The European Central Bank published Working Paper No. 3208 on March 26, 2026, presenting on-chain evidence that governance in four major DeFi protocols — Aave, MakerDAO, Ampleforth, and Uniswap — is concentrated in a small number of wallets. The top 100 token holders control more than 80% of gove...
"DAOs, as we've been running them, are extraordinarily difficult, and not in the way that building hard things is difficult. They're difficult in the way of fighting your own organizational structure every single day." — Stani Kulechov, Founder and CEO, Aave
The European Central Bank published Working Paper No. 3208 on March 26, 2026, presenting on-chain evidence that governance in four major DeFi protocols — Aave, MakerDAO, Ampleforth, and Uniswap — is concentrated in a small number of wallets. The top 100 token holders control more than 80% of governance token supply across all four protocols. Voting power is tighter still: the top 20 delegates in Ampleforth hold 96% of proxy voting rights.
The findings arrive three months before MiCA's July 1, 2026, deadline for crypto-asset service provider (CASP) authorization across the European Union. Under MiCA's Recital 22, protocols "provided in a fully decentralised manner without any intermediary" are exempt from licensing. The ECB paper argues that the concentration it documented makes identifying reliable "regulatory anchor points" difficult — and implicitly challenges whether any of the four protocols qualify for the exemption. The European Commission's formal DeFi assessment report is due by mid-2026, and the ECB data will likely inform that analysis.
The paper, titled "Who to regulate? Identifying actors within DeFi's governance," was authored by ECB economists Alexandra Born, Zakaria Gati, Claudia Lambert, Mahvish Naeem, and Antonella Pellicani. The analysis uses on-chain data snapshots from November 2022 and May 2023 across four protocols selected for their governance maturity and on-chain transparency: Aave, MakerDAO (now rebranded as Sky), Ampleforth, and Uniswap.
The researchers mapped token ownership, delegation structures, and voting activity to assess whether governance power is distributed broadly enough to constitute functional decentralization. The paper explicitly frames its findings in the context of MiCA's "fully decentralised" carve-out, noting that concentration "remained stable over time" between the two snapshots — suggesting the patterns are structural, not transient.
The methodology combined wallet clustering (linking addresses to known entities), delegation chain analysis, and vote-event reconstruction. The authors acknowledge limitations: pseudonymous addresses cannot always be resolved, and off-chain coordination (Snapshot votes, forum deliberation) falls outside the dataset.
Across all four protocols, the top 100 governance token holders controlled more than 80% of total supply. At the top of the distribution, concentration was sharper:
| Protocol | Top 5 Holders (% of Supply) | Top 100 Holders (% of Supply) | |---|---|---| | Ampleforth | ~59% | >80% | | Aave | ~47% | >80% | | Uniswap | ~47% | >80% | | MakerDAO | 36–47%* | >80% |
*MakerDAO range reflects variation between the November 2022 and May 2023 snapshots.
Ampleforth exhibited the highest concentration among the top five wallets, with approximately 59% of governance token supply controlled by a handful of addresses. Aave and Uniswap showed similar top-heavy distributions, each with roughly half the supply held by five wallets.
Token ownership alone understates the concentration. DeFi protocols employ delegation mechanisms — token holders assign voting rights to delegates who vote on proposals on their behalf. The ECB found that delegation further concentrates power rather than dispersing it:
| Protocol | Top Delegates | Share of Delegated Votes | |---|---|---| | Ampleforth | Top 20 | 96% | | MakerDAO | Top 10 | 66% | | Uniswap | Top 18 | 52% |
Ampleforth's figure is notable: 20 delegates controlled 96% of all proxy voting power. At MakerDAO, 10 delegates held two-thirds of votes. Uniswap's power was comparatively dispersed — yet the top 18 delegates still held a majority.
The paper identified Andreessen Horowitz (a16z) as Uniswap's single largest delegate by voting power, with 125 addresses delegating their votes to the venture firm as of May 2023. This aligns with a16z's known $165 million Uniswap investment and governance participation, but raises questions about venture capital firms functioning as de facto protocol controllers.
The authors characterize the situation as "form over substance" — protocols that present themselves as governed by distributed token-holder consensus are, in practice, directed by a small cohort of delegates, many of whom are venture firms, protocol insiders, or unidentifiable entities.
Approximately one-third of top voters across the four protocols could not be publicly identified using available on-chain and off-chain data. Among identifiable top voters, the ECB categorized them as:
The anonymity of significant governance participants compounds the regulatory challenge. MiCA's framework assumes identifiable counterparties. When 33% of top voters are pseudonymous and unresolvable, regulators cannot assign accountability — a precondition for any licensing or enforcement action.
The ECB found that "around half or more" of governance token holdings could be traced to either the protocols themselves (treasuries, founders, developer allocations) or to centralized and decentralized exchanges. Binance was identified as the largest centralized exchange holder across all four protocols, with holdings ranging from approximately 2% to 15% depending on the protocol.
This presents a circular governance problem. Protocol foundations and their insiders hold governance tokens allocated at launch. Centralized exchanges hold tokens deposited by retail users — tokens whose governance rights are effectively dormant or exercised at the exchange's discretion, not the depositor's. Neither category represents the broad, independent participation that decentralization implies.
MiCA Recital 22 excludes from regulation crypto-asset services "provided in a fully decentralised manner without any intermediary." The regulation does not define "fully decentralised" with hard thresholds. The European Securities and Markets Authority (ESMA) has adopted a "spectrum of decentralization" assessment methodology, but no concrete metrics have been published.
The ECB paper does not make regulatory recommendations, but the data it presents maps directly to the functional decentralization criteria regulators are expected to apply. These include:
A protocol with 50,000 token holders but three wallets controlling 60% of votes would fail functional decentralization standards, according to the assessment framework described in reporting by SpendNode. The ECB data suggests all four studied protocols may fall short of this threshold, though the paper does not make that conclusion explicitly.
Only 13 of 27 EU member states have fully transposed MiCA as of March 2026, and the CASP authorization deadline is July 1, 2026. The European Commission's DeFi assessment report — expected by mid-2026 — will likely draw on the ECB's findings to inform whether a bespoke regulatory framework for DeFi is warranted.
The ECB's concentration findings exist within a broader pattern of low DAO participation. According to industry data compiled by SQ Magazine and CoinLaw, average DAO voter participation stood at approximately 17% in 2025, with most protocols reporting turnout below 18%. Leading protocols like Aave and MakerDAO achieved approximately 22% turnout on critical votes.
Over 6.5 million addresses globally held DAO governance tokens in 2025, yet active participation rates remained below 20%. More than 100 DAOs have adopted delegated voting mechanisms to address low turnout — but as the ECB paper demonstrates, delegation has concentrated power further rather than distributing it.
The MakerDAO-to-Sky rebrand illustrates the governance challenge at a protocol level. When the rebranding vote was held, just four entities accounted for nearly all the votes to proceed, according to The Block. The MKR-to-SKY token migration (at a 1:24,000 ratio) imposed a 1% conversion penalty starting September 2025, increasing every three months — a parameter set by a small governance cohort affecting all token holders.
Aave founder Stani Kulechov acknowledged the structural difficulty publicly, calling for DAO evolution toward "streamlined execution" models that reduce the friction of "weeks of forum posts, temperature checks, and multiple votes." The proposed "Aave Will Win" governance plan, introduced in January 2026, aims to restructure decision-making authority within the protocol.
80%+ concentration is structural, not transient. The ECB found consistent token concentration across two snapshots six months apart, indicating the pattern is baked into protocol design and token distribution, not a temporary market condition.
Delegation amplifies, not mitigates, concentration. Mechanisms designed to boost participation instead funneled voting power to a smaller number of professional delegates. In Ampleforth, 20 delegates control 96% of voting power.
a16z is Uniswap's de facto governance leader. With 125 addresses delegating to the venture firm, a16z commands the largest single voting bloc in Uniswap governance — a protocol with over $4 billion in total value locked.
One-third of top voters are unidentifiable. Pseudonymity in governance creates a regulatory dead zone. MiCA requires identifiable counterparties; the current DAO structure does not provide them.
MiCA's "fully decentralised" exemption may be functionally empty. If the four most established governance-token protocols cannot meet the threshold, it is unclear which protocols, if any, would qualify.
Binance holds 2–15% of governance tokens across all four protocols. Centralized exchange custody of governance tokens creates a concentration vector that is neither protocol-controlled nor user-directed.
July 2026 is the regulatory cliff. The CASP authorization deadline and the Commission's DeFi assessment report converge in mid-2026. The ECB paper provides the data infrastructure for regulatory action.
ECB Working Paper No. 3208 is a central-bank-grade dataset confirming what on-chain analysts have observed anecdotally: DeFi governance is concentrated in practice, regardless of how tokens are distributed in theory. The top 100 holders control more than 80% of governance tokens across Aave, MakerDAO, Ampleforth, and Uniswap. Delegation compresses that power further. One-third of top voters cannot be identified.
The regulatory implications are direct. MiCA's decentralization exemption assumes that some protocols operate "without any intermediary." The ECB's data suggests intermediaries exist — they are delegates, venture firms, core teams, and exchanges. The question for 2026 is whether European regulators will apply the functional decentralization test rigorously or accept the current ambiguity. The Commission's mid-2026 DeFi assessment report will provide the answer. The ECB has supplied the evidence.