LlamaRisk Insights: aUSDC/aUSDT as USDe Backing Assets
Summary
This analysis reviews the implications for Aave of Ethena’s newly approved plan to use aUSDC and aUSDT as collateral backing for USDe, allowing up to 10 percent of Ethena’s total stable-coin reserves—about $260 million—to be supplied to Aave. The immediate benefit for Aave is deeper USDC and USDT liquidity, which should temporarily reduce borrow rates and attract additional users.
Several risks accompany this integration. First, it would intensify asset-supply concentration: Ethena’s position, combined with other large holders, could leave a substantial share of aUSDT—and, to a lesser extent, aUSDC—in the hands of only a few entities. Second, Ethena’s deposits are likely pro-cyclical; the protocol is incentivized to provide liquidity when utilization is low and withdraw when utilization and perpetual-funding rates are high, potentially reducing exit liquidity when Aave needs it most. Third, a re-hypothecation layer is introduced because USDe would be backed by assets held on Aave; a severe USDe de-peg could trap that collateral and trigger a feedback loop of liquidity stress for both protocols.
A regression-based withdrawal-responsiveness analysis suggests that the aUSDC market deleverages more quickly than the aUSDT market after large withdrawals, indicating greater resilience to sudden outflows.
Given these findings, continuous monitoring of Ethena’s supply share, overall concentration dynamics, and utilization levels is essential. LlamaRisk will track these metrics and report developments to the Aave DAO as they arise.
Context
A March 2025 governance proposal to add aUSDC and aUSDT to USDe’s backing framework has been approved, clearing both Risk-Committee review and a Snapshot vote (see the original forum post and Snapshot result).
The Ethena Risk Committee imposed three key guard rails:
- Ethena’s combined aUSDC and aUSDT deposits must stay below 10 percent of the two tokens’ aggregate supply on Aave (about $600 million at current levels).
- The allocation may not exceed 10 percent of Ethena’s total stable-coin backing (about $260 million).
- Positions must be actively managed, with size adjusted as yields, utilization, and exit-liquidity thresholds evolve.
Because LlamaRisk sits on the Risk Committee, we recused ourselves from issuing recommendations; independent members Blockworks Research and Untangled Credio authored the official risk parameters.
Assessment
Increased Liquidity for USDT and USDC
The most direct implication of this adoption is an expected increase in liquidity for USDC and USDT on Aave. The final amount allocated is subject to the exposure limits set by Ethena’s Risk Committee, which initially would be at 260M, representing approximately 4.5% of all the USDT and USDC supplied combined. Initially, this would reduce utilization rates, leading to increased borrow demand and attracting new users due to lowered rates.
We have recently observed such instances, particularly when HTX deposited 1B USDT in 3 days, lowering utilization rates and attracting new borrows. This was executed without apparent negative consequences to the market’s stability. Therefore, expected inflows from Ethena, which would be lower than the observed deposits of HTX, would be largely neutral to the broader stability of Aave’s Core market.
Source: Arkham, May 13th, 2025
Concentration Risk
While Ethena’s deposits would bolster liquidity, they also have the potential to become a large share of Aave’s total supply of aUSDT and aUSDC and could lead to considerable concentration.
It is prudent to consider the potential impact on Aave’s liquidity should Ethena need to rapidly unwind these positions due to its own risk management requirements or market conditions affecting USDe. During periods of high borrow demand, the utilization rate hovers around the targeted 92% for both assets, with aUSDC showing a slightly higher utilization (12-month average 81% against 77% for aUSDT). This could make it difficult to quickly withdraw a large percentage of the assets.
Aave has experienced periods when the utilization ratio is low, leaving much room for withdrawals. At the same time, there have also been instances when the utilization is high, making it very difficult to process large withdrawals in a short timeframe, offering different levels of liquidity at different times.
Source: LlamaRisk, May 15th, 2025
Source: LlamaRisk, May 15th, 2025
In addition, HTX currently has around 35% of the supply of aUSDT, which is already an elevated concentration level. If Ethena makes larger deposits in the future, we could see two entities holding over 40% of the supply of aUSDT. It cannot be predicted if the entities would withdraw their assets simultaneously, but if that were to happen, it could potentially lead to liquidity crunches in the aUSDT market until borrowers repay their debt.
Source: Arkham, May 14th, 2025
The experience with HTX holding over $1.5B in aUSDT and making withdrawals of up to $1B in a single day shows that large withdrawals can be processed without apparent stability consequences. However, it is important to note that these withdrawals occurred during low utilization ratios, as seen in the graph below. For all the withdrawals, the utilization was around 60%.
HTX major withdrawals:
| 2025-03-19 | 2025-03-30 | 2025-04-12 | 2025-04-21 | 2025-05-07 |
|---|---|---|---|---|
| 1.0B | 1.0B | 0.9B | 0.9B | 1.0B |
Source: LlamaRisk, May 15th, 2025
Procyclicality of Deposits
It is also notable that Ethena could provide liquidity to Aave when it is less needed and withdraw it when the demand for borrowing is highest. Both protocols benefit from overall higher long-leverage demand, which translates to higher utilization rates for Aave and higher funding rates for Ethena. This could prompt Ethena to allocate more stablecoins to perpetual futures hedging positions, withdrawing the stablecoin deposits from Aave.
Source: Ethena, May 12th, 2025
Source: Aave, May 12th, 2025
Without considering potential periods of market stress that might cause Ethena to redeem its supplied assets, Ethena may seek to redeem its assets during periods of high utilization ratios, potentially leading to liquidity crunches; this can be due to higher funding rates or Ethena’s active management strategy, which involves withdrawing assets to protect against exit liquidity falling below 1.25 times the supplied capital.
Rehypothecation Risk
Due to Ethena having part of USDe backing assets on Aave, in a hypothetical situation where USDe becomes severely de-pegged or uncollateralized, potentially creating bad debt on Aave, it would not be possible for Ethena to withdraw the assets until bad debt is repaid or covered. This could worsen the situation for Ethena, as some of its USDe backing assets might be lost. This would also negatively affect Aave due to the ~$1.4B of USDe and USDe derivatives deposited on Aave, which could potentially lose more value, creating a negative feedback loop.
Source: Dune, May 12th, 2025
Nonetheless, this scenario would be highly unlikely with the present risk parameters for the USDe markets, its limited supply caps, and Ethena’s active management of the assets deposited. However, a new risk vector would be introduced if Ethena deposits part of its USDe backing into Aave. Therefore, this consideration should not be overlooked.
Withdrawal Responsiveness Analysis
Setup
To better understand how Aave’s aUSDC and aUSDT markets might react to large capital movements, particularly withdrawals, we conducted a quantitative analysis where we employed responsiveness analysis to estimate the responsiveness of borrowing demand to short-term changes in deposits. In particular, we used these parameters to perform the modeling:
- Deposit Change Window (Independent Variable): Change in total asset supply over the preceding 3 and 24 hours.
- Borrow Change Window (Dependent Variable): Change in total asset borrows over the next 3, 24 and 72 hours.
Responsiveness Across Borrowing Windows
The following table summarizes how Aave’s Core market borrowers respond to deposit withdrawals. The table shows beta coefficients (β₁) indicating the $ change (reduction if β₁ is positive) in borrow amount per $1 of deposit withdrawal. Only statistically significant (p < 0.05) beta coefficients are indicated.
All scenarios are for an initial 3-hour deposit change window, except the last column, which refers to a 24-hour deposit change window.
| Asset | Withdrawal Scenario | BorrowΔ Next 3h (β₁) | BorrowΔ Next 24h (β₁) | BorrowΔ Next 72h (β₁) | BorrowΔ Next 72h (for 24h DepositΔ) (β₁) |
|---|---|---|---|---|---|
| USDC | |||||
| Any Withdrawal (> $0) | 0.0926 | 0.2673 | 0.4364 | 0.4337 | |
| Withdrawal >= $10M | 0.0881 | - | - | 0.5942 | |
| Withdrawal >= $20M | 0.1658 | 0.3352 | - | 0.4890 | |
| Withdrawal >= $30M | 0.2093 | 0.3828 | 0.9857 | – | |
| USDT | |||||
| Any Withdrawal (> $0) | 0.0177 | 0.1384 | 0.2182 | 0.1297 | |
| Withdrawal >= $10M | - | 0.1236 | 0.2335 | 0.1935 | |
| Withdrawal >= $20M | - | 0.0945 | 0.2126 | 0.2139 | |
| Withdrawal >= $30M | -0.0073 | 0.0841 | 0.1846 | – | |
| Withdrawal >= $100M | – | – | – | 0.1658 | |
| Withdrawal >= $250M | – | – | – | 0.3342 |
Key Findings
The aUSDC market is more responsive to reductions in deposited assets; this, combined with its lower concentration, makes the aUSDC market more attractive than the aUSDT market from a risk management perspective, potentially making it easier to withdraw assets, especially in periods of high utilization, even though it is unclear how it might react to larger withdrawals due to lack of statistically significant data. The overall indications were:
- USDC withdrawals during high utilization (>95%) lead to rapid and sustained reductions in borrowing. Within 3 hours, 9–21% of the withdrawn amount is offset, and for withdrawals of $30M or more, nearly all is offset within 72 hours (Beta ≈ 0.99). In similar scenarios, USDT is only about one-third to one-half as responsive as USDC.
- USDC borrowing becomes more sensitive as withdrawal size increases. Any withdrawal reduces borrows (Beta ≈ 0.44 over 72h), but for withdrawals ≥$30M, the effect nearly doubles (Beta ≈ 0.99), showing strong deleveraging within 72 hours.
- USDT borrowing contracts after withdrawals across all sizes and timeframes, but less than USDC (Betas 0.08–0.23). Its 3-hour responsiveness is minimal (Beta ≈ 0.02), and it shows unusual behavior for very large withdrawals, likely due to more frequent extreme events in the dataset.
Disclaimer
This review was independently prepared by LlamaRisk, a community-led decentralized organization funded in part by the Aave DAO. LlamaRisk is an Ethena’s Risk Committee member and an independent attestor of Ethena’s PoR. LlamaRisk did not receive 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.