1) Executive summary
CoW Protocol $COW is an intent-based trading layer where users sign orders describing the outcome they want and professional solvers compete in batch auctions to deliver the best possible execution. Instead of routing each trade through a fixed path, the project turns every order into a competition across nearly every major public and private liquidity source, with MEV protection built into the settlement design.
The problem it addresses is structural: onchain traders face MEV extraction, fragmented liquidity across a growing number of chains and venues, and execution complexity that ordinary users should not have to manage. As trading spreads across more networks, wallets, and applications, the gap between the price a user expects and the price they receive becomes a product surface of its own.
CoW Protocol's answer is delegated execution. Users sign intents; solvers calculate gas, manage price risk, and submit transactions on their behalf, absorbing execution failures. The business model follows the same flow: users pay fees collected on trade surplus and volume, through integration partners and as network fees covering gas. CoW DAO retains a portion of those fees as revenue, and what remains after solver incentives is gross profit.
The opportunity is to be the execution layer behind other products rather than a single venue, and Q2 '26 extended that positioning in two directions. Atomic Bundles, launched in May, gave any project a reusable template for building multi-step, atomic DeFi workflows on top of CoW Protocol's execution layer, with Euler as the first production implementation for actions such as collateral swaps and one-click looping. At the same time, embedded distribution widened: Lido Fast Swaps routed staking exits through CoW Protocol, Summer.fi used it for automatic vault conversions, CoW Protocol joined MetaMask Swaps as an execution provider, auto.fun added protected swaps, and tokenized stocks and ETFs became tradable through xStocks on BNB Chain and cbMEGA on Base. CoW DAO, formed in 2022, continued to develop the project across research, smart contract engineering, and design.
The quarter paired a market-wide contraction in DEX trading with a diverging usage picture: financial metrics declined from Q1 but held well above year-ago levels, and June closed the quarter with the strongest monthly activity, led by first-time users on BNB Chain. April marked five years since CoW Swap's 2021 launch, and also brought the quarter's hardest event, a social-engineering attack that hijacked the cow.fi domain; the team reports protocol contracts were not compromised, and it recovered the domain, published a public post-mortem, and paid every verified claim in full under CIP-86. Governance stayed active on economics, with value distribution proposals developed with Aragon, solver rewards moving under the CIP-85 framework, and continued solver-market reform.
🔑 Key metrics (Q2 2026)
Trading volume: $9.9 B (-28.7% QoQ, -51.0% YoY)
Fees: $5.8 M (-40.3% QoQ, +42.6% YoY)
Revenue: $3.9 M (-42.2% QoQ, +15.7% YoY)
Gross profit: $2.2 M (-37.1% QoQ, +68.0% YoY)
Treasury: $94.6 M (-14.0% QoQ, -38.3% YoY)
Monthly active users: 56.1 K (-13.0% QoQ, +46.5% YoY)
Fully diluted market cap: $174.3 M (-18.4% QoQ, -46.5% YoY)
👥 CoW DAO team commentary
"Q2 is the visible tail of a deliberate transition, not a soft quarter. Volume and revenue fell again, but those headline figures lag the choice that produced them: a decision we made at the end of 2025 to stop competing for flow that doesn't pay for itself. The leading indicators carry more weight, and three stand out:
Partner-attributed revenue held its share of the total through a down market, which says the integration base is resilient as a proportion of the whole even as absolute numbers move with the cycle.
The quarter's internal path, a May trough followed by a sharp June rebound led by new users, suggests the contraction is bottoming rather than accelerating;
The quarter's most consequential work sits deliberately outside this quarter's P&L, in the quality of solver competition and in moving value distribution from research to a concrete, quantified mechanism, both designed to compound over future quarters rather than this one.
The strategy itself also evolved. Q1 was about shedding unprofitable volume; Q2 was about changing where our flow originates, shifting toward wallet-embedded demand that monetizes at healthier margins than price-sensitive aggregator routing. The contraction in volume and revenue is the cost of that transition, and its intended return shows up in margin quality, a broader user base, and the token mechanisms now entering governance.
In Q3 we move into a heavy R&D phase. The team is prioritizing platform improvements aimed at unlocking B2B growth - faster quoting and shorter auctions among them - alongside sustained progress on CoW DAO's expansion to Solana."
2) Trading volume
Trading volume measures the total value of trades executed through CoW Protocol's batch auctions, where signed orders are settled by competing solvers. Volume totaled $9.9 B in Q2 '26, down 28.7% QoQ and 51.0% YoY, its lowest quarterly level since Q3 '24.
The decline was largely an industry-wide condition: DEX aggregator volume fell roughly 18% QoQ sector-wide, and total spot DEX volume fell roughly 35% over the same period, placing CoW Protocol's move between the two. Within the quarter, volume totaled $4.0 B in April, $2.5 B in May, and $3.4 B in June, so the quarter ended with a recovery from its May low.
Cumulative trading volume, the running total of all volume settled through the project since launch, crossed $200 B during the quarter and ended Q2 '26 at $202.8 B. The milestone quarter coincided with CoW Swap's fifth anniversary in April.
The chain mix re-concentrated toward Ethereum even as the user mix broadened. Ethereum settled 92.6% of Q2 '26 volume (+2.4 pp QoQ, +0.2 pp YoY), with Base at 3.0% and Arbitrum One at 2.6%, both down from Q1. BNB Chain reached 1.0% of volume, up from 0.7%, while Polygon, Gnosis, and Avalanche each held 0.4% or less. Large trades still concentrate where liquidity is deepest.
The top 10 asset pairs accounted for $5.0 B, or 50.9% of Q2 '26 volume. USDC-WETH led at $1.5 B, followed by USDC-USDT at $1.1 B and USDT-WETH at $591.9 M. Stablecoin-to-stablecoin pairs remain prominent: USDC-USDT, PYUSD-USDC, USDC-USDe, and USDT-USDe together totaled roughly $2.0 B, a weight consistent with the reduced fees on correlated and stable pairs introduced in February. The rest of the list is dominated by majors, including USDC-WBTC at $374.7 M and the liquid staking pair WETH-stETH at $207.3 M.
👥 CoW DAO team commentary
"The pattern tells more than the level. The sequential decline decelerated markedly from the prior quarter, which points to stabilization rather than a continued slide. The June rebound had two distinct drivers: a large cohort of new, smaller-ticket users, and a concentrated burst of trading over the first weekend of June driven by macro conditions and ETH and BTC price action.
Those two drivers deserve different weight. The new-user cohort is the structural signal - it indicates the acquisition engine is working and the base is broadening - while the volatility-driven activity is episodic and we would not extrapolate from it. Neither flatters the optics: a broader base of smaller trades depresses average trade size even as it strengthens the franchise. Volume will remain a noisy and understated read on the business for as long as this transition runs."
3) Fees
Fees measure the total value paid by users to trade through CoW Protocol, spanning fees on trade surplus and volume, integration partner fees, network fees covering gas, and MEV Blocker fees. Fees totaled $5.8 M in Q2 '26, down 40.3% QoQ but up 42.6% YoY.
Q2 '26 fees equal roughly 5.8 bps of trading volume, down from 7.0 bps in Q1 '26 but well above the 2.0 bps of Q2 '25. The YoY gain continues to reflect the fee policies introduced over the past year; sequentially, fees fell faster than volume as the mix shifted through the quarter. June was the quarter's strongest fee month at $2.5 M, alongside the late-quarter pickup in trading activity.
Ethereum accounted for 84.4% of Q2 '26 fees (+1.5 pp QoQ, -3.8 pp YoY), its first quarterly share increase in a year. Base contributed 5.4%, BNB Chain 4.7% (up 3.4 pp QoQ), Arbitrum One 3.4%, and Polygon 1.3%. BNB Chain's 4.7% of fees on 1.0% of volume is consistent with smaller, retail-sized trades carrying higher fee intensity.
👥 CoW DAO team commentary
"Fees, not volume, are the truer read of the business this quarter. We deliberately decoupled the two, earning more from each unit of flow we choose to serve.
Mix is what drives the result. The shift toward wallet-sourced demand matters because that flow is structurally more margin-enabling than aggregator-routed volume, so it monetizes at a higher and more durable intensity.
We expect fees to hold up better than volume through the transition, and we would treat any quarter where that relationship holds as evidence the strategy is working rather than as an anomaly. The relevant question for the periods ahead is not the absolute fee level but whether fee intensity - fee bps - continues to rise as the flow mix tilts further toward embedded, B2B-driven demand."
4) Revenue
Revenue measures the portion of fees retained by CoW DAO, derived from trade-level fees, partner revenue sharing, RPC rebates, and MEV-related activities. Revenue totaled $3.9 M in Q2 '26, down 42.2% QoQ but up 15.7% YoY.
Revenue equaled roughly 68% of fees, close to Q1 '26's roughly 71%, so the sequential decline tracks the fee base rather than a change in retention.
Ethereum accounted for 88.4% of Q2 '26 revenue (+1.2 pp QoQ), with Base at 4.9% and BNB Chain at 3.3%, up 2.3 pp QoQ and ahead of Arbitrum One's 2.3%. BNB Chain's revenue share sits well above its 1.0% volume share, the same fee-intensity pattern visible in the fee mix.
Gross profit measures what CoW Protocol retains after solver incentives, the COW rewards paid to solvers for executing orders. Gross profit totaled $2.2 M in Q2 '26, down 37.1% QoQ but up 68.0% YoY.
Gross profit is the report's bottom-line view of onchain unit economics: of the $5.8 M users paid in fees, $3.9 M was retained as revenue and $2.2 M remained after compensating the solver network. All three layers declined QoQ and grew YoY, in the same direction as fees.
Treasury measures the value of assets held in onchain addresses controlled by CoW DAO, including the main safe and the Managed Treasury. The treasury averaged $94.6 M in Q2 '26, down 14.0% QoQ and 38.3% YoY.
The decline is consistent with the repricing of the native token, given that COW makes up three quarters of holdings and FDV fell 18.4% QoQ. Treasury resources fund operations, contributor incentives, and ecosystem growth, the same economics the value distribution proposals address.
COW accounted for $71.0 M, or 75.0% of the Q2 '26 treasury average. The non-COW balance of roughly $23.6 M sits largely in stablecoin and yield-bearing positions: sDAI at $8.2 M, USDC at $5.7 M, the kpk_USDC_v2 vault at $3.9 M, stETH at $1.3 M, syrupUSDC at $880.6 K, kpkEURC at $818.5 K, and sGho at $675.6 K. This diversified sleeve represents the DAO's spending power independent of COW's market price.
👥 CoW DAO team commentary
"Revenue fell in absolute terms with the market cycle and the volume it rests on. The ratios held. Partner-attributed revenue kept its share of the total at roughly 35%, essentially unchanged quarter over quarter, through a down market - resilience that is proportional rather than nominal.
Underneath that stability, the quality of revenue improved in ways the headline does not show. Lending remains the largest category by both volume and revenue. The newer contribution comes from consumer wallets, which increased this quarter and are registering meaningfully for the first time. Wallet flow monetizes at healthier margins than aggregator-routed volume, so its growing share improves revenue quality even while lending continues to anchor the base.
We have lost a meta-aggregator listing following the April domain incident, which removed a partner that had been a material contributor - with discussions ongoing to be re-enabled. Adjusting for that, the underlying revenue base is steadier than the top-line fall implies.
Volume mechanically drives the near-term gross-profit print, so this quarter's figure says little about the trajectory. The mechanics matter more, because it targeted the structural determinant of margin rather than the cyclical one.
We upgraded how solver consistency is scored, rewarding credible bids and reliable settlement rather than raw activity. That change targets the efficiency of the competition that ultimately sets unit economics. Related efforts to widen participation in that competition, without loosening the trust model, point in the same direction.
We expect these changes to support margin structurally as they mature, while the quarter-to-quarter figure continues to move with volume."
5) Monthly active users
Monthly active users measures unique addresses that execute revenue-generating trades through CoW Protocol over a 30-day window, averaged across the quarter. MAU averaged 56.1 K in Q2 '26, down 13.0% QoQ but up 46.5% YoY, the first quarterly decline in a year.
The quarterly average understates how the quarter ended. On a month-end basis, MAU read 45.2 K in April, 48.7 K in May, and 98.5 K in June, at the time an all-time high, surpassing the prior peak of 80.4 K from February 2026. Because the increase came late in the quarter, it lifts the Q2 average only partially and instead sets the entry point for Q3.
The user base is far less Ethereum-concentrated than the volume base, and Q2 widened that gap. Ethereum's MAU share fell to 32.7% (-2.6 pp QoQ, -25.5 pp YoY), while BNB Chain jumped to 22.3% from 6.7% in Q1, making it the second-largest user base ahead of Base at 17.2% and Arbitrum One at 12.4%; Polygon held 8.2% and Gnosis 5.3%. Two thirds of users now trade on chains other than Ethereum while Ethereum still settles nine tenths of volume: breadth in users, depth in value. The team attributes much of the June increase to BNB Chain and reports that 88% of the chain's June wallets were first-time users.
👥 CoW DAO team commentary
"The decoupling of users from volume is the intended output of the strategy. Rising active users alongside falling volume shows the protocol broadening its base away from a historically whale-weighted concentration, and smaller average trade size is the consequence of that shift.
The second-order effect compounds. A larger, more diverse set of users increases the odds that solvers can match orders directly against one another - coincidences of wants - which improves the price users receive. Better prices in turn strengthen the case for the next user. That loop, not any single quarter's user count, is what our ambition to become the price-finding venue for on-chain trading rests on."
6) Fully diluted market cap
Fully diluted market cap measures the project's valuation assuming full dilution, calculated as the COW price multiplied by total supply under current token economics. FDV averaged $174.3 M in Q2 '26, down 18.4% QoQ and 46.5% YoY.
The divergence noted in Q1 extended through Q2: fees, revenue, gross profit, and users all remained above year-ago levels while FDV declined. Governance moved closer to this gap in May, when the core team published value distribution proposals developed with Aragon covering solver bond safes, a treasury burn trial, and a flexible buyback mandate.
👥 CoW DAO team commentary
"Rising usage and a lagging valuation point to a market that has not yet repriced our fundamentals, not to a deteriorating business.
The value-distribution framework now entering governance is the mechanism intended to close that gap, and it rests on a track record that is already in the data rather than in forecast. Buybacks have exceeded solver emissions by roughly 40% on the quarter, leaving the protocol solver operations structurally deflationary. The framework is designed to build on that base and give protocol performance a clearer path through to the COW token."
7) Outlook
CoW Protocol exits Q2 2026 with volume, fees, and revenue lower than in Q1, in a quarter when the whole DEX market contracted, but well above the same period last year. June was the quarter's strongest month for volume, fees, revenue, and users, so the first thing to watch is whether that late rebound carries into Q3. Early data points that way: month-end MAU crossed 100 K in the first three weeks of July.
The second thread is how the DAO connects the project's earnings to the COW token. In May, the core team published proposals, developed with Aragon, that would burn treasury-held COW to offset tokens paid out through the end of 2026 (an estimated 60 M to 85 M COW), allow up to 100% of weekly revenue to be spent buying back COW when market conditions warrant, and route a fifth of solvers' weekly rewards into dedicated safety deposits. The proposals remained under DAO discussion as of mid-July.
Work also continued on the solver market, the competition among the agents that execute trades. A framework that rewards solvers for consistent performance rather than one-off wins (CIP-85) passed in April, a stricter method for scoring that consistency followed in June, and proposals to lower the barriers to becoming a solver were open at quarter-end.
Distribution is the third thread. Atomic Bundles, the May launch that lets other projects build multi-step transactions on CoW Protocol, found its first production user in Euler, and the affiliate program launched in April passed $100 M in referred trading volume by May, per the team. After the quarter closed, the team reported a 23.5% share of trading among intent-based DEXs by its own methodology (July 8), joined the Encrypt the Mempool Coalition, an industry group working to shield pending Ethereum transactions from front-running (July 10), and returned its Fast Swaps feature to Lido's staking interface (July 13).
Entering Q3, the open items are whether June's momentum persists, how the DAO decides on value distribution, and whether the growing set of integrations turns into fee-paying volume.
8) Definitions
Metrics:
Trading volume: measures the total value of trades executed through CoW Protocol's batch auctions, quarterly total.
Fees: measures the total value paid by users to trade through CoW Protocol, including fees on trade surplus and volume, integration partner fees, network fees covering gas, and MEV Blocker fees, quarterly total.
Revenue: measures the portion of fees retained by CoW DAO, from trade-level fees, partner revenue sharing, RPC rebates, and MEV-related activities, quarterly total.
Gross profit: measures what CoW Protocol retains after solver incentives (the COW rewards paid to solvers for execution), calculated as revenue minus solver incentives, quarterly total.
Treasury: measures the value of assets held in onchain addresses controlled by CoW DAO, including the main safe and the Managed Treasury, quarterly average.
Monthly active users: measures the number of unique addresses executing revenue-generating trades through CoW Protocol over a 30-day window, quarterly average.
Fully diluted market cap: measures the project's valuation assuming full dilution, calculated as the COW price multiplied by total supply under current token economics, quarterly average.
9) About this report
This report is published quarterly and produced leveraging Token Terminal's end-to-end onchain data infrastructure. All metrics are sourced directly from blockchain data. Charts and datasets referenced in this report can be viewed on the corresponding CoW Protocol Q2 2026 Report dashboard on Token Terminal.