An Aave governance proposal filed as an ARFC—a detailed precursor to a binding on-chain Aave Improvement Proposal—calls for shutting down V3 markets across six chains and deprecating dozens of token listings. The cleanup covers $98.1 million in supplied assets and $15.6 million in active debt, with balances measured on July 28.
LlamaRisk, Aave's risk management service provider, coordinated with other Aave service providers to draft the recommendation. The proposal targets 50 low-use reserves and 21 matured Pendle principal token listings spread across 11 deployments, alongside a full exit from Sonic, Scroll, zkSync, Metis, Soneium and Aptos.
The Aptos exit is the sharpest reversal in the batch. Aave launched its V3 market on Aptos just 11 months ago, and available liquidity has since fallen 94 percent over six months. Quarterly revenue from the deployment came in below $1,000, according to LlamaRisk's assessment—a figure that barely registers against the operational overhead of maintaining an active market.
Most of the targeted chains were already partially shuttered before this proposal landed. Every reserve on Scroll, zkSync, Metis and Soneium had been frozen prior to the ARFC. Sonic and Aptos are the two remaining active deployments; LlamaRisk recommends freezing both as part of the wind-down sequence.
The governance groundwork for this cleanup predates the current ARFC by months. A temp check on Aave's multichain strategy closed Dec. 5, 2025, drawing 923,400 votes in favor—with under 1 percent opposed—to raise the reserve factor on underperforming instances and shut down the markets on zkSync, Metis and Soneium. That vote also established a $2 million annual revenue floor as a threshold for any new instance deployment going forward.
Scroll entered the deprecation queue afterward through an accelerated path. In April, LlamaRisk filed a direct-to-AIP proposal—bypassing the normal ARFC stage—to freeze every Scroll reserve and raise selected reserve factors. The filing described the action as completing Scroll's deprecation following a rapid deterioration in both network liquidity and Aave market activity on the chain.
Aave published an updated risk framework on June 9 covering asset risk, bridge risk, monitoring criteria and chain risk, along with explicit thresholds for winding down reserves or full deployments. The current ARFC effectively operationalizes those rules for the first time at scale.
Aave founder Stani Kulechov addressed the proposal directly, saying the consolidation will "reduce Aave's economic and technical risk surface as part of the new Aave Risk Framework and Technical Asset Listing Framework." Kulechov was clear the move does not represent a reversal of Aave's multichain expansion strategy. "Aave will continue applying continuous risk assessment for all assets across all deployments," he said.
The timing sits alongside continued expansion elsewhere. Aave launched on Avalanche earlier this month, and the protocol brought V3 lending and its GHO stablecoin to Monad in a separate deployment. The exits and the new launches reflect the $2 million annual revenue threshold established in the December governance vote: chains that clear the bar get capital; chains that don't get wound down.
The Pendle principal token component of the ARFC deserves its own read. Pendle PTs are fixed-yield instruments that expire at a set maturity date—once matured, they no longer generate yield or trading activity, making them dead weight in a lending market. Retiring 21 matured PT listings clears balance-sheet clutter that produces no protocol revenue and carries residual smart-contract exposure.
As an ARFC, the current proposal still requires progression through Aave's full governance process before any on-chain execution. A snapshot vote and then an on-chain AIP vote follow before reserves are formally frozen or liquidity withdrawn. Given that the December 2025 temp check passed with 923,400 votes and under 1 percent opposition, the direction of travel is established—the remaining steps are procedural rather than contested.
