Uniswap's Burn Was Built for the Protocol It Used to Be

(Source: UniswapLabs' Dune Dashboard)

Excluding the one-off 100mn treasury burn, the supplied series records 7.07mn UNI burned through 20 Jul. Ethereum generated 54.9% of that amount, followed by Base at 28.9% and Arbitrum at 11.3%. Those three chains accounted for 95.1% of the recurring burn, leaving the system considerably more concentrated than Uniswap's broader multichain footprint.

As of 21 Jul, the recurring burn was worth roughly $26.2mn, representing 1.1% of the 625.3mn circulating supply. The trailing 30-day average runs 38.7k UNI a day, which annualizes to 14.1mn UNI – roughly double what has been burned to date and broadly in line with the 14.8mn implied by the full-period average.

That symmetry is the headline number most desks will quote, and it is the number most likely to be wrong. It describes a pipeline that currently touches only v2 and v3, on a subset of chains, at a moment when the majority of Uniswap's fee generation has moved somewhere else.

Governance chose caution, then chose scale

UNIfication did not enable protocol fees everywhere at once. Uniswap followed a deliberately staged rollout, beginning with v2 and a limited set of high-volume v3 pools on Ethereum mainnet. Trading fees on those pools flow to a per-chain vault contract before being permanently burned, and the initial activation served as a live test of the mechanism's effect on liquidity – confirming both that the burn executed as designed and that liquidity providers (LPs) did not walk.

After those deployments showed no material deterioration in protocol health, governance moved toward full v3 coverage through a tier-based fee adapter, replacing pool-by-pool activation with default fees across entire LP fee tiers. The next phase extended the same framework beyond Ethereum, including chains such as Arbitrum, Base, Optimism or BNB, with fees collected locally and the corresponding UNI bridged back to Ethereum for burning.

The activation sequence is legible in the burn data itself. Ethereum opened the pipeline on 28 Jan, Unichain followed within days, Base, Arbitrum and Optimism came online together in March, and the long tail – World Chain, BNB and Polygon – arrived only in the final eight weeks. Each addition is visible as a step function rather than a ramp, which is why the monthly burn totals climbed from 864k UNI in February to 1.81mn in June, the strongest month on record.

Then July broke the pattern. Twenty days into the month, only 704k UNI has been burned, a daily pace of 35.2k against June's 60.3k – a 42% decline in burn rate during the single best fee-generating month Uniswap has printed all year. That divergence is not a failure of the mechanism. It is a map of what the mechanism does not yet cover.

Ethereum still burns the most, Robinhood prints the most

Robinhood Chain went from nothing to a quarter of Uniswap's volume in eight days, a trajectory Sandmark tracked from the chain's first days, when the model validated faster than the token economics did. In the week from 29 Jun, the chain cleared $28.2mn, 0.4% of network volume. In the week from 6 Jul it cleared $3.10bn – 24.8% of the total – and in the week from 13 Jul, $4.44bn, or 29.2%, second only to Ethereum mainnet's $5.10bn, even dwarfing the DeFi behemoth in intraweek volume on a few days of strong network activity.

(Source: DeFiLlama)

Volume share understates it. In the same week from 13 Jul, Robinhood Chain generated $24.98mn of gross fees against a network total of $34.04mn – 73.4% of every fee dollar generated through Uniswap and one of the highest single-chain fee prints since December 2024, from 29.2% of the volume. The partial week from 20 Jul holds the pattern at 74%. The blended rate on Robinhood Chain runs 56.3bps, against 8.1bps on Ethereum mainnet, a seven-fold spread that has nothing to do with routing efficiency and everything to do with what is trading: high-fee-tier memecoin pools where tokenized equities remain a rounding error relative to the speculative flow.

The past week's $34.04mn print accounts for Uniswap's largest weekly fee total in the series across all chains, roughly 37% above the prior peak set in early February and the strongest since October 2025. Cumulatively, across the 18 days from 3 Jul to 20 Jul, Robinhood Chain has produced $8.13bn of volume and $42.96mn of fees – against $11.26bn and $8.94mn on Ethereum mainnet over the same weeks. That is 72 cents for every dollar of mainnet volume, and nearly five times its fee generation.

The network-wide consequence is visible in the blended take from fees to revenue. Uniswap's effective fee rate across all chains ran 5.7bps in the week from 29 Dec; in the week from 13 Jul it ran 22.5bps, a near-quadrupling of monetization per dollar traded in under seven months. Uniswap is not moving more volume. It is charging materially more for the volume it moves.

That distinction is significant for the burn because gross fees and protocol revenue are not the same thing. Every swap pays a gross fee into the pool, and the overwhelming majority of it is retained by LPs. Revenue is only the slice the protocol captures rather than leaving with LPs, and it is that slice – accruing today from v2 and v3 pools alone – that buys and burns UNI. That is where the arithmetic starts to break down.

The two gaps the burn cannot reach yet

The UNIfication rollout still leaves two conspicuous gaps. Uniswap v4 remains outside the protocol-fee framework, while Robinhood Chain – now processing a significant share of Uniswap's volume and fee generation – has yet to be integrated into the existing v2 and v3 fee collection system. Both are now live governance proposals in their final onchain voting stage. Until those measures pass and execute, however, v4 activity and every swap on Robinhood Chain remain outside the UNI burn pipeline.

(Source: DeFiLlama)

The scale of the v4 gap has widened faster than most of the commentary around it. Across January to July, v4 has averaged 49.7% of monthly volume, peaking at 61.2% in May, and 46.9% of monthly fee share, peaking at 61% in the same month. Excluding v4 from the burn is not trimming the edges of the pipeline; it is leaving roughly half of it disconnected.

The volume story and the fee story have decoupled, which is the more interesting structural point. Monthly volume has fallen to $36.4bn across the first 21 days of July from $60.5bn in January. Fees have gone the other way without qualification, printing $72.6mn in 21 days of July against $59.6mn in the whole of February, the previous record.

Take rates fill in the rest. For v2 the mechanics are clean – a trader pays 0.30%, LPs retain 0.25%, the protocol captures 0.05% as revenue, implying a 16.67% cut of gross fees. Realized capture has been considerably thinner: the implied rate averages 6.4% across January to June, reflecting the share of v2 volume still sitting in unactivated pools. V3 is less uniform by design, with protocol share varying between 16.7% on higher fee tiers and 25% on lower ones depending on tier and governance configuration; realized capture averages 18.3% over the same window, and has trended up from 10.3% in January to 25.6% in June as tier-level activation replaced pool-by-pool selection.

V4 requires a different architecture again. Its hooks permit dynamic pricing and an effectively unlimited range of LP fee configurations, which makes the pool-by-pool approach used for v3 impractical. The proposal instead introduces a rules-based controller that classifies pools into families and computes their protocol fee on demand, preserving governance overrides for individual pools or token pairs. Modelling that precisely is not possible at this stage, so the conservative floor is the honest assumption: 16.67%, the bottom boundary of the v3 tier range and the closest available analogue on the protocol.

Robinhood Chain, meanwhile, is not a modelling problem at all. It is a routing issue – the pools exist, the fee logic exists, and the only missing element is the vote.

Three versions, one chain and a rotation already underway

Robinhood Chain's version mix inverted in under three weeks. Volume detonated on 8 Jul, when v3 alone jumped to $330mn in a day, and kept climbing through 11 Jul as v3 hit an all-time daily peak of $573.7mn – the clear inflection point where memecoin trading took hold on the chain. Aggregate daily volume topped out the same day at $874mn.

(Source: DefiLlama)

V2 has been the most front-loaded and the most volatile, peaking at $234mn on 11 Jul before decaying into a $50–120mn daily range by mid-to-late July, consistent with hype-driven listings cycling through cheaper v2 pools first. V4 tells the opposite story. It opened as the smallest venue on the chain – barely $1–2mn a day in the first week – and has gained share steadily since, running $170–230mn daily by mid-July and outpacing v2 outright on several sessions between 15 Jul and 19 Jul. V3 has dominated throughout and remains the largest venue by a wide margin, but its lead over v4 has narrowed sharply in the most recent prints.

The fee split diverges from the volume split in a way that sharpens the point. V4 carries 22.6% of Robinhood volume but 36.5% of its fees; v2 carries 20.9% of volume and only 12.7% of fees. Over the final six sessions, the volume gap has closed further still, with v4 at 34.5% against v2's 13.5%. Liquidity and routing are migrating toward v4 as the chain matures, and v4 is monetizing that flow at a materially higher rate. On Robinhood Chain, the fastest-growing segment is v4 – meaning the two gaps in the burn are compounding on each other, not simply adding up.

The burn rate that would quadruple overnight

Integrating v4 and Robinhood Chain would have burned 3.05mn UNI in 18 days. Applying the conservative 16.67% floor to v4 revenue and the respective realized take rates to Robinhood's v2, v3 and v4 fee generation produces $11.56mn of protocol revenue between 3 Jul and 20 Jul, converted at the running UNI volume-weighted average price (VWAP) into 3.05mn UNI of additional burn.

(Source: CoinMetrics, DeFiLlama, estimates compiled by Sandmark)

That single 18-day window equals 43.1% of everything burned since the fee switch went live, and 60.2% of it comes from Robinhood Chain alone. Against the 602k UNI actually burned over the same 18 days, the combined figure is fivefold larger.

Annualized, the estimated incremental burn runs 61.8mn UNI, or 9.9% of circulating supply. Stacked on the existing pipeline, the combined run-rate reaches 74mn UNI a year – 11.8% of circulating supply against the 14.1mn, or 2.3%, that the current configuration delivers. A 4.4x step change, executed by two votes, rather than any change in user behaviour.

Strip out Robinhood's v4 revenue, which is not part of the standing proposal focused on v2 and v3, and the estimate still lands at 2.36mn UNI over the window and 47.8mn annualized – 7.7% of circulating supply, and 3.4x the current pace. The gap between the two scenarios, roughly 14mn UNI a year, is itself almost exactly what the live pipeline is on track to burn.

Set against a $2.28bn circulating market capitalization and a $3.26bn fully diluted value, the fuller scenario implies something close to $225mn of annual supply retirement at current levels – the kind of figure that changes how the token is modelled rather than how it is described.

A ceiling worth bearing in mind

Robinhood Chain's fee generation is overwhelmingly speculative flow through high-fee-tier pools, and speculative flow does exactly one thing reliably, which is stop. The 56bps blended rate that makes the chain so valuable to the burn is a function of what is trading, not of who is trading it, and tokenized equities – the strategically durable piece of the venue – remain immaterial next to the memecoin book. Extrapolating an 18-day window that contains one of the single largest weekly fee prints in Uniswap's history is, on its own terms, an exercise in extrapolating a peak.

Treat 61.8mn UNI annualized as a ceiling rather than a forecast, and the analysis still holds. Even a normalization that halves Robinhood's contribution leaves the combined pipeline burning at more than double its current rate because the v4 gap is structural rather than cyclical – v4 has taken half of Uniswap's volume and fee generation across every chain, in every month since April, entirely independent of what is happening on Robinhood.

The governance votes are in their final onchain stage. The interesting question is no longer whether the fee switch works, but whether the coverage catches up to the protocol before the flow that makes the coverage valuable moves on. Uniswap has built a burn mechanism precisely calibrated to the version of itself that existed in 2024. The version that exists now is generating roughly twice the fees it can reach.