Uniswap DAO

The Uniswap DAO governs Uniswap, the largest decentralized exchange (DEX) on Ethereum and one of the most-used protocols in DeFi. Control over the protocol's parameters, its treasury, and its on-chain fee mechanism rests with holders of the UNI governance token, who decide by token-weighted on-chain vote. With one of the largest token-holder bases of any DAO, Uniswap governance is widely treated as the archetypal protocol DAO.

Uniswap is also one of the few DAOs whose whole record is legible from the chain: the ballots, the quorum rule, the treasury, the fee switch and the burn are all on Ethereum mainnet and all readable without an API key. The sections below distinguish what the governance record actually shows from what it is usually reported to show. Every number is read at Ethereum mainnet block 25,717,629 (2026-08-09T13:09:47Z) or taken from deployed, verified source.

Origins and the UNI airdrop

UNI launched in September 2020 with a retroactive airdrop of 400 tokens to every address that had ever used the protocol – an early, widely-copied template for distributing governance power to users rather than selling it (see airdrops and token distribution). The airdrop was one half of an answer to SushiSwap, which had migrated about $840 million of Uniswap's liquidity into its own pools a week earlier; the other half was a liquidity-mining programme of Uniswap's own, 5 million UNI to each of four ETH pairs, running from 18 September to 17 November 2020.

It has never been restarted at the protocol level, and its last day cost Uniswap 38% of its total value locked in twenty-four hours. The token conferred the right to govern the Uniswap protocol, its governance-controlled treasury, and a community fund, with day-to-day stewardship of grants and ecosystem work delegated over time to the Uniswap Foundation. (Uniswap governance overview)

The fee switch debate

For years Uniswap's defining governance question was the "fee switch": the protocol earned no revenue for the DAO because trading fees went entirely to liquidity providers, while UNI conferred votes but no claim on cash flow. Whether – and how – to divert a fraction of fees to the protocol and its token holders was debated repeatedly on the governance forum, entangled with the legal risk that turning UNI into a value-accruing asset might strengthen the case that it is a security. The general pattern is covered on fee switches and value accrual.

DUNI: a legal person for governance (2025)

Before the fee switch, the DAO took a legal form. On 12 September 2025 Uniswap governance executed proposal 90, which organised Uniswap Governance as DUNI, a Wyoming decentralized unincorporated nonprofit association (see DAO legal structures). It passed with 52,968,177 UNI for and less than one UNI against, over a 40m quorum. The proposal states that the change does not alter the protocol, the UNI token or on-chain governance; it adds a legal person that can enter contracts, retain service providers and meet tax and regulatory obligations, while its members are not personally liable for its debts.

It sent $16.5m of UNI, 1,583,493 UNI at a 30-day average price of $10.42, to a DUNI-owned wallet to fund legal defence and tax compliance, and put the tax owed for prior years at under $10m including interest and penalties. Cowrie, a Wyoming firm, received $75,000 of UNI to act as compliance administrator (Uniswap governance).

UNIfication (2025–2026)

The debate was resolved by the UNIfication proposal, which passed on 25 December 2025 – the most consequential vote in the protocol's history. It activates a protocol fee that routes a portion of trading fees into a contract (the "token jar") and lets anyone burn UNI to withdraw an equivalent value (via a "fire pit" burn mechanism), and it executed a retroactive burn of 100,000,000 UNI – about $590 million at the time – approximating what would have accrued had the fee been live since launch. The package also wound down the Uniswap Foundation, moving most of its staff and responsibilities to Uniswap Labs. (DL News; The Defiant)

In early 2026 the DAO carried the fee switch beyond Ethereum. A follow-up package – whose final on-chain votes closed on 4 March 2026 – activated protocol fees across eight additional chains (Arbitrum, Base, OP Mainnet, Celo, Soneium, X Layer, World Chain and Zora) from 8 March, routing a share of fees into the same burn mechanism. It also introduced a tier-based fee adapter that automates collection across all Uniswap v3 pools, removing the need for a separate governance vote per pool, for an estimated ~$27 million in additional annualized revenue. By July 2026 protocol fees were live on ten chains beyond mainnet and the mechanism had recorded a single-day high of 186,000 UNI burned; the rollout is tracked on fee switches and value accrual. (The Block; DL News)

What the chain records about UNIfication

UNIfication is proposal 93 on the Governor. Read straight out of the Governor's proposals(93) struct, the record is:

The tallies circulated in the press are accurate to the token. What the reporting leaves out is the shape of them. The most consequential decision in the protocol's history drew 742.50 UNI of opposition against 125,342,016.81 in favour – a ratio of about 168,812 to 1 – and 1,500,006.69 UNI that showed up specifically to abstain, more than two thousand times the entire "against" side. Total participation was 126,842,766 UNI: 12.7% of supply, on the vote that redirected the protocol's revenue and dissolved its foundation.

Execution landed 48 seconds after the timelock eta, in a transaction sent by an ordinary address – anyone may call execute once the delay has run, and someone was waiting for it.

The burn is a transfer, not a reduction in supply

UNI's totalSupply() at Ethereum mainnet block 25,717,629 (2026-08-09T13:09:47Z) is 1,000,000,000.000000000000000000 – exactly the figure minted at deployment in September 2020, to the wei. The 100 million UNI that UNIfication "burned" did not leave that number, because the UNI contract has no burn function. Its full external surface is transfer, transferFrom, approve, permit, delegate, mint, setMinter and the views; there is no burn and no burnFrom (verified source).

So the burn is a transfer. In the execution transaction the Timelock emitted Transfer(0x1a9C…35BC → 0x000…dEaD, 100000000000000000000000000) – 100,000,000 UNI to an address whose private key nobody has and which was never generated from one. The address's UNI balance moved from 49,576.62 to 100,049,576.62 in that single step, which is how the figure can be stated exactly rather than as "roughly 100 million".

The fire pit has kept running since. At Ethereum mainnet block 25,717,629 (2026-08-09T13:09:47Z) the burn address holds 108,347,579.97 UNI – a further 8,298,003.35 UNI sent there in the seven and a half months since execution, by holders exchanging tokens for a share of the fee jar. That is 10.83% of all UNI parked at an address that cannot spend, delegate or vote it, while every "percent of supply" statistic about Uniswap still divides by the unchanged 1,000,000,000.

This is not an accounting quibble unique to Uniswap; it is the normal state of "burns" on tokens that were not written with a burn path, and it is worth checking before repeating a supply figure. What distinguishes Uniswap is scale: the tenth of the supply at the dead address is larger than the entire circulating supply of most DAOs in this directory.

The mint key is held by the contract that lit the fire pit

UNI can never be destroyed, but it can be created. The token's mint() is live, and reading its four state variables at Ethereum mainnet block 25,717,629 (2026-08-09T13:09:47Z) gives the whole rule:

Two of these deserve care. First, minter() returns the same address as the Governor's timelock(): the contract that executed the burn is the contract that holds the mint key. The route that removed 100 million UNI from circulation and the route that could issue 20 million more per year are the same governance process, differing only in which proposal passes.

Second, mintingAllowedAfter is 1 January 2024, set as a constructor argument at deployment and never advanced since – and it can only advance by being minted against, because mint() sets it to block.timestamp + minimumTimeBetweenMints on every call. That it still reads as a date two and a half years past is therefore proof of a negative: no UNI has ever been minted, which the unchanged supply independently confirms. The commonly-repeated "four-year mint lock" expired earlier than usually stated, and has been open ever since without being used.

Put the two mechanics together and UNI's supply is one-directional. mint() raises totalSupply; nothing lowers it. The 108,347,579.97 UNI at the burn address is about 5.4 years of mint headroom at the 2%-a-year cap – which is not a prediction that governance will re-issue it, but it is the reason "deflationary" is the wrong word for what UNIfication built. Uniswap's holders did not remove the ability to inflate; they voted to move a tenth of the float somewhere unspendable while leaving the issuance switch where it was.

The burn quietly raised the bar for every future vote

Governor Bravo's quorum is not a percentage. It is a compile-time constant in the deployed implementation:

uint public constant quorumVotes = 40000000e18; // 40,000,000 = 4% of Uni

The comment was true when it was written and is still arithmetically true, because total supply never moves. But 40,000,000 UNI is only 4% of a supply that includes the 108,347,579.97 UNI at the burn address, which cannot vote. Measured against the 891,652,420.03 UNI that can, the quorum is now 4.486% – and every fire-pit burn nudges it higher, with no vote taken and no parameter changed. A mechanism designed to return value to holders is also, as a side effect, slowly tightening the quorum that those holders must clear to govern.

The margin is not theoretical. Proposal 71 closed with 39,102,815.18 UNI in favour against 2,509,749 opposed and failed anyway, 897,184.82 UNI short of the constant. Four proposals in the Governor's history have failed and all four failed the same way, which is the subject of the next section.

Ninety-two proposals, and none of them was voted down

The Governor's proposalCount() is 100 and its initialProposalId is 8, so ids 9–100 – 92 proposals – are its own record (ids 1–8 belong to the predecessor Governor Alpha). Reading all 92 structs directly gives a census rather than an impression:

In all four failures the "for" side beat the "against" side by a wide margin and the proposal died on turnout: id 49 (21,774,717 for / 250,123 against), id 71 (39,102,815 / 2,509,749), id 88 (24,653,700 / 3,024) and id 91 (23,286,362 / 55,016). No proposal in the Governor's history has ever ended with more UNI against it than for it. The binding constraint on Uniswap governance has never been disagreement; it has been getting anyone to show up.

Among the 72 that passed, the median "against" total is 159.77 UNI – roughly two thousand dollars' worth of opposition on the median successful governance action at a multi-billion-dollar protocol. Fifteen of the 72 drew less than 1 UNI against; 41 drew less than 1,000. The single largest opposition ever recorded is proposal 31, with 28,466,755.23 against and 55,880,027.76 for. The most recent, proposal 100 (executed July 2026), drew 1,274,746.48 against 46,604,582.79 for – the first seven-figure opposition in over a year, and a fee-parameter vote rather than a spending one.

This is the empirical shape of token-weighted voting at scale, and it is easy to misread in both directions. It is not evidence that proposals are rubber-stamped: the filtering happens earlier, on the forum and in the temperature checks, so contested ideas mostly never reach an on-chain ballot (see proposal lifecycle). But it does mean the on-chain vote is close to a ratification step, and that the number worth watching is turnout against a fixed quorum, not the for/against split.

The other half of the same transaction: the growth budget on-chain

The execution transaction that burned 100 million UNI carried a second token event that received little coverage: an Approval of 40,000,000 UNI from the Timelock to 0xCa046A83…dAc9e1D2, a contract whose verified source names it UNIVesting – "a vesting contract that releases UNI tokens quarterly to a designated recipient". This is the on-chain form of the 20 million UNI annual growth budget that UNIfication created alongside the burn: five million a quarter, four quarters a year, with two years of it approved up front. The proposal's deflationary headline and its issuance commitment were authorised in the same transaction (see fee switches and value accrual).

Its terms are in the source and its state is readable at Ethereum mainnet block 25,717,629 (2026-08-09T13:09:47Z):

The three quarterly tranches due on 1 January, 1 April and 1 July 2026 have all been drawn: 15,000,000 UNI has moved, and 13,999,999.04 of it is still sitting at the recipient address. Five more tranches remain within the current approval. Against the 108,347,579.97 UNI at the burn address, the budget is small – but it runs in the opposite direction, on a schedule, without a further vote.

Two design details are worth noting for anyone reading DAO vesting contracts. withdraw() is callable by anyone, not just the recipient, which is a convenience rather than a control. And updateRecipient() is callable by the owner or the recipient, so the beneficiary can reassign the stream to another address without a governance vote – the DAO controls the size of the drip and the allowance behind it, but not, on its own, who ends up receiving it.

Why it matters

UNIfication marks a broader 2025–2026 shift across DeFi from pure governance tokens toward value-accruing ones – tying token worth directly to protocol revenue. It also illustrates a recurring tension in large DAOs: a foundation and a closely-associated development company (here, Uniswap Labs) end up holding much of the operational and governance weight, raising the same centralization questions that play out across the DAO landscape.

The on-chain record sharpens that reading. A vote with 742 UNI against it moved a tenth of the supply to an unspendable address, opened a 40 million UNI stream to one account, and left the mint key exactly where it was – while quietly raising the turnout every future proposal must clear. None of that is hidden; all of it is one eth_call away. It is simply not what the headline number describes.

How Caper approaches this

Uniswap had to retrofit value accrual onto a token that originally carried none, and could only reach a burn by transferring tokens somewhere unreachable. A caper settles both questions at construction rather than by vote. Its governance token is minted once, in full, with no mint role and no burn role – the supply is fixed at creation and, per the contract's own comment, "even a fully-malicious future logic can never mint or burn the governance token". There is no mint key to leave lying open and no need for a dead address, because there is nothing to reduce.

Caper does take Uniswap's escrow route in one place, and names it as such: tokens migrated from an origin token are deposited into a vault with deliberately no withdrawal primitive – a functional burn, described in the source as exactly that rather than as a supply reduction. The one thing a caper genuinely burns is the ballot: vote tokens are destroyed when a member exits, with the burn role restricted to the immortal state component. And funding flows through an on-chain bonding curve into a shared treasury from day one, so a member's economic claim is realized at exit rather than depending on a later vote to switch revenue on. The mechanics are described neutrally on those Caper pages.

See also

• Curve DAO – the other archetype of DEX governance, where influence is priced in lock duration rather than raw balance.
• Convex Finance – what happens when a third party accumulates a governance majority in someone else's DAO.
• SushiSwap (SUSHI DAO) – forked Uniswap V2 in 2020 and drained its liquidity in the original vampire attack.
• Compound DAO – origin of the Governor Bravo contract Uniswap runs.
• Quorum and threshold design · Voter apathy · Fee switches and value accrual

References

• Governor Bravo (0x408E…24C3) and its implementation source – proposal structs, quorumVotes, initialProposalId.
• UNI token (0x1f98…F984) – verified source for mint(), totalSupply and the absent burn path.
• Execution of proposal 93 – the 100,000,000 UNI transfer and the 40,000,000 UNI approval, in one transaction.
• UNIVesting (0xCa04…e1D2) – verified source and live state of the quarterly stream.
• Uniswap Governance Forum – where proposals originate.
• Uniswap docs: Governance.
• DL News – Uniswap DAO to activate the fee switch and burn 100M UNI.
• The Defiant – Uniswap passes the UNIfication fee-switch proposal.