Skip to content
Data provided by CoinGecko
Ethereum News

Aave V4 USDC sweep is a forum plan, not live yield

Glass tank of coins in a vault beside a glowing ring

Aave has not turned idle USDC on its Ethereum Core Hub into a live yield sleeve. On 20 November 2026 at 22:32 UTC, TokenLogic published an Aave Request for Comments that asks the DAO to deploy the first LiquidityBufferController and sweep spare USDC into Circle Gateway. The specification lists that controller as TBD. Until an on-chain proposal sets an address, no supplier is earning this Gateway yield, and no borrower is waiting on the new keeper.

Ethereum traded at $1,682.68 in the CoinGecko snapshot fetched at 2026-11-22T00:01:19+00:00, up 3.92% over the prior 24 hours. Bitcoin was at $60,114 in the same snapshot. The forum post does not move those prints. It asks governance to relocate a slice of idle USDC and still count that slice as supply.

What happened

The post is a single forum item, number 26469, from the TokenLogic account. It proposes the controller for one market: USDC on the Ethereum Core Hub at 0xCca852Bc40e560adC3b1Cc58CA5b55638ce826c9. The asset line is the usual Ethereum USDC contract, 0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48. A public Ethereum node returned contract bytecode for the hub address when this desk checked at 2026-11-22T00:09:07+00:00. That only shows the hub contract is deployed. It does not show a controller. The specification cell for LiquidityBufferController still says TBD. The hub is on Etherscan, and the request is on the Aave governance forum.

TokenLogic says USDC supply on that hub rose from $9.0 million to $59.3 million over the three months before the post, with a peak of $65.1 million on 29 October. Idle USDC averaged 9.2% of supply, about $2.3 million. The ten largest suppliers held 32% of supply. The ten largest borrowers held 44% of borrows. The largest supplier alone held 13%, or $6.9 million, which is more than the hub usually keeps idle.

The off-chain buffer floor is 5% of supply. The on-chain liquid buffer is 9%. Exposure is capped at 10 million USDC and at 25% of supply, whichever is lower. Sweeps wait at least a day (86,400 seconds) and need at least 100,000 USDC of excess. Recalls are capped at 10 million USDC, at least 30 minutes apart, and the keeper checks every 5 minutes. The table sets a max fee of 2 USDC. Funds may move only from the hub to Gateway and back.

Next steps in the post follow the ordinary Aave process. Collect feedback. If the forum agrees, go to Snapshot. If Snapshot returns YAE, which is the forum’s own word for a yes, write an AIP that sets the controller and the on-chain parameters. Nothing in that list has a transaction hash yet.

Context

Borrowers pay interest only on USDC they actually borrow. Idle balances sit in the hub and earn nothing. Aave V4 lets a LiquidityBufferController sweep idle tokens into a strategy while the hub still counts those tokens as supply. TokenLogic writes utilization as drawn divided by (liquidity + drawn + swept). A sweep does not change that ratio, so the borrow rate stays put. Only the unswept balance can be withdrawn or borrowed until a recall lands.

Circle’s Gateway overview, page date 11 November 2026, says deposits sit in non-custodial wallet contracts, the same asset can mint in under 500 milliseconds once a balance exists, and a 7-day trustless withdrawal pulls funds from the Gateway Wallet without Circle’s API. TokenLogic uses both doors. The keeper path is a signed burn intent, a Circle attestation, then a mint back to the hub. If attestation is down, an admin can pause sweeps and use the on-chain withdrawal, which the forum post clocks at about 7 days, or 50,400 blocks. Those match the two speeds in the Circle docs. They are not one speed. The Circle filings guide covers the issuer. Counterparty risk is the right label for why “same issuer” is not “same timing.”

TokenLogic says Gateway adds no new credit name, because hub USDC already depends on Circle. The risks they add are operational: attestation, the keeper, the Gateway contracts, and the wait for cash. The open question is whether the hub can pay a withdrawal while funds are mid-recall.

The sizing test replays withdrawals and borrows since 1 August and assumes idle USDC starts at the 5% floor, with everything above it already swept. A start fails if idle hits zero. At a 5% floor and a 9% liquid buffer they report a 0.46% failure share, 17 events, inside a 0.5% limit they chose. A continuous replay shows 21 sweeps and 20 recalls, and they say no withdrawal would have reverted. That is one backtest, not a live keeper log.

Revenue is small on purpose. At about $59.0 million of supply, utilization near 91%, and an assumed 2.9% Gateway rate, TokenLogic puts net annual revenue around $13,900 in the comparison and about $14,000 in the recommendation, after fees and gas. The replay did not hit either exposure cap. They say the hub is highly utilized, so little USDC sits above the 9% buffer. If utilization were 85% to 88%, the table rises to about $47,000 to $93,000 a year at the same supply. Those rows are sensitivity, not cash in the DAO.

DefiLlama’s Aave V4 endpoint, fetched at 2026-11-22T00:09:07+00:00, showed about $483.8 million of liquidity on Ethereum and about $303.2 million borrowed there. That is all of V4 on Ethereum, not this hub. The Ethereum fee overview at the same reading put Aave V4 at about $160,410 over 24 hours and about $385,331 over seven days, against about $2.04 million for Aave V3 over 24 hours. All-protocol Aave liquidity was about $13.76 billion on the main Aave endpoint. The forum’s $59.3 million is one reserve in that book. Liquidity here means cash the hub can pay now. V4 figures are from DefiLlama’s Aave V4 page and that API pull.

Our read

This is a request to park idle USDC in Circle Gateway under caps, with a keeper that is not on chain yet. It is not a yield product. The base-case $14,000 a year is about two hours of the $160,410 Aave V4 Ethereum fee total DefiLlama recorded for the prior day (14,000 / 160,410 × 24). Even the optimistic $93,000 row is a fraction of one day of that fee print. The sweep will not be why Aave V4 makes money.

The part that can matter is the interest-rate signal. Because swept USDC still counts as supply, reported utilization can stay under the kink while the hub is actually dry. TokenLogic’s own example: if Gateway holds 10% of supply and every remaining idle dollar is borrowed, reported utilization is 90%. They say the Core Hub USDC kink, called uOptimal, is 92%. Borrow rates would not hit the steep slope that usually pulls in repayments and new supply. Under a working keeper this should be brief. If Circle’s attestation service sticks, the only exit they describe is the 7-day withdrawal, and they say the rate model may need a lower kink. That is the risk worth underlining. The $14,000 is not.

TokenLogic also discloses that it is an active service provider to the Aave DAO and the beneficiary of stream 100086. Read the parameter choices as a vendor proposal, then check them against the replay.

Falsifiable claim: no LiquidityBufferController is set for USDC on this hub. The claim fails if, before 00:00 UTC on 6 December 2026, the Ethereum Core Hub at 0xCca852Bc40e560adC3b1Cc58CA5b55638ce826c9 shows a non-zero LiquidityBufferController for USDC at 0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48.

What to watch next

First, the process. Feedback on thread 26469, then a Snapshot, then an AIP only if that vote is YAE. A long forum thread is not an execution.

Second, the address cell. It should stay TBD until the AIP. A non-zero controller is the moment the backtest becomes an operation.

Third, the 10 million USDC cap. TokenLogic says that cap starts to leave yield behind once supply is about $200 million to $250 million, if utilization holds. Raising it means a full unwind needs several 30-minute recalls. They say the floor and the buffer should be recalibrated before that raise.

Fourth, the failure mode they already named. If recalls slip, or the attestation service goes dark, watch for an interest-rate change that lowers uOptimal by the share sitting in Gateway. Also watch the three-month review they promised, which is supposed to compare real keeper delays with the 5-minute check and the 30-minute gap used in the note. Later votes belong with Ethereum news.