In a July 29 forum-stage proposal, Aave risk service provider LlamaRisk recommended winding down the decentralized lender’s V3 deployments on Sonic, Scroll, zkSync, Metis, Soneium, and Aptos. The plan would put $4.1 million of debt on a staged exit path that keeps existing positions open during the initial step.
The Aave Request for Final Comments, or ARFC, covers 25 lending reserves with $12.8 million supplied, based on LlamaRisk data dated July 28. The forum thread still showed the request under discussion on July 31. Aave’s proposal lifecycle places an ARFC before a community Snapshot and any executable on-chain Aave Improvement Proposal.
LlamaRisk’s economic case rests on support costs exceeding revenue. It said Sonic, Scroll, and zkSync each generate less than $5,000 in quarterly protocol revenue at current balances, while Metis, Soneium, and Aptos each generate less than $1,000. The proposal cites oracle, monitoring, and operational support costs but does not quantify the shortfall.
The same ARFC separately targets 50 individual reserves and 21 matured Pendle principal tokens across 11 deployments, with $85.3 million supplied and $11.5 million borrowed. Those balances sit outside the six-market totals.
How the six-market exit would work
For the six full-market exits, every reserve would be frozen and its supply and borrow caps cut to 1. Reserves carrying debt would receive a 99% reserve factor and a 5% interest rate model base variable rate; unborrowed reserves would not receive those two changes.
The two settings act on different sides of the market. The 5% figure is the base-rate component applied to borrowing. The 99% reserve factor determines how interest revenue is divided, directing nearly all interest paid by borrowers to the Aave treasury and leaving little for supplier yield. LlamaRisk expects lower yield to encourage withdrawals, which raises utilization and gives borrowers a stronger incentive to repay.
A freeze stops new supply, new borrowing, and use as fresh collateral. Positions already open would remain open, and the proposal says any additional unwind would be considered case by case. The starting point also varies: every listed Sonic and Aptos reserve was active in the July 28 tables, while every listed reserve on Scroll, zkSync, Metis, and Soneium was already frozen.
What users could face next
The staged approach is designed to minimize immediate liquidation risk while preserving stronger levers if balances remain. LlamaRisk said later steps could raise interest rate curves or gradually reduce liquidation thresholds for selected collateral. After positions unwind further, deployment oracles could be replaced with fixed-price adapters under the method described in a companion oracle proposal.
Users remaining after the first step would therefore face possible rate, collateral, and oracle changes as the wind-down progresses, while the initial freeze itself leaves their positions open.