Proposal: Onboarding Aave's aUSDC and aUSDT as USDe Backing Assets

[Stani]

Summary

This proposal seeks to onboard Aave’s aUSDC and aUSDT as eligible backing assets for USDe. Both assets offer deep liquidity and competitive return opportunities, enhancing the resilience and capital efficiency of USDe. Their inclusion aligns with Ethena’s goal of diversifying collateral sources while maintaining robust risk management practices.

Asset Overview

Issuer	Token	Underlying Asset	Yield Source	3m Avg APY	6m Avg APY	Supply (Ethereum)
Aave	aUSDC	USDC	Interest accrued from borrowers on Aave	8.6%	6.8%	$880M
Aave	aUSDT	USDT	Interest accrued from borrowers on Aave	7.4%	5.8%	$1.4B

Capital Efficiency

As part of the backing management of USDe there are often large holdings of liquid USDC and USDT in the USDe backing to facilitate withdrawals. Allocating to aUSDC and aUSDT with a portion of the USDe backing helps improve the capital efficiency of these non rewarding assets.

The yields for both assets regularly outperform treasury rates thanks to increased demand for borrow on Aave. Of particular interest to Ethena are the spikes in yields when borrow demand increases, as we saw in December when yields for both assets spiked above 20% consistently. In these scenarios, the protocol could benefit from yield spikes without taking on any additional risk. In an environment where interest rates on T-Bills are likely to be cut next year, aUSDC and aUSDT could be compelling options.

Ethena also has allocated over $1.3bn to USDS, which is reducing its rates as market rates fall. Replacing some of this allocation with aUSDC and aUSDT might make sense from a diversification point of view, reducing idiosyncratic risk to any one issuer.

Implementation

If approved by the Ethena Risk Committee, Ethena will integrate aUSDC and aUSDT into USDe’s backing asset framework up to a maximum of 30% of the total supply of both aUSDC and aUSDT respectively on Aave. The allocation will be periodically reviewed based on liquidity conditions, risk factors, and yield efficiency.

Conclusion

Onboarding aUSDC and aUSDT as USDe backing assets could enhance protocol diversification, improve yield efficiency on non-rewarding USDT and USDC, and strengthen USDe’s liquidity profile. This proposal aligns with Ethena’s long-term vision of maintaining a robust and adaptive collateral base for USDe.

We welcome feedback from the community and look forward to discussions regarding the integration of these assets into USDe’s backing.

[muntangled]

Introduction

In response to the proposal to onboard Aave’s aUSDC and aUSDT as USDe backing assets, we, Untangled Credio, present an analysis of the liquidity and yield dynamics of these assets on the Aave protocol, to inform the community’s evaluation of the proposal.

This analysis is independent of the one on USDtb posted under that thread. For a combined view of the Ethena Risk Committee across all Aave stablecoin supply proposals (USDC/T and USDtb), go here.

Supply caps

Our simulation engine, trained on 30 days of historical data and projecting over a 7-day horizon, analyzed 10,000 liquidity trajectories for aUSDC and aUSDT at Ethereum block 22247319. Key findings include:

• Liquidity up to Optimal Utilization Threshold: To avoid pushing Aave’s utilization rate above its optimal threshold (92%), aUSDC maintains a mean liquidity of $570M (1st percentile: $372M), and aUSDT shows $733M (1st percentile: $148M).

• Total Available Liquidity: aUSDC offers a mean available liquidity of $576M, with a 1st percentile (worst-case) of $372M. For aUSDT, the mean is $935M, with a 1st percentile of $275M. This suggests substantial withdrawal capacity, even in stressed scenarios.

We recommend a combined supply cap of $500 - $600m - Staying below these thresholds ensure stable APRs and reduces the risk of liquidity crunches. However, aUSDT’s lower 1st percentile liquidity suggests it may face tighter constraints under extreme conditions, warranting cautious allocation.

Yield Potential

We estimated the supply APR (yield) for liquidity providers based on Aave’s interest rate model and simulated utilization rates. The average supply APR over the 7-day forecast reflects stable returns, driven by Aave’s piecewise interest rate structure (base rate: 0%, slope 1: 6%, slope 2: 35%, reserve factor: 10%). While exact APRs vary with utilization, our simulations show that yields remain competitive and predictable, aligning with USDe’s goal of stable backing asset returns. High utilization scenarios (>92%) could spike APRs, but Aave’s model typically restores equilibrium by incentivizing repayments or new liquidity.

Backtest results

We backtested our simulation engine across 100 rolling 30-day windows, starting from December 13, 2024 (block 21395862). The realized minimum liquidity for aUSDC and aUSDT consistently fell within the predicted distributions.

Other Considerations

While aUSDC and aUSDT exhibit large liquidity, Ethena should consider:

• Liquidity Volatility: aUSDT’s higher standard deviation ($330M vs. aUSDC’s $80M) suggests greater liquidity fluctuations, which could impact USDe’s stability in certain scenarios.
• Utilization Spikes: Exceeding the 92% optimal utilization threshold could temporarily elevate APRs and reduce available liquidity, though Aave’s design mitigates prolonged imbalances.
• Protocol Risks: As with any DeFi protocol, Aave faces smart contract and market risks, which should be monitored for USDe’s backing portfolio.

Recommendation

Our analysis supports the onboarding of aUSDC and aUSDT as USDe backing assets due to their deep liquidity and reasonable yields. We recommend a combined supply cap of ~ $500m - $600m for USDC/T.

We advise ongoing monitoring of utilization rates and liquidity metrics to ensure alignment with USDe’s peg stability objectives. Also the total exposure to Aave across all stablecoin markets USDC/T and USDtb should be capped and monitored.

We welcome community feedback. For detailed methodology and results, refer to our full analysis (link-to-whitepaper).

[BlockworksAdvisory]

Introduction

Following up on this proposal — as well as the proposal regarding Ethena supplying USDtb to Aave — Blockworks Advisory conducted a detailed risk analysis on the broader integration of Aave into USDe’s backing framework.

The full report can be found here.

Key Points: Risks & Recommendations

• Cap Capital Supply to Mitigate Multi-Faceted Risks: Limit the aUSDT and aUSDC supply on Aave to between 5% and 10% of their total aUSDT and aUSDC supply combined (which would currently translate to between $313.5M and $627M). This restricts exposure to redemption, token-specific, insolvency, and governance risks while also managing the inverse relationship between yield and available exit liquidity.
• Set a Minimum Yield Threshold: Ensure that any deployed capital in USDC and USDT markets earns at least a 25 basis points premium over the higher of either the risk-minimized rate (approximately 4.5%) or the risk-free rate (about 4.4%).
• Adopt an Active Management Strategy: Rather than maintaining passive positions, dynamically manage the supplied capital based on the highest yield available. Monitor the exit liquidity and withdraw capital when the market hits an exit liquidity of 1.25 times the supplied capital, thereby protecting against adverse liquidity shifts.
• Aave Backing Collateral Cap: Ethena should limit its allocation of stablecoin backing to Aave to at most 10% of the total stablecoin backing (which would currently be $332M).
• Control Looping by Retaining a Predominant Stake: To mitigate the risk that looped USDtb positions could indirectly dominate sUSDe supply (which should stay below 10% of the sUSDe total supply), Ethena should initially supply at least 50% of the USDtb market’s total supply. The market’s total supply cap should start with a conservative market supply cap of around $40M to minimize risks, which is in line with historical initial caps on Aave.

Onboarding Aave’s aUSDC and aUSDT as USDe Backing Assets

To recap, the proposal seeks to onboard aUSDC and aUSDT as backing assets for USDe because they offer a higher yield compared to both the risk-minimized rate provided by sUSDS (currently at 4.5%) and the risk-free rate offered by US3M (4.41%), currently being captured through the USDtb portion of USDe backing; additionally, it presumes that if interest rates on T-Bills are likely to be cut next year, aUSDC and aUSDT could be compelling backing asset options.

Research Questions

1. Yield Analysis: Do the yields for both USDC & USDT markets on Aave regularly underperform, track, or overperform current sUSDS / US3M rates?
2. Token-Specific Risk: How does peg stability of USDC & USDT affect supplying collateral to those markets on Aave?
3. Redemption Requests: What risk is associated with redemption requests for USDC & USDT markets on Aave?
4. Market Insolvency: What is the risk of bad debt in USDT & USDC markets on Aave?
5. Governance Risks: What are the Aave DAO governance risks incurred by supplying liquidity on Aave to USDC & USDT markets?
6. Expected Returns: What are the expected returns for supplying capital to USDC & USDT markets in a bear and bull scenario on Aave?
7. Supplied Collateral Caps: What should the max supply cap of aUSDT and aUSDC be?
8. Minimum Acceptable Yield: What minimum yield should Ethena accept relative to the risk associated with both USDC & USDT markets on Aave?

While ethaUSDC and ethaUSDT are some of the most liquid markets on Aave and return meaningful yield, the return from supplied capital is relative to the utilization rate of these markets, and so too is the redemption risk. As yield grows, available exit liquidity decreases. The inverse relationship of yield and available exit liquidity is backdropped by token-specific risk (peg stability), market insolvency (bad debt), and governance risk. Due to the inverse relationship between yield and available exit liquidity—compounded by token-specific, insolvency, and governance risks—ethaUSDC and ethaUSDT positions on Aave must be actively managed to balance risk-reward relative to risk-minimized or risk-free rates.

Yield Analysis

Question: Do the yields for both USDC & USDT markets on Aave regularly underperform, track, or overperform current sUSDS / US3M rates?

• Null hypothesis: Excluding outliers, the median yields for both USDC & USDT assets regularly fall in line or below current sUSDS/treasury rates thanks to moderate utilization ratios

• Alternative hypothesis: Excluding outliers, the median yields for both USDC & USDT assets regularly outperform sUSDS/treasury rates thanks to increased utilization ratios

Overall: After removing statistical outliers, the alternative hypothesis is rejected, and the null hypothesis is upheld. As a result, Ethena should pursue active management of ethaUSDT and ethaUSDC based on the minimum required yield and available exit liquidity.

Methodology Details

Defining Statistical Outlier: We excluded any data points that deviated more than 1.5 times the interquantile range, i.e. below Q1 − 1.5 × IQR and above Q3 + 1.5 × IQR from the full time series. We applied this statistical filter to the supply APY data to reduce the impact of outliers and better reflect the underlying trend in supply APY conditions.

Additional Information: The IQR method helps identify and remove extreme outliers by using the middle spread of the data, ensuring that the analysis is less skewed by unusually high or low values that don’t reflect typical market conditions. The resulting dataset, “USDT Filtered vs Raw Supply APY Over Time”, retains the core structure of the original while improving interpretability for downstream analysis. The IQR filter removes the following percentage of data points for USDC (7%) and USDT (5%) from the full data set.

USDT Utilization Rate & Filtered Data

Overall: The first-of-its-kind spike in utilization for the USDT market created a temporary liquidity crunch, causing the associated lending and borrowing rates to surge well above historical norms and marking Q4 2024 as a distinct statistical anomaly in the overall dataset.

USDT Utilization Rate

• Record High Utilization: In November 2024, USDT utilization hit a record high of 95.71%
• Trailing Utilization Rate: The trailing 6-month average USDT utilization rate is 77.97% and the trailing 12-month average USDT utilization rate is 78.61%
• Logical Conclusion: The utilization rates exhibited in November 2024 for USDT are statistical outliers

USDT Filtered vs Raw Supply APY

• Raw APY (Grey) In November Is An Outlier: The supply APY in November 2024 is visually a statistical outlier relative to every point in time except the beginning of the USDT market on Aave V3
• Logical Conclusion: To observe the regular supply APY of the USDT market, material deviations from the mean should be removed, and a filtered supply APY should be used (red).

USDC Utilization Rate & Filtered Data

Overall: The unprecedented surge in USDC market utilization led to a temporary liquidity crunch, driving lending and borrowing rates to climb sharply beyond historical averages and distinguishing Q4 2024 as a notable statistical outlier in the overall dataset.

USDC Utilization Rate

• Record High Utilization: In November 2024, USDC utilization hit an unprecedented high of 99.74%
• Trailing Utilization Rate: The trailing 6-month average USDT utilization rate is 81.53%, and the trailing 12-month average USDT utilization rate is 82.88%
• Logical Conclusion: The utilization rates exhibited in November and December 2024 for USDC are statistical outliers

USDC Filtered vs Raw Supply APY

• Raw APY (Grey) From October ‘24 To January ‘25 Is An Outlier: The supply APY from October 2024 to January 2025 is visually a statistical outlier relative to every point in time since the beginning of the USDC market on Aave V3
• Logical Conclusion: To observe the regular supply APY of the USDC market, material deviations from the mean should be removed, and a filtered supply APY should be used (red).

Supply APY Distribution & Extreme Outlier Analysis

Including Outliers: Supply APY Distribution & Extreme Outlier Analysis

Analysis and Charts

Overall: Before excluding outliers, data reveals that both USDT and USDC show much lower mean, median, and standard deviations when considering data since 2023 than 6-month figures. Additionally, USDT exhibits better capture of extreme APY events due to its relatively higher kurtosis; the most recent six months of supply APY are statistically uncharacteristic for both tokens, and USDC’s median supply APY does not materially outperform either the risk-minimized or risk-free rates.

Distribution Since Aave v3 USDT Genesis:

• USDT Higher Risk Premium Than USDC: Both Borrow APR and Supply APY for USDT distributions exhibit less uniformity than USDC, possibly signaling greater volatility or/and a higher risk premium in USDT markets on Aave.

Summary Statistics Since Aave v3 USDT/USDC Market (Genesis):

Summary Statistics USDT/USDC Market (6-Month):

Summary Statistic Analysis:

1. Evaluating 6-Month Summary In Isolation: Taking the 6-month summary statistics in isolation would lead to the conclusion that both USDT – mean 6.1% and median 4.5% – and USDC – mean 14.2% and median 7.12% – offer higher average/median yields than the risk-minimized sUSDS rate (4.5%) and the US3M (4.4%). It would also cause the observer to believe that the volatility of USDC supply APY is substantially higher than that of USDT supply APY, and the relatively low kurtosis of USDC signals USDT captures better extreme APY events. The conclusion would be that reallocating a portion from the sUSDS allocated backing ($1.3B) to aUSDT and aUSDC could increase expected returns.
2. Evaluating Both Summary Statistic Periods: Zooming out from the 6-month summary statistic to the genesis statistics, it is self-evident that the regular supply APY mean (USDT: 5.2% USDC: 7.5%), median (USDT: 4.2% USDC: 4.8%), stdev (USDT: 4.5% USDC: 9.1%) are much lower. Upholding the 6-month summary statistics, the relatively low kurtosis of USDC indeed signals USDT captures better extreme APY events.
3. Logical Conclusion: Even before removing the extreme outliers from the supply APY of both USDT and USDC markets, the overall picture shows that the previous six months of supply APY were statistically uncharacteristic of both. Furthermore, the median supply APY of USDC is below the risk-minimized (4.50%) and risk-free rate (4.41%). While the averages are higher, they are driven by infrequent spikes and not consistent outperformance, indicating the median is a more consistent measure.

Excluding Outliers: Supply APY Distribution & Extreme Outlier Analysis

Analysis and Charts

Overall: After removing statistical outliers, USDC consistently outperforms USDT with an average APY of 5.2% compared to 4.5%, though both averages hover around the sUSDS benchmark (4.5%). The median analysis reinforces this stability, showing that USDC’s median yield (4.47%) nearly meets risk-free benchmarks, while USDT’s median (4.03%) falls short. Though it is true that USDC’s mean remains notably higher than benchmark rates by about 159bps, the median yields for both assets generally align with or fall below risk-minimized rates, indicating that passive holding would not yield additional benefits. Consequently, these insights suggest Ethena should adopt an active management strategy rather than a passive holding approach to maximize risk-to-return relative to benchmark.

All-Time Supply APY Distribution Comparison:

• Including outliers: both USDT and USDC supply rates mostly cluster in the low single-digit APY range. USDT exhibits narrower distribution with higher infrequent supply APYs, displaying less day-to-day variation.
• Excluding outliers: while both USDT and USDC supply APYs on Aave cluster in roughly the same range, USDC has a slightly broader distribution with higher potential yields toward the upper end.

Summary Statistics Since Aave v3 USDT/USDC Market (Genesis) (Ex\ Outliers):

Summary Statistics USDT/USDC Market (6-Month) (Ex\ Outliers):

Sidenote: USDC and USDT filtered supply APY was used to produce the visualized summary statistics.

Summary Statistic Analysis:

1. USDC APY Leads USDT: After removing statistical outliers, USDC (5.18%) maintains a stronger average yield than USDT (4.52%). Both rates hover near or slightly above the sUSDS benchmark of 4.5%, yet they remain noticeably lower than previous summary figures — with mean supply APYs since genesis at 5.2% for USDT and for 7.3% USDC and fall even further beneath the six-month averages of 7.4% for USDT and 14.2% for USDC.
2. USDC Median Yield Stays Near Benchmarks: The median APY for USDT (4.0%) is short of the benchmark risk-minimized yield (4.5%) and risk-free yield (4.41%), while USDC (4.5%) nearly matches it.
3. Medians Lag Risk-Free Rates: When excluding statistical outliers, it is apparent that median supply APY from genesis (USDT: 4.0% USDC: 4.5%) and the previous six months (USDT: 4.5% USDC: 4.7%) have not materially outperformed the risk-minimized sUSDS rate (4.5%) and the risk-free US3M (4.41%). Notably, this conclusion does change when looking at the mean for USDT (5.17%) and USDC (6.09%), which is 69bps and 159bps above the risk-minimized rate. However, it is thought that the utilization rates for both markets during the previous six-month period were uncharacteristic and non-recurring.
4. Passive Holding Isn’t Justified—Active Strategy Needed. Based on the all-time and six month median, we can state confidently that Ethena would not want to hold these assets passively because they would not regularly outperform the risk-minimized and risk-free rate. As a result, the alternative hypothesis is rejected, and the null hypothesis is upheld. Excluding outliers, the median yields for both USDC & USDT assets regularly fall in line or below current sUSDS/treasury rates thanks to moderate utilization ratios. Instead of passively holding, Ethena should pursue active management. Additionally, Ethena’s portion of aUSDT or aUSDC will inevitably also have an impact on borrowing dynamics, especially if it adds up to a large percentage of total supply. While extra supply is expected to lower yield in the short-term, medium to long term a deeper liquidity can attract more consistent heavy borrowing, changing the analyzed dynamic.

Peg Stability

Overall: The primary defense against these risks is to limit the exposure to assets that could potentially suffer from peg instability. By capping the allocation to Aave’s aUSDC and aUSDT at a level that reduces systemic interdependency, Ethena can minimize the adverse effects of any sudden liquidity crisis or depegging event.

Question - Token-Specific Risk: How does peg stability of USDC & USDT affect supplying collateral to those markets on Aave?

USDC and USDT aim to hold a 1:1 parity with the U.S. dollar, but temporary depeggings can and have occurred before. When the peg deviates, confidence in the stablecoin weakens, market liquidity diminishes, and unexpected losses may occur for Aave lenders holding aUSDC or aUSDT. Specifically, if redemptions or conversions become less reliable:

• Increased Borrowing Costs and Volatility: Temporary depegging can lead to sharper swings in collateral values, utilization rates, and escalate borrowing costs, exposing both lenders and borrowers to sudden market shifts.
• Systemic Risk Amplification: A deviation undermines the overall stability of the protocol and may trigger a cascade of defensive actions, further destabilizing the market.

Even though USDC, USDT, and other similar tokens have a strong track record of stability, it is prudent to consider extreme scenarios for a complete analysis of risk. For instance, in March 2023, USDC de-pegged for a few hours in the wake of the Silicon Valley Bank collapse, inciting market panic. In a scenario like this it is possible that:

• Many Aave depositors rush to withdraw USDC to avoid exposure,
• Speculators aggressively borrow USDC to short the asset,
• Utilization rates spike to extreme levels (approaching 100%), and
• New deposits almost completely disappear, leaving Aave with virtually zero withdrawable USDC until the situation stabilizes.

In such a stressed environment, if Ethena acts as a major supplier of these assets, it risks being unable to withdraw a substantial portion of its holdings precisely when it is most needed. This creates a scenario where extreme market events not only disrupt the peg but also freeze liquidity within the Aave ecosystem. It is paramount for Ethena to avoid this. To do so, Ethena must limit how much capital is at risk and how exposed it is to external events or market panic.

Redemption Risks Of aTOKENs

Overall: When utilization rate is high, supply APY becomes very volatile, and exit liquidity diminishes. Ethena should mitigate its exposure to capturing outlier supply APY because it corresponds with increased redemption risk and, thereby, elevated exposure to depeg scenarios. To capture the greatest yield with the minimum risk, the amount of capital supplied should be based on available exit liquidity and the minimum required return relative to benchmark.

Question: What risk is associated with redemption requests for USDC & USDT markets on Aave?

Background On aTOKEN Redemption Process

Within the Aave protocol, aTokens are issued on a one-to-one basis with their underlying assets—depositing 1 USDC yields 1 aUSDC, which can later be redeemed for your original USDC plus accrued interest. The value of these aTokens is maintained at a 1:1 ratio with the corresponding supplied asset, and yield is continuously credited by tracking changes in the liquidity index between deposit and withdrawal. The redemption risk here refers to the possibility that, when you withdraw, the protocol may not be able to fully convert your aTokens back into the underlying asset. This risk is not a matter of legal perfection or priority but is instead determined by the market’s available liquidity, protocol’s solvency, and the security of its governance mechanisms.

USDT Exit Liquidity:

• Record Lows For Exit Liquidity: During November 2024 utilization rates hit their highest level ever at 95.71%, dropping available exit liquidity to around $200m for the period and to 4.29% of total supplied capital.
• Comparing Exit Liquidity Across USDT & USDT: Perhaps the reason for higher available liquidity of USDT relative to USDC was that slope 2 for USDT was set to 100% versus USDC’s slope 2 of 60% (Source).
• Logical Conclusion: Extremely high utilization rates lead to elevated supply APY periods, while simultaneously reducing the liquidity cushion for suppliers, demonstrating an inverse relationship between yield and exit liquidity.

USDC Exit Liquidity:

• Record Highs For Utilization Rate: From October to December 2024 utilization rates hit an all-time high of 99.74%, reducing available exit liquidity to around $50m for the period and to 2.75% of total supplied capital.
• Aave Increases The Cost Of Borrowing: In response to both record utilization levels, on 12/15/24 Aave DAO increased slope 1 for all stablecoin assets to raise the cost of borrowing under Uoptimal and decreased slope 2 for all stablecoin assets (except USDT, USDC, and DAI) to lower the cost of borrowing over Uoptimal. Later in February of 2025 Aave DAO decreased borrow slope 1 for all stablecoin assets – still remaining above the slope 1 rates before the 12/15/24 raise.
• Logical Conclusion: The increase to borrowing costs in response to increased utilization suggests that supply was not increasing at a faster rate than borrowing demand and action needed to be taken. USDC high utilization rates boosted supply APYs but reduced the liquidity cushion, displaying an inverse relationship between yield and exit liquidity.

It’s also important to zoom in on how Aave’s utilization ratio and available liquidity have historically behaved during periods when the supply APY was above the benchmark, as these are the scenarios most relevant to Ethena. For USDC, the supply APY exceeded 4.75% on 370 out of 786 days (47.07%), and for USDT, on 308 days (39.19%). Below, we show how utilization and exit liquidity were distributed during those periods. While Ethena’s participation would likely improve liquidity conditions somewhat, historically, available liquidity for withdrawals has mostly stayed below $200M, with utilization consistently high around 90%.

Risk Of Bad Debt

Overall: A new system to cover bad debt (Umbrella), paired with the past precedent of Aave DAO acting as the lender of last resort ($100m in treasury) suggests that impactful bad debt is a low probability.

Question: What is the risk of bad debt in USDT & USDC markets on Aave?

Time and again Aave DAO has historically been the Lender Of Last Resort (LOLR) with Marc Zeller stating, “The few excess debt events in Aave protocol history have been addressed by governance through the mobilization of “cash” DAO treasury funds” (Aavenomics TEMP CHECK). This past precedent implicitly affirms that Aave DAO is willing and able to clear bad debt.

Furthermore with the addition of Umbrella (a new integration on Aave that enables users to stake aTokens and earn rewards by taking on the responsibility of covering bad debt) the risks of bad debt accruing to Aave DAO are further reduced.

Risk Of Governance Influence

Overall: Since protocol emergency admin functions are reasonably scoped and members of the Guardian Committee are doxed, the risk of negative governance influence on lending positions is minimal.

Question: By supplying liquidity on Aave to USDC & USDT markets what are the risks incurred by Aave DAO?

The three primary admin function by which Aave DAO can pose a risk to lenders/supplied of any market are:

• Emergency Admin: are a member of the guardian multisigs, one, is a failsafe emergency actor for cross-chain messaging in Emergency Mode, and, two, is able to “veto” an onchain payload if it is deemed malicious
• Risk Admin: manage protocol risk parameters through verified on-chain processes and pool admin functions.
  • Pool Admin: Used Aave protocol V3’s new PoolConfigurator contract to enable or disable assets in isolation mode, set debt ceilings, adjust supply and borrow caps, modify liquidation fees, create new eMode categories, and update various protocol fees.

Emergency Voting Mechanisms (Guardian Committee): Emergency Voting Mechanisms, particularly the Guardian multisig, serve as a critical safeguard against takeover attempts of the Aave Protocol. By allowing a 5-of-9 multisig veto to cancel malicious proposals or pause/unpause important functions, the Guardian provides a last-resort defense to protect the protocol and its users—especially when a large amount of AAVE resides on centralized exchanges.

Protocol Emergency Guardian:

• Chaos Labs (risk service provider).
• Llamarisk (risk service provider).
• Karpatkey (finance service provider).
• Certora (security service provider).
• Tokenlogic (finance service provider).
• BGD Labs (development service provider).
• ACI (growth and business development service provider)
• Ezr3al (Aave DAO delegate).
• Stable Labs (Aave DAO delegate).

Governance Emergency Guardian:

• Seb (Zapper)
• Mounir (Paraswap)
• Gavi Galloway (Standard Crypto)
• Nenad (Defi Saver)
• Fernando (Balancer)
• Roger (Chainlink community)
• Mariano Conti (DeFi OG)
• Marin (Lido)
• Certora (security service provider)

Expected Returns Based On Supply Cap

Overall: Even under reasonable bull utilization levels of 85% for USDT and 89% for USDC, the expected supply APY is 3.82% and 4.12%, respectively. It is not until utilization rates are above optimal that USDT and USDC markets on Aave outperform the risk-minimized sUSDS rate (4.5%) and risk-free (4.4%). In times of low utilization, the opportunity cost of capital to allocate to either market increases while capital efficiency decreases relative to benchmark.

Question: What are the expected returns for supplying capital to USDC & USDT markets in a bear and bull scenario on Aave?

Bear Case

USDT Calculations

• Current Supply $3,369,617,651
• Current Borrowed $1,884,586,986
• Current UR: 55.93%
• Current Supply APY: 1.83%
• Current Borrow APR: 3.65%
• 30% Proposed max supply: $1,444,121,850
• 15% Proposed max supply: $594,638,409
• 5% Proposed max supply: $177,348,297

The average difference between the Borrow APR and Supply APY, excluding outliers using the filtered APY is 1.72%. We’re currently at the lowest utilization levels for each market since their genesis, indicating we’re in a bear scenario.

Assuming the full allocation is made today, the reserve factor is constant, and the borrowed amount remains the same, the pool’s supply APY would decrease by:

• 30% Proposed max supply:
  • Total Supply = (Supplied + Proposed max supply)
    • $4,813,739,501 = $3,369,617,651 + $1,444,121,850
  • Projected_UR = 39.14% = $1,884,586,986 / $4,813,739,501
  • Borrow APR with current curve: 2.54%
  • Projected Supply APY = 0.82%
  • Projected Decrease = 1.01%
• 15% Proposed max supply:
  • Total Supply = (Supplied + Proposed max supply)
    • $3,964,256,060 = $3,369,617,651 + $594,638,409
  • Projected_UR = 47.53% = $1,884,586,986 / $3,964,256,060
  • Borrow APR with current curve: 3.10%
  • Projected Supply APY = 1.38%
  • Projected Decrease = 0.45%
• 5% Proposed max supply:
  • Total Supply = (Supplied + Proposed max supply)
    • $3,546,965,948 = $3,369,617,651 + $177,348,297
  • Projected_UR = 53.13% = $1,884,586,986 / $3,546,965,948
  • Borrow APR with current curve: 3.46%
  • Projected Supply APY = 1.74%
  • Projected Decrease = 0.09%

USDC Calculations

• Current Supply $2,689,001,558
• Current Borrowed $1,786,314,057
• Current UR: 66.43%
• Current Supply APY: 2.59%
• Current Borrow APR: 4.30%
• 30% Proposed max supply: $1,152,429,239
• 15% Proposed max supply: $474,529,686
• 5% Proposed max supply: $141,526,397

The average difference between the Borrow APR and Supply APY, excluding outliers using the filtered APY is 1.61%

Assuming the reserve factor and borrowed amount remain constant the pools supply APY would decrease by:

• 30% Proposed max supply:

  • Total Supply = (Supplied + Proposed max supply)
    • $3,841,430,797 = $2,689,001,558 + $1,152,429,239
  • Projected_UR = 46.50% = $1,786,314,057 / $3,841,430,797
  • Borrow APR with current curve: 1.42%
  • Projected Supply APY = 1.31%
  • Projected Decrease = 1.28%

• 15% Proposed max supply:

  • Total Supply = (Supplied + Proposed max supply)
    • $3,163,531,244 = $2,689,001,558 + $474,529,686
  • Projected_UR = 56.45% = $1,786,314,057 / $3,163,531,244
  • Borrow APR with current curve: 3.68%
  • Projected Supply APY = 2.01%
  • Projected Decrease = 0.58%

• 5% Proposed max supply:

• Total Supply = (Supplied + Proposed max supply)

  • $2,830,527,955 = $2,689,001,558 + $141,526,397

• Projected_UR = 63.11% = $1,786,314,057 / $2,830,527,955

• Borrow APR with current curve: 4.14%

• Projected Supply APY = 2.52%

• Projected Decrease = 0.07%

Bull Case

USDT Calculations

The trailing 6-month average USDT utilization rate is 77.97% and the trailing 12-month average USDT utilization rate is 78.61%. A bull scenario could lead to a utilization of 85% which would be an increase of 55% from current levels.

The average difference between the Borrow APR and Supply APY, excluding outliers using the filtered APY is 1.72%

Assuming the full allocation is made today, the reserve factor is constant, and the borrowed amount increases to match average utilization levels, the pool’s supply APY would increase by:

• Projected Supply APY With AVG UR:
  • Projected_UR = 85%
  • Borrow APR with current curve: 5.54%
  • Projected Supply APY = 3.82%

USDC Calculations

The trailing 6-month average USDT utilization rate is 81.53%, and the trailing 12-month average USDT utilization rate is 82.88% USDC. A bull scenario could lead to a utilization of 89%, which would be an increase of 27%.

The average difference between the Borrow APR and Supply APY, excluding outliers using the filtered APY is 1.61%

Assuming the full allocation is made today, the reserve factor is constant, and the borrowed amount increases to match average utilization levels, the pool’s supply APY would increase by:

• Projected Supply APY With AVG UR:
  • Projected_UR = 89%
  • Borrow APR with current curve: 5.80%
  • Projected Supply APY = 4.19%

Recommended Supply Cap & Required Minimum Yield

Overall: Ethena should cap the amount of aUSDT to aUSDC supplied between or at 5-10% of total supply to account for the inverse relationship between yield and available exit liquidity, token-specific risks, insolvency risks, and governance risks. The position should be actively managed because – as displayed by the Yield Analysis section – passively managed positions do not accurately balance the risk-reward relative to risk-minimized or risk-free rates. By requiring that the supplied capital to either market accrue a minimum 25bps premium on top of benchmark and the exit liquidity remains above 1.25x(supplied capital), Ethena accrues outsized returns for reasonable risks.

Questions

Supplied Collateral Caps: What should the max supply cap of aUSDT and aUSDC be?

For other venues where Ethena deploys capital, the Risk Committee has established a soft 10% cap on exchange open interest per asset backing USDe. This cap is applied differently depending on the venue type: when opening perpetuals positions, the cap is based on the share of open interest Ethena contributes; when deploying capital to lending platforms, it should be based on the share of total supply Ethena contributes. In this case, when utilizing Aave or similar lending protocols, Ethena should use the cap relative to the total borrowable supply, aligning with its role as a lender in the system.

Given the redemption, token-specific, insolvency, and governance risks, Ethena should not account for more than 5%-10% of the combined aTOKEN total supply of USDC and USDT markets. Presently, the combined aTOKEN total supply of both markets is $6,060,000,000, resulting in a maximum allocation of $303,000,000 for a 5% cap. Assuming the 5% cap was reached, aUSDT and aUSDC backing collateral would comprise 7.87% of the Liquid Stables category at the time of writing – enhancing diversification.

Minimum Acceptable Yield: What minimum yield should Ethena accept relative to the risk associated with both USDC & USDT markets on Aave?

It is not until utilization rates are above optimal that USDT and USDC markets on Aave outperform the risk-minimized sUSDS rate (4.5%) and risk-free (4.4%). In an environment where both markets are at their optimal utilization rate — USDT uOptimal (92%) USDC uOptimal (92%) — demand for borrowing is higher than average. To expose Ethena to the yield from utilization rates increasing past optimal levels it should consider allocating capital just before the optimal utilization rate is met.

For all the risk incurred by allocating to both markets, a minimum 25bps improvement on the MAX(risk-minimized, risk-free) yield should be considered; to achieve the maximum yield, Ethena could dynamically adjust to whichever market has a higher supply APY. Balancing value extraction with redemption risk is key. Ethena should consider removing deployed capital from the market when 1.25x(supplied capital) remains in available exit liquidity in that specific market.

Notably, if Ethena’s supply cap were higher, it would increase the exposure to all formerly detailed risks but also heavily reduce the utilization ratio when supplying capital at or slightly before uOptimal for both markets. As a result, the supply APY would drop below the minimum required return previously outlined and borrowing demand would need to meet the outsized allocation to return to uOptimal levels. Keeping the supply cap at a small percentage of the total supply of both markets makes this scenario unlikely.

[EthenaLabs_Research]

A proposal has been published for ENA and sENA holders to vote on Ethena lending in Aave

By voting For, users are voting for Ethena to begin supplying USDtb, USDC and USDT on Aave, within the parameters defined by the Risk Committee.

https://snapshot.box/#/s:ethenagovernance.eth/proposal/0x98421ae57d4c3d8584b32d247fb9bf92f8912de03aa517abce73f992a0ab01ec

[BlockworksAdvisory]

The Risk Committee is aware of the current Aave exposure in USDe backing and is conducting a deeper assessment to establish updated limits, with a focus on maintaining adequate redemption times even under periods of stress.

[BlockworksAdvisory]

Ethena Aave Integration: Dynamic Safe Cap Model

Executive Summary

This analysis presents a comprehensive stress testing framework for determining optimal Aave lending caps for USDe’s liquid backing assets. The model dynamically calculates maximum safe lending amounts under various redemption stress scenarios while maintaining adequate liquidity for user withdrawals.

Key Takeaways

• Redemptions are predictable and largely contained. Historical redemptions have been modest relative to total supply and are largely driven by Pendle PT expiries (scheduled) and occasional market shocks, not by yield fluctuations. Negative USDe yield spreads correlate with increased redemptions, but with lag and smoothing effects due to sUSDe’s cooldown and sticky capital.
• Structural scenarios can constrain lending. While all past redemption levels are manageable even under extreme Aave stress, theoretical structural shocks, like simultaneous redemptions of floating USDe + near-maturity Pendle, can reduce safe Aave caps.
• Dynamic stress-responsive framework: The model scales historical redemption data to current supply, applies time-based weighting to Pendle risk, and flexibly adjusts recommended caps as supply composition shifts.
• Recommended allocation range: Under current conditions, a 49% cap on liquid stables in Aave is advised. The range of 35-60% remains robust across realistic input variations. This is dependent on a large buffer of liquid stables being readily available to honor redemptions.
• When to adjust caps: Decrease allocations if floating USDe rises >$4B, Pendle maturities shorten (<5d), liquid backing drops < $6B, or Aave utilization exceeds 90%.
  Increase if more USDe moves to sUSDe, Pendle maturities extend, or liquid backing exceeds $8B.

Redemption Stress

There are several important considerations when assessing redemption stress. Ethena must always maintain a sufficient buffer of liquid stablecoins to honor redemptions quickly and reliably. This is critical to avoid pressure on USDe’s peg. To size this buffer appropriately, we need to understand how large redemptions tend to be, and what drives them. Identifying these drivers allows us to build them into our models and adjust expectations dynamically as conditions change.

Redemptions and Yield Dynamics

Hypothesis: Lower yields lead to higher redemptions.

Since the primary use case for USDe is to earn yield via sUSDe, periods of low yields can incentivize users to redeem and reallocate their capital elsewhere. To capture this effect, we benchmark USDe yields against the median sDAI yield, and observe how redemptions behave when the USDe–sDAI yield spread turns negative.

Looking at the chart above, it’s clear that redemptions are rare during periods when USDe yields outperform the benchmark (non-shaded areas). However, redemptions don’t align perfectly with yield shifts, for two main reasons:

1. Cooldown period: Yield is captured via sUSDe, which has a 7-day unstaking cooldown. Sophisticated users may anticipate falling yields, but there’s typically a lag between when yields become less attractive and when redemptions actually occur.

2. Sticky capital: Large holders often have strategic relationships with Ethena and may tolerate short periods of uncompetitive yields. They usually only redeem after more prolonged periods of unattractive returns, rather than reacting immediately.

This behavior means redemption activity is partly lagged and smoothed, rather than reacting mechanically to yield changes in real time.

The chart above plots redemptions alongside the USDe–sDAI yield spread, with orange shading marking periods when the spread had been negative seven days earlier to account for the sUSDe cooldown period. While this approach isn’t perfectly precise (as it only uses sDAI as a benchmark and doesn’t capture all alternative yield opportunities), it still highlights the pattern well: when the yield spread is elevated, as it was from May to July 2024, and again from November 2024 through January 2025, redemptions remain minimal.

That said, there are also several redemption spikes that don’t align with low-yield periods, suggesting that other factors were at play. These outliers warrant further investigation to identify their underlying causes.

Redemptions and Pendle Expiries

Pendle expiries appear to be one of the primary drivers of large redemption spikes, and unlike yield-driven outflows, they occur in a scheduled and predictable manner rather than being triggered by market conditions.

Recent Pendle expiry dates include:

• 2024-04-04 (USDe only)
• 2024-04-25 (sUSDe only)
• 2024-07-25 (USDe + sUSDe)
• 2024-09-26 (sUSDe only)
• 2024-10-24 (USDe + sUSDe)
• 2024-12-26 (sUSDe + USDe)
• 2025-02-27 (sUSDe only)
• 2025-03-27 (USDe + sUSDe)
• 2025-05-29 (eUSDe + sUSDe)
• 2025-07-31 (sUSDe + USDe)
• 2025-08-14 (eUSDe only)

The chart below overlays dotted vertical lines marking these expiry dates and shows redemptions as a percentage of total USDe supply rather than absolute amounts.

When viewed alongside yield spread data, most redemption spikes can be explained by either Pendle expiries or periods of low (or negative) yield spreads. Very few redemptions fall outside these two categories, suggesting that Pendle maturities and yield conditions are the key predictors of redemption activity.

Redemptions - Top 10 analyzed

The ten largest redemption events together accounted for $1.13B, averaging $113M per event, and represent the most significant moments of liquidity outflow in USDe’s history. These redemptions often clustered around two types of conditions: Pendle PT expiries and sharp market dislocations.

Pendle expiries were the dominant driver. Six out of the ten events occurred within a week of a major Pendle maturity date, suggesting that redemptions were primarily driven by investors rotating out of maturing PT positions. The largest single event, on March 1, 2025, saw $267.9M (4.9% of supply) redeemed just two days after a major Pendle tranche expired. This was immediately preceded by another large redemption on February 28 ($125M, 2.2%) and followed earlier by February 21 ($123.3M, 2.1%), both also near the same expiry window. Notably, this aligned with the Bybit hack, which briefly shook confidence among USDe holders, and was most likely the main driver of these redemptions. Other Pendle-driven outflows included July 25, 2024 ($95.5M, 2.9%), March 27, 2025 ($80.9M, 1.5%), and December 20, 2024 ($80M, 1.35%).

By contrast, the remaining redemptions appear to have been driven by market stress. The remaining four redemptions were linked to broader market volatility and negative sentiment rather than yield arbitrage. On August 5, 2024, $95.8M (3.1%) was redeemed as BTC plunged 15%, breaking below $50K for the first time in months. Similarly, September 2, 2024 saw $88.8M (3.3%) redeemed amid a sustained negative yield spread. More recently, June 25, 2025 ($88M, 1.6%) and June 18, 2025 ($86M, 1.5%) occurred during low or negative spreads and heightened geopolitical risk, as broader markets were bleeding.

Yield conditions shaped the environment in which these redemptions occurred. Interestingly, most of these major outflows (7 of 10) occurred during periods of positive yield spreads, when redemption was less financially attractive. This reinforces the idea that timing of Pendle maturities, rather than purely yield-driven incentives, was the stronger catalyst. Negative-spread periods did coincide with some redemptions, especially September 2, June 25, and June 18, which were not around Pendle maturities but were more likely triggered by external market shocks.

Date	Amount	% of Supply	Yield Spread	Near Pendle Expiry?	Context / Notes
2025-03-01	$267.9M	4.903%	4.3896%	Yes — 2d after 2025-02-27	Largest redemption on record; clustered around Pendle expiry and around 8 days after Bybit hack
2025-02-28	$125.0M	2.174%	4.5412%	Yes — 1d after 2025-02-27	Follow-up redemption during same expiry window and around 7 days after Bybit hack
2025-02-21	$123.3M	2.079%	2.0523%	Yes — 6d before 2025-02-27	Bybit hack (and pre-expiry unwind)
2024-08-05	$95.8M	3.078%	5.6204%	No	BTC dropped 15% below $50K — risk-off market
2024-07-25	$95.5M	2.885%	3.0083%	Yes — same day (2024-07-25)	Major Pendle expiry
2024-09-02	$88.8M	3.271%	−2.9237%	No	Negative yield spread persisted
2025-06-25	$88.0M	1.597%	−0.9822%	No	Negative spread; geopolitical turmoil
2025-06-18	$86.0M	1.481%	−1.0870%	No	Negative spread; markets broadly bleeding
2025-03-27	$80.9M	1.527%	4.1842%	Yes — same day (2025-03-27)	Pendle expiry
2024-12-20	$80.0M	1.350%	0.0387%	Yes — 6d before 2024-12-26	Pendle expiry

Model Overview

Core Problem

Ethena deploys USDe’s liquid stablecoin backing (USDC/USDT) across multiple venues, including Aave lending markets. The challenge is determining how much can be safely lent on Aave while preserving the ability to meet redemption demands under stress conditions.

Key Question

What is the maximum amount of liquid backing that can be lent on Aave while maintaining sufficient liquidity to handle redemption shocks?

Mathematical Framework

Core Formula

The model is built around a fundamental liquidity constraint:

Available Liquidity Under Stress ≥ Redemption Shock

Available Liquidity = (Ltotal - A) + h × A = Ltotal - (1-h) × A

Amax = (Ltotal - S) / (1-h) Where:

• A = Amount lent on Aave

• Ltotal = Total liquid stablecoin backing

• S = Redemption shock amount

• h = Aave withdrawable fraction under stress (0 ≤ h ≤ 1)

• Amax = Maximum safe Aave lending amount

Key Insight: The model recognizes that under stress conditions, only a fraction h of Aave deposits may be quickly withdrawable, while off-Aave liquidity remains fully accessible.

Model Inputs and Assumptions

1. USDe Supply Structure

Component	Amount	Description
Total Supply	$12.48B	Total USDe in circulation
sUSDe (Staked)	$5.64B	Excluded from immediate redemption due to 7-day cooldown
Pendle Total	$4.18B	Time-weighted based on maturity
Floating USDe	$2.66B	Immediately redeemable

2. Liquid Backing Composition ($6.9B Total)

Component	Amount	Description
Current Aave Deposits	$2.8B	Currently lent on Aave
Copper Buffer	$0.8B	Custody buffer
Mint/Redeem Contracts	$30M	Operational contracts
Replenisher Wallet	$0.5B	Protocol operations
Unspecified Off-Aave	$2.6B	Other liquid venues (USDtb, idle USDC, etc)

3. Stress Parameters

Aave Withdrawable Fractions (h)

The model tests eight stress levels representing the percentage of Aave deposits withdrawable during crisis:

Fraction	Description	Stress Level
0.0 (0%)	Complete Aave liquidity freeze	Maximum stress
0.01 (1%)	Extreme stress (99% haircut)	Extreme stress
0.05 (5%)	Severe stress (95% haircut)	Severe stress
0.1 (10%)	High stress (90% haircut)	High stress
0.2 (20%)	Moderate-high stress (80% haircut)	Moderate-high stress
0.4 (40%)	Moderate stress (60% haircut)	Moderate stress
0.6 (60%)	Low stress (40% haircut)	Low stress
0.8 (80%)	Minimal stress (20% haircut)	Minimal stress

Redemption Shock Scenarios

Historical Shocks (Based on actual redemption data as % of supply):

• P95 Historical: $115.8M (95th percentile daily redemption scaled to current supply)
• P99 Historical: $270.0M (99th percentile daily redemption scaled to current supply)
• Max Historical: $611.9M (Maximum historical daily redemption scaled to current supply)

Structural Shocks:

• All Floating USDe: $2.66B (All immediately redeemable USDe)
• Structural Combined: $3.84B (Floating USDe + Effective Pendle Demand)

4. Pendle Position Treatment

Step Function Risk Assessment

Pendle Principal Tokens (PTs) containing USDe are weighted by time-to-maturity using a discrete step function reflecting realistic user redemption behavior:

Days to Maturity	Redemption Probability	Rationale
≤ 5 days	30%	Immediate pressure – users actively redeem
6–10 days	20%	High pressure – preparation for redemption
11–20 days	10%	Moderate pressure – some early redemptions
> 20 days	0%	Minimal pressure – users wait closer to maturity

Current Pendle Position: $4.18B total with effective demand of ~$858M (20%)

Rationale:

• Eliminates daily volatility in redemption estimates
• Based on realistic user behavior patterns
• Only near-maturity positions contribute meaningful immediate risk
• Provides operational predictability and risk management clarity

This heuristic was chosen because historical data did not provide conclusive or actionable insights on user behavior near maturity. Over the past two months, USDe in PTs has grown so rapidly that relying on historical patterns was deemed inappropriate, as illustrated in the chart below.

5. Key Assumptions

1. Historical Scaling: Past redemption patterns (as % of supply) are indicative of future stress scenarios
2. Stress Independence: Aave liquidity stress and redemption shocks can occur simultaneously
3. Off-Aave Liquidity: Non-Aave liquid assets remain fully accessible under stress
4. Pendle Behavior: Time-to-maturity is the primary driver of redemption likelihood
5. Operational Continuity: Model assumes continued operation under stress (no protocol shutdown)

Scenario Results Analysis

Historical Scenarios (Manageable Stress)

All historical shock scenarios ($116M - $612M) show minimal impact on Aave lending capacity:

• Even under severe Aave stress (0% withdrawable), caps remain >90% of liquid capacity
• Demonstrates that historical redemption patterns are well within operational bounds

Sensitivity Analysis Across Input Variations

To validate robustness, the model was tested across six different input scenarios representing realistic changes in USDe composition and market conditions:

Scenario	Key Changes	Aave Cap Range (% at 1% withdrawable)	Robustness
Base Case	Current actual values	49% capacity	Baseline
High Floating USDe	+$2B more floating USDe	34% capacity	More constrained
Low Floating USDe	More locked in sUSDe/Pendle	90% capacity	Less constrained
High Pendle Exposure	+$2B Pendle, shorter maturity	38% capacity	More constrained
Low Liquid Backing	-20% total liquid backing	34% capacity	More constrained
Conservative Structure	More sUSDe, less Pendle	63% capacity	Less constrained

Key Sensitivity Insights:

• Base allocation range: 30-50% works across most realistic scenarios under extreme Aave stress
• Conservative scenarios allow 60-90% allocation when USDe structure is favorable (more sUSDe, less floating)
• Worst-case scenarios limit to ~35% when multiple stress factors combine (high floating USDe + reduced backing)
• Most vulnerable to: Higher floating USDe, shorter Pendle maturities, reduced liquid backing
• Most resilient to: More conservative USDe composition (higher sUSDe proportion)

Structural Scenarios (Significant Stress)

All Floating USDe ($2.66B shock):

Aave Withdrawable	Cap Amount	% of Liquid Total
0%	$4.24B	61%
20%	$5.30B	77%
40%+	$6.90B	100% (No constraint)

Structural Combined ($3.84B shock):

Aave Withdrawable	Cap Amount	% of Liquid Total
0%	$3.06B	44%
20%	$3.83B	55%
60%+	$6.90B	100% (No constraint)

Model Strengths

1. Dynamic Framework: Adapts to changing supply levels and market conditions
2. Comprehensive Stress Testing: Covers both historical and structural shock scenarios
3. Realistic Assumptions: Incorporates time-based risk weighting for complex positions
4. Granular Analysis: Tests wide range of stress severities
5. Data-Driven: Uses actual historical redemption patterns scaled to current protocol size

Model Limitations

1. Correlation Assumptions: May underestimate correlations between Aave stress and redemption demand
2. Historical Dependence: Future stress may exceed historical patterns
3. Static Pendle Weighting: Actual redemption behavior may vary from modeled decay function
4. Operational Assumptions: Assumes continued access to off-Aave liquidity under extreme stress
5. Binary Liquidity View: Reality may involve partial/delayed access rather than binary availability

Key Conclusions

1. Historical Redemptions Are Manageable

Even the largest historical redemption patterns (scaled to current supply) create minimal constraints on Aave lending capacity, suggesting robust operational resilience.

2. Structural Risks Require Careful Management

Scenarios involving significant floating USDe redemptions combined with Pendle position unwinding can meaningfully constrain Aave lending capacity, particularly when Aave itself experiences liquidity stress.

3. Critical Risk Threshold

The model identifies 60% Aave withdrawable fraction as a threshold below which structural shocks need to be watched closely. Below this level, Aave lending capacity could become constrained.

4. Pendle Maturity Risk

Time-to-maturity analysis shows that Pendle expiries (21 days) can contribute meaningful stress, highlighting the importance of monitoring maturity concentration.

5. Operational Buffer Importance

The difference between “All Floating USDe” and “Structural Combined” scenarios ($1.2B) demonstrates the significant incremental risk from Pendle positions, supporting the value of operational buffers.

6. Recommended Aave Allocation

Based on the model results and sensitivity analysis, we currently recommend allocating a maximum of 49% of liquid stables to Aave under current conditions. The range of 35-60% remains robust across realistic input variations.

Risk Profile	Recommended Allocation	Rationale	Sensitivity-Adjusted
Conservative	30-40% of liquid stables	Safe even under extreme Aave stress (1% withdrawable). Handles worst-case scenarios.	Works in all tested scenarios
Moderate	40-60% of liquid stables	Assumes moderate Aave stress resilience (10-20% withdrawable).	Safe unless extreme structural changes
Aggressive	60-80% of liquid stables	Assumes Aave maintains reasonable liquidity (40%+ withdrawable). Higher yield but requires favorable USDe structure.	Only suitable with conservative USDe composition

Dynamic Adjustment Framework:

• Decrease allocation by 10-15% if:
  • Floating USDe increases significantly (>$4B)
  • Pendle positions approach shorter maturities (<14 days)
  • Total liquid backing decreases (<$6B)
  • Aave utilization rates exceed 90%
• Increase allocation by 5-10% if:
  • More USDe moves to sUSDe (longer lock-up)
  • Pendle maturities extend (>30 days)
  • Liquid backing grows substantially (>$8B)
  • Aave markets remain consistently liquid

Monitoring Triggers for Reallocation:

• Structural changes: >20% change in floating USDe composition
• Pendle concentration: >50% of Pendle exposure maturing within 14 days
• Market conditions: Aave utilization >95% or major liquidity events
• Growth scenarios: Total supply growing >25% quarterly

Additionally, to help manage risk for Aave as well, Ethena should commit to a strategy for the gradual and careful unwinding of its Aave exposure, particularly as large PT pools approach maturity. A phased approach helps avoid sudden liquidity shocks on Aave, gives borrow rates time to adjust naturally, and ensures Ethena maintains enough liquid assets to cover redemptions, ultimately lowering systemic risk across both protocols. Llama Risk’s analysis on Aave’s governance forum goes deeper into this point, as well as the broader question of managing Ethena’s Aave exposure.