Cryptoeconomics

News, data and analysis on tokenized assets, market design and digital economic systems.

Concept

Protocol revenue and value accrual

Between the fee a user pays and any income a token holder receives sit liquidity providers, node operators, loss buffers and a governance vote that can redirect the whole flow, which is why gross fees and holder income differ by an order of magnitude.

Why it matters

A protocol can carry billions of dollars of deposits, charge a fee on every transaction and still deliver nothing to the holders of its token. From 2020 to 2024 that was the normal case: fees went to the people supplying liquidity or running infrastructure, and the token conferred a vote and an option on a future decision. From late 2025 the largest protocols began taking that decision, and the question became how much, by what mechanism, and at whose expense.

The vocabulary is imprecise in a way that flatters. "Fees" usually means gross payments by users; "revenue" sometimes the share retained by the protocol and sometimes the same gross figure; "earnings" whatever the publisher has chosen. Aave was reported at $140m of protocol revenue for 2025 and $402m annualised in June 2026. Neither is what reaches a token.

How it works

Start with the fee and follow it down the stack. On an automated market maker the user pays a swap fee, historically all of it to liquidity providers. On a lending market the borrower pays interest, most of it to depositors, with a reserve factor retained by the protocol. On an oracle network the consuming application pays for data and node operators are paid to supply it; on a stablecoin the holder pays nothing and the issuer earns the reserve return. At each step there is a residual. Value accrual is the set of arrangements routing that residual to a token rather than to a treasury, a labour pool or a committee. Four mechanisms are in use: a buyback purchases the token on the open market; a burn destroys supply, directly or by requiring a burn to withdraw fees; a distribution pays holders or stakers in the fee asset; and a reserve accumulates the token onchain, which converts revenue into a balance sheet item rather than income.

Uniswap chose the burn route. The UNIfication proposal, published on 10 November 2025 and executed by token holders on 28 December 2025 by 125.3m votes to 742, set a 0.05% protocol fee on v2 pools while cutting the liquidity provider share from 0.30% to 0.25%, took between a sixth and a quarter of provider fees on v3 tiers, routed proceeds to a contract called TokenJar that can only be emptied by burning UNI, and burned 100m UNI from the treasury outright as a retroactive equivalent of fees not previously collected.

Aave chose buybacks, and then removed the discretion. A programme from April 2025 bought roughly $1m of AAVE a week, acquiring 205,000 tokens, about 1.28% of supply. In March 2026 governance cut the annual budget from about $50m to $30m, citing a 25% fall in borrowing fees. Aavenomics 3.0, in effect from 27 June 2026, made the buybacks automatic, with protocol and GHO revenue flowing to holders without a committee setting the amount.

Economic mechanism

Follow the money and the incentive problem appears: the residual is taken from someone who was previously receiving it, and that party can leave.

Uniswap's fee switch is the cleanest illustration. The 0.05% protocol fee on v2 was not added to the user's cost; the provider's share fell from 0.30% to 0.25%. The protocol taxed its suppliers rather than its customers, on the reasoning that pool depth is set by expected fee income against inventory risk and there was slack. Because the code is open and forking permitted, the fee is bounded by the cost of switching venue, and whether $4tn of cumulative volume can be taxed without losing order flow is now an observable question.

Aave's dispute was about who holds the right to the residual, not its size. Aave Labs, the original author of the code, shifted swap-related fees away from the community treasury in December 2025. The proposal that passed on 12 April 2026 redirected 100% of revenue from Aave-branded products to the DAO and paired that with a $25m stablecoin grant and 5,000 AAVE to Aave Labs. The economics are those of a licence between a brand owner and the operator of the brand: the surplus is contested, the party controlling the deployment has leverage, and the vote is the enforcement mechanism.

Chainlink's reserve, announced on 7 August 2025, is a different design again. Payment abstraction converts fees paid in stablecoins or gas tokens into LINK and accumulates them in a contract behind a multi-day timelock, funded in part by 50% of fees from staking-secured Smart Value Recapture services that previously paid node operators. Nothing is distributed: the node operator's share falls, and the holder gains through supply withdrawn from circulation rather than cash.

Where these mechanisms break is the buffer. Fees funding a loss-absorbing reserve are the first defence against bad debt, oracle failure or an exploit, and every dollar routed to a buyback is a dollar not held against those losses. A protocol committed to automatic distribution has given up the ability to retain earnings in a bad quarter without a governance vote taken under stress.

Participants

Users pay the fee. Liquidity providers, depositors and node operators supply the input and receive most of it. A treasury or foundation holds the retained share and funds development, audits and incentives, and governance decides the split. Development companies (Aave Labs, the Uniswap Foundation) sit between the code and the token holders and negotiate for a share. Where a protocol earns a spread on reserves rather than a fee, as Circle does on USDC, the split is contractual: Coinbase receives 50% of residual reserve income as principal distributor.

Examples

Uniswap's UNIfication is the largest fee switch executed to date, and the first to pair a new charge with a cut to supplier compensation and a treasury burn.

Aave's April 2026 vote is the clearest case of a protocol asserting ownership of revenue generated under its brand against the firm that wrote the code, after a unilateral diversion of swap fees four months earlier. Aavenomics 3.0 then converted a committee-set buyback into an automatic one from 27 June 2026, at roughly $30m a year against $2.21bn of all-time fees.

Sky shows accrual without a distribution. It reported $14.01bn of collateral behind USDS and a governance-set Sky Savings Rate of 3.52% on sUSDS in July 2026: the surplus is paid to stablecoin savers as a policy rate, and what is left for SKY is the residual.

Risks and limitations

The most substantive disagreement is whether buybacks and burns are value accrual at all. One side treats them as the onchain equivalent of a share repurchase, returning surplus in a form that avoids the securities-law and tax consequences of a dividend. The other argues that a burn funded by a variable, competitive fee is a promise to keep spending rather than a claim on earnings, and that a holder with no enforceable right to payment holds an expectation, not an asset. The record is too short to settle it: Uniswap's fee dates from December 2025, Aave's automatic buyback from June 2026.

Reported figures are not comparable across protocols, or often across periods for one protocol, because reserve factors, buffer allocations and application-layer revenue are included inconsistently. Aave's $10m to $20m of application-layer revenue from swaps was the item the 2026 dispute turned on, and it sat outside the headline figure.

Durability is the third question. Aave cut its buyback budget by 40% in March 2026 after a 25% fall in borrowing fees: the right response, and a demonstration that the flow is cyclical. The legal characterisation of a token receiving protocol income is also unresolved in most jurisdictions, and a mechanism built to look less like a dividend is built around that uncertainty.

Key metrics

Track gross fees paid by users; the share retained after supplier compensation; the amount actually converted into buybacks, burns or distributions in a stated period; and the loss-absorbing buffer against outstanding credit exposure. Ratios of market capitalisation to annualised holder income mean something only where the income is contractually specified. Supply destroyed, in token units, is the least ambiguous figure a burn produces.

The UNIfication document of 10 November 2025 and the executed proposal are the primary record of a fee switch. The Aave governance record from December 2025 to June 2026 documents the dispute over who owns the revenue. Chainlink's announcement of 7 August 2025 sets out the accumulation model and the reallocation that funds it.

The weekly read on onchain market economics

What issued, what settled, what the supervisors changed, with the numbers behind it and a note on what the numbers do not show. One email, Thursday mornings.

No tracking pixels. Unsubscribe in one click. We do not sell or share the list.