Lending as Infrastructure: How Kamino and JupLend Created Solana USDe Market

In the week of May 13–18, 2026, two Solana lending protocols — Kamino Lend and JupLend — launched isolated USDe/USDG markets within days of each other. Each market accumulated approximately $530 million in total market size in less than a week, generated $949.7 million in USDG DEX volume, and produced $663 million in USDe DEX volume — the latter representing a near-complete creation of a previously non-existent market. The scale of lending adoption is further illustrated by USDe’s onchain concentration: of USDe’s approximately $570 million total market cap on Solana, roughly 96% is currently deposited as collateral across these two lending markets. The lending protocol has become, in effect, the primary venue for USDe on Solana. This article examines the structural design of both markets, the onchain activity data one week from launch, and the key risks facing participants.

Market structure

Both protocols offer structurally similar products. Users deposit USDe, Ethena’s delta-neutral synthetic dollar as collateral and borrow USDG, Paxos’s Global Dollar stablecoin, against it. Both markets operate at 92% LTV with a 94% liquidation threshold and are fully isolated from their respective protocols’ broader lending pools.

The differences lie in the liquidity architecture and product parameters.

Kamino’s USDG pool is sourced via the Ethena Prime Vault managed by Sentora, while JupLend accepts USDG deposits directly, and additionally supports SOL as collateral for borrowing USDe or USDG at 80% LTV — a use case not available on Kamino’s Ethena market. The SOL collateral option is relevant for interpreting JupLend’s user count, as a portion of its wallets are borrowing against SOL rather than running the USDe leverage strategy.

Both platforms offer a one-click leverage feature called Multiply. The mechanism executes in a single atomic transaction: a flash loan of USDG is swapped into USDe, deposited as collateral, and additional USDG is borrowed against that collateral to repay the flash loan.

The USDe rewards APY offered by each protocol has moved in parallel but not lockstep. Both started at approximately 3.5% on May 12. Kamino stepped up to 4.04% on May 14 while JupLend moved more modestly to 3.92%, before both pulled back to near-parity by May 21 at 3.747% and 3.75% respectively. The divergence during the May 14–21 window reflects different incentive adjustment timing rather than a structural difference between the two protocols.

Each Multiply transaction generates a DEX swap — USDG to USDe on a lever-up, USDe to USDG on a deleverage. This is the direct mechanism behind the observed DEX volume.

Market size and utilization

Both protocols reached similar market size endpoints by May 20: JupLend at $530.7 million (USDe + USDG combined), Kamino at $529 million. Both markets reached 100% utilization within the launch week and have remained there since.

The hourly utilization data reveals that the two markets filled at materially different speeds. Kamino accelerated sharply from the outset, crossing 50% by May 14 morning and reaching 100% by May 14 evening, while JupLend was still at approximately 48% at that same moment. JupLend’s ramp was more gradual, reaching 99%+ only by the afternoon of May 15, roughly 24 hours after Kamino had fully filled. This trajectory is consistent with the net borrow data: Kamino’s $140.9 million single-day spike reflects concentrated capital arriving at once, while JupLend’s distributed borrowing across May 13–15 produced a smoother curve.

One notable anomaly occurred on May 16, when JupLend’s utilization briefly dropped from ~99.5% to a trough of 75.5% — due to a $11 million drop in total borrow from $200 million to $189 million — before recovering to 100% within approximately seven hours. Kamino’s utilization only dipped briefly to 95.24% during the same window before recovering. From May 17 onward, both markets have remained at effectively 100%.

Interest rate behavior

The utilization trajectories translated directly into interest rate dynamics, and the borrow curve data reveals that the two protocols use genuinely different rate model designs.

Both use a three-point kinked curve with a 90% utilization breakpoint. The difference is what happens above that kink. JupLend’s rate accelerates sharply above 90%, reaching approximately 2.5% at 100% utilization. Kamino’s curve is configured differently: above the 90% kink, the rate is flat. Kamino’s curve parameters were adjusted multiple times during the observation period, with the borrow rate at 100% utilization stepping upward on three occasions: 1.98% when the pool first filled on May 15, then 2.00% on May 17, and 2.51% on May 18. Each step reflects a manual cap update rather than an automatic response to utilization.

JupLend’s rate moved continuously and automatically throughout the same period without any parameter intervention. The May 16 utilization dip on JupLend from ~99.5% to 75.5% produced a sharp supply rate drop from 2.19% to 1.10%, consistent with falling below the kink point on JupLend’s accelerating curve. Kamino’s utilization only dipped to 95.24% during the same window, remaining above the 90% kink and therefore staying within the flat-cap range — which is why its rate was largely unaffected. At steady state, both protocols converged to similar observed borrow rates — JupLend at 2.52% and Kamino at 2.51%.

User activity and borrowing patterns

From May 12 to May 19, JupLend recorded 1,660 multiply transactions from 676 unique wallets. Kamino recorded 512 multiply transactions from 260 unique wallets. JupLend generated 3.2 times as many loop transactions and 2.6 times as many unique participants.

JupLend’s activity was front-loaded, with the majority of transactions concentrated in the first three days. The decline from May 15 onward reflects both the pool approaching capacity and the multiply strategy being listed as filled. Kamino’s trajectory was slower to start but showed a secondary cluster on May 16, suggesting a different participant profile arriving on a delayed basis.

Net daily USDG borrowing reveals the concentration of Kamino’s activity more clearly. Kamino recorded a single-day net borrow of $140.9 million on May 15, compared to a peak of $95.6 million for JupLend on the same day. Kamino’s subsequent daily net borrows were negligible — below $15,000 — indicating that the bulk of its positions were established in one session rather than accumulated gradually.

This pattern, combined with Kamino’s lower wallet count, is consistent with a smaller number of large participants deploying capital in concentrated tranches. JupLend’s borrowing was distributed more evenly across May 13–16, suggesting broader participation across more independently acting wallets.

Of the 676 JupLend wallets and 260 Kamino wallets, 35 interacted with both protocols. By May 21, 2026 at 7AM UTC, only 25 out of 35 accounts held active positions on both protocols simultaneously. Among these wallets, the leverage distribution skews toward maximum utilization. On Kamino, 52% of active overlapping wallets were levered at 11.5x or above against a 12.5x maximum; on JupLend, 57% were at 11.5x or above against a 12.3x maximum. The median leverage for overlapping wallets was 11.45x on Kamino and 11.74x on JupLend, both close to their respective protocol maximums. This suggests the overlapping wallet cohort — which by definition is sophisticated enough to operate across two protocols simultaneously — is predominantly running near-maximum leverage on both.

DEX volume impact

The most significant secondary effect of the launch was a substantial expansion of DEX volume for both USDG and USDe on Solana. USDG already had an established presence, averaging approximately $661 million in monthly volume from January through April. The launch week from May 13 to 18 generated $949.7 million in USDG volume alone, exceeding any single prior month.

The effect on USDe was more consequential. Prior to May 13, USDe had generated approximately $2.5 million in cumulative DEX trading volume on Solana across the entire year. The Multiply mechanism, by requiring a DEX swap on every loop transaction, converted lending activity directly into trading volume. By May 14, USDe single-day DEX volume reached $256 million, approximately 102 times the prior cumulative total. Combined USDG and USDe volume for the period reached approximately $1.27 billion.

Orca’s USDG-USDe pool dominated routing during the peak activity period, capturing over 90% of USDe swap volume on May 13–15. This concentration reflects the pool having the deepest liquidity for the specific pair required by the Multiply loop. Meteora held 70% share on May 12, the pre-launch day, when volumes were minimal and liquidity was thin across all venues.

The share picture shifts materially from May 17 onward. As Multiply loop activity collapsed and total volume fell to $5–10 million per day, Aquifer’s share rose from near-zero to 43% on May 17 and 28% on May 18. This suggests that the residual daily volume — position maintenance and small new entrants — routes differently from the high-volume loop transactions, likely reflecting different swap aggregator routing decisions at smaller transaction sizes or a shift in where arbitrageurs are sourcing liquidity to maintain the USDG/USDe peg.

Volume declined sharply after May 16, consistent with the Multiply transaction count decline as pool capacity was reached. The May 17–18 run rate of $5–10 million per day in each token represents the ongoing maintenance volume of the existing leveraged position base — periodic rebalances and small position adjustments rather than new position opens.

The DEX liquidity established during the launch week persists. DEX pairs for USDe/USDG that did not exist in meaningful form before May 13 are now active markets with established depth.

Risk considerations

Depeg and liquidation risk. Both protocols hardcode the USDe oracle at 1:1: JupLend to USDG and Kamino to USDT. This means market price fluctuations in USDe do not directly trigger liquidations. However, liquidation risk remains relevant in other scenarios, including oracle failures or manipulation. At high leverage, simultaneous liquidations across the combined $1 billion position base could amplify selling pressure on USDe, and the internal feedback loop between collateral value and forced selling remains a structural concern. Both markets are isolated from broader lending pools, which limits contagion to other assets.

Negative carry risk. USDe yield is variable. If the USDe rewards APY declines while USDG borrow costs remain the same or increase, the position’s net APY can fall or turn negative — meaning the cost of borrowing exceeds the yield on collateral. Because Multiply applies leverage, both positive and negative carry are amplified by the selected multiplier. Both protocols adjusted their rewards APY multiple times within the observation window, and the USDG borrow rate has also stepped upward through parameter adjustments, illustrating that the carry can compress without warning.

Execution and liquidity risk. Opening or closing a Multiply position may be affected by slippage, price impact, and market liquidity. In stressed conditions, or for large positions, exits may be slower or more expensive than under normal conditions. This is particularly relevant given that both pools are currently at full capacity — new positions cannot be opened without existing borrowers first deleveraging.