01Headline Finding
The June reading across Ethereum's six largest lending protocols is consolidation under yield compression at the sector level. The Real Yield Spread (the blended stablecoin lending rate minus the 4-week U.S. Treasury bill yield) closed June 30 at −37.1 bps, a 36.8-bps deepening from the May 31 print and the deepest month-end inversion since March 2026.
Capital did not leave the sector in response. The six protocols absorbed $1.33B of net deposits at constant prices (holding each token's price fixed at the snapshot date and counting only the change in token quantity), with Aave V3's $845M the largest single-protocol inflow in the captured series and Spark's $415M a distant second.
Aave V3's $845M did not arrive in USDC. Its own USDC book shed $162M of net constant-price capital across the month. What arrived was $452M of wstETH, $143M of cbBTC, $104M of USDTB, $99M of USDT, and smaller positives in USDE, SUSDE, WBTC, WETH, and WEETH.
Aave V3's headline USDC supply APY at 3.19% ran 41 bps below the 4-week Treasury bill through the month, but the depositor flow that landed on the protocol arrived in ETH-family and BTC-family collateral and in stablecoin alternatives to USDC.
Fluid, whose USDC supply APY at 6.41% ran 281 bps above the T-bill and was the sector's only positive real yield spread on USDC, received $16M of net inflow. Five of the six protocols saw positive flow. The exception was Euler V2, and its $21M outflow was small and distributed across many unrelated vaults rather than a coordinated depositor event.
USDC depositors on Aave V3 were not rewarded for staying, and net USDC flow was negative in June. The yield-seekers moved, and Fluid absorbed part of that outflow. The $845M that arrived was collateral: wstETH, cbBTC, USDTB, USDT. These aren't yield-seeking supply. They're borrow-side positions flowing to the sector's largest lending book.
Our May report's rotation thesis terminates. The June reading is that when sector yields compress below the risk-free rate, yield-focused capital moves to where the rate is competitive, and collateral continues to accumulate where the borrow depth is deepest. The two effects run in opposite directions at Aave V3 simultaneously.
The nominal supplied contraction (−$3.41B sector-wide) sits next to a positive constant-price net flow across the sector. Nominal figures use each token's actual spot price at the snapshot date, so quantity moves and price moves both show up in the dollar figure. The wedge is mark-to-market on collateral (the same positions revalued at current spot prices): spot ETH fell 21.3% in June, and the four largest LRTs (liquid restaking tokens, wrapped stake positions that also secure additional Ethereum-based services) fell in near-lockstep.
All six protocols saw constant-price net flow captures in June: Aave V3 +$845M, Spark +$415M, Compound V3 +$59M, Morpho +$20M, Fluid +$16M, Euler V2 −$21M. The nominal supplied deltas by contrast reflect mark-to-market effects on collateral: each protocol's nominal contraction scales with how much LRT and ETH-family exposure sits in its collateral base. The contraction is real at nominal prices and partially or fully reversed at constant prices, depending on each protocol's collateral composition.
Euler V2 is the exception: its collateral is not primarily LRT or ETH-family, so the 20.2% nominal supplied contraction is likely closer to true depositor flow than mark-to-market effect. The LRT reprice section unpacks the LRT arithmetic in detail.
Source: on-chain reads against each protocol's core contracts, DefiLlama, and the U.S. Treasury 4-week T-bill via FRED. Compound V3 and Euler V2 row-card readings substitute on-chain Comet and EVK aggregations for DefiLlama's chain-TVL figures, which over-count both protocols.
02By the Numbers
Snapshot date: June 30, 2026, 23:59 UTC. All month-over-month comparisons against May 31, 2026.
| Metric | May 31 | June 30 | Δ |
|---|---|---|---|
| Total Supplied | $32.64B | $29.23B | −$3.41B (−10.4%) |
| Active Borrows | $13.13B | $11.98B | −$1.15B (−8.7%) |
| Sector LDR | 40.22% | 40.99% | +0.77 pp |
| Real Yield Spread | −0.3 bps | −37.1 bps | −36.8 bps |
| Aave V3 net deposit flow (constant price) | +$352M | +$845M | +$493M (+140%) |
| Morpho Curator HHI (V1 + V2 combined) | 2,144 | 2,095 | −49 |
| Morpho top-3 curator share (V1 + V2 combined) | 77.3% | 74.6% | −2.7 pp |
| Protocol | Supplied May 31 | Supplied June 30 | Supplied Δ | Borrowed May 31 | Borrowed June 30 | Borrowed Δ |
|---|---|---|---|---|---|---|
| Aave V3 | $18.61B | $16.59B | −$2.02B (−10.9%) | $7.84B | $7.26B | −$0.58B (−7.3%) |
| Morpho | $5.96B | $5.09B | −$0.87B (−14.6%) | $2.16B | $1.84B | −$0.32B (−14.9%) |
| Spark | $5.38B | $5.03B | −$0.34B (−6.4%) | $1.75B | $1.65B | −$0.10B (−5.5%) |
| Fluid | $0.96B | $0.82B | −$0.14B (−14.4%) | $0.53B | $0.45B | −$0.09B (−16.2%) |
| Compound V3 | $1.23B | $1.30B | +$0.07B (+5.4%) | $0.42B | $0.44B | +$0.02B (+4.7%) |
| Euler V2 | $0.50B | $0.40B | −$0.10B (−20.2%) | $0.43B | $0.34B | −$0.09B (−20.4%) |
| Sector total | $32.64B | $29.23B | −$3.41B (−10.4%) | $13.13B | $11.98B | −$1.15B (−8.7%) |
| Protocol | May 31 | June 30 | Δ |
|---|---|---|---|
| Euler V2 | 85.23% | 85.00% | −0.23 pp |
| Fluid | 55.49% | 54.37% | −1.12 pp |
| Aave V3 | 42.13% | 43.80% | +1.67 pp |
| Morpho | 36.20% | 36.10% | −0.10 pp |
| Compound V3 | 33.96% | 33.68% | −0.28 pp |
| Spark | 32.55% | 32.86% | +0.31 pp |
| Sector LDR | 40.22% | 40.99% | +0.77 pp |
03The Real Yield Spread, Deep Inversion
The mechanical decomposition is the inverse of May's. Stable APYs fell while T-bill yields held flat. The blended stablecoin supply rate, TVL-weighted across USDC, USDT, DAI, and USDS on Aave V3, Spark, Morpho, and Fluid, fell from 3.60% at the May close to 3.23% at the June close, a 37-bps compression. RYS is scoped to the four protocols with material stablecoin books; Compound V3 and Euler V2 stablecoin markets sit below 5% of sector stablecoin supply and would not shift the blended figure by more than 1-2 bps if included. This scope matches the methodology used across Issue 002 and the captured historical series. The 4-week U.S. Treasury bill yield held at 3.60% across the same window, unchanged to two decimal places.
The spread re-inverted because on-chain rates fell, not because TradFi rates moved. That is the inverse of May's mechanism, where stable APYs lifted to meet T-bills.
What June added to the picture is duration of inversion. The Real Yield Spread last printed positive on June 4, 2026, at +42.0 bps. Through June 30, that prior positive print sits 25 days back, and the intervening series has been continuously negative. The June 4 reading was a one-day blip; the run before it was already negative. The functional sequence within the month is a brief positive print on day four, followed by a deepening inversion through the remaining 26 days.
Against the captured historical series, June's −37 bps is not the deepest reading on record. The 13-month trajectory has the deepest inversion at February 2026's −151 bps, with March 2026 at −118 bps and January 2026 at −64 bps, all deeper than June. April 2026 lifted to −28 bps and May closed at parity before June re-inverted.
Against the prior cycle, the May to November 2025 month-end series ranged from August 2025's −19 bps to June 2025's −90 bps, with two positive prints in between (July 2025 +26 bps, September 2025 +38 bps). June 2026's reading places this month near the mid-point of the prior cycle's inverted regime: deeper than October 2025's −41 bps, shallower than November 2025's −63 bps. June's print is consistent with the spread reverting to the pre-rally regime, rather than continuing the recovery that May suggested.
June 5, 2026 marked the month's largest liquidation event across the sector. Sector-wide collateral seized on the day reached $128.31M across 1,766 individual events, ranking first among the month's thirty trading days and running 10.25 times the trailing seven-day median of $12.51M. Aave V3 carried $93.98M of the day's seized collateral, Fluid $15.04M, Morpho $11.08M, Compound V3 $7.22M, and SparkLend $964K, with Euler V2's residual $21K rounding out the day.
The multi-protocol distribution indicates a spot-price shock that traveled across the sector rather than a single-protocol operational failure. Alongside the mark-to-market compression on LRT and ETH-family collateral, the June 5 event is one of the two proximate mechanical drivers of the sector-wide $3.41B nominal-supply contraction across the month, and it lands squarely in the middle of the yield-compression window this section tracks.
Whether July sees the inversion persist or close depends primarily on whether T-bill yields fall to meet on-chain rates, or on-chain rates lift to meet T-bills. The next Federal Open Market Committee decision lands in late July; absent a cut, the prior cycle suggests the inversion holds near current levels.
04Sector Overview
ix protocols, $29.23 billion of total supply at the June close, and $1.33 billion of net constant-price depositor inflow across the month. Both halves of that sentence describe consolidation: the first is continuation of a ten-month sector contraction at nominal prices, and the second is the counter-signal that emerged in June despite the Real Yield Spread reopening to −37 bps against the 4-week T-bill.
The sector aggregate fell 10.4% in nominal supply across June, from $32.64B at May 31 to $29.23B at June 30. Active borrows fell 8.7% to $11.98B. The sector loan-to-deposit ratio (active borrows divided by total supply, a protocol utilization measure) ticked up from 40.22% to 40.99%. At constant prices, depositor quantity rose $1.33B across the same window. The wedge between the two readings is mark-to-market on collateral, dominated by the ETH-family repricing The LRT reprice section unpacks in detail. The dollar value fell; the depositor quantity grew.
| Protocol | Net flow (constant price) | Share of sector inflow | Supplied at June 30 (nominal) |
|---|---|---|---|
| Aave V3 | +$845M | 63.4% | $16.59B |
| SparkLend | +$415M | 31.1% | $5.03B |
| Compound V3 | +$59M | 4.4% | $1.30B |
| Morpho | +$20M | 1.5% | $5.09B |
| Fluid | +$16M | 1.2% | $0.82B |
| Euler V2 | −$21M | −1.6% | $0.40B |
| Sector total | +$1.33B | 100% | $29.23B |
The distribution shape distinguishes June's inflow from the May reading. Every protocol except Euler V2 saw net positive constant-price flow. Aave V3 was the largest inflow, and it was also the largest existing protocol by supply.
May's shape was more mixed: Aave V3 gained $352M of net constant-price deposits in that month, while SparkLend and Morpho each absorbed roughly $752M and the sector-wide flow ran less concentrated. Our May report's rotation thesis, the framing that Aave V3 was losing supply share to migrations, held up on nominal supply readings but does not survive on constant-price readings. In constant-price terms, June's flow accelerated a concentration pattern that was already forming in May.
Aave V3 captured 63.4% of the sector's net constant-price inflow while holding 56.7% of the sector's nominal supply at June 30, an over-index of roughly 7 pp. SparkLend captured 31.1% against a 17.2% share of supply, over-indexing by 14 pp. Together the two largest protocols absorbed 94.5% of net inflow while holding 73.9% of stock. The remaining four captured 5.5%.
That's the concentration mechanism in one ratio: existing incumbents absorbed new capital at rates that widened rather than narrowed the gap between top and tail, in a month where the on-chain yield on the flow's largest destination sat 41 bps below the risk-free rate.
Three of June's four material single-day net outflows on Aave V3 were intra-sector or intra-protocol asset rotations rather than depositor exits. The June 12 outflow of $89.5M was dominated by USDC leaving Aave V3 alongside wstETH inflows on the same day. June 11's $85.6M outflow ran the same shape. Only June 3's $70.1M outflow carried genuine cross-asset departure at any material scale.
Sector-wide, the flow signal is that depositors added quantity on twenty-seven of thirty June trading days across Aave V3 alone, and on aggregate positive-flow days across every other protocol except Euler V2.
05The LRT reprice
ssue 002's LRT section framed May's LRT contraction as depositor exit. The sector lost $1.62 billion of LRT collateral at actual prices across May, of which $1.17 billion was constant-price outflow and $443 million was price decline. June extended the contraction at the actual-price level but inverted the mechanism. June's LRT collateral fell almost entirely because collateral repriced, not because depositors withdrew.
The four largest liquid restaking tokens fell in near-identical proportions in June. weETH fell 21.04% per token, from $2,193.22 at May 31 to $1,731.73 at June 30. rsETH fell 21.00%. ezETH fell 21.28%. osETH fell 19.99%. Spot ETH fell 21.25% across the same window, from $2,004.62 to $1,578.55.
The four largest LRTs moved almost mechanically together with ETH, which is what the collateral base of a well-collateralized LST wrapper should do. The mechanism is bond math on the underlying stake: when ETH falls 21%, wrapped stake positions fall by roughly the same percentage after accrual, and any LRT running a small basis premium moves within 1 pp of ETH's direct move.
The Aave V3 weETH book anchors the arithmetic. At May 31, weETH on Aave V3 Core carried $2.11 billion of supply. At June 30, $1.69 billion. That is a $419 million nominal contraction on one asset in one market.
Priced against the May 31 spot of $2,193.22, the May 31 balance implied roughly 961,000 weETH tokens sitting on the market. Priced against the June 30 spot of $1,731.73, the June 30 balance implied roughly 975,000 weETH tokens. Depositor quantity on Aave V3 Core's weETH book grew by approximately 14,000 tokens across June, worth about $34 million at the June 30 price. The remaining $453 million of the nominal contraction is entirely per-unit price effect.
Aave V3 Prime and Aave V3 Horizon carried no weETH at either endpoint, so the Core read is the full Aave V3 read.
Aave V3 Core weETH depositors added roughly 14,000 tokens across June while the position's dollar value fell $419 million.
Extended across the sector, the June LRT picture is a $3.73 billion sector-wide LRT collateral position at May 31 versus $2.92 billion at June 30, an $807 million actual-price contraction across the four largest LRTs. Priced at June 30 spot rates against May 31 token quantities, the per-unit repricing on weETH, rsETH, ezETH, and osETH accounts for roughly $781 million of the $807 million move. The residual, approximately $26 million, is the constant-price depositor flow component across the four LRTs combined.
Depositors added weETH quantity modestly, removed rsETH and ezETH quantity in smaller absolute magnitudes, and held osETH roughly flat. That $26 million net-out figure is a fraction of May's $1.17 billion constant-price LRT outflow. June's contraction is roughly 97% price effect and 3% depositor flow.
The May thesis Issue 002's LRT section built out held rigorously for May. Depositors exited LRT collateral in real quantity across the month, with $1.17 billion of constant-price outflow that was a genuine behavioral shift concentrated on weETH. June did not repeat that pattern.
The Aave V3 weETH position, which absorbed most of May's outflow, saw depositors ADD tokens in June. Depositors who stayed through May stayed through June, and depositors on the Aave V3 Core book added modestly to their positions across the month even as the dollar value of what they held fell substantially. The post-May-14 LRT-exit thesis Issue 002's LRT section documented remains correct for May. The mechanism did not extend to June.
The forward question sits at the intersection of LRT depositor behavior and ETH price direction. If July's LRT spot prices stabilize or recover, the depositors who held through the two-month drawdown will be sitting on positions worth substantially more in dollar terms without having added or withdrawn. If prices continue declining at June's rate, the same depositors will see another 20% haircut on the dollar value of unchanged collateral. Whether that second haircut catalyzes exit at the magnitude May absorbed is not visible in the captured LRT data alone. The signal that would answer the question is a July constant-price flow reading against a further LRT price decline, which the sector overview will pick up when the next month's data lands.
06.1Protocol Deep Dive: Aave V3
ave V3 closed June 30 with USDC supply APY at 3.19% against a 4-week U.S. Treasury bill yield of 3.60%. Depositors accepted a 41-bps penalty per dollar supplied and added $845 million of net constant-price deposits across the month. That is the largest single-protocol inflow in the captured series, and it arrived while the marginal dollar was compensated less on Aave V3 than in short-dated Treasuries.
At June 30, Aave V3 carried $16.59 billion of total supply on Ethereum against $7.26 billion of active borrows, a loan-to-deposit ratio of 43.8%. The nominal supply fell $2.02 billion from May 31, a 10.9% contraction; the constant-price series lifted $845 million across the same window. The gap of roughly $2.87 billion is mark-to-market on collateral, dominated by the 21.04% per-unit decline in weETH and the 21.25% decline in spot ETH between May 31 and June 30.
Both readings are correct at their own definitions. The dollar value of the depositor stake shrank; the underlying depositor quantity grew.
The asset-level composition of the growth is not uniform. Three days in June printed material net outflows: June 3 at −$70 million, June 11 at −$86 million, June 12 at −$90 million. Across those three days USDC accounted for approximately $156 million of gross outflow, USDT $18 million, and WETH $16 million. Inflows on the same days concentrated in wstETH ($98 million across the three days), cbBTC ($20 million), and USDtb ($19 million).
What ran off Aave V3 on the days when anything ran off at all was stablecoins. What replaced it was a mix of staked-ETH, wrapped-BTC variants, and a small stablecoin with a specific carry story.
On the twenty-seven other days in June, the net was cumulatively about +$1.09B and broadly distributed. The month's inflow was not a single trade at scale; it was steady accumulation punctuated by three days of stable-token rotation.
Depositors accepted a 41-bps penalty and added $845 million of net constant-price deposits anyway.
As The LRT reprice section documents, Aave V3 Core weETH depositors added roughly 14,000 tokens across June while the position's dollar value fell $419 million. The nominal decline masks a small net inflow of depositor quantity under a $450 million per-unit price effect. This reverses May's LRT-exit trajectory documented in Issue 002's LRT section, at least at the Aave V3 Core boundary.
Two candidate mechanisms explain why depositors accepted the rate penalty. The first is depth. Aave V3's supply book at $16.59 billion is roughly 3.3 times the size of the second-largest protocol in coverage, SparkLend at $5.03 billion, and 3.3 times the size of Morpho at $5.09 billion. When sector yields compress and the marginal dollar is choosing between protocols primarily on redemption depth rather than headline rate, the largest pool retains the flow by default.
The second is protocol-specific carry structure. Aave V3 hosts specific collateral relationships (E-mode pairs, isolation-mode positions, LRT-loop residues) that do not exist in the same configuration on other protocols. Depositors already inside those positions face non-trivial migration costs even if a headline stablecoin APY is 100 bps higher elsewhere.
Neither mechanism resolves cleanly from June's data alone. The distinction between "depth as inertia" and "depth as active preference" is what next month's flow data can begin to disambiguate.
Aave V3's June inflow is the largest single-protocol expression of a sector-wide consolidation, not a story of Aave V3 growing at other protocols' expense. SparkLend added $415 million of net constant-price flow across June; Compound V3, Morpho, and Fluid each added smaller positives; only Euler V2 saw net outflow, small and distributed across many unrelated vaults rather than a coordinated depositor event.
The sector added $1.33 billion in aggregate across six protocols, and Aave V3 captured 63% of that total by virtue of already holding 57% of sector supply. The consolidation is protocol-level and sector-level simultaneously. Aave V3 leads by size rather than by being the only game in town. The Morpho deep dive picks up the same consolidation logic playing out one level down, at the curator layer inside Morpho.
06.2Protocol Deep Dive: Morpho
orpho's curator concentration at June 30, 2026 is less concentrated than Issue 002 reported for May 31, not more. The correct combined view across both vault contract systems gives an HHI of 2,095 (the Herfindahl-Hirschman index, a concentration measure summing each participant's squared market share; higher means more concentrated) and a top-three curator share of 74.6%.
Issue 002's May 31 reading of HHI 3,103 and top-three share 93.9% captured only V1. V2 was already the larger side of the curator market at May 31 and has grown further since.
At June 30, Morpho carried $5.09 billion of nominal supply on Ethereum against $1.84 billion of active borrows, a loan-to-deposit ratio of 36.1%. Constant-price net flow across the month was positive $20.3 million. The $870 million nominal contraction is roughly $890 million of mark-to-market on Morpho's LRT-heavy curator vault collateral, mirroring the pattern the Aave V3 deep dive documents for its weETH book. Depositors did not leave Morpho in June; the collateral they hold repriced.
The finding that requires correction is the curator layer. MetaMorpho, referred to as V1, and Vault V2, launched earlier in 2026, are two separate contract systems Morpho operates in parallel; a curator is the party that sets each vault's risk parameters and market allocations. Morpho's curator market at June 30, measured across both systems, holds $2.12 billion of curated total value locked. V1 accounts for $672.56 million and V2 for $1.59 billion. V2 is 2.4 times the size of V1 at this snapshot, and V2 has been operational and accumulating capital since at least March 2026.
Twenty curators share that $2.12 billion. Sentora ranks first at 31.2%, or $661.76 million, all of it in V2 across three flagship vaults (Sentora RLUSD Main deployed 2026-03-04, Sentora PRIME Main deployed 2026-05-08, and a PayPal-branded PYUSD vault Sentora curates on the same infrastructure). Steakhouse Financial ranks second at 29.2%, or $618.15 million, split roughly evenly across V1 ($316 million on the long-standing 0xBEEF and 0xbeef vanity vault addresses) and V2 ($302 million). Gauntlet ranks third at 14.2%, or $300.83 million, still V1-heavy. The remaining 25.0% distributes across seventeen curators including Sky Money, Galaxy Curation, SparkDAO, Armitage by Wintermute, Waterline, AlphaPing, Clearstar, and others, none holding more than 5.3% individually.
The concentration story Issue 002 told holds at the V1 layer. It does not translate to the protocol layer once V2 is included.
Issue 002's May 31 curator reading was a manual capture of our dashboard's concentration panel, which at the time read only Morpho's V1 vault system. That reading recorded three named curators (Sentora, Steakhouse, Gauntlet) at 93.9% combined share and an HHI of 3,103.
The V1-only reading was accurate on its own terms. It answered the question "how concentrated is Morpho V1 among curators of V1 vaults?" The question the prose posed was broader: how concentrated is Morpho's curator layer as a whole? On that question, V2 was already the larger side of the market at May 31, and by June 30 the combined-view answer is 2,095, roughly the ordinary-competitive band for a curator ecosystem with a distinct leader and a healthy second tier.
Two things follow from the correction.
First, the curator market at Morpho did not concentrate further in June. Sentora's V2 book was already the largest at May 31. It grew by a share of June flows comparable to what other established V2 curators did. Steakhouse's V2 book grew alongside. The finding is not that concentration increased. It's that Morpho's curator market was less concentrated than the V1-only reading suggested all along, because half the market was invisible to it.
Second, forward Morpho curator readings need the combined V1+V2 methodology. The V1-only May 31 reading (3,103) is not comparable to combined readings; re-measured on the combined basis, May 31 comes to 2,144 against June 30's 2,095. A fuller backfilled historical series on the combined methodology is a follow-up for Issue 004 prep.
What the June 30 reading does support at the protocol level is that Morpho's book grew in the way flexible-curator lending would predict when sector yields compressed. Depositors did not run to Steakhouse specifically or away from Sentora specifically. Both grew their books. Smaller curators grew alongside them.
The curator ecosystem behaves like a competitive market with distinct operators serving distinct depositor bases, a different structural read from the near-monopoly the V1-only reading suggested. The protocol level consolidated; the curator level, measured correctly, did not.
06.3Protocol Deep Dive: Euler V2
uler V2 was the only protocol in coverage with net constant-price outflow in June, at $21 million. The magnitude is small in absolute terms and small relative to Euler V2's own May reading of $339 million net outflow.
Where May was one operator exiting one platform in one month, June was distributed residual movement across many vaults at once. The structural signal is that the protocol's 85% loan-to-deposit ratio held through both the yield compression that drove the sector and the aftershocks of May's Sentora unwind.
At June 30, Euler V2 carried $401 million of supplied capital on Ethereum across its active EVK vaults (the Euler Vault Kit, Euler V2's per-vault architecture where each vault sets its own risk parameters independently), against $341 million of active borrows and $60 million of unborrowed liquidity, for an 85.0% loan-to-deposit ratio. On the on-chain override basis this issue uses across Compound V3 and Euler V2, the May 31 comparison points were $503 million supplied, $428 million borrowed, and 85.23% LDR.
Supplied fell $101 million across June at nominal prices, roughly 20%, while borrows fell $87 million at 20% as well. Utilization (the fraction of supply currently borrowed) held to two decimal places. Structurally, the protocol contracted proportionally on both sides of the book.
The June outflow does not concentrate on any single operator. The four largest single-vault outflows were EVK Vault eRLUSD-7 at $3.0 million, EVK Vault eUSDC-70 at $1.1 million, EVK Vault ewstETH-2 at $0.6 million, and EVK Vault eAUSD-7 at $0.4 million, together $5.1 million. The remaining $16 million of June's protocol-level net outflow distributed across dozens of smaller vaults with no operator or asset dominating. That is a materially different shape from May, when four vaults under a single Sentora governor accounted for $128 million of the protocol's contraction and the concentration on one operator's decision was the section's central finding.
May was one operator exiting one platform. June was small distributed movement across many vaults at once.
The residual Sentora presence on Euler V2 continued to draw down in June, but at low absolute magnitudes. The two Sentora-governed vaults from Issue 002's table (eRLUSD-7 and eUSDC-70) together shed roughly $4 million of TVL across the month, compared to $85 million between them in May.
Sentora's operational scale on Euler V2 has now compressed to a fraction of the $172.9 million May-1 starting position across their four Euler vaults, without shifting the protocol's aggregate LDR downward or triggering any comparable outflow elsewhere. May's Sentora unwind on Euler V2 has finished. The Morpho side of that event is covered in the Morpho deep dive.
Euler V2's structural LDR persistence is the June-specific finding. Across a month where sector yields compressed below the risk-free rate, three of the six covered protocols saw LDR fall (Fluid −1.12 pp, Morpho −0.10 pp, and the compound-level readings dominated by mark-to-market accounting), and Euler V2 held. The vault-architecture LDR advantage documented in Issue 002's Fluid deep dive (isolated markets running per-vault utilization that pool-based lenders can't sustain because they need idle-capital buffers for liquidation slippage) remained intact through June's macro compression.
Euler V2's June is neither the biggest-venue flow story (that belongs to Aave V3) nor a methodology correction (that belongs to Morpho). It is the smallest of the six protocols running the sector's highest utilization by architecture, without operator-driven volatility for the first time in this report's coverage.
06.4Protocol Deep Dive: Fluid
luid's supplied capital fell 14.4% at nominal prices across June, from $963 million to $824 million. At constant prices, the protocol added $16 million of net depositor flow. The nominal contraction is entirely mark-to-market on Fluid's ETH-family collateral, which repriced along with the sector's LRT and wstETH base. Depositors on Fluid were slightly positive across the month, in the same shape The LRT reprice section documents for the LRT sector generally.
At June 30, Fluid carried $824 million of supplied capital on Ethereum against $448 million of active borrows and $376 million of unborrowed liquidity, for a 54.37% loan-to-deposit ratio. The May 31 comparison points were $963 million supplied, $534 million borrowed, and 55.49% LDR. Both sides of the book contracted at nominal prices while utilization fell 1.12 pp.
The protocol continued to run the second-highest LDR of the six covered protocols after Euler V2, well above the pool-based cluster of Aave V3, Morpho, Compound V3, and SparkLend that clusters between 32.9 and 43.8%. The vault architecture LDR story Issue 002's Fluid deep dive documented held through June's compression at the same margin it held through May.
The distinctive June signal on Fluid is the rate side, not the flow side. Fluid's USDC supply APY at June 30 was 6.41%, up from 5.93% at May 31. It ran roughly double the next-highest protocol (Compound V3 at 3.21%, Aave V3 at 3.19%, Euler V2 at 2.83%) and roughly 281 bps above the 4-week U.S. Treasury bill at 3.60%.
Fluid's USDC rate lifted 48 bps across June while the sector's blended stablecoin supply APY fell 37 bps, from 3.60% at May close to 3.23% at June close. Where the Real Yield Spread section documents a sector-wide inversion of Real Yield Spread on the blended stablecoin basket, Fluid's USDC market was one of the specific venues that ran opposite to the sector trend. The rate leadership Fluid held at May 31 widened across June rather than compressing toward the T-bill along with the rest of the sector.
Fluid's USDC supply APY at June 30 was 6.41%, roughly double the next-highest protocol and 281 bps above the 4-week Treasury bill.
The GHO cross-venue reading first flagged in late June shifted mildly across the rest of the month. Aave V3 GHO borrowed contracted $22.9 million across the trailing thirty days, a 15.8% decline. Fluid GHO supplied ticked up $1.7 million across the same window, a 22.8% lift on the smaller Fluid book. The absolute magnitudes are modest, and the ratio between the Aave decline and the Fluid rise is roughly 13-to-1.
The interpretation stands: this is a slow-motion re-allocation rather than a fresh mint story. Aave V3 acts as GHO's facilitator (its mint authority; GHO is created when users borrow it against collateral), and that side of the relationship is shrinking materially while Fluid's GHO supply side receives a fraction of the same magnitude. Whether the pattern continues into July is a question next month's data can begin to answer.
Fluid was the smallest positive net constant-price inflow among the five protocols with positive flow in June, at $16 million against Aave V3's $845 million, SparkLend's $415 million, Compound V3's $59 million, and Morpho's $20 million.
On the flow side, the absolute figure sits below the sector consolidation story the sector overview documents. On the rate side, Fluid's June is the sharpest counter-signal to the yield-compression story the Real Yield Spread section tracks: the sector's highest USDC supply APY at 6.41%, extending the rate leadership Issue 002's rate section documented at May close rather than compressing toward the T-bill along with the sector median. That combination of small positive flow and widening rate leadership sits at odds with the depth-driven consolidation the Aave V3 deep dive describes. When yields compress below the risk-free rate at the sector median, capital consolidates at the largest venue by size (the Aave V3 finding) and also finds the specific markets that maintain positive spreads (this Fluid finding). Both mechanisms ran in June.
06.5Protocol Deep Dive: SparkLend
parkLend was the second-largest constant-price inflow in the sector in June at $415 million, one third of the sector's total net depositor flow. That reading is materially different from Issue 002's May finding, which framed SparkLend as the migration beneficiary receiving capital that Aave V3 was losing.
In June, both protocols were positive. SparkLend was no longer the destination for departures; it was the second-tier destination in a month when the largest venue absorbed the majority.
At June 30, SparkLend carried $5.03 billion of supplied capital on Ethereum against $1.65 billion of active borrows and $3.38 billion of available liquidity, for a 32.86% loan-to-deposit ratio. The May 31 comparison points were $5.38 billion supplied, $1.75 billion borrowed, and 32.55% LDR. Both sides of the book contracted at nominal prices (supplied down 6.4%, borrows down 5.5%) while utilization ticked up 0.3 pp.
The nominal contraction is mark-to-market: SparkLend's collateral base weighted heavily toward wstETH, which fell along with the rest of the ETH-family collateral base at roughly 21%. At constant prices, SparkLend added $415 million of net depositor quantity across June.
The distinguishing feature of SparkLend's June inflow is what it does NOT map to. Issue 002's April-May reading was a rotation story: Aave V3 shed $503 million while SparkLend absorbed $752 million and Morpho absorbed a nearly identical $759 million. SparkLend's growth ran opposite to Aave V3's contraction in Issue 002, with the migration mechanism visible in the asset composition: wstETH exiting Aave V3's LRT-adjacent risk and landing on SparkLend's wstETH-heavy structure.
In June, that mechanism did not repeat. Aave V3 absorbed $845 million of net inflow, not a net outflow, and SparkLend absorbed $415 million alongside it. Both protocols received depositor flow. Neither was the destination of an exit from the other. SparkLend's June growth arrived alongside Aave V3's growth, not from Aave V3's contraction.
SparkLend's growth in June arrived alongside Aave V3's growth, not from Aave V3's contraction. The April-May rotation thesis has terminated.
The asset composition of SparkLend's June inflow is broad-based to the extent the captured data resolves it. The weETH-specific component of SparkLend's June net inflow was $25 million on the constant-price series, or roughly 6% of the $415 million total. The remaining 94%, roughly $390 million, sits across the rest of SparkLend's collateral and lending markets.
The captured daily flow data does not decompose the non-weETH balance at the asset level for SparkLend, so the specific mix (wstETH continuation, USDS via the SPK Farming Pool incentive structure, sUSDS deposits, or a mix of all three) is not resolvable from this snapshot. What can be said with confidence is that the June inflow was not weETH-driven, and it was not a single-asset rotation of the kind Issue 002's Morpho deep dive documented for April's Aave-to-Spark migration.
The rate context supports a partial mechanism read. Issue 002's Morpho deep dive documented Spark's USDS market as running at 5.89% supply APY at May 31, the top of the rate matrix by a wide margin. If SparkLend's rate leadership on USDS continued to draw incentive-seeking capital across June, that would explain part of the non-weETH inflow, but the June rate-dispersion capture in this issue does not resolve SparkLend USDS at the same granularity Issue 002 did. The observation is directional rather than measured: SparkLend's structural rate advantage on USDS remained intact from Issue 002, and SparkLend received the second-largest net inflow of any protocol in June. Whether the two are causally linked in the specific proportions the flow data can support is a follow-up for July's data.
SparkLend's position in the June consolidation hierarchy is unambiguously second-tier. Aave V3 absorbed 63% of the sector's net inflow, SparkLend absorbed 31%, and the remaining four protocols split 5%. The pattern is consistent with a sector where depositors under yield compression concentrated at established venues in roughly proportion to how established each venue is.
SparkLend is Aave V3's fork, has run stably on Ethereum since 2023, and holds the second-largest supplied position among the six covered protocols. It received the second-largest inflow. Issue 002 read SparkLend as unique: the only protocol positive on both depositor flow and fees. That held only for May. In June, positive depositor flow is a sector-wide pattern except at Euler V2, and whether SparkLend's fee expansion continued is an open question this issue does not resolve.
06.6Protocol Deep Dive: Compound V3
ompound V3's supplied capital grew 5.4% at nominal prices across June, from $1.23 billion at May 31 to $1.30 billion at June 30 (on the on-chain Comet aggregation this issue uses as ground truth). Constant-price net depositor flow was $59 million, the third-largest positive in the sector behind Aave V3 and SparkLend and an order of magnitude smaller than either. The protocol's LDR held at 33.7%, roughly unchanged from 34.0% at May close.
At June 30, on-chain reads across Compound V3's six Ethereum Comet markets (a Comet is Compound V3's term for each isolated single-asset base market) returned $604 million of base-token supply plus $828 million of collateral, for a $1.43 billion combined Ethereum footprint. The three at-scale Comets carried nearly all of the volume: USDC base at $328 million supply plus $483 million collateral, USDT base at $192 million plus $262 million collateral, and WETH base at $82 million plus $80 million collateral, together accounting for approximately 99% of the protocol's Ethereum book.
The three smaller-base Comets (USDS, wstETH, WBTC) remained sub-scale, with USDS at $1.9 million base supply, wstETH at $259 thousand, and WBTC at literal dollars. The USDS base ran the highest base supply APY of any Comet on the protocol at 4.65%, still above the T-bill even as the sector's median stable APY fell below it.
The June depositor-flow signal is quiet. The +$59 million constant-price inflow is broad-based across the three at-scale Comets, with no single Comet driving the aggregate.
On the borrow side, the WETH Comet carried 85.3% utilization at $70 million of active borrows against $82 million of base supply, which is high enough to be worth flagging: the WETH Comet's per-market utilization sits at levels only Euler V2 runs at the protocol level, and the base supply APY on it at 1.71% reflects the low-yield end of the ETH lending market rather than the high-utilization dynamic. Base and collateral both moved in the same direction (up modestly) across June, consistent with the protocol receiving a small share of the sector's consolidation flow without any single market anchoring the growth.
Compound V3 continued to run the sector's highest oracle monoculture at 93% Chainlink dependency; the smaller USDS Comet continued to run the only Compound market with a supply APY above the T-bill.
Compound V3's position in the June consolidation hierarchy is third-tier: fourth-largest supplied base ($1.30 billion, ahead of Fluid and Euler V2 but behind Aave V3, SparkLend, and Morpho) and third-largest positive inflow at $59 million. The pattern is consistent with the frame the sector overview documents: a small, trusted, established venue receiving modest positive flow without exit pressure or an idiosyncratic driver.
Because each Comet sets its rate independently, the smaller USDS market ran the only positive spread over T-bills of any Compound market on Ethereum at June close. The market's size ($1.9 million base supply) limits the economic significance of that spread. Compound V3's June is the sector consolidation pattern at small scale: what would matter at a larger protocol becomes signal without weight at a protocol running 4.5% of sector supply.
07Looking Ahead
The June data supports six forward calls for July. Each is stated with the mechanism it depends on and the July-data reading that would discriminate cleanly. Issue 004 can honestly correct any call that misses.
The first call sits on the Real Yield Spread. At the July 31 close, expect the spread to land somewhere between −20 and −60 bps. The mechanism is that T-bill yields likely hold at 3.60% through the Federal Open Market Committee's late-July meeting absent a surprise cut, while on-chain rate compression persists at the current sector deposit trajectory. If the FOMC cuts, expect the spread to close from the rate side (T-bill down, on-chain rates roughly flat) rather than from the on-chain side. The discrimination is the July 31 spread print combined with the actual FOMC decision.
The second call is Aave V3's dominance of sector inflow. Expect Aave V3's share of net constant-price sector inflow in July to remain above 50%, most likely in the 55 to 70% range. The mechanism is the consolidation-under-yield-compression thesis the headline documents: if the Real Yield Spread stays inverted, capital keeps concentrating where borrow depth is deepest, and depositors keep accepting the yield penalty in exchange for redemption depth at the largest venue. Fluid's June rate leadership at 6.41% USDC could pull marginal share, but the 50-to-1 Aave-to-Fluid June flow ratio suggests the pull will not be structural. Discrimination sits in the July daily flow data across all six covered protocols.
The third call is Fluid's rate divergence persistence. Expect Fluid USDC supply APY to stay above 5% through July, and Fluid's constant-price net inflow to grow modestly but stay under $50 million for the month. The mechanism is that Fluid's USDC rate is a function of borrow demand on its lending markets, which have been growing through the second quarter of 2026 per the LDR trajectory documented across Issues 001 and 002. Depositor response to the rate stays small because the June 50-to-1 ratio suggests the preference for depth over headline rate is real rather than a one-month artifact. Discrimination sits in the July rate-dispersion reading combined with Fluid's daily flows.
The fourth call is Euler V2 quiet residual continuation. Expect Euler V2 constant-price flow to land within plus or minus $30 million of zero for July, and the loan-to-deposit ratio to hold within two points of 85%. The mechanism is that May's single-operator volatility (Sentora's $128 million Euler V2 unwind documented in Issue 002) has finished at the Euler V2 leg, and June's residual $21 million outflow distributed across many small vaults rather than one operator. Absent a new operator event, the residual is structurally stable. Discrimination sits in Euler V2's daily and per-vault flows at July 31.
The fifth call is LRT depositor stability across a further price move in either direction. If LRT spot prices stabilize or recover in July, weETH quantity on Aave V3 Core grows by at least 5,000 tokens on top of June's 14,000-token addition. If LRT prices fall further, weETH quantity stays flat or grows slightly, and dollar values fall proportionally. The mechanism is that June proved depositor behavior is stable across LRT price drawdowns. The depositor base that held through May and June is unlikely to exit in July absent a protocol-specific event. Discrimination sits in the July Aave V3 weETH-by-market reading, on token quantity rather than dollar value.
The sixth call is inside Morpho's curator layer. Expect Steakhouse Financial to grow its combined V1+V2 share from 29.2% at June 30 toward 30 to 35% by July 31. Expect Sentora to hold or grow their V2 share from the current 31.2%. The mechanism is that the V1-to-V2 vault migration ran on a longer timescale than one month, and new V2 vaults launched in the second quarter of 2026 continue to attract curator attention. Discrimination sits in the July curator HHI reading with the V1/V2 split preserved. If Steakhouse's combined share falls or Sentora's V2 share drops, the call is wrong.
The consolidation-under-yield-compression thesis Issue 003 documents is falsifiable through July's data. If Aave V3's share of sector inflow falls below 50% and Fluid's constant-price flow rises above 75 million dollars, the thesis needs revision. If the Real Yield Spread closes from the rate side while flow patterns hold, June's consolidation was cyclical rather than structural. The Issue 003 reading is that both mechanisms are structural, and the July close provides the discriminating datapoint that either confirms the reading or requires the correction Issue 004 will publish.
Next issue
The next State of DeFi Lending issue lands on the 7th of next month. Drop your email below to get it in your inbox.
Support the work
DatumLabs builds these dashboards and the monthly State of DeFi Lending issues independently, with no paywall and no sponsorships. If the work has been useful for a treasury decision, a research note, or a product call, contributions help fund the next month of analysis.
Feedback
Spotted an error in the data, or want a metric added? You can DM @joel_obafemi on X or email joelobafemii@gmail.com. Both go to the author.