Live
Moonwell Brief · Joel ObafemiIssue 007 · July 2026

Moonwell Brief

The Calm and the Clones

Moonwell's supply held flat at $94.5M for the first time in 2026, with the year's best depositor retention. The only real disturbance came on-chain, where 160 clone proposals spammed Moonriver before being cancelled en masse. Otherwise, a quiet month.

BaseOP MainnetMoonbeamEthereum
11 min readCite this issue

01Headline finding

une ended asking whether Base Core would find a floor after handing back May's gain. July's answer was a qualified yes: the book went flat. Combined supply closed at $94.5M, essentially unchanged on the month and the first time in 2026 the number has not fallen.

Underneath the flat line, two currents cancelled out. Depositors kept leaving, but more slowly, and rising collateral prices covered the difference: the constant-price outflow eased to $4.4M from June's $5.7M. Base drove both the exit and the slowdown, its $3.3M outflow roughly half of June's, while Moonbeam's sunset wind-down added a steady $470K.

The depositors who stayed are the stickiest cohort measured this year: 90.3% of June's active base was still supplying at the July close, the year's best one-month wallet retention. Constant-price dollar retention was a strong 91.5%, though May's cohort held more of its value, at 97.2%.

The liquidation tape confirms the calm. June saw 1,089 events seize $939K; July produced 429 on Base and OP Mainnet seizing $54K, 94% less and quieter than even May's floor. Borrowers spent the month keeping their positions healthy, and the book stayed unstressed.

The one disturbance was an on-chain anomaly. In a single July week, roughly 160 copycat proposals, each a verbatim clone of a legitimate reserve-withdrawal proposal and each submitted from a different, previously unseen address, hit the legacy Moonriver governor's proposal queue. All were cancelled without executing, and the Foundation followed within days with a security-hardening spell that moved the MOVR market's administration to a dedicated security contract.

The calm has edges, and this issue keeps them in view. The vaults fell a seventh straight month, WELL set another low, and two July proposals put Moonbeam on its official path out. Sections 06 and 10 take those up.

02Overview

Supply by component, December to July
Core lendingVaults
$0$55.6M$111.2M$166.9M$222.5MDecJanFebMarAprMayJunJul
Core lending and vaults stack to combined supply. July went flat: core rose to $76.3M, its first increase of 2026, while the vaults fell a seventh straight month to $18.2M. Hover any month for the split.
Source: Event-history indexer, oracle-priced at each month-end
Month over month, June 30 to July 31, 2026
MetricJun 30Jul 31Change
Combined supplied$94.4M$94.5Mflat
Core lending$74.9M$76.3M+$1.4M
Vaults TVL$19.5M$18.2M-$1.3M
Active suppliers14,71114,400-311
Protocol revenue$44.3K$29.2K-$15.1K
Liquidation events (Base+OP)1,089429-660
Moonwell by chain, as of July 31, 2026
ChainCore suppliedVaults TVLCombinedNet flow (const-price)
Base$70.1M$18.2M$88.3M-$3.3M
Ethereum$3.5M$0$3.5M+$366K
OP Mainnet$2.1M$0$2.1M-$1.0M
Moonbeam$620K$0$620K-$471K
Moonwell total$76.3M$18.2M$94.5M-$4.4M

One convention up front: USD levels are oracle-priced at month-end, and flows are constant-price, meaning the change in token units over the month valued at July 31 prices. That separation is what lets a flat headline show its moving parts.

The parts moved in opposite directions. Depositors still withdrew a net $4.4M in July, down from June's $5.7M, while a rising tape on cbBTC, WETH, and the LSTs lifted the value of what stayed by roughly the same amount. Ethereum drew $366K of genuine net inflow for a second consecutive month, and Moonbeam roughly halved as the sunset proposals began draining reserves and positions.

Depositors kept leaving, but more slowly, and the ones who stayed set the year's depositor-retention record: the book that remains is the book that wants to be here.

03Markets and real yield

USDC stayed the anchor at $15.6M supplied against $13.2M borrowed, and its rate story completed the round trip: base supply APY settled at 3.95%, below June's 4.47% and just under the 3.78% one-month T-bill. For the first time since April, plain USDC lending on Base pays roughly the risk-free rate, and what premium remains is the WELL incentive layer.

The yield action moved to the smaller stables. EURC spiked to 16.6% at the close on a burst of borrowing, and USDS printed 8.7%.

Stablecoin base supply APY vs the risk-free rate, Dec to Jul
USDCEURCUSDS1-mo T-bill
0%4.47%8.94%13.41%17.87%DecJanFebMarAprMayJunJul
Base supply APY at each month-end against the 1-month US Treasury yield. USDC finished the round trip from May's 8.4% spike to 3.95%, at par with the 3.78% T-bill. EURC took the spike baton at 16.6%.
Source: On-chain capture at month-end block; T-bill = FRED DGS1MO (Jul 31: 3.78)
Largest Base Core markets by supply, as of July 31, 2026
MarketSuppliedBorrowedUtilizationBase supply APYBase borrow APY
USDC$15.6M$13.2M84.7%3.95%5.22%
cbBTC$11.5M$1.8M15.9%0.14%0.98%
MORPHO$11.3M$333K2.9%0.01%0.68%
AERO$8.7M$2.3M26.0%1.02%6.18%
WETH$8.6M$7.1M82.4%0.86%1.16%
LBTC$3.5M$258K7.4%0.03%0.46%
cbXRP$2.5M$838K32.9%0.00%1.01%
wstETH$2.4M$341K13.9%0.10%0.86%

Ethereum deserves its own table this month. The deployment reached $3.5M supplied across its four markets, and MIP-E02, executed July 29, steepened the USDC and USDT rate curves so sharply that both printed above 20% base borrow APY at month-end on under 28% utilization. Either borrowing compresses or suppliers get an unusually loud invitation; August will show which.

Ethereum markets, as of July 31, 2026
MarketSuppliedBorrowedUtilizationBase supply APYBase borrow APY
WETH$1.62M$830K51.2%0.56%1.29%
cbBTC$805K$267K33.2%0.64%2.28%
USDT$557K$155K27.8%4.78%20.51%
USDC$537K$150K27.9%4.81%20.56%

04Wallets and risk

The deposit base kept thinning, at half June's pace and with the year's best depositor retention. Active suppliers fell to 14,400 from 14,711, with 1,099 new wallets against 1,426 churned. 90.3% of June's active base was still supplying at the July close, the year's best one-month wallet retention; constant-price dollar retention reached a strong 91.5%, behind only May's 97.2% and April's 91.7%.

Concentration crept up again as the tail thinned: the top 10 wallets held 35.4% of supplied value and the top 50 held 60.2%, continuing the year-long grind higher. July did not win depositors back, but the base that remains increasingly looks like the durable core rather than the exit queue.

Cohort retention triangle: do depositors stay?
Colour byeach cell: W = wallets · $ = dollars
CohortM0M1M2M3M4M5M6M7
Dec 2025
20,382 wallets
100%
W 100%
85.4%
W 85.7%
56.5%
W 77%
49%
W 73.1%
46.7%
W 67.2%
42.9%
W 63.7%
42.4%
W 57.4%
40.4%
W 56.2%
Jan 2026
19,880 wallets
100%
W 100%
63.6%
W 83.7%
56.5%
W 79.2%
48.9%
W 72%
48.1%
W 68.3%
47.4%
W 61.4%
45.1%
W 60.1%
Feb 2026
18,282 wallets
100%
W 100%
87.7%
W 88.4%
79%
W 79%
73%
W 75.2%
72.4%
W 68.7%
68.4%
W 67%
Mar 2026
18,266 wallets
100%
W 100%
87.4%
W 83.8%
81%
W 79.1%
78.5%
W 71.1%
75.9%
W 69%
Apr 2026
16,907 wallets
100%
W 100%
91.7%
W 89.7%
90%
W 79.1%
87.1%
W 76.3%
May 2026
16,316 wallets
100%
W 100%
97.2%
W 84.5%
91.7%
W 81.1%
Jun 2026
14,727 wallets
100%
W 100%
91.5%
W 90.3%
Jul 2026
14,400 wallets
100%
W 100%
Each row is a monthly cohort (the active deposit base of wallets ≥ $10 at that month-end); columns are months since the cohort started. Each cell shows dollar retention (NRR, large) and wallet retention (W). The pre-February cohorts that lived through the oracle incident still show it; the June cohort's first month, with 90.3% of wallets retained, is the year's best first-month wallet retention (its 91.5% dollar retention trails May's 97.2%). This triangle is re-derived on the re-synced indexer; cohort sizes shift about 2% against earlier issues.
Source: Envio event-history indexer: per-wallet SupplyDelta + RateSample, constant-price at Jul-31, $10 floor
Supply concentration: top-10 wallet share, Dec to Jul
0%9.56%19.12%28.67%38.23%DecJanFebMarAprMayJunJul
The top 10 wallets' share of supply edged up to 35.4% as the small-depositor tail kept thinning faster than the whales.
Source: Event-history indexer (combined core + vaults)

Risk work continued on schedule: Anthias Labs' July package (MIP-X63) carried the monthly cap and reserve tuning, and the Ethereum deployment got its first dedicated rate-curve intervention in MIP-E02.

05Liquidations

The June wave ended as fast as it began, and July read like a book under control. 429 liquidation events on Base and OP Mainnet seized $54.0K of collateral against $49.2K of debt repaid, 6% of June's volume; even May's quiet floor saw more dollars move. Borrowers kept their positions healthy through a calmer tape, and the seize numbers show it.

What little there was concentrated in cbXRP, one leveraged position accounting for most of the $23.6K taken there, with a long tail of dust-sized WETH events. Moonbeam added 228 small events totalling $1.6K as the sunset wind-down worked through remaining positions.

Liquidated collateral by asset, July 2026 (Base and OP Mainnet)
CollateralSeizedEvents
cbXRP$23.6K16
USDC$13.1K72
cbBTC$11.5K32
EURC$1.5K4
AERO$1.5K27
WETH$1.3K200
All collateral$54.0K429

Fifty-four liquidators competed for a sixth of June's flow. The top address took $23.4K across 11 events, roughly 43% of the seized value, and it was June's fourth-place bot. June's dominant liquidator sat July out almost entirely, which is what a month without whale-sized undercollateralization looks like.

06Vaults

The vaults fell a seventh straight month, to $18.2M from $19.5M, and this time the biggest vault led the decline. Flagship USDC dropped to $8.1M from $9.7M while Flagship ETH rose to $6.8M on price and modest inflow. The cumulative drawdown now stands at roughly 64% of January TVL.

One forward note the data allows: by the second week of August the live vault total had recovered to about $19.6M, with Flagship USDC back above $9.5M, so July's close may prove the trough. August's issue will say.

Moonwell-Morpho vaults, as of July 31, 2026
VaultTVL
Flagship USDC$8.1M
Flagship ETH$6.8M
Frontier cbBTC$2.1M
Flagship EURC$1.1M
Ecosystem USDC$74K
All vaults$18.2M

07Financials

Revenue normalized. June's number was a liquidation story, with $28.2K of its $44.3K protocol take coming from the 3% seize share of a heavy month. July's seize share was $1.7K because there was almost nothing to seize.

The underlying lending engine improved. Borrower interest rose 21% to $150.3K across the five chains, and the protocol's reserve-factor share of it rose to $24.1K.

The vaults chipped in as well: $6.8K of performance fees, measured on-chain from the fee-share mints and split evenly with the curator, Anthias Labs, for a $3.4K protocol share. Total protocol revenue came to $29.2K, a leaner month than June but built on interest rather than distress.

Fees and revenue, July 2026
LineAmount
Gross fees: borrower interest$150,300
Gross fees: liquidation bonus$5,100
Gross fees: Morpho vault performance$6,800
Total gross fees$162,200
Protocol revenue: reserve-factor interest$24,100
Protocol revenue: liquidation seize share (3%)$1,700
Protocol revenue: vault fee share (50%)$3,400
Total protocol revenue$29,200
Annualized take rate0.37%

Two bookkeeping notes. First, the borrower-interest comparison is ledger-to-ledger: the on-chain recompute that repaired June's scanner gap also restated June's reserve-factor interest to $15.8K from the $11.5K originally published, and July's 21% growth is measured against the restated figure. Second, Base carried $132.7K of July's interest, and Moonriver, in wind-down, still produced $9.2K as its reserves were withdrawn through the market rails.

08OEV

Third month at target, smallest sample yet. The active Chainlink wrappers handled 32 liquidations in July and recaptured $1,263 for the protocol against $541 of liquidator bonus: a 70.0% capture rate, the third consecutive month within a point of the MIP-X56 design target. The sample is two orders of magnitude smaller than June's because the liquidations themselves were, but the split held exactly, which is what a mechanical fee split is supposed to do.

Protocol OEV by wrapper, July 2026
WrapperChainEventsProtocol OEV
cbXRP/USDBase11$999
BTC/USDBase3$158
AERO/USDBase10$45
ETH/USDBase3$44
WELL/USDBase4$17

Routing held its ground in dollar terms. 55.7% of Base and OP Mainnet liquidation volume ($30.1K of $54.0K) ran through the wrappers, down from June's 68% but still a solid majority of value in a month dominated by dust events too small to route.

The searcher side got healthier. Fourteen addresses executed OEV liquidations and the top one took 50.6%, an HHI near 3,000, against June's single bot at 64% and an HHI of 4,200. One quiet month does not make a competitive market, but the monoculture eased.

09Governance

Governance ran a routine month. Seven proposals executed, all with support above 98%: the two-part formalization of Moonbeam's exit (MIP-M45 "Moonbeam Sunset" on July 8 and MIP-X64 "Before Sunset" on July 23), the recurring incentive program (MIP-X62), Anthias Labs' monthly risk package (MIP-X63), the Ethereum rate-curve adjustment (MIP-E02), and on Moonriver a reserve-withdrawal spell (MIP-R40) followed by security hardening (MIP-R41).

The one anomaly happened around MIP-R40. Within days of the legitimate proposal executing, roughly 160 verbatim clones of it hit the legacy Moonriver governor's proposal queue, each submitted from a different, previously unseen address, alongside two proposals whose entire description was a bare IPFS link. None executed: the two IPFS proposals were voted down and the clone wave was cancelled en masse on July 17.

The Foundation cleared the noise within days. MIP-R41 transferred administration of the MOVR market to a dedicated security contract and cleared the path to move the protocol-controlled MOVR balance to the Foundation wallet. The legacy governor's low proposal threshold made the spam cheap; the hardening makes it pointless.

Executed on-chain proposals, July 2026
ProposalTitleSupport
MIP-M45Moonbeam Sunset98.9%
MIP-X62Automated Liquidity Incentive Proposal98.4%
MIP-R40Moonriver Safe Reserve Withdrawals100%
MIP-R41Moonriver Governance Security Hardening100%
MIP-X63Anthias Labs Monthly Risk Parameter and Reserve Recommendations99.6%
MIP-X64Moonbeam Sunset: Before Sunset Phase99.9%
MIP-E02USDC and USDT IRM Adjustment and cbETH Remediation99.2%

A measurement milestone landed alongside: this is the first issue whose WELL holder count is measured on-chain rather than estimated. A Transfer-event scan of the Base and Moonbeam token contracts counts every address with a positive balance: 148,460 holders on July 3 rising to 148,774 by July 31, a flat month in a declining market. Earlier issues carried an interpolated series; treat month-to-month comparisons against them as approximate.

10Cross-deployment and token

Moonwell held third place among Base lending venues, and this month its book moved in the right direction: $89.3M supplied by DefiLlama's measure, up from $87.4M, while the gap to fourth place stayed a comfortable 2.6x. Aave's Base deployment rebounded strongly to $733M and Morpho Blue reached $4.9B; the tier below Moonwell stayed bunched between $27M and $34M.

Base lending venues by supplied USD, July 31, 2026
RankVenueSupplied
1Morpho Blue$4.91 billion
2Aave V3$733 million
3Moonwell$89.3 million
4Fluid$33.9 million
5Compound V3$30.6 million
6Euler V2$27.1 million

WELL kept sliding: $0.002947 at the July close, a $13.4M market cap, down 14.9% from June's $0.003462. The token continued to underperform its own protocol, and the divergence is now the token chart's defining feature: supply up, holders flat, revenue normalizing, price at another low. Staked WELL eased to 1.20B from 1.23B, a small unwind but the first notable one this year.

Methodology

Supply, flows, wallets, and vault TVL are frozen at July 31 on the event-history indexer: a per-wallet signed ledger of every mint, redeem, transfer, and liquidation across Base, OP Mainnet, Moonbeam, and Ethereum, re-synced from genesis after an infrastructure migration and valued with DeFiLlama historical prices at each month-end cutoff. Flows are constant-price (the change in token units over the month valued at July 31 prices), which separates depositor behavior from price movement; the same convention prices every cohort cutoff in the retention triangle at report-month prices, so dollar retention reads as net revenue retention rather than a price chart. Because of the re-sync, cohort sizes in section 04 re-derive roughly 2% higher than earlier issues while the retention percentages themselves are consistent with what June published. Financials are the dashboard ledger; June comparisons use the restated post-recompute figures. Morpho vault performance fees are measured on-chain from the MetaMorpho fee-share mints (AccrueInterest events, cross-checked against the ERC-20 mints to each vault's fee recipient), converted to underlying at the July 31 block and priced at the month-end cutoff; the protocol books half under its 50/50 split with the curator, Anthias Labs. Month-end APYs are supplyRatePerTimestamp / borrowRatePerTimestamp at each chain's July 31 snapshot block, annualized. OEV capture is measured on active (v2) wrappers; searcher identity (tx.from) was backfilled for July on August 12. Holder counts are the new Transfer-event scanner (Base + Moonbeam, positive balance); prior issues used an interpolated series. Comparators are DefiLlama chain TVL plus borrowed at July 31. The risk-free reference is the 1-month US Treasury yield (FRED DGS1MO, 3.78% on July 31). The Moonriver spam analysis identifies clone proposals by verbatim description match and one-proposal-per-address submission pattern; clone calldata was not decoded, so the report characterizes the wave as copycat spam rather than asserting an attempted theft.

11Looking ahead

August's questions are mostly continuations. Does the slowing outflow finally cross zero and turn a month green, or does the book keep grinding sideways while price does the work? Does Ethereum's growth curve steepen now that MIP-E02 has made its stablecoin markets the highest-yielding in the protocol, and do those 20% borrow prints compress before they attract the wrong kind of attention?

Do the vaults confirm July as the trough, as the early-August live read of roughly $19.6M suggests? On the wind-down chains, Moonbeam enters the "After Sunset" phase with under $700K of positions left, and the hardened Moonriver governor gets to prove the spam wave stays beaten. And WELL: at some point a book that has stopped shrinking and a token making new lows have a conversation.

Found any inaccurate data?DM on X|Email