[ARFC] Umbrella Coverage Expansion

[LlamaRisk]

Summary

LlamaRisk proposes an update to the Aave V3 Core Umbrella target liquidity levels, prompted by a roughly 54% increase in total debt since the mechanism launched 4 months ago. The significant growth in USDC and USDT borrowing has rendered the initial coverage levels outdated. To align with current market conditions, the recommendation is to increase the target liquidity for USDT to 130 million and decrease the target for GHO to 8 million. In contrast, the targets for USDC and WETH will remain unchanged. This addresses the observed shifts in borrow activity and ensures coverage levels are appropriate for the current risk exposure.

In addition to adjusting existing modules, we propose expanding Umbrella’s scope to include three high-impact assets: wstETH, USDe, and WBTC. This expansion would bring assets accounting for 98.12% of total borrows under Umbrella’s protection. In the medium term, the associated increase in annual emissions would be offset by the ongoing depreciation of legacy staking modules.

Introduction

It has been four months since the launch of Umbrella, which has proven successful, with all existing modules reaching their target coverage levels. However, since the initial insurance caps were recommended, market dynamics have shifted, particularly for USDC and USDT; both have seen significant supply and borrow activity growth on the Aave V3 Core instance.

Source: LlamaRisk, September 29, 2025

Beyond these assets, others such as wstETH and USDe represent a substantial share of Aave’s TVL, highlighting the need to expand Umbrella’s coverage to include them. As a result, the scope is to re-evaluate the existing insurance cap recommendations for currently covered assets, in light of updated liquidity and risk conditions, and to propose new coverage paths for additional high-impact assets on the Aave V3 Core instance. This expansion would support the gradual phase-out of stkAAVE coverage while maintaining overall coverage levels and optimizing the efficiency of the emissions.

Existing Umbrella Modules

Coverage Evolution

Since its launch, the total debt on the Aave V3 Core instance has grown significantly, from $13.8B to $21.3B, marking a ~54% increase. The four assets currently covered under Umbrella, USDC, USDT, WETH, and GHO, now account for $18.7B, or 87.7% of total borrows.

Source: LlamaRisk, September 29, 2025

While it may appear that most of Aave’s borrow exposure is already secured through Umbrella, it’s important to recognize that the initial insurance recommendations were made based on significantly lower borrow caps set nearly five months ago. Given the substantial increase in total debt since then, the original caps no longer reflect current market conditions.

The individual growth trajectories of the covered assets since Umbrella’s launch are detailed here:

Source: LlamaRisk, September 29, 2025

WETH and GHO have maintained relatively stable supply and borrow levels since the launch of Umbrella. In contrast, USDC and USDT have experienced significant growth in both supply and borrow activity. In response to this demand, their respective caps were raised multiple times:

• USDC supply cap increased from $4.8B to $7.5B, and the borrow cap from $4.32B to $7B
• USDT supply cap rose from $5.25B to $9.5B, and borrow cap from $4.72B to $8.8B.

Adjusted Coverage Estimations

Following the changes, new liquidity targets for each asset under Umbrella were determined using our VaR methodology, which simulates extreme market stress using historical price data to generate synthetic shock scenarios. It models user liquidations under these shocks to estimate each asset’s liquidation capacity (the maximum debt that can be liquidated profitably for liquidators). Based on the simulation run done on September 29, 2025, we present the liquidation capacity changes overlaid with the asset borrow cap changes:

Asset	Liquidation Capacity Change	Borrow Cap Change
USDC	50,400,000 USDC → 117,700,000 USDC	4,320,000,000 USDC → 7,000,000,000 USDC
USDT	12,200,000 USDT → 42,730,000 USDT	4,720,000,000 USDT → 8,800,000,000 USDT
WETH	6,900 WETH → 697 WETH	2,700,000 WETH → 2,900,000 WETH
GHO	5,000,000 GHO → 18,220,000 GHO	180,000,000 GHO → 180,000,000 GHO (unchanged)

Over the four months since the initial simulation run on May 15, 2025, the liquidation capacity for USDC, USDT, and GHO has increased substantially, while WETH capacity has declined. These shifts are primarily explained by changes in borrower distribution: when a larger share of debt is concentrated in a few large borrowers, liquidity capacity decreases as fewer liquidatable positions fall within the inlier set defined by the Interquartile Range (IRQ) method.

In the case of WETH, the decline in liquidation capacity was further exacerbated by market conditions. Increased price volatility of ETH resulted in a shortage of WETH liquidity in secondary markets. At the same time, the long validator exit queue created a parallel shortage of LST/ETH liquidity on DEXs. This is particularly relevant since the two largest WETH borrowers, accounting for ~47% of WETH borrows, engage in LST/LRT looping strategies at low health factors of 1.02 and 1.06. As a result, the availability of profitable liquidation opportunities in WETH diminished.

Source: LlamaRisk, September 29, 2025

Moreover, shock simulations indicate that GHO, USDC, and USDT were among the most stable assets, experiencing minimal negative price movements. WETH, while showing a mild average drawdown of approximately -0.057% under stressed scenarios, still demonstrates relatively stable and predictable behavior. This stability suggests that, although their current liquidation capacity is low, WETH and GHO remain manageable from a risk perspective due to their robust shock resilience during market stress.

To put the overall numbers into perspective, the changes in total debt and borrow cap coverage since the initial deployment are as follows:

Asset	Initial Total Debt Coverage*	Current Total Debt Coverage*	Initial Borrow Cap Coverage	Current Borrow Cap Coverage
USDC	135,400,000 USDC	183,368,556 USDC	3.13%	2.62%
USDT	150,250,000 USDT	153,486,582 USDT	2.46%	1.74%
WETH	31,900 WETH	39,294 WETH	1.18%	1.35%
GHO	17,000,000 GHO	32,351,731 GHO	9.44%	17.97%

*Total debt coverage is the sum of liquidity capacity and the current Umbrella Module deposits.

Based on the details presented above, new target liquidity recommendations for the existing Umbrella Modules are as follows:

Asset	Current Target Liquidity	Current Liquidity	Recommended Target Liquidity
USDC	66,000,000 USDC	65,668,556 USDC	66,000,000 USDC (unchanged)
USDT	104,000,000 USDT	110,756,582 USDT	130,000,000 USDT
WETH	25,000 WETH	38,597 WETH	25,000 WETH (unchanged)
GHO	12,000,000 GHO	14,131,731 GHO	8,000,000 GHO

For USDC, the increase in liquidity capacity has proportionally tracked the borrow cap expansion, and a similar pattern is observed for GHO. In contrast, WETH coverage has been maintained primarily through substantial deposits into its Safety Module. The main exception is USDT, where current borrow cap coverage has declined. Based on these observations, we recommend increasing USDT’s target liquidity while reducing GHO’s target liquidity by 33%, as its existing liquidity capacity is already sufficient. To accommodate the higher USDT target liquidity, we propose that the DAO adjust incentive emissions for the USDT Umbrella modules.

With these changes, the maximal emissions would be increased by 920K aUSDT per year, partially offset by a reduction of 400K GHO per year in GHO module emissions. This adjustment would preserve the current APY ranges, maintain spend efficiency for the DAO, and ensure continued attraction of sufficient staked liquidity.

No changes are recommended for the WETH module at this time, as its deposits remain comfortably above the target coverage level, and neither a target liquidity increase nor additional incentives are required to sustain participation.

Expansion of Umbrella Modules

There are currently 32 assets on Aave V3 Core with non-zero borrow balances that are not yet included under Umbrella coverage. However, only a subset of these, six assets, have borrows exceeding $50M, making them the most relevant candidates for near-term inclusion. These include wstETH ($1.1B), USDe ($882M), WBTC ($220M), DAI ($133M), RLUSD ($88M), and USDtb ($54M).

Source: LlamaRisk, September 29, 2025

Based on observed growth trends, several stablecoin borrow markets, notably USDe, RLUSD, and USDtb, have shown meaningful increases in 2025. As part of the next phase of Umbrella expansion, we propose adding wstETH, USDe, and WBTC to the coverage framework, while continuing to monitor the other emerging stablecoins for future inclusion.

Following the addition of these three assets, the Umbrella Safety Module would encompass assets that collectively account for approximately 98.12% of total borrows on Aave V3 Core, equivalent to $20.9B in outstanding debt. After this expansion, only $390M in borrowings would remain outside Umbrella’s scope. The table below outlines our target liquidity recommendations for the proposed assets.

Asset	Current Borrow Cap	Liquidation Capacity	Recommended Target Liquidity	Total Debt Coverage	Coverage of Borrow Cap
wstETH	480,000 wstETH	2,950 wstETH	9,050 wstETH	12,000 wstETH	2.5%
USDe	2,500,000,000 USDe	0 USDe	43,920,000 USDe	43,920,000 USDe	1.76%
WBTC	28,000 WBTC	454 WBTC	250 WBTC	704 WBTC	2.51%

According to our methodology, the liquidation capacity for USDe is effectively zero, partially due to insufficient market liquidity to liquidate USDe borrow positions profitably under shock scenarios. Albeit, USDe exhibited zero shock deviation, suggesting stable pricing behavior even in stressed conditions. In contrast, wstETH showed the highest simulated price shock at -0.058%, which still represents a relatively mild drawdown. WBTC demonstrated greater resilience, with a smaller shock value of -0.04%, indicating stronger stability during market stress situations.

To support the inclusion of these new assets into the Umbrella Safety Module, we estimate the following maximum annual emissions required to maintain sufficient risk-adjusted yield at target liquidity levels: 50 awstETH (~$255K), 773K aUSDe ($780K), and 0.5 aWBTC ($57K), targeting Umbrella APY of 0.55%, 1.76%, and 0.2% at target liquidity. This would result in an additional annual cost of approximately $1.09M to the DAO.

Conclusion & Recommendations

We recommend the following adjustments and additions to the Umbrella Safety Module to better align with current market conditions and borrow activity. Once implemented, these changes will increase the total annualized Umbrella emissions from $8.74M to $10.35M, while expanding coverage by an additional $192M. Reserves amounting to 98.12% of total borrows on Aave V3 Core would become covered by Umbrella.

Ultimately, this increase in emissions would be offset by the ongoing deprecation of legacy staking modules, which currently cost the DAO approximately $42.6M per year. Therefore, this transition reflects a more capital-efficient and scalable approach to risk management within the Aave protocol.

Asset	Current Target Liquidity	Recommended Target Liquidity	Current Max Emissions	Recommended Max Emissions
USDT	104,000,000 USDT	130,000,000 USDT	3,670,000 aUSDT/year	4,587,500 aUSDT/year
GHO	12,000,000 GHO	8,000,000 GHO	1,200,000 GHO/year	800,000 GHO/year
wstETH	-	9,050 wstETH	-	50 awstETH/year
USDe	-	43,920,000 USDe	-	773,000 aUSDe/year
WBTC	-	250 WBTC	-	0.5 aWBTC/year

These amounts are subject to a review from @BGDLabs and @TokenLogic, after which the recommendations will be updated accordingly.

Next Steps

• Incorporate feedback and escalate ARFC to a Snapshot vote.
• If the ARFC Snapshot outcome is YAE, publish an AIP vote for final confirmation and enforcement of the proposal.

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.

[bgdlabs]

Thanks for the proposal @LlamaRisk .

Some points of feedback/questions from our side:

• The Target Liquidity/s on the new proposed wstETH and USDe - or even the eligibility for inclusion of wstETH - seem pretty high for us in comparison with the existing assets covered.
  The rationale is that collateral backing wstETH is mainly comprised of assets directly correlated, oracle-wise, to ETH via CAPO, so deficit accrual would only occur when the value of the exchange rate negatively dislocates. This type of deficit, while important to handle in the future, is not really something so trivial to parametrise at the moment.
  The same happens with USDe, but in lower magnitude, as there really are non-correlated collaterals backing USDe (WETH, wstETH, WBTC, amongst others), but the majority are price-correlated (sUSDe, PTs). Probably the recommended for USDe Target Liquidity should take that into account and be lower, as definitely the non-price-correlated collateral for USDe is way lower than, let’s say, USDC.
• Could it be possible to include a more exhaustive list of parameters proposed on the system (still with comparison to the previous), akin to the one we included in the initial Umbrella activation HERE?
  Even if, for configuration purposes, some of those could be inferred, others, like Deficit Offset, cannot. And definitely having references of APYs would help the community on the voting stage.

[LlamaRisk]

We appreciate the feedback from @BGDLabs. In response, we’ve revised the proposed parameters, removing wstETH from inclusion under Umbrella, as 97.2% of its borrow collateral originates from LRTs via E-modes. Given the oracle setup, the probability of a realized deficit is entirely dependent on an LRT slashing event, which remains very unlikely. For USDe, we’ve reduced the target liquidity since a large portion (69.4%) of its collateral base comprises correlated assets such as PT tokens and sUSDe, while still recognizing the non-zero systemic and liquidity risks associated with the remaining non-correlated collateral.

Looking ahead, we plan to formalize a transparent framework to better define and quantify what Umbrella covers, helping the community clearly understand its scope and coverage principles.

wstETH

Source: LlamaRisk, October 22, 2025

wstETH on Aave V3 Core is primarily utilized in leveraged LRT looping strategies, which currently account for 97.2% of the total collateral backing wstETH borrows ($926M). Within this composition, rsETH represents $540M, followed by ezETH at $255M and weETH at $104M.

The liquidation and bad debt of these positions are closely tied to how ETH LRTs are priced on Aave. Each LRT is valued through an LRT/ETH exchange rate feed layered on top of the base ETH/USD feed. This configuration results in correlated oracle pricing between wstETH and the respective LRTs. Consequently, bad debt can materialize only in the event of a significant negative deviation in the LRT/wstETH exchange rate (which would materialize only in the case of LRT slashing).

Source: Dune, October 22, 2025

Both weETH and rsETH exchange rates have slightly depreciated against wstETH. However, looping with these assets remains profitable as most of their rewards are distributed in the form of points, unlike ezETH, which auto-compounds its rewards, resulting in greater price appreciation.

On April 23, 2024, ezETH experienced a secondary market depeg of 78%, triggering over $150M in liquidations across platforms like Gearbox and Morpho. However, under Aave’s current oracle setup, no liquidations would have occurred, as the internal exchange rate remained stable. Therefore, given the low likelihood of realized deficits, we do not see a need to include wstETH under Umbrella.

USDe

Source: LlamaRisk, October 22, 2025

Since June 2025, the majority of USDe borrows have been made against price-correlated collateral, primarily sUSDe and Ethena-related PT tokens under E-modes, which together represent 69.4% of the total collateral backing USDe borrows. This high degree of correlation lowers the risk of bad debt accrual, while the remaining non–price-correlated collateral amounts to approximately $215M. Borrow cap utilization has also remained mild, implying that even with a lower coverage, the share of currently borrowed amounts under protection would remain proportionally high. Therefore, we propose to revise the previously recommended USDe target liquidity to 20M USDe, reflecting both the predominance of correlated collateral and USDe’s relatively lower liquidity capacity.

Revised Recommendations

Staked asset	Covered asset	Target Liquidity	Max emission (rewards at target liquidity)	Umbrella APY range (up until excess liquidity)	Total APY (Aave + Umbrella)*	Cooldown/unstake window	Deficit offset
aUSDe	USDe	20,000,000 USDe	550,000 aUSDe/year	1.83%-5.5%	6.14%-9.81%	20/2 days	100,000 USDe
aWBTC	WBTC	250 WBTC	1.5 aWBTC/year	0.39%-1.2%	0.42%-1.23%	20/2 days	0.5 aWBTC
aUSDT	USDT	104,000,000 USDT → 130,000,000 USDT	3,670,000 → 4,587,500 aUSDT/year	2.35%-7.05%	7.37%-12.07%	-	-
GHO	GHO	12,000,000 GHO → 8,000,000 GHO	1,200,000 → 800,000 GHO/year	6.66%-20%	6.66%-20%	-	-

* Aave APY reflects the 1-year average Supply APY.

Based on the revised parameters, the updated borrow coverage for the newly onboarded assets, USDe, and WBTC, is as follows:

Asset	Current Borrow Cap	Total Debt Coverage (Liquidation Capacity + Target Liquidity)	Coverage of Borrow Cap
USDe	2,500,000,000 USDe	20,000,000 USDe	0.8%
WBTC	28,000 WBTC	704 WBTC	2.51%

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 advic

[ChaosLabs]

Summary

Chaos Labs supports the revised Umbrella parameters on Aave V3 Core. The rationale for the changes in Umbrella coverage stems from the growth in non-correlated debt for USDe and WBTC, and the growth in borrow size of the existing Umbrella markets.

Motivation & Technical Analysis

As the risk that Umbrella aims to address is the bad debt accrual, and that risk is highly exacerbated by the presence of borrowers funded by non-correlated collateral, where cross-asset price movements and frictions in the liquidations process can produce deficits, Chaos Labs supports the growth of the Umbrella coverage, primarily focusing on assets that maintain a limited asset correlation.

Some of the additions proposed in the original recommendations, specifically USDe and wstETH, are partial or full exceptions to this logic, with their borrows being 14.45% and <1% backed by uncorrelated assets. As such, we support the changes proposed by BGD on reducing the target coverage of these assets to limit the target to the debt presenting meaningful liquidation risk.

Most USDe borrowing is collateralized by PT and sUSDe instruments that share the same underlying economic exposure as USDe itself. Given these positions co-move by nature of their backing, adverse price actions are inherently limited to duration risk, making the accrual of bad debt structurally unlikely. The window where dislocations can happen is temporally limited, as PTs approach maturity, their pricing tends to that of their underlying, which in this situation is presented by USDe itself. In practice, this means Umbrella delivers little incremental protection for the correlated bulk of USDe positions.

However, as WBTC shows a meaningful non-correlated component that benefits from umbrella coverage, we fully support its introduction with the proposed 250 WBTC target.

Additionally, the uncorrelated debt of the assets that already have an Umbrella coverage pool has increased significantly since the initial parameterization of the target liquidity of the pools. The primary example is USDT, which has shown a growth in uncorrelated asset debt from $ 2.2B to $ 3.3B, hence warranting an increase in its coverage pool size.

Specification & Recommendation

As previously mentioned, we support the newly recommended parameters for the Umbrella changes in the LlamaRisk proposal

Staked asset	Covered asset	Current Target Liquidity	Recommended Target Liquidity	Current Max emission (rewards at target liquidity)	Recommended Max emission (rewards at target liquidity)
aUSDT	USDT	104,000,000 USDT	130,000,000 USDT	3,670,000 aUSDT/year	4,587,500 aUSDT/year
GHO	GHO	12,000,000 GHO	8,000,000 GHO	1,200,000 GHO/year	800,000 GHO/year
aUSDe	USDe	-	20,000,000 USDe	-	550,000 aUSDe/yr
aWBTC	WBTC	-	250 WBTC	-	1.5 aWBTC/yr

Disclosure

Chaos Labs has not been compensated by any third party for publishing this recommendation.

Copyright

Copyright and related rights waived via CC0.

[ACI]

The current proposal has been escalated to ARFC Snapshot as part of Skywards.

Vote will start tomorrow, we encourage everyone to participate.