Aave Labs’ proposed September 14, 2026 arrangement would create a single isolated Aave V4 Hub and Spoke for institutions borrowing stablecoins, without changing existing Aave markets or reserves, according to the Aave Governance Forum proposal. It would lend against assets held and enforced outside the protocol, rather than place another asset into an existing pool, using dedicated accounting and synchronization infrastructure to represent their value onchain.
The collateral would remain with Anchorage for the life of the loan, outside any blockchain wallet controlled by the protocol. Chainlink’s CRE infrastructure would reconcile Anchorage’s custody records with the onchain borrowing position and maintain a Proof of Reserve record.
Collateral that is neither tokenized nor deposited into a smart contract may be usable in the arrangement. The corresponding tradeoff is reliance on custodian action, contractual recourse and systems that align offchain records with the onchain loan instead of some of the direct, token-native enforcement associated with onchain collateral.
Anchorage collateral would be verified onchain without being held onchain
Under the proposed design, Anchorage would custody the underlying assets. Neither Aave nor Chainlink would hold them. Chainlink CRE would connect Anchorage’s collateral-management system to the onchain position, reconciling the two and maintaining the Proof of Reserve record.
The borrower’s protocol position would be represented by a non-transferable Custodied Collateral Token, or CoCT. That token is deliberately narrower than a conventional collateral token: it is an internal accounting unit, with minting and burning controlled exclusively by CustodySync. It does not give its holder a direct legal claim over the asset held at Anchorage.
This distinction defines the proposal’s risk architecture. In a conventional tokenized-collateral arrangement, the onchain asset and the smart-contract rules governing it are closely connected. Here, the CoCT communicates the collateralized position to Aave’s system, but it is not the collateral itself and cannot independently establish ownership. The relevant legal recourse would run through an Account Control Agreement.
That separation may allow a broader range of institutionally held assets to support borrowing than a model in which every asset must be placed onchain. But an accurate onchain representation is necessary, not sufficient, for enforcement: the accounting record, custody arrangements and legal agreement must work together both while the loan is healthy and, more critically, when it is not.
The proposed isolated Hub and Spoke is therefore not simply a technical wrapper around an Anchorage account. It is the place where Aave would recognize borrowing capacity against custody-held assets, while the actual possession and legal control of those assets remain outside the protocol. The design makes onchain lending contingent on an offchain control framework rather than eliminating that framework.
Liquidation moves from an automated protocol event to an Anchorage OTC sale
The clearest change appears in liquidation. Standard DeFi liquidations are generally designed around onchain collateral and automated transactions. A borrower falls below a required threshold, and liquidation mechanisms can sell or transfer the relevant token under protocol rules. The Anchorage proposal would use a materially different path.
If a position must be liquidated, Anchorage would sell the underlying collateral through an over-the-counter transaction and then settle the resulting debt onchain. The proposal consequently introduces risks tied to custodian execution, settlement timing and available market liquidity that do not exist to the same extent in an automated AMM liquidation.
That does not make the arrangement inherently unworkable; OTC execution may be better suited to certain institutional assets or transaction sizes than an onchain market. But it changes what lenders and Aave governance must assess. The decisive question is no longer only whether a token’s oracle price and a smart contract’s liquidation incentive can protect a position. It also encompasses the ability to execute a sale, complete settlement and apply contractual rights through the Account Control Agreement in the required circumstances.
The CoCT structure reinforces that point. Because it is an internal, non-transferable accounting representation without a direct legal claim on the underlying asset, it cannot be treated as an independently saleable substitute for the custodied collateral. The token records a position in the system; Anchorage’s sale of the underlying asset supplies the liquidation proceeds.
This is the central exchange Aave is proposing. The protocol could accommodate collateral that remains entirely offchain, but the enforcement chain contains more operational and legal links. Synchronization infrastructure is needed to maintain the relationship between the custody account and the loan. Custodian performance is needed to execute the sale. Legal arrangements determine recourse. The system’s resilience in stress would depend on those linked processes as well as on the onchain debt settlement.
The proposal itself identifies the shift, distinguishing its liquidation process from automated AMM liquidation. For an isolated institutional market, that may be an intentional design choice rather than a flaw—but it makes execution quality and settlement mechanics part of the credit framework, not peripheral service-provider considerations.
Isolation contains a new risk model, but governance has not priced it yet
Aave’s proposed containment measures are designed to ensure that this new model does not directly spill into the protocol’s existing markets. The collateral would sit in a dedicated Hub and Spoke, and CoCT draw caps would be permanently set to zero. That means the receipt tokens could not themselves be borrowed out of the Hub.
Those restrictions address one specific concern: preventing a representation of offchain collateral from becoming a reusable lending asset elsewhere in Aave. They do not, however, settle the economic terms that will determine how much risk the isolated market can assume.
The proposal is not a live lending market. Collateral factors, liquidation incentives, supply and borrow caps, interest-rate strategy and oracle configuration remain subject to recommendations from Aave’s risk providers, followed by a subsequent AIP vote. These are not implementation details. In a structure where liquidation depends on an OTC sale and settlement process, the relationship among collateral factors, incentives, liquidity and oracle design will determine the degree of room available before enforcement must begin.
Aave has already established an institutional lending channel through Horizon, which launched in August 2025 for qualified borrowers using tokenized real-world assets as collateral. Aave Labs said Horizon had surpassed $440 million in deposits by February 2026, offering a prior-period indication of institutional demand for the broader strategy.
Horizon is a relevant benchmark, but not a direct proof point for the Anchorage plan. Horizon used tokenized collateral. The proposed Anchorage structure would keep the asset entirely offchain and rely more heavily on a custodian, legal agreements and synchronization infrastructure. Its appeal may lie in expanding the possible collateral base beyond assets that can be made token-native, while its risk profile is correspondingly less dependent on smart contracts alone.
Isolation gives Aave governance room to test that distinction without altering current reserves or markets. Still, the structure’s practical conservatism will not be known from the architecture by itself. It will emerge in the later decisions on collateral factors, liquidation incentives, caps, rates and oracle configuration—and in whether governance approves the subsequent AIP needed to move the proposal beyond design.
