[ARFC] Low Adoption Asset Deprecation on Aave V3

[LlamaRisk]

Summary

LlamaRisk, together with the Aave service providers working on risk surface reduction, recommends offboarding a broad set of low-activity Aave V3 reserves along with six whole deployments. The following specification lists the current configuration of each reserve and the parameter changes required to wind it down.

The individual removals cover 50 reserves and 21 matured Pendle PTs across eleven deployments, holding $85.3M of supply and $11.5M of debt. The whole-market deprecations add 25 reserves across Sonic, Scroll, zkSync, Metis, Soneium and Aptos, holding $12.8M of supply and $4.1M of debt. A large share of the set is already in motion, with borrowing disabled, reserves frozen, or caps already reduced to 1 on most of it.

A further set of V3 reserves flagged for Chainlink price feed risk is being offboarded through the companion oracle deprecation ARFC. Those reserves receive a freeze, caps of 1 and a fixed-price oracle there, so they are listed in the cross-reference section below and excluded from the tables and specification of this document.

Motivation

This recommendation is part of a portfolio-level effort to reduce Aave’s risk surface across deployments, applying the Aave Risk Framework rather than reacting to a problem with any single asset. Most reserves in scope are assets whose usage on Aave has remained below, or declined to, the level the framework requires for a standalone listing. Each listed reserve carries a fixed operational load regardless of its size: an oracle to maintain, risk parameters to monitor, and a liquidation path that must function reliably. Where a reserve’s activity no longer justifies that load, it is wound down.

The scope also includes several structural cases. Bridged tokens such as USDC.e and USDbC are removed where the native version is listed and retained, so the same asset is not carried twice. MaticX is being sunset by its issuer. A group of Pendle Principal Tokens has passed maturity, after which the reserve serves no ongoing function. On six smaller deployments the assessment applies to the whole market rather than individual reserves: aggregate activity has declined to a level where the revenue the deployment generates does not cover the cost of supporting it, so the entire market is wound down at once.

Wind-down mechanics

Each reserve is wound down so that exposure is removed while users exit in an orderly way and liquidation risk is minimised.

The default action on every reserve is to freeze it, reduce its supply and borrow caps to 1 and, on reserves that carry a borrow, raise the RF. The freeze blocks new supply, borrow, and use as fresh collateral. Raising the RF directs more of the borrow interest to the treasury, so suppliers earn less yield and withdraw their assets. As supply leaves, utilisation rises and borrowers are pushed to repay. For assets that are currently used as collateral, freezing the market stops new activity, but the positions already in place can remain open. Any effort to unwind those positions is discussed on a case-by-case basis. The next steps listed per reserve below reflect this default action and the current state of each reserve.

Beyond the default action, further levers may be applied depending on how each asset behaves, so they are deliberately not part of the per-reserve next steps:

• Where borrowers do not repay despite the raised reserve factor, the IRM curves are increased to make carrying a borrow more expensive.
• When it is deemed risky to keep the exposure, the Liquidation Threshold can be gradually reduced to deleverage the market, applied per asset when that is necessary to remove a lingering collateral position.

The whole-market deprecations apply the same approach to every reserve at once: each reserve is frozen with caps reduced to 1 and, where it is borrowed, the RF is raised to 99% and the IRM base rate is set to 5%, matching the initial rate step used in the oracle deprecation ARFC. We will reassess periodically whether further measures, such as raising the IRM further, are needed.

Overlap with the oracle deprecation ARFC

The reserves of Aave V3 below qualify for this scope on adoption grounds but also carry a Chainlink price feed assessed at elevated risk. They are handled in the oracle deprecation ARFC, which freezes each reserve, reduces its caps to 1 and replaces the live feed with a fixed-price adapter. They are excluded from the specification of this document to avoid double-specifying the same reserves.

Asset	Instance	Feed tier	Supplied	Borrowed
LUSD	Ethereum	Very High	$2.0M	$665k
RPL	Ethereum	High	$608k	$212k
BAL	Ethereum	Very High	$46k	$3k
FRAX	Ethereum	Very High	$38k	$29k
KNC	Ethereum	High	$6k	$1k
FXS	Ethereum	High	$1k	$17
LUSD	Arbitrum	Very High	$192k	$75k
FRAX	Arbitrum	Very High	$170k	$48k
MAI	Arbitrum	Very High	$16k	$13
MAI	Avalanche	Very High	$21k	$4k
FRAX	Avalanche	Very High	$7k	$6k
LUSD	Optimism	Very High	$26k	$18k
MAI	Optimism	Very High	$9k	$3k
sUSD	Optimism	Very High	$26k	$12k
GHST	Polygon	Very High	$31k	$311
miMATIC	Polygon	Very High	$9k	$11
BAL	Polygon	Very High	$1k	$450
STG	Ethereum	Medium	$92	$2

USDm on Celo and SCR on Scroll are also covered by the oracle deprecation ARFC. SCR additionally falls under the Scroll whole-market deprecation below, where the deployment-wide parameter changes still apply.

Scope by market

The first chart sets the in-scope supplied value of each live deployment against that market’s total supplied value. On the large general-purpose markets the changes touch a small fraction of overall size.

Source: LlamaRisk, July 28th, 2026

The six whole-market deprecations cover their deployments in full and hold $12.8M of supply between them: Sonic $7.6M, Scroll $2.2M, Aptos $1.7M, zkSync $0.8M, Metis $0.3M and Soneium $0.2M. The chart below shows how much each deployment, live and fully deprecated alike, contributes to the roughly $98M of supplied value in scope.

Source: LlamaRisk, July 28th, 2026

Live markets: individual reserve removals

Ethereum Core

The Ethereum Core scope is dominated by two BTC liquid-staking wrappers, FBTC and eBTC, which together hold about $16.3M of supply against $63k of borrowing. Both were listed for collateral demand that has not materialised at scale, and their supplied balances have fallen from roughly $38.9M and $33.5M respectively over the past six months as depositors migrated out. CRV and UNI carry the largest remaining borrow balances in scope, with supply roughly halving over the same window, CRV from $4.2M to $2.2M and UNI from $4.1M to $1.7M. The remainder is a long tail of reserves below $250k, most of them already frozen, borrow-disabled or capped to 1, where this proposal formalises a wind-down that is already underway.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
FBTC	$11.1M	$63k	50%	Active	No	General + E-Mode	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 75%
Matured PTs (15)	$86k	$0	-	Active, supply cap of 1	No	E-Mode only	Matured	Freeze all and set caps to 1
CRV	$2.2M	$238k	35%	Active	No	General	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
UNI	$1.7M	$29k	20%	Active	No	General	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
crvUSD	$185k	$78k	20%	Active, supply cap of 1, borrow cap of 1	Yes	Non-collateral	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
MKR	$175k	$2k	20%	Frozen, supply cap of 1, borrow cap of 1	Yes	General	Limited adoption	Raise RF to 50%
ETHx	$159k	$650	15%	Active, supply cap of 1	No	General + E-Mode	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
1INCH	$154k	$6k	20%	Active	No	General	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
ezETH	$111k	$0	-	Active	No	General + E-Mode	Limited adoption	Freeze reserve and set caps to 1
ENS	$72k	$5k	20%	Active	No	General	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
SNX	$59k	$13k	95%	Active, supply cap of 1	No	General	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 99%
sDAI	$50k	$0	-	Active, supply cap of 1	No	General	Limited adoption	Freeze reserve and set caps to 1
eUSDe	$17k	$0	45%	Active, supply cap of 1	No	General + E-Mode	Limited adoption	Freeze reserve and set caps to 1

Source: LlamaRisk, July 28th, 2026

Ethereum Prime

The Ethereum Prime instance is built around wstETH-collateralised WETH leverage, and the three reserves in scope never found a role in that structure. ezETH supply has fallen from $11.7M in late January to $337k as looping demand consolidated on other deployments. USDS holds $217k against $42k of borrowing and already has both caps at 1. sUSDe holds $46k and was never borrowable. USDS and sUSDe remain fully served by their Ethereum Core listings, where liquidity is materially deeper, and ezETH is already in the Ethereum Core scope of this proposal. The Prime listings therefore add operational load without adding usage.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
ezETH	$337k	$0	15%	Active	No	General + E-Mode	Limited adoption	Freeze reserve and set caps to 1
USDS	$217k	$42k	25%	Active, supply cap of 1, borrow cap of 1	Yes	Non-collateral	Limited adoption, deeper market on Ethereum Core	Freeze reserve, set caps to 1 and raise RF to 50%
sUSDe	$46k	$0	10%	Active	No	General + E-Mode	Limited adoption, deeper market on Ethereum Core	Freeze reserve and set caps to 1

Source: LlamaRisk, July 28th, 2026

Arbitrum

On Arbitrum the largest position is DAI, at $3.6M supplied and $2.5M borrowed, flagged for limited on-chain exit liquidity rather than inactivity, with supply down from $6.1M six months ago. rETH and tBTC hold a combined $3.7M with almost no borrowing against them. USDC.e is the bridged predecessor of native USDC and is removed as a duplicate listing, its supply declining from $1.8M to $1.1M as users migrate to the native version.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
DAI	$3.6M	$2.5M	25%	Active, borrow cap 4,410,000 DAI	Yes	General + E-Mode	Limited liquidity	Freeze reserve, set caps to 1 and raise RF to 50%
rETH	$2.2M	$35k	15%	Active	No	General	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
tBTC	$1.4M	$0	20%	Active	No	General + E-Mode	Limited adoption	Freeze reserve and set caps to 1
USDC.e	$1.1M	$943k	50%	Active	No	General + E-Mode	Bridged USDC, native version retained	Freeze reserve, set caps to 1 and raise RF to 75%
ezETH	$150k	$0	15%	Active	No	General + E-Mode	Limited adoption	Freeze reserve and set caps to 1
EURS	$12k	$6k	20%	Frozen, borrow cap 65,000 EURS	Yes	General + E-Mode	Limited adoption	Set caps to 1 and raise RF to 50%

Source: LlamaRisk, July 28th, 2026

Plasma

Plasma’s in-scope balance is dominated by matured Pendle PTs at $32.2M, which no longer accrue yield and only await withdrawal. The two live reserves reflect the deployment’s post-launch contraction: WETH supply has fallen from $47.2M to $2.1M and weETH from $277.5M to $1.1M over six months as launch-phase looping capital exited.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
Matured PTs (6)	$32.2M	$0	-	Active, supply cap of 1	No	E-Mode only	Matured	Freeze all and set caps to 1
WETH	$2.1M	$861k	15%	Active, supply cap of 1, borrow cap of 1	Yes	General	Limited adoption and liquidity	Freeze reserve, set caps to 1 and raise RF to 50%
weETH	$1.1M	$0	20%	Active	No	General + E-Mode	Limited adoption and liquidity	Freeze reserve and set caps to 1
wstETH	$195	$1	35%	Frozen, borrow cap 5,000 wstETH	Yes	General + E-Mode	Limited adoption and liquidity	Set caps to 1 and raise RF to 50%

Source: LlamaRisk, July 28th, 2026

Base

The Base scope is small: tBTC at $421k, with supply down from $836k over six months, the bridged USDbC removed as a duplicate of native USDC, and a dust ezETH reserve.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
tBTC	$421k	$47k	20%	Active	No	General + E-Mode	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
USDbC	$172k	$125k	50%	Active	No	General	Bridged USDC, native version retained	Freeze reserve, set caps to 1 and raise RF to 75%
ezETH	$17k	$0	15%	Active	No	General + E-Mode	Limited adoption	Freeze reserve and set caps to 1

Source: LlamaRisk, July 28th, 2026

Polygon

Polygon’s scope is led by the bridged USDC.e at $3.4M supplied and $2.9M borrowed, removed as a duplicate of native USDC. EURS holds $1.9M and already sits at an RF of 99% from earlier deprecation steps. MaticX is wound down because Stader is sunsetting the token. The remaining six reserves are frozen dust positions below $40k each.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
USDC.e	$3.4M	$2.9M	60%	Active	No	General + E-Mode	Bridged USDC, native version retained	Freeze reserve, set caps to 1 and raise RF to 85%
EURS	$1.9M	$252k	99%	Active, supply cap of 1	No	General + E-Mode	Limited adoption and liquidity	Freeze reserve and set caps to 1
MaticX	$654k	$5k	20%	Active	No	General + E-Mode	Issuer sunsetting the asset	Freeze reserve, set caps to 1 and raise RF to 50%
DPI	$38k	$4k	35%	Frozen, borrow cap 779 DPI	Yes	General	Limited adoption	Set caps to 1 and raise RF to 50%
stMATIC	$16k	$0	-	Frozen	No	General + E-Mode	Limited adoption	Set caps to 1. Monitor, residual immaterial
SUSHI	$17k	$6k	20%	Frozen, borrow cap 180,000 SUSHI	Yes	General	Limited adoption	Set caps to 1 and raise RF to 50%
CRV	$9k	$416	35%	Frozen, borrow cap 300,000 CRV	Yes	General	Limited adoption	Set caps to 1 and raise RF to 50%
EURA	$7k	$2k	20%	Frozen	No	E-Mode only	Limited adoption	Set caps to 1 and raise RF to 50%
jEUR	$4k	$3k	20%	Frozen, borrow cap 100,000 jEUR	Yes	E-Mode only	Limited adoption	Set caps to 1 and raise RF to 50%

Source: LlamaRisk, July 28th, 2026

Avalanche

The Avalanche scope consists of three bridged tokens, WBTC.e, LINK.e and AAVE.e, carrying minimal borrowing and exposure. For example, WBTC.e supply sits at $2.6M, down from $4.1M six months ago.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
WBTC.e	$2.6M	$122k	20%	Frozen, borrow cap 1,100 WBTC.e	Yes	General	Limited adoption	Set caps to 1 and raise RF to 50%
LINK.e	$824k	$15k	20%	Active	No	General	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
AAVE.e	$154k	$0	-	Active	No	General	Limited adoption	Freeze reserve and set caps to 1

Source: LlamaRisk, July 28th, 2026

Optimism

On Optimism the bridged USDC.e, at $1.2M supplied and down from $2.0M six months ago, is removed as a duplicate of native USDC, while DAI and rETH carry a combined $1.1M with usage below the framework’s thresholds.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
USDC.e	$1.2M	$303k	50%	Active	No	General + E-Mode	Bridged USDC, native version retained	Freeze reserve, set caps to 1 and raise RF to 75%
DAI	$617k	$517k	25%	Active, borrow cap 900,000 DAI	Yes	General + E-Mode	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
rETH	$518k	$1k	15%	Active	No	General + E-Mode	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%

Source: LlamaRisk, July 28th, 2026

Gnosis

GNO is excluded from this scope for now and will be treated separately. The remaining Gnosis reserves in scope are listed below.

WETH is the material Gnosis reserve at $3.2M supplied and $1.8M borrowed, with supply down from $7.8M over six months and borrowing already disabled. The legacy bridged USDC reserve is frozen with caps at 1 and only awaits the RF step.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
WETH	$3.2M	$1.8M	15%	Active	No	General + E-Mode	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
USDC	$142k	$13k	80%	Frozen, supply cap of 1, borrow cap of 1	Yes	General	Limited adoption	Raise RF to 99%

Source: LlamaRisk, July 28th, 2026

BSC

The BSC scope holds wstETH, FDUSD and Cake, together $3.6M of supply. wstETH supply has fallen from $5.2M to $2.1M over six months and carries under $1k of borrowing, while FDUSD’s $582k borrow is the only material debt in scope.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
wstETH	$2.1M	$941	15%	Active, supply cap of 1	No	General + E-Mode	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
FDUSD	$858k	$582k	20%	Active	No	General	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
Cake	$636k	$13k	20%	Active	No	General	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%

Source: LlamaRisk, July 28th, 2026

MegaETH

WETH is excluded from this scope for now and will be treated separately. The two remaining in-scope reserves hold about $4k between them, so no supply history chart is included.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Reason	Next steps
USDT0	$4k	$3k	10%	Active, borrow cap 9,000,000 USDT0	Yes	Non-collateral	Limited adoption	Freeze reserve, set caps to 1 and raise RF to 50%
ezETH	$6	$0	20%	Active, supply cap of 1	No	E-Mode only	Limited adoption	Freeze reserve and set caps to 1

Whole-market deprecations

Each deployment is wound down in full: every reserve is frozen with caps reduced to 1 and, on reserves that carry a borrow, the RF is raised to 99% and the IRM base rate is set to 5%. We will reassess periodically whether additional measures, such as raising the IRM further, are needed.

Sonic

Deposits on the Sonic deployment have fallen from $28.9M to $7.6M over the trailing six months, a decline of 74%, and $2.7M remains borrowed. At current balances, rates and reserve factors the deployment generates under $5k per quarter in protocol revenue, which does not cover the cost of maintaining oracles, monitoring and operational support for the market.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Next steps
USDC	$2.9M	$1.6M	10%	Active, borrow cap 1,050,000 USDC	Yes	General	Freeze reserve, set caps to 1, raise RF to 99% and set IRM base rate to 5%
wS	$1.7M	$879k	15%	Active, borrow cap 36,100,000 wS	Yes	General	Freeze reserve, set caps to 1, raise RF to 99% and set IRM base rate to 5%
stS	$1.5M	$0	10%	Active	No	General	Freeze reserve and set caps to 1
WETH	$1.5M	$249k	15%	Active, borrow cap 97 WETH	Yes	General	Freeze reserve, set caps to 1, raise RF to 99% and set IRM base rate to 5%

Source: LlamaRisk, July 28th, 2026

Scroll

Deposits on the Scroll deployment have fallen from $16.1M to $2.2M over the trailing six months, a decline of 86%, and $422k remains borrowed. At current balances, rates and reserve factors the deployment generates under $5k per quarter in protocol revenue, which does not cover the cost of maintaining oracles, monitoring and operational support for the market.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Next steps
WETH	$1.5M	$169k	50%	Frozen, supply cap of 1, borrow cap of 1	Yes	General + E-Mode	Raise RF to 99% and set IRM base rate to 5%
weETH	$312k	$11k	85%	Frozen, supply cap of 1	No	General + E-Mode	Raise RF to 99% and set IRM base rate to 5%
USDC	$311k	$226k	85%	Frozen, supply cap of 1, borrow cap of 1	Yes	General	Raise RF to 99% and set IRM base rate to 5%
wstETH	$93k	$17k	85%	Frozen, supply cap of 1	No	General + E-Mode	Raise RF to 99% and set IRM base rate to 5%
SCR	$14k	$52	85%	Frozen, supply cap of 1	No	Non-collateral	Raise RF to 99% and set IRM base rate to 5%

Source: LlamaRisk, July 28th, 2026

zkSync

Deposits on the zkSync deployment have fallen from $7.2M to $844k over the trailing six months, a decline of 88%, and $235k remains borrowed. At current balances, rates and reserve factors the deployment generates under $5k per quarter in protocol revenue, which does not cover the cost of maintaining oracles, monitoring and operational support for the market.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Next steps
WETH	$383k	$104k	15%	Frozen, borrow cap of 1	Yes	General + E-Mode	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
ZK	$213k	$138	20%	Frozen, borrow cap of 1	Yes	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
USDC	$102k	$112k	10%	Frozen, borrow cap of 1	Yes	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
weETH	$106k	$0	45%	Frozen	No	General + E-Mode	Set caps to 1. Monitor, residual immaterial
USDT	$22k	$19k	10%	Frozen, borrow cap of 1	Yes	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
wstETH	$18k	$234	5%	Frozen, borrow cap of 1	Yes	General + E-Mode	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
wrsETH	$71	$0	10%	Frozen, supply cap of 1	No	General + E-Mode	Monitor, residual immaterial
sUSDe	$127	$0	20%	Frozen	No	General	Set caps to 1. Monitor, residual immaterial

Source: LlamaRisk, July 28th, 2026

Metis

Deposits on the Metis deployment have fallen from $1.4M to $297k over the trailing six months, a decline of 79%, and $31k remains borrowed. At current balances, rates and reserve factors the deployment generates under $1k per quarter in protocol revenue, which does not cover the cost of maintaining oracles, monitoring and operational support for the market.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Next steps
Metis	$77k	$216	15%	Frozen, borrow cap of 1	Yes	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
m.USDT	$82k	$8k	10%	Frozen, borrow cap of 1	Yes	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
m.USDC	$74k	$21k	10%	Frozen, borrow cap of 1	Yes	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
WETH	$42k	$2k	15%	Frozen	No	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
m.DAI	$21k	$394	25%	Frozen	No	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%

Source: LlamaRisk, July 28th, 2026

Soneium

Deposits on the Soneium deployment have fallen from $3.2M to $173k over the trailing six months, a decline of 95%, and $38k remains borrowed. At current balances, rates and reserve factors the deployment generates under $1k per quarter in protocol revenue, which does not cover the cost of maintaining oracles, monitoring and operational support for the market.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Next steps
WETH	$124k	$8k	15%	Frozen, borrow cap 720 WETH	Yes	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
USDC.e	$42k	$30k	10%	Frozen, borrow cap 7,200,000 USDC.e	Yes	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%
USDT	$7k	$536	10%	Frozen, borrow cap 4,500,000 USDT	Yes	General	Set caps to 1 and raise RF to 99% and set IRM base rate to 5%

Source: LlamaRisk, July 28th, 2026

Aptos

Available liquidity across the Aptos reserves has moved from $18.0M to $1.0M over the trailing six months (-94%, DefiLlama pool TVL, supply net of borrows). The market carries about $1.7M of supply and $719k of debt, and at current balances and reserve factors it generates under $1k per quarter in protocol revenue, which does not cover the cost of maintaining the market.

Asset	Supplied	Borrowed	RF	State	Borrowable	Collateral	Next steps
USDT	$1.2M	$672k	10%	Active, borrow cap 448,000 USDT	Yes	General + E-Mode	Freeze reserve, set caps to 1, raise RF to 99% and set IRM base rate to 5%
APT	$351k	$8k	20%	Active, borrow cap 8,370 APT	Yes	General	Freeze reserve, set caps to 1, raise RF to 99% and set IRM base rate to 5%
USDC	$156k	$40k	10%	Active, borrow cap 33,000 USDC	Yes	General + E-Mode	Freeze reserve, set caps to 1, raise RF to 99% and set IRM base rate to 5%
sUSDe	$5k	$0	20%	Active	No	General + E-Mode	Freeze reserve and set caps to 1

Source: LlamaRisk, July 28th, 2026

Special cases

• MaticX on Polygon is in scope for issuer wind-down, with Stader sunsetting the token. Its Aave oracle is an exchange-rate adapter that values MaticX at its fixed MATIC redemption rate times the MATIC/USD feed, so it already tracks the asset’s redemption value, and roughly $0.6M of MATIC-correlated debt is backed by MaticX collateral that stays correctly valued as MATIC moves. MaticX is already at LTV of 0 and borrow-disabled, and is wound down on the standard basis.

Next Steps

For the whole-market deprecations the sequence continues beyond this proposal: once the RF and rate changes have forced the remaining positions to unwind further, the oracles on those deployments will be deprecated in the same way as in the oracle deprecation ARFC, fixing each feed so the market can be retired completely.

For the individual reserve removals no further parameter work is expected beyond IRM adjustments, applied later only where borrowers have not responded to the initial changes or if the reserve becomes stressed.

Specification

Whenever a reserve is frozen its supply and borrow caps are also reduced to 1, and already-frozen reserves have their caps reduced to 1 where they are not already, so every deprecated reserve ends in the same terminal state as in the companion oracle deprecation ARFC.

Aave V3 Ethereum Core

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
FBTC	No	Yes	175 / 1	1 / 1	50%	75%
ezETH	No	Yes	3,330 / 1	n/a / 1	n/a	n/a
eUSDe	No	Yes	1 / 1	1 / 1	45%	n/a
ETHx	No	Yes	1 / 1	1 / 1	15%	50%
Matured PTs (15)	No	Yes	1 / 1	1 / 1	n/a	n/a
CRV	No	Yes	11,000,000 / 1	1 / 1	35%	50%
UNI	No	Yes	1,500,000 / 1	1 / 1	20%	50%
1INCH	No	Yes	7,000,000 / 1	1 / 1	20%	50%
crvUSD	No	Yes	1 / 1	1 / 1	20%	50%
MKR	Yes	Yes	1 / 1	1 / 1	20%	50%
ENS	No	Yes	50,000 / 1	1 / 1	20%	50%
SNX	No	Yes	1 / 1	1 / 1	95%	99%
sDAI	No	Yes	1 / 1	n/a / 1	n/a	n/a

Aave V3 Ethereum Prime

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
ezETH	No	Yes	150 / 1	1 / 1	15%	No change
USDS	No	Yes	1 / 1	1 / 1	25%	50%
sUSDe	No	Yes	3,000,000 / 1	1 / 1	10%	No change

Aave V3 Arbitrum

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
DAI	No	Yes	4,900,000 / 1	4,410,000 / 1	25%	50%
rETH	No	Yes	1,300 / 1	1 / 1	15%	50%
tBTC	No	Yes	35 / 1	1 / 1	20%	n/a
USDC.e	No	Yes	1,700,000 / 1	1,530,000 / 1	50%	75%
ezETH	No	Yes	66 / 1	1 / 1	15%	No change
EURS	Yes	Yes	80,000 / 1	65,000 / 1	20%	50%

Aave V3 Plasma

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
Matured PTs (6)	No	Yes	1 / 1	1 / 1	n/a	n/a
WETH	No	Yes	1 / 1	1 / 1	15%	50%
weETH	No	Yes	10,000 / 1	1 / 1	20%	n/a
wstETH	Yes	Yes	20,000 / 1	5,000 / 1	35%	50%

Aave V3 Base

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
tBTC	No	Yes	8 / 1	1 / 1	20%	50%
ezETH	No	Yes	23 / 1	1 / 1	15%	No change
USDbC	No	Yes	500,000 / 1	450,000 / 1	50%	75%

Aave V3 Polygon

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
USDC.e	No	Yes	4,390,000 / 1	3,950,000 / 1	60%	85%
EURS	No	Yes	1 / 1	1 / 1	99%	99%
MaticX	No	Yes	9,330,000 / 1	1 / 1	20%	50%
jEUR	Yes	Yes	120,000 / 1	100,000 / 1	20%	50%
stMATIC	Yes	Yes	61,000,000 / 1	n/a / 1	n/a	n/a
EURA	Yes	Yes	300,000 / 1	250,000 / 1	20%	50%
DPI	Yes	Yes	1,417 / 1	779 / 1	35%	50%
SUSHI	Yes	Yes	299,320 / 1	180,000 / 1	20%	50%
CRV	Yes	Yes	1,400,000 / 1	300,000 / 1	35%	50%

Aave V3 Avalanche

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
WBTC.e	Yes	Yes	2,000 / 1	1,100 / 1	20%	50%
LINK.e	No	Yes	155,000 / 1	1 / 1	20%	50%
AAVE.e	No	Yes	7,200 / 1	n/a / 1	n/a	n/a

Aave V3 Optimism

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
USDC.e	No	Yes	1,800,000 / 1	1 / 1	50%	75%
DAI	No	Yes	1,000,000 / 1	900,000 / 1	25%	50%
rETH	No	Yes	450 / 1	1 / 1	15%	50%

Aave V3 Gnosis

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
WETH	No	Yes	3,500 / 1	2,400 / 1	15%	50%
USD//C on xDai	Yes	Yes	1 / 1	1 / 1	80%	99%

Aave V3 BSC

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
wstETH	No	Yes	1 / 1	1 / 1	15%	50%
FDUSD	No	Yes	1,200,000 / 1	1,080,000 / 1	20%	50%
Cake	No	Yes	600,000 / 1	1 / 1	20%	50%

Aave V3 MegaETH

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF
ezETH	No	Yes	1 / 1	1 / 1	20%	No change
USDT0	No	Yes	10,000,000 / 1	9,000,000 / 1	10%	50%

Whole-market deployments

Every reserve on each of these deployments is frozen with caps reduced to 1. Reserves that carry a borrow have the RF raised to 99% and the IRM base variable rate set to 5%.

Aave V3 Sonic

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF	Current IRM base rate	Recommended IRM base rate
USDC	No	Yes	2,170,000 / 1	1,050,000 / 1	10%	99%	0%	5%
wS	No	Yes	59,100,000 / 1	36,100,000 / 1	15%	99%	0%	5%
stS	No	Yes	50,300,000 / 1	1 / 1	10%	n/a	0%	n/a
WETH	No	Yes	556 / 1	97 / 1	15%	99%	0%	5%

Aave V3 Scroll

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF	Current IRM base rate	Recommended IRM base rate
WETH	Yes	Yes	1 / 1	1 / 1	50%	99%	0%	5%
weETH	Yes	Yes	1 / 1	1 / 1	85%	99%	1%	5%
wstETH	Yes	Yes	1 / 1	1 / 1	85%	99%	0%	5%
USDC	Yes	Yes	1 / 1	1 / 1	85%	99%	0%	5%
SCR	Yes	Yes	1 / 1	1 / 1	85%	99%	0%	5%

Aave V3 zkSync

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF	Current IRM base rate	Recommended IRM base rate
WETH	Yes	Yes	600 / 1	1 / 1	15%	99%	0%	5%
weETH	Yes	Yes	200 / 1	1 / 1	45%	n/a	0%	n/a
wstETH	Yes	Yes	500 / 1	1 / 1	5%	99%	0%	5%
wrsETH	Yes	Yes	1 / 1	1 / 1	10%	n/a	0%	n/a
ZK	Yes	Yes	45,000,000 / 1	1 / 1	20%	99%	0%	5%
USDC	Yes	Yes	800,000 / 1	1 / 1	10%	99%	0%	5%
USDT	Yes	Yes	150,000 / 1	1 / 1	10%	99%	0%	5%
sUSDe	Yes	Yes	110 / 1	1 / 1	20%	n/a	0%	n/a

Aave V3 Metis

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF	Current IRM base rate	Recommended IRM base rate
Metis	Yes	Yes	80,000 / 1	1 / 1	15%	99%	0%	5%
m.USDT	Yes	Yes	250,000 / 1	1 / 1	10%	99%	0%	5%
m.USDC	Yes	Yes	400,000 / 1	1 / 1	10%	99%	0%	5%
WETH	Yes	Yes	150 / 1	1 / 1	15%	99%	1%	5%
m.DAI	Yes	Yes	25,000 / 1	1 / 1	25%	99%	0%	5%

Aave V3 Soneium

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF	Current IRM base rate	Recommended IRM base rate
WETH	Yes	Yes	800 / 1	720 / 1	15%	99%	0%	5%
USDC.e	Yes	Yes	8,000,000 / 1	7,200,000 / 1	10%	99%	0%	5%
USDT	Yes	Yes	5,000,000 / 1	4,500,000 / 1	10%	99%	0%	5%

Aave V3 Aptos

Asset	Current frozen	Recommended frozen	Supply cap (current / new)	Borrow cap (current / new)	Current RF	Recommended RF	Current IRM base rate	Recommended IRM base rate
USDT	No	Yes	1,040,000 / 1	448,000 / 1	10%	99%	0%	5%
APT	No	Yes	415,000 / 1	8,370 / 1	20%	99%	0%	5%
USDC	No	Yes	144,000 / 1	33,000 / 1	10%	99%	0%	5%
sUSDe	No	Yes	2,600 / 1	n/a / 1	20%	n/a	0%	n/a

Disclosure

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.

Changelog:

2026-08-11 EtherFi has committed to grow the eBTC reserve significantly in the coming weeks, therefore, it has been decided to remove the asset from the deprecation list on Aave Core market

[menaskop]

I believe this proposal is entirely reasonable. The current crypto-winter has demonstrated that optimizing all forms of transaction costs is one of the most effective ways to improve not only a project’s tokenomics but also its overall economic sustainability.

Moreover, efficiency should always be evaluated not only at the asset level but also at the protocol and blockchain levels. I refer to this as the CPD-framework: Chain, Protocol, dApp. From this perspective, there is little value in overloading the Aave interface by maintaining support for networks that see little to no meaningful usage.

Previously, there was also a more comprehensive proposal regarding the optimization of the process for onboarding new networks and assets. I believe this deserves separate discussion, as it represents an important and substantial topic on its own.

It is also worth noting that following the non-technical rsETH incident, the issue of exposure to different tokens and coins has become significantly more important and has provided valuable lessons for the Aave DAO community. As a result, the rationale behind this proposal has become even stronger.

[Achthar]

But would it not make more sense to first look whether there are other players that would be willing to run the V3 instances (e.g. like on HyperEVM) that are to be deprecated.

The DAO would still generate (sometimes very minor) revenue streams but not have the overhead.

One would just need to find teams that would be willing to take this over (e.g. allow for a few months - if nobody is found by then - it can be shut down).

[menaskop]

How would this work technically in practice? I mean, what would the actual implementation look like? (You can use any of the examples mentioned above.)

And my second question: who would be responsible for the security of users’ assets and the protocol overall in that case?

[Achthar]

It is just an idea - of course, defining every step in detail would require much much more research, but potentially one could do what Maker/Sky has done via their SubDAOs

1. Post that the DAO is looking for a team to manage the Aave deployment on chain X - this should be helped by the network itself (if not, the idea dies right here) if they think it is in their own interest to keep a major lender on their chain
2. User’s opinions matter here and polls should be made to get a sentiment on whether they would just withdraw if someone else would take over (if they would, the idea is dead here)
3. Infrastructure considerations: oracle provider change/handover needs to be defined (reach out to providers and find a working solution for continuation if needed)
4. Once a initial team is found, create a new DAO that will have a the similar role as the HyperLend team has on their deployment
5. One can issue the new governance token to users and initial backers plus a large chunk to the Aave DAO - this would set up the initial governance
6. Aave’s role would be reduced to a sole code license provider / lending infra provider and the deployment in question would be rebranded

This would apply to the examples where the TVL is not too small and there are specific economic opportunities (e.g. leveraged staking that does not exist elsewhere) where Aave V3 is stronger than most other solutions.

I do not have the knowledge about how easy it would be to find teams or collaborators to make this work - this requires comprehensive research and outreach. Due to the limited amount of assets (especially derivatives) on the chains in question, I would assume that a rather lean team can manage them.

I think Aave benefits more from authorized forks and active collaborations rather than other teams starting to fork the Aave V3 code.

[menaskop]

I understand your point, but I think this is more the scope of a standalone proposal than a comment. It would also require at least some of the research to be done upfront - for example, identifying which networks and asset classes such an approach could actually make economic sense for.

Otherwise, I understand the idea. Let’s see whether anyone is interested in taking it further.

[kinetickoala]

i get what you are trying to do, but are the costs really that high?

hopefully this is given enough time so people can make new markets elsewhere

[LlamaRisk]

This ARFC has been moved to the voting stage on Snapshot.

Snapshot

[Abel189]

One aspect I find particularly important is the monitoring phase after the initial deprecation parameters are applied. For markets with meaningful outstanding debt, such as DAI and USDC.e on Arbitrum or WETH on Gnosis, it would be useful to understand what specific metrics or thresholds will trigger the next step of the wind-down (higher IRM, further LT reductions, etc.). A clearly defined monitoring framework could make the transition more predictable while still allowing the risk team to react to borrower behaviour and remaining liquidity.