1. Summary
TokenLogic has proposed updating the stablecoin interest rate curves across Aave deployments in Risk Stewards: August 2026 Stablecoin Interest Rate Adjustments. The proposal raises Slope1 by 50 bps on 20 stablecoin reserves (25 bps on Base USDC, which already prices above the other deployments) and realigns the USDe borrow curve with Ethena’s native staking rate through a 5.25% base rate.
In line with that recommendation, we will execute the Slope1 increases and gradually start the USDe base rate adjustments through the Risk Steward process. The Slope1 increases proceed in 10 bps steps, with each step assessed before the next rounds. Our analysis below supports this approach: both Ethereum Core stablecoin reserves now sit at or above their optimal utilization, borrower rate sensitivity is low, and higher supply rates have historically attracted deposits. The complete parameter specification is provided at the bottom of this report.
2. Utilization has been climbing over the past month
Source: LlamaRisk, August 26, 2026
Since early May, USDC utilization has been stabilized around its 92% kink, i.e., it’s already at the target utilization the DAO would want before a rate hike bites, and over the past week it has held above it at 93-94%. USDT decreased for two months after the April’s liquidity crisis, but then began a recovery period since mid June: 76.2% → 92.5% by August 26, crossing its 92% kink in the final week of August.
Taking a deeper look at what’s driving the USDT climb, we observe that over 650M of net new borrows were originated since mid-June, with a roughly 55-45 mix between majors (BTC and ETH and their derivatives) and a whale-heavy USDe carry trade. In particular, over 2,500 addresses have net borrowed ~$360M vs. majors, proving there’s still borrowing demand at these prices. This trend carried the reserve to its kink in the final week of August.
3. Borrowers’ sensitivity is low
A per-address study of every BTC/ETH-collateral borrower who experienced a real rate rise of> 0.5 pp relative to when they opened their position found no net directional signal: 68% of borrowers held flat or grew their position, while only 13.1% partially reduced and 18.4% closed or heavily repaid. The group’s median change was 0%. It took roughly 15 weeks for rates to increase by 50 bps for each borrower during the period analyzed.
4. Supply side: higher rates attract deposits
The past month’s flows show where marginal stablecoin liquidity currently sits. Core USDC absorbed 77M of net deposits over the past 30 days (+4.0%), while Base USDC (-16M), Plasma USDT (-12M) and both Arbitrum reserves were flat or negative. Core USDT shed 80M over the month but took in 501M over the past quarter. Stablecoin supply is consolidating into Core rather than the satellite instances.
Higher supply rates do attract deposits. The supply APY correlates positively with subsequent supply growth on every instance we checked, with the effect building over 4 to 8 weeks: +0.16 to +0.41 on Core USDC and USDT, and +0.35 to +0.76 on Base, Arbitrum and Plasma, where liquidity is more mercenary. A Slope1 hike raises the supply APY at unchanged utilization, so part of its effect is to pull in new supply, which pushes utilization back down. For USDC at the kink this rebuilds the liquidity buffer, which is the intended outcome. With USDT now at the kink, the same mechanism works in the program’s favor: each 10bps step attracts new supply into a reserve whose demand has already cleared the target, rebuilding the liquidity buffer on both reserves.
Source: LlamaRisk, August 26, 2026
5. Market yields and utilization move together
To test the relationship between broader market yields and Core stablecoin demand, we assembled 20 months of daily data (January 2025 to August 2026): USDC and USDT utilization and borrow APR sampled on-chain six times per day and averaged (a borrower pushes USDC utilization to ~100% around midnight every night, so end-of-day snapshots are unusable), sUSDe and sUSDS realized 7-day APY computed from on-chain share prices, and BTC/ETH perpetual funding rates from Binance, annualized. The sample covers the hot market of early 2025 (USDC borrow APR above 10%), the mid-2025 cooldown, the Kelp incident and the current recovery, so it spans several distinct demand regimes.
Source: LlamaRisk, August 26, 2026
The hypothesis that utilization is positively correlated with external yield holds, but the two yield families behave differently. Perp funding correlates with utilization in levels (+0.24 USDC and +0.23 USDT for BTC, +0.27 and +0.31 for ETH). The yield-bearing stablecoin rates instead correlate with Aave borrow rates (+0.66 to +0.79 for sUSDe APY and the Sky Savings Rate) rather than with utilization. That is the shared easing cycle: stablecoin yields and Aave stablecoin rates all fell together from the double-digit levels of January 2025 to roughly 4% today, so their level correlation reflects a common macro factor, not a causal pull on Core utilization.
6. Borrowers’ reaction is delayed
We measure transmission against forward debt growth. It is a quantity outcome, so both legs of the carry spread (external yield minus borrow APR) are behavioral: a high borrow APR slows borrowing (+0.10 to +0.17) and a high external yield accelerates it.
The yield sources separate cleanly:
- Funding rates lead borrowing. ETH funding correlates +0.21 to +0.32 with debt growth over the following 1 to 8 weeks on both reserves, BTC funding similarly at short horizons. The effect builds for roughly 3 to 4 weeks after a funding move.
- The funding carry spread prices the effect: each percentage point of (funding minus borrow APR) maps to about +0.4 to +0.6% debt growth over the next week and +0.8 to +1.0% over the next month (correlations +0.17 to +0.31, R² is low, the relationship is noisy but consistent across reserves and horizons).
- sUSDe APY and the Sky Savings Rate do not lead borrowing on this sample. Yield-stable carry positions respond to spread compression when it happens (Section 7), but the level of stablecoin yields is not a usable leading indicator for Core demand.
Source: LlamaRisk, August 26, 2026
Source: LlamaRisk, August 26, 2026
For the proposed hike this means a 50bps Slope1 increase is, in isolation, a 0.5pp compression of every carry spread, which history maps to roughly 0.5% slower monthly debt growth. That is small next to the ~650M of organic borrow inflow since mid-June and consistent with the low per-address sensitivity found in Section 3. The practical implication is about timing rather than size: steps that land while funding is elevated will be absorbed, steps that land into a funding drought will bite hardest, and since the demand response builds over 3 to 4 weeks, spacing steps about a month apart is the shortest cadence that still lets the DAO observe the true response before the next step.
7. Which positions repay first when rates rise
We classified every address that borrowed or repaid Core USDC or USDT since January 2025 (53k addresses, with the 600 largest closed accounts reclassified from their historical supply mix) by dominant collateral, and compared each cohort’s mean weekly net flow (borrows minus repays) in the 1-2 weeks after a rate rise (>0.35pp over 14 days) against all other weeks. The sample contains 20 such post-rise weeks for USDC and 24 for USDT.
Source: LlamaRisk, August 26, 2026
| Cohort | USDC baseline | USDC post-rise | USDT baseline | USDT post-rise |
|---|---|---|---|---|
| Majors (BTC/ETH and LSTs) | +16.2m/wk | -11.1m/wk | +29.2m/wk | -25.4m/wk |
| Yield stables (sUSDe, PTs, sUSDS) | +1.3m/wk | -12.6m/wk | -12.3m/wk | +41.1m/wk |
| Other stablecoin collateral | -1.6m/wk | +3.1m/wk | -3.1m/wk | +2.1m/wk |
Three observations:
- Majors-collateral borrowers repay first. On both reserves they swing from steady net borrowing to net repayment after a rise, a swing of 27M (USDC) to 55M (USDT) per week. This is aggregate flow dominated by large accounts, and it is compatible with Section 3’s finding that most addresses hold: the median borrower does nothing, while a few large positions supply the week-scale rate response.
- USDC yield-stable loopers also cut (a 14M/wk swing), and relative to their footprint (15B of gross flow vs 75B for majors) they are about 2.5x as rate-sensitive as majors borrowers.
- USDT yield-stable loopers show the opposite sign because causality runs the other way there: the USDT rate rises of the sample were produced by the carry trade’s own expansion (the whale USDe positions from Section 2), and the inflow persisted after the rate had risen. This cohort keys off the carry spread, not the rate level, so it will unwind when the spread compresses regardless of which side moves.
8. Specification
| Instance | Asset | Current Slope1 | Recommended Slope1 |
|---|---|---|---|
| Aave V3 Ethereum Core | USDC | 4.00% | 4.10% |
| Aave V3 Ethereum Core | USDT | 4.00% | 4.10% |
| Aave V3 Plasma | USDT0 | 4.00% | 4.10% |
| Aave V3 Base | USDC | 4.50% | 4.60% |
| Aave V3 Arbitrum | USDT | 4.00% | 4.10% |
| Aave V3 Avalanche | USDC | 4.00% | 4.10% |
| Aave V3 Avalanche | USDt | 4.00% | 4.10% |
| Aave V3 Monad | USDC | 4.00% | 4.10% |
| Aave V3 Monad | USDT0 | 4.00% | 4.10% |
| Aave V3 MegaETH | USDm | 4.00% | 4.10% |
| Aave V3 BNB Chain | USDC | 4.00% | 4.10% |
| Aave V3 BNB Chain | USDT | 4.00% | 4.10% |
| Aave V3 Gnosis | USDC.e | 4.00% | 4.10% |
| Aave V3 Optimism | USDT | 4.00% | 4.10% |
| Aave V3 Linea | USDC | 4.00% | 4.10% |
| Aave V3 Linea | USDT | 4.00% | 4.10% |
| Instance | Asset | Current Base | Recommended Base |
|---|---|---|---|
| Aave V3 Ethereum Core | USDe | 0.00% | 1.00% |
| Aave V3 Avalanche | USDe | 0.00% | 1.00% |
| Aave V3 Mantle | USDe | 0.00% | 1.00% |
| Aave V3 Monad | USDe | 0.00% | 1.00% |
| Aave V3 Plasma | USDe | 0.00% | 1.00% |
9. Next Steps
We will implement these changes via the Risk Steward process. Between steps we will monitor borrowed amounts, deposit growth, and utilization on every in-scope reserve, and pause the schedule for any reserve that shows a contraction beyond its normal variability.
Disclaimer
This review was independently prepared by LlamaRisk, a DeFi risk service provider funded in part by the Aave DAO. LlamaRisk is not directly affiliated with the protocol(s) reviewed in this assessment and did not receive any compensation from the protocol(s) or their affiliated entities for this work.
The information provided should not be construed as legal, financial, tax, or professional advice.