Aave is walking away from six chains. A governance filing drafted by risk manager LlamaRisk on July 30 winds down the protocol's deployments on Sonic, Scroll, zkSync, Metis, Soneium and Aptos, and retires 50 low adoption asset reserves at the same time. Stani Kulechov confirmed it the same day. Almost every write up of it has led with $98 million, and that number describes something else entirely.
The six chains hold $12.8 million between them
The $98.1 million figure covers the whole cleanup package. Of that, $85.3 million sits in the 50 individual reserves and 21 matured Pendle principal tokens spread across eleven separate Aave deployments, most of them on chains nobody is closing. The six chains actually being shut down account for $12.8 million in supply and $4.1 million in debt.
The deposit slide behind the decision is real enough. Over six months Soneium fell from $3.2 million to $173,000, Aptos from $18.0 million to $1.0 million, zkSync from $7.2 million to $844,000, Scroll from $16.1 million to $2.2 million, Metis from $1.4 million to $297,000, and Sonic from $28.9 million to $7.6 million. Sonic is the awkward one there, because the Sonic Foundation seeded that launch with $15 million, 50 million S tokens and $800,000 in stablecoins in May 2025.
Nine chains sat under the floor when Aave wrote it
Aave did not invent the $2 million bar this week. A TEMP CHECK posted to the governance forum on November 19, 2025 established a $2,000,000 annual revenue floor for new instance deployments, and it came with a table of what every existing chain was earning. Nine of them annualized below the floor. Five of those nine are on this week's list.
| Chain | Annualized revenue, Aave forum, Nov 2025 | Aave TVL now, DefiLlama | On the wind down list |
|---|---|---|---|
| BNB Chain | $731,472 | $151.81M | No |
| Optimism | $546,454 | $41.68M | No |
| Scroll | $515,842 | $1.72M | Yes |
| Gnosis | $319,898 | $51.82M | No |
| Sonic | $202,198 | $6.69M | Yes |
| Celo | $58,351 | $5.04M | No |
| Soneium | $51,940 | $155,254 | Yes |
| zkSync | $20,464 | $807,588 | Yes |
| Metis | $3,346 | $304,231 | Yes |
Aptos carried no line in that November table, which leaves four chains measured under Aave's own floor and left running: BNB Chain, Optimism, Gnosis and Celo. Celo is the sharpest case. It earned $58,351 on Aave's numbers against Sonic's $202,198, holds $5.04 million in TVL today against Sonic's $6.69 million, and keeps its market while Sonic loses one.
Size is doing the deciding, not revenue
The TVL column explains more. The survivors under the floor carry real balances, $151.81 million on BNB Chain and $41.68 million on Optimism, and unwinding a market that size means finding an exit for thousands of borrowers. The chains going away are all sub $7 million. That is a defensible way to sequence a retreat. It is not what a $2 million revenue floor says on paper, and the gap between the two is worth watching as the DAO moves through later phases.
There is also the question of whether the survivors still clear anything. DefiLlama's 30 day protocol revenue for Aave now annualizes Polygon near $1.6 million and Arbitrum near $1.7 million, both of which sat comfortably above $2 million in the November table. The two sets of figures use different accounting, so treat that as a direction rather than a verdict. Phase 1 of the strategy landed as an ARFC on January 29. This week's filing is what came after, and the forum table it was built on still has names on it.
For anyone with funds on the six chains, the mechanics matter more than the headline. Borrowers get pushed out through rate settings and frozen reserves, which takes weeks. Watch the deposit outflows Aave has absorbed before for a sense of pace. The same measurement problem shows up when two trackers report the same layer 2 balances $29 billion apart, and chain level economics arguments like Solana's burn rate proposal turn on which number the author picked.
