01Headline finding#

The lending side turned. As of May 31, Moonwell held $100.0 million supplied, down 3.3 percent on the month, but the mix underneath had flipped: for the first time all year, Core lending grew, to $79.2 million, on a genuine constant-price net inflow, with Base Core adding $3.2 million. The combined total slipped only because the vaults fell for a fifth straight month, down 19.9 percent to $20.8 million. Net dollar retention reached 91.8 percent, its best of the year, and bad debt eroded a little further to $5.0 million.

After February's deleveraging and a two-month floor, May was the first month suppliers put money into Core lending faster than they pulled it out. USDC utilization climbed to 90.5 percent and its supply yield jumped well above the risk-free rate, yet the protocol earned less from it than ever: with borrowing volumes still low, the take rate slipped to 0.29 percent. And in the last days of the month Moonwell opened its first new deployment since the Base and OP Mainnet era, launching on Ethereum through MIP-E00 with only negligible day-one liquidity.

02Overview#

These levels are oracle-priced at each month-end, so the change reflects both flows and token price drift; the constant-price net flow is given in the headline.

Core lending rose 2.3 percent while the vaults fell 19.9 percent, so the combined total eased just 3.3 percent. OP Mainnet kept expanding, its utilization climbing to 56.5 percent, and Ethereum, launched only at month-end, carried no measurable supply at the May 31 snapshot. All USD figures are constant-price, read from the Moonwell oracle at the end-of-month block on each chain.

03Markets and real yield#

USDC ran hot: $15.2 million supplied against $13.8 million borrowed at 90.5 percent utilization, which drove its base supply APY to 8.43 percent. Behind it, MORPHO reached $12.0 million, having grown every month of the year, and AERO climbed to $8.6 million.

For the first time in the series, USDC suppliers earned a clear premium over the risk-free rate from real lending: 8.43 percent base plus 1.25 points of WELL for 9.67 percent all-in, against the 3.72 percent T-bill. EURC, with a small, highly-utilized book, ran even higher. The recovery in dollar yield was real, but it sat on a small borrow base.

04Wallets and risk#

Concentration reached its high for the year even as the lending side recovered, because the fresh Core dollars came from larger holders rather than the return of small depositors: the top 10 wallets now held 33.3 percent of supplied value and the top 50 held 57.5 percent, with active suppliers down to 16,000. Risk kept healing alongside it, bad debt easing to $5.0 million from $5.3 million as cbETH worked down to $1.5 million, though the same three illiquid markets still held nearly all of it.

Anthias Labs moved from tuning parameters to clearing bad debt. Its late-May recommendations proposed a $550,120 reserve-funded repayment of debt held by 12 addresses that supply no collateral, concentrated in WETH, AERO, wstETH, USDC, and cbBTC, and a cbETH-incident remediation step that raises affected borrowers' recovery from about 12 to 17 percent.

05Liquidations#

The calmest month of the period: 127 events seizing $17,500 of collateral against $16,600 of debt repaid, the lowest in the series and a fraction of February's $7.1 million. The deleveraging was firmly over.

06Vaults#

The vaults were the one part of Moonwell that did not turn, falling to $20.8 million from $25.9 million, a fifth consecutive decline that has now erased roughly 59 percent of their January TVL even as Core lending steadied. Flagship USDC and ETH still held the bulk at $9.4 million and $8.6 million.

07Financials#

The thinnest month of the year for protocol economics. Gross fees fell to $129,800 as borrowing volumes stayed low, and protocol revenue was just $24,600 at a 0.29 percent annualized take rate, the lowest in the series. The paradox of May is that suppliers earned more (USDC near 9 percent) while the protocol earned less, because the high USDC rate sat on a thin borrow base and most of what borrowers paid passed straight through to suppliers.

08OEV#

OEV recaptured was $14,800 across 11 events. The more consequential development was MIP-X56, executed this month, which improves the accuracy of the OEV fee split between the protocol and liquidators. The resulting protocol capture share is omitted pending the corrected calculation.

09Governance#

May was the Ethereum month, eight proposals headlined by the new deployment: recognizing WELL on Ethereum, establishing governance there, and setting launch parameters.

The Ethereum launch carried with broad support, though MIP-X58 drew 7 million WELL against, the most opposition since March's contested Moonriver vote. MIP-E00, the first proposal executed on Ethereum, drew 182 distinct voters across the Base, OP Mainnet, and Ethereum vote collectors.

10Cross-deployment and token#

Ranked against the other Base lending venues, Moonwell was the third-largest by core lending supply at the May close, behind Morpho Blue and Aave V3 and ahead of Compound V3, Fluid, Euler, and Seamless. Its $73.7 million of Base core lending held third through a quarter that nearly halved Aave's Base book, which fell to $718 million from $1.50 billion in January, while Morpho Blue reached $4.20 billion. The five Moonwell vaults are Morpho Blue deposits, counted under Morpho's total, so they are left out of the ranking to avoid double-counting.

WELL closed May at $0.004165, an $18.9 million market cap. Staked stkWELL reached 28.4 percent of circulating supply, a fourth straight monthly increase, the governance base deepening through the whole drawdown.

11Looking ahead#

June, the close of the second quarter, is the first full month of the four-chain era. Watch whether Ethereum attracts real liquidity beyond launch dust, whether the vaults finally stop bleeding after losing nearly 60 percent of their January TVL, whether Core lending builds on May's first net inflow, whether the stubborn cbETH bad debt finally clears, and whether MIP-X56 starts to move the OEV capture rate.