks in on-chain vs off-chain governance: the binding artifact and the deliberative record are held by two different systems, and only one of them was used.

Weight is counted twice over, on two chains

The space runs eight strategies, which read as duplicates until the network field is included. They are four token positions – ETHFI and sETHFI, on Ethereum and on Arbitrum – each counted by two different accounting methods, erc20-balance-of for tokens sitting in a wallet and erc20-votes for weight that has been delegated on-chain. ETHFI is 0xfe0c…c0eb on Ethereum and 0x7189…dc27 on Arbitrum; sETHFI is 0x86B5…0161 on both. Aggregating an electorate across chains this way is the pattern covered under cross-chain governance.

Reading both methods, rather than ERC20Votes alone, is what makes the tally work. The largest voter above holds 1,862,678.93 ETHFI on Ethereum, and its delegates() returns the zero address: it has never delegated, not even to itself, so its getVotes() reads zero. A space configured only on ERC20Votes would have scored its single largest participant at nothing. This is the activation gap the wiki documents under voting power activation, and here it is answered at the venue rather than left for holders to fix.

Supply, against the announcement

The March 2024 announcement quoted above states a fixed 1,000,000,000 ETHFI with “no further issuance.” Read directly, totalSupply() on Ethereum is 998,535,999 ETHFI and on Arbitrum 7,506,364; sETHFI's Ethereum supply is 91,687,309. The announced figure is a cap and a no-issuance commitment, so the live Ethereum number can only fall below it, and it has – by 1,464,001 tokens. The DAO has passed two buyback programs in that window, in April and October 2025, both listed in the table above.

How Caper approaches this

ether.fi shows the standard shape of a large DeFi DAO: an off-chain Snapshot vote sets intent, and a Foundation multisig is trusted to carry it out. Caper narrows that gap in two places. Vote weight is earned through taking part as well as through holding, so a large bag alone cannot capture a vote. And the tally settles on-ledger, where execution can only carry out the winning option and anyone can trigger it, so no multisig stands between the result and the action.