[ARFC] Remove USDS as collateral and increase RF across all Aave Instances

[ARFC] Remove USDS as collateral and increase RF across all Aave Instances

Author: ACI

Date: 2025-11-17


Summary

This ARFC proposes removing USDS as collateral and increasing RF across all Aave instances.

USDS generates negligible revenue while its issuance model introduces asymmetric risks that could impact Aave’s stability.

The proposal implements the following:

  • Set LTV to 0%.
  • Set reserve Factor to 25%.
  • Removal from all stablecoin e-Modes.

Motivation

Motivation for these changes is driven by both reducing revenue from USDS borrows, and changing risk profile bringing into doubt the suitability of the asset as collateral.

1. Low Profitability

Data shows a decline in revenue each month, with $543,528 since Q1 2025, and current annualised revenue of $135,260 based on revenue since the beginning of Q4. With quarterly revenue data showing a sharp decline since the beginning of the year, we believe that current revenue is far below the scale required to justify current risk exposure.

Instance Q1 Q2 Q3 Q4
Cove V3 $87,948 $127,768 $96,556 $17,417
Prime V3 $156,205 $47,999 $9,276 $359
Total $244,153 $175,767 $105,832 $17,776

To combat this decline in revenue we propose raising RF to 25% which would increase Aave revenue by +150% on the same borrow volume.

2. Potential Risks from Issuance Model

With the new Stars being created within the Sky ecosystem, the risk profile of USDS is changing. Given these changes characterised by credit lines at low or zero cost, and the risk surface area presented by the yield strategies being utilised, we believe that the risk profile has increased to a degree which is outside of what is acceptable for Aave collateral.

Alongside the increase in RF we propose setting LTV to 0% in order to remove Aave’s exposure to USDS as collateral. Combined, we believe these measures bring the risk/reward of USDS back in line with other assets on Aave.

Specification

Pending confirmation by Risk Service Providers we propose the following parameter updates:

Instance Current LTV Proposed LTV Current RF Proposed RF
V3 Core 75 % 0 % 10% 25%
V3 Prime 0% 0% 10% 25%

Alongside the above we propose that USDS is removed from all e-Mode configurations across all instances.

Proposal will be updated with latest Risk Service Providers feedback.

Disclaimer

The current proposal has been created by ACI independently. ACI has not received compensation for the creation and review of this proposal.

Next Steps

  1. Publication of a standard ARFC to continue collecting community & service provider feedback before escalating proposal to ARFC snapshot stage.
  2. If the ARFC snapshot outcome is YAE, publish an AIP vote for final confirmation and enforcement of the proposal.

Copyright

Copyright and related rights waived under CC0

4 Likes

seems prudent. I don’t know why anyone would hold USDS TBH

Summary

LlamaRisk supports removing USDS as collateral due to the change in its risk profile, as more Sky Agents via Stars (and future Institutional Primes) are allocated USDS collateral to generate risk-adjusted yield, expanding the risk surface of the collateral backing USDS. We also support increasing the RF from 10% to 25%, along with the proposed removal of USDS from E-modes on Prime and Core.

Architecture

According to the Asset Allowlist Classification (AAcA), USDS is a fiat-pegged stablecoin managed through a collateralized debt position (CDP). As of November 21, 2025, USDS has a circulating supply of $5.5B on Ethereum. Together with $4.4B DAI, it is backed by a diversified $12.1B collateral portfolio at a 122% collateralization ratio, with the underlying assets shown below.


Source: USDS Collateral, Sky, November 21, 2025

Sky allocates the USDS collateral to Sky Agents to generate yield for USDS, and these are as follows:

Stars

Stars within the Sky Ecosystem are protocols with autonomy to deploy USDS collateral through Sky Primitives into various DeFi and TradFi opportunities, generating “risk-adjusted yield.” Sky Primitives provides several benefits for Sky Agents (i.e., Stars), including the Allocation System, Boosted Distribution Reward, Integration Boost, SkyLink setup, Governance Access Reward, and the Pioneer System.

This structure allows Sky to manage the tail risks associated with Stars operating under the Sky Agent Framework, while Stars focus on innovation, adoption, growth, and profitability. Current active Stars are Spark, Grove, and Obex. In October 2025, Sky introduced the initial concept for the Star Allocation Framework, which plans to expand the ecosystem to additional Stars, capped at seven. This expansion increases the overall risk profile of USDS, as each new Star protocol introduces its own tail risks that can affect Sky.

The full list of future Stars has not yet been published, but the known entities so far are:

  • Spark: The first Star under the Sky Agent Framework, currently allocated 33.65% ($4.03B) of the total USDS collateral. Their asset holdings are as follows:


Source: Top 10 Spark Allocations, Spark, November 21, 2025

  • Grove: Manages 15.12% ($1.81B) of the total collateral, with focus on institutional and RWA allocation evident via their biggest holding being Janus Henderson JAAA ($1B).


Source: Grove Finance, November 21, 2025

  • Obex: An incubator for institutional-grade stablecoin projects with a $2.5B mandate, they’re the latest addition to Stars, currently managing 0.75% ($90M) of the total collateral. The first allocation from Sky was made this week, on November 18, 2025. They intend to increase their SyrupUSDC allocation to $250M.
  • Keel: A new addition to the Stars, designated as a “Pioneer” for the Solana ecosystem, though it has not yet been allocated collateral. Its initial areas of focus include Stablecoin Lending, RWAs and Tokenized Yield, Stablecoin Swap Liquidity, RWA Redemption and Secondary Liquidity, Liquidity and TVL Bootstrapping, and Savings across Solana, with an initial program of up to $2.5B. Its entire balance sheet will consist of funds borrowed from Sky.
  • Others (yet to be announced)

The Allocation Framework sets a Base Rate, which functions as the cost of credit extended by Sky. Stars must generate returns above this rate to retain any value, pushing them toward more innovative and potentially riskier strategy deployments, as noted for each agent above.

The introduction of additional Stars diversifies the allocation of underlying collateral across more venues, thereby significantly expanding USDS’s risk surface. This expansion occurs as the backing becomes increasingly exposed to assets that Aave has either not onboarded or has isolated in separate instances to segregate their risk (such as Horizon for RWAs).

Institutional Primes

A future new category of Sky Agent, alongside Stars, was proposed on the Sky Forum: Institutional Primes. In concept, they are similar to Stars, with the key distinction that Institutional Primes have a fully KYC-restricted token holder base, unlike Stars such as Spark, whose tokens are freely tradable.

Institutional Primes also do not receive access to the Pioneer System under Sky Primitives, which provides rewards and incentives for Agents expanding to new chains. Unlike Stars, their number is not capped and can effectively be unlimited. USDS collateral will also be allocated to Institutional Primes through Sky’s credit lines.

Parameter Changes

Collateral Disabled

On Core, $9.1M worth of assets have been borrowed against USDS. Given the changed risk profile of USDS described above, we support disabling USDS as collateral and setting its LT/LTV parameters to zero, which would prevent additional suppliers from borrowing against it. On V3 Prime, USDS is already disabled as collateral.

Increase Reserve Factor (RF)


Source: USDS Supply APR Change, LlamaRisk, November 21, 2025

With the proposed increase in the USDS Reserve Factor (RF) from 10% to 25%, the supply APR would decrease by 28 basis points on Core and 24 basis points on Prime at their respective current utilization levels. This change would also result in a 150% increase in protocol revenue if utilization remains unchanged.

Removal from E-modes

On Core, USDS is currently borrowable only within the sUSDe, USDe, eUSDe, and PT-sUSDe-27NOV2025 E-modes and cannot be used as collateral. Of the $15.4M in total USDS borrows, 48.8% is backed by collateral originating from these E-modes, primarily sUSDe and USDe.

Given the low risk associated with USDS’s borrow-only status under E-modes, and considering current activity levels (shown below), we are comfortable with either keeping or removing USDS from E-modes on Prime and Core based on DAO preference, as this choice does not introduce any incremental risk.


Source: LlamaRisk, November 21, 2025

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.

Change log

Dec 1st, 2025: removal of USDS as a borrow asset in E-modes.

The current proposal has been escalated to ARFC Snapshot.

Vote will start tomorrow, we encourage everyone to participate.

1 Like

Overview

This analysis evaluates the collateral usage of USDS and DAI across Aave markets and finds that both assets provide limited economic value in their current configuration. Their use as collateral is minimal and does not generate meaningful borrowing activity or revenue for the protocol. In light of this imbalance between risk and contribution, we recommend disabling their collateral functionality across all instances where they remain enabled.

The analysis also reviews supply and borrow caps and provides updated recommendations that better reflect observed market usage and risk considerations. In addition, it evaluates the impact of increasing the USDS Reserve Factor and the configuration of USDS within E-modes, aligning the parameter set with the realized usage profile and revenue contribution of the asset.

Rationale

USDS and DAI show very limited usage as collateral across Aave markets. Their presence as collateral does not create meaningful borrowing demand and therefore generates negligible revenue for the protocol. In addition, on several chains, these assets have become relatively illiquid, diminishing via DAI’s effective legacy status. Maintaining collateral eligibility in this context introduces risk that is not justified by the minimal economic value these assets contribute.

Asset Instance Aggregate Collateral Utilization Estimated Annual Revenue (@5% Yearly Borrow Rate) % of Total Supply Used as Collateral
DAI Optimism $45,796 $176 4.13%
DAI Polygon $174,030 $670 3.18%
DAI.e Avalanche $163,020 $628 3.30%
DAI Arbitrum $203,117 $782 3.01%
DAI Ethereum Core $943,503 $3,632 0.59%
USDS Ethereum Core $8.88M (70% PYUSD) $34,663 18.04%
m.DAI Metis $1,980 $8 5.89%

Given this imbalance between risk and reward, it is appropriate to disable their collateral functionality. This adjustment reduces unnecessary collateral-related exposure without affecting meaningful protocol activity or revenue generation.

At the same time, USDS continues to generate borrow-driven revenue on Ethereum Core and, to a much smaller extent, on Ethereum Prime. Parameter changes on the revenue side, particularly the Reserve Factor, can therefore improve the protocol’s compensation.

Reserve Factor Adjustment for USDS

USDS currently operates with a Reserve Factor of 10% on its primary instances, whereas DAI is configured at 25% across all its deployed instances.

Over approximately the last three months, the realised utilisation and rate environment of the main USDS pools can be summarised as follows.

Instance Period Average Utilization Average Borrow APR Average Supply APR (RF = 10%)
Ethereum Core (USDS) Last 3 months 53.6% 6.12% 2.93%
Ethereum Prime (USDS) Last 3 months 78.6% 6.52% 4.76%

Holding realised utilisation and borrow rates constant and only adjusting the Reserve Factor from 10% to 25% therefore produces a mechanical reduction in USDS supply APR and a corresponding increase in protocol revenue.

Over the same three-month window, the impact of such a change is:

Instance Average Supply APR (RF = 10%) Projected Average Supply APR (RF = 25%) Annualized Protocol Revenue at RF = 10% Annualized Protocol Revenue at RF = 25%
Ethereum Core (USDS) 2.93% 2.48% ≈ $216k ≈ $539k
Ethereum Prime (USDS) 4.76% 3.97% ≈ $10.9k ≈ $27.3k

Given the recent increase in complexity of the underlying backing of DAI and USDS, and given current configuration, USDS leaves a meaningful amount of interest margin uncaptured by the protocol. As such, Chaos Labs supports an increase in the Reserve Factor.

E-mode Configuration for USDS

USDS does not function as collateral inside its most commonly used E-modes and it is exclusively used as a borrowed asset, primarily to leverage yield strategies. On Ethereum Core, the dominant use of USDS within E-Mode is as debt against USDe and sUSDe within Ethena-related E-mode, where users borrow USDS to increase exposure to the Ethena yield. This corresponds to approximately 4.12m USDS borrowed, or 30% of the outstanding USDS debt, with the remaining 70% sitting outside them.

Maintaining USDS as a borrow-only asset within these E-modes does not create additional tail-risk pathways for Aave relative to treating it as a standard borrowable asset outside E-modes. However, if the DAO wishes to remove USDS from E-modes as part of a broader initiative, such a change can be implemented without creating risk concerns, as it would not affect collateral risk nor introduce liquidity stress in the USDS market.

Supply and Borrow Caps

This analysis also reviews supply and borrow caps for USDS and DAI across Aave markets. The recommended adjustments reflect the usage of the asset and align supply and borrow limits with observed demand and risk considerations.

Specification

Instance Asset Current Max LTV Recommended Max LTV
Optimism DAI 63% 0%
Polygon DAI 63% 0%
Avalanche DAI.e 63% 0%
Arbitrum DAI 63% 0%
Ethereum Core DAI 63% 0%
Ethereum Core USDS 75% 0%
Metis m.DAI 63% 0%
Asset Chain Current Supply Cap Recommended Supply Cap Current Borrow Cap Recommended Borrow Cap Current RF Recommended RF
USDS Ethereum Core 800,000,000 160,000,000 360,000,000 144,000,000 10% 25%
USDS Ethereum Prime 200,000,000 40,000,000 180,000,000 36,000,000 10% 25%
DAI Ethereum Core 338,000,000 200,000,000 271,000,000 180,000,000 - -
DAI OP Mainnet 10,000,000 2,000,000 7,000,000 1,800,000 - -
m.DAI Metis 1,000,000 200,000 900,000 180,000 - -

Disclaimer

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

Copyright

Copyright and related rights waived via CC0

1 Like

The current ARFC Snapshot ended yesterday, reaching out both Quorum and YAE as winning option with 560.2K votes.

Therefore [ARFC] Remove USDS as collateral and increase RF across all Aave Instances has PASSED.

Next step will be the publication of an AIP for final confirmation and enforcement of the proposal.

Also, we will include in the AIP latest feedback provided by Chaos.

Hey everyone, first of all it’s of course completely up to Aave governance to decide to remove USDS and DAI as collateral given they had relatively low usage. Sky Ecosystem and the composition and risk of the collateral portfolio and allocation system currently requires a lot of specialized knowledge to properly understand, and we’re working very hard to make it more standardized and easy to access - in the short term through individual data rooms such as Spark Data Hub and https://data.grove.finance/ and updated risk frameworks, and long term through the sentinel network (warning: wall of text https://forum.sky.money/t/sky-agent-framework-sky-sentinel-network/27473) and other future proofed scalability solutions.

I am writing just to respond to one particular allegation which is confusing to me, the claim that Stars borrow from sky through “credit lines at low or zero cost”. Given that Stars pay the base rate to Sky, which is currently 4.55%, I am wondering what examples you are referring to of these zero or low cost credit lines. This may be a misunderstanding of how the fee payments are settled, which happens through monthly settlements, such as the one that was just completed this monday as a part of the 2nd November Executive Vote: https://vote.sky.money/executive/template-executive-vote-launch-starguards-october-monthly-settlement-cycle-and-treasury-management-function-october-ranked-delegate-compensation-atlas-core-development-compensation-payment-to-gnosis-add-allocator-obex-a-to-sp-beam-prime-agent-proxy-spells-november-27-2025

Overall I think it’s unfortunate that it won’t be as easy for Aave to tap into the returns that Sky generates through sUSDS, but regardless Sky will of course continue to underwrite and deploy into Aave instances, such as RLUSD in Aave Horizon, and also have upcoming Stars that will be aiming to work with Aave v4 instances which is a feature that synergizes very well with the modular Star model.

In the future, as Sky’s transparency and scalability initiatives pay off, I believe it will once again make sense to use it as collateral in v3, and also find opportunities for it in Aave v4

4 Likes

The challenge I see is that the risk of using USDS or DAI as collateral outweighs the actual benefits or demand for using these assets in that role. I don’t see any compelling use cases for USDS or DAI as collateral, nor are the potential advantages significant enough to justify maintaining the status quo.

I’m open to reassessing this position if stronger, mutually beneficial opportunities emerge between the Sky ecosystem and the Aave ecosystem. But at the moment, Sky cannibalizes Aave with one star and offers only crumbs with the other. This approach is not conducive to creating meaningful synergies between the two DAOs.

1 Like