[ARFC] Aave Institutional

Summary

Aave Institutional is the DAO’s offchain over-collateralised lending business. It lends dollar stablecoins to institutional borrowers against BTC and ETH held at qualified custodians, at rates materially above what the same dollars earn onchain.

Aave Labs asks the DAO to approve two funding sources for this new business, running in parallel:

  1. A new GHO facilitator with an initial bucket capacity of 25M GHO managed by the GHO Stewards.
  2. The use of DAO balance-sheet assets, pledged as collateral to borrow up to $25M of USDC or USDT.

Loans are advanced primarily in USDC or USDT, so GHO issuance is paced to what GHO can support at the time. The balance sheet carries the first loans so that origination can begin while GHO is grown to support it, and the mix shifts toward GHO over time.

Borrowers pay 6.0% to 8.0% APR against a funding cost of approximately 4.5% on either route. The full net interest margin, 1.50% to 3.50%, accrues to the DAO. Aave Institutional currently holds approximately $300M of deployable borrow demand, and the lead facility is $20M against BTC.

Every funding authorisation requires approval from the GHO Stewards in accordance with current DAO policy (two-of-three multisig between Aave Labs, TokenLogic and LlamaRisk). No GHO is advanced against offchain collateral without the GHO Stewards agreeing to it at the time, on live Stability Module information. Every conversion between GHO and the lending currency is routed and executed jointly with TokenLogic.

Motivation

GHO’s constraint is not mint capacity. It is productive demand that generates yield from outside the protocol at attractive rates. Aave Institutional is that demand.

Aave Institutional opens a market the protocol does not serve today. Institutions want dollar financing against BTC and ETH held with a qualified custodian, on bilaterally negotiated terms, and they will pay above onchain rates for it. It is an established lending market with documented demand, and GHO is well placed to fund it.

Three things make this new business particularly attractive for the DAO:

Attractive Margins: Borrowers pay 6.0% to 8.0% APR, a cost of capital above the rates currently available on Aave Protocol, against a funding cost of approximately 4.5%. The full net interest margin, 1.50% to 3.50%, accrues to the DAO.

Conservative Collateral and Risk Profile: Loans are advanced at a typical 60% to 75% loan-to-value against BTC and ETH. Collateral is held and monitored by a regulated qualified custodian that issues the margin calls and enforces them. Any liquidations are automated via the regulated trading desk of the relevant qualified custodian, rather than sitting with Aave Institutional.

Documented Demand: Aave Institutional currently holds approximately $300M of indicated borrow demand. The lead facility is sized at $20M against BTC at approximately 60% loan-to-value, evergreen with a 90-day notice period on which either party may call the facility or adjust its rate, with tri-party custody and title transfer on default.

Aave already mints GHO against volatile crypto collateral through the V3 facilitator. This proposal applies the same mechanism to collateral of comparable quality at a lower loan-to-value, with a qualified custodian appointed to monitor and enforce it, and at materially better economics for the DAO.

Specification

1. Background

Field Detail
Title Aave Institutional
Mechanism A dedicated facilitator minting GHO against a book of over-collateralised institutional loans secured on BTC and ETH held at qualified custodians
Operator GHO Stewards
Network Ethereum mainnet

2. Lending capacity and funding

GHO funded route:

Field Detail
Capacity 25M GHO under the new facilitator
Use of proceeds Origination of secured institutional loans
GHO Borrow Cap GHO Stewards
Indicative borrower pricing 6.0% to 8.0% APR
Loan tenor 30-day rolling to 12 months
Lending currencies USDC and USDT primarily, GHO where the borrower is amenable
Revenue to the DAO The full net interest margin, 1.50% to 3.50% per annum on the amount drawn

DAO Balance Sheet funded route:

Field Detail
Funding source Over-collateralised borrowing against existing DAO balance-sheet assets on Aave V3 markets
Phase Primary funding source until GHO can support the loans, with facilitator-minted GHO phased in thereafter
Collateral pledged WETH and WBTC, with AAVE capped as set out below
Limit on AAVE as collateral AAVE may not exceed 50% of the collateral pledged to this route, measured at the time of each pledge. Management of subsequent drift in the non-AAVE share sits with the Aave Finance Committee, led by TokenLogic.
Debt asset USDT and USDC, drawn in the currency of the borrower’s facility
Borrowing cost The venue’s prevailing borrow rate, expected 4.0% to 5.0% APR
Size Sufficient collateral to fund up to $25M USD equivalent. This is additional to the facilitator’s 25M GHO bucket capacity, not part of it

3. Peg impact and conversion

Aave Labs considers the effect on the GHO peg the material question on this proposal.

Stability Module redemption inventory across the GHO deployment stood at $59.9M on 24th September 2026, held in two funded instances.

Instance Network Underlying Redeemable
GSM stataUSDT (0x8822…f5e3) Ethereum waEthUSDT 19.2M USDT
GSM stataUSDT0 (0xd061…ebf37) Plasma waPlaUSDT0 40.7M USDT
Total 59.9M USDT

The requested initial capacity of 25M GHO is approximately 42% of current redemption inventory. The USDC instances hold negligible redeemable balance and are excluded. GHO is fungible across networks through Chainlink CCIP, so redemption inventory is a property of the deployment rather than of any single network. DAO service providers are in active conversations with depositors to grow GSM balances and expand capacity under this route.

Minting GHO against offchain collateral carries a legitimate objection. It increases GHO in circulation while placing the backing outside the protocol’s immediate reach, and converting that GHO into stablecoin dollars through the GSM draws on the same inventory that supports the peg.

Two things address this. The first is the authorisation control in Specification 5, which places assessment of Stability Module impact with the GHO Stewards at the point of every funding decision rather than fixing it in advance by a term that could not anticipate market conditions. The second is that loans funded from DAO balance-sheet assets involve no exchange between GHO and the lending currency and draw no Stability Module inventory at all. That is why they carry the initial phase.

Where GHO is the funding source, it is exchanged into the lending currency at drawdown and back into GHO at repayment. Every exchange of GHO is routed and executed jointly with TokenLogic. In practice, this means four things:

  1. The route is agreed before each conversion is executed, in this order of preference: Matched sGHO inflows first, since they are net neutral to the Stability Module by construction and, at matched duration, remain so for the life of the loan. Secondary market liquidity second. The Stability Module last, and only with TokenLogic’s agreement at the time.
  2. Size is tranched and timed rather than executed in a single clip where market depth requires it.
  3. A maximum acceptable deviation from peg is agreed in advance, and a conversion is deferred if it cannot be executed inside that limit.
  4. Every conversion is reported afterwards with size, route, realised price, and any Stability Module inventory used.

Repayment runs the same process in reverse and buys GHO. Across a full loan cycle the directional effect on the peg largely offsets, and the residual cost is execution slippage on each leg rather than a permanent addition to GHO supply.

Aave Institutional also anticipates growing sGHO deposits to support further scaling, and is in discussions with a number of parties to effect this.

4. Mechanism and risk

Every facility is over-collateralised, with collateral value exceeding the loan at origination and maintained above a defined threshold through the life of the loan. Each facility is bilaterally negotiated, so pricing, tenor and loan-to-value are set per counterparty with oversight from the GHO Stewards.

Borrower collateral is held at a leading qualified custodian, which is also appointed as collateral manager for the facility. The custodian monitors collateral value against the agreed thresholds, issues margin calls on a deterioration in collateral price, and liquidates where a call is not met. Each loan is governed by a Master Loan Agreement with legal security interests taken over the collateral, alongside an Account Control Agreement between the lending entity, the borrower and the custodian giving the lender the right to direct the sale of collateral and repay the Aave debt in full on default. Collateral title transfers on default and is not rehypothecated at any point. Monitoring and enforcement therefore sit with a regulated third party rather than with Aave Institutional or the borrower.

Risk Mitigation
Effect on the GHO peg from new supply Every funding authorisation requires GHO Steward approval.
Credit loss on an individual loan Loan-to-value of 60% to 75%, individual underwriting, a defined margin-call cadence, title transfer on default, no rehypothecation, and monitoring and liquidation performed by a qualified custodian rather than by the lender
Slower enforcement than onchain liquidation Loan-to-value is set more conservatively than in the V3 GHO market and tenors are capped at 12 months. Enforcement is executed by the custodian that already holds the collateral, so there is no transfer step between decision and sale
GHO price impact from conversion Every conversion routed, sized and timed jointly with TokenLogic against an agreed maximum deviation from peg, with matched sGHO inflows used first and the Stability Module last. Repayment conversions buy GHO and offset the drawdown leg
Reflexive exposure from AAVE pledged as collateral The conditions in which BTC-backed loans come under stress are the conditions in which AAVE falls, so the collateral weakens at the moment enforcement is tested and a forced sale into a falling market is self-reinforcing. AAVE is capped at 50% of the collateral pledged as part of the balance sheeting funded route, with the balance in WETH and WBTC and position health monitored by TokenLogic
Collateral concentration Scope limited to BTC and ETH. Any extension returns to governance
Custodian Qualified custodians only, with an Account Control Agreement in place with each

5. Governance controls

Control Held by
Bucket capacity, including reduction to zero GHO Stewards, within existing mandate
Facilitator removal Aave DAO
Extension of the collateral set beyond BTC and ETH Aave DAO
Limit on AAVE as balance-sheet collateral Aave Finance Committee
Mint execution GHO Stewards
Expansion of the GHO Borrow Cap GHO Stewards, within existing mandate
Funding authorisation GHO Stewards
Execution of GHO to lending currency conversion Aave Labs jointly with TokenLogic
Collateral management, margin calls and liquidation Qualified custodian appointed for the facility, under the Account Control Agreement
Facilitator contract upgradeability Immutable, with bucket capacity governed by the Aave DAO

6. Facilitator implementation

TokenLogic will deploy a dedicated facilitator contract, registered on the GHO token by governance under FACILITATOR_MANAGER_ROLE, with bucket capacity set under BUCKET_MANAGER_ROLE. Minting is constrained by the standard bucket check, so that bucketLevel plus the minted amount may not exceed bucketCapacity. GHO is minted under GHO Stewards authority, and its use for an eligible institutional loan requires authorisation from them.

Reporting and Offboarding

Aave Labs will publish a regular loan tape to this forum, covering outstanding balance, collateral composition, loan-to-value distribution, margin-call events, realised losses, and the current funding position.

On offboarding, Aave Labs wishes to be explicit about a constraint the DAO should understand before voting rather than discover afterwards. removeFacilitator() requires bucketLevel to be zero, so a lending facilitator creates a position that cannot be fully unwound by governance action alone while loans remain outstanding. Reducing bucket capacity prevents new minting immediately, but it does not retire GHO already issued against live loans.

Three commitments bound that exposure. Term loans written under this facility mature within 12 months maximum. Evergreen facilities carry a call right and a rate-change right in the lender’s favour, typically on 90-day notice, so no loan can outlive a wind-down decision by more than its call period. New origination ceases immediately on a DAO signal, without requiring a vote to compel it.

The worst case is a wind-down over 12 months from the signal, and on the current book it would be materially shorter.

Conclusion

GHO’s constraint is not mint capacity. It is demand that pays for itself, and this is it.

The ask is 25M GHO under a new facilitator, and DAO balance-sheet assets pledged to borrow up to $25M of the lending currency. Together they fund over-collateralised loans to institutional borrowers at 6.0% to 8.0% APR against a funding cost of approximately 4.5% on either route, with the full net interest margin accruing to the DAO.

The yield is external rather than recycled. The collateral is BTC and ETH at 60% to 75% loan-to-value, held at qualified custodians and never rehypothecated. Every loan carries an exit inside 12 months.

Every control that matters stays with the DAO and the GHO Stewards. No draw is funded without the GHO Stewards approving it at the time on live Stability Module information, the first loans draw no Stability Module inventory at all, and AAVE is capped at 50% of the collateral pledged from the balance sheet.

The downside is bounded and governed. The upside is a new, durable external revenue line for GHO and an institutional channel the protocol can scale into.

Next Steps

  1. Gather community ARFC feedback.
  2. If sentiment is favourable, escalate to the Snapshot stage.
  3. Following a positive Snapshot outcome, submit an AIP carrying both authorisations.
    a. Register the facilitator on the GHO token and set initial bucket capacity at 25M GHO.
    b. Authorise the pledging of DAO balance-sheet assets on Aave V3 to borrow up to $25M of USDC/USDT for loan funding, subject to the collateral limits in Specification 2.

Disclaimer

Aave Labs is presenting this proposal in its capacity as a contributor to the Aave ecosystem and is an interested party in its outcome. Aave Institutional is operated by Aave Labs, which would hold one of the three funding authorisation signatures, with the full net interest margin accruing to the Aave DAO. No third parties co-authored this proposal. Pricing, loan-to-value ranges, pipeline size and wind-down timelines are indicative and forward-looking, Stability Module figures are stated as of 24th September 2026, and nothing here constitutes financial, legal or investment advice or an offer to transact.

Copyright

Copyright and related rights waived via CC0.

8 Likes

The proposal paints a bullish vision, but before approving any funding, Labs needs to address an ongoing alignment gap: token holders cannot remain the last piece of the puzzle.

With equity out of the picture and $AAVE as the protocol’s true center of gravity, we need a clear, committed roadmap and timeline detailing how and when the ‘Aave Will Win’ framework directly aligns with and rewards token holders. Aave will not win if its holders’ economic alignment continues to be sidelined.

Thank you for your attention.
Apu Mallku

6 Likes
  1. Who are the parties to the Master Loan Agreements and the Account Control Agreements? Is this a loan from the DAO to Aave Labs, or directly to the borrowers?
  2. Is the funding cost of approximately 4.5% = sGHO rate?
  3. Who is the custodian?
1 Like

Thank you for the questions, @samj.

Parties to the agreements. The Master Loan Agreement is entered into between an Aave Labs entity and the borrower. The Account Control Agreement is a three-party agreement between the qualified custodian, the Aave Labs entity, and the borrower.

Funding cost. This depends on the funding route. On the DAO balance sheet route, the funding cost is the prevailing interest rate on Aave V3 for the USDC or USDT borrowed against the pledged assets. On the GHO funded route, the funding cost is the current sGHO rate.

Custodians. We are working with a number of custodians, all of them major qualified custodians.

1 Like

Thank you to Aave Labs for bringing forward the Aave Institutional ARFC. We support the proposal and appreciate the care taken to address the operational and peg considerations directly.

The facility would add a genuinely new category of backing to GHO. Its loans are secured by BTC and ETH held at qualified custodians, with conservative loan-to-value ranges, active margining, no rehypothecation, and custodian-handled collateral enforcement. That gives GHO access to institutional borrowers and externally generated yield with minimal smart contract risk and strong demand for the same collateral base. It is a channel with room to grow if the initial facilities perform as expected.

The first loans are funded by borrowing against DAO balance-sheet assets, without drawing on the Stability Module. This is necessary because the Stability Module’s redemption inventory is not sufficient to carry a loan of this size. A loan of this duration asks more of GHO’s backing than it can provide unless the exposure is sized and managed against the liquidity available to support it. This applies equally to every other component of GHO’s backing, each of which has its own duration and must be sized against the liquidity the rest of the book can provide. TokenLogic is developing a holistic approach to managing GHO’s backing, including duration risk, which we will set out in a forthcoming post.

The Aave Finance Committee, led by TokenLogic, will manage the balance-sheet position, while TokenLogic monitors its health, including the limit on AAVE within the pledged collateral. We will provide additional information on the initial collateral selection on a Funding Update closer to the origination of the loan.

The economics are attractive. At 25M of lending, the ARFC’s indicative borrower pricing of 6.0% to 8.0% would generate roughly $0.6M to $1.1M more gross annual revenue than the roughly 3.5% currently earned by supplying USDT on Aave Core. Custody, operating, execution, and credit costs still need to be paid from that amount. The funding cost differs by route. The GHO route carries the rate paid to sGHO depositors, whose savings sit behind the GHO minted and lent. The balance-sheet route carries the variable rate the DAO pays to borrow USDC or USDT on Aave V3, expected at 4.0% to 5.0%. Each route earns the loan rate less its funding cost. The balance-sheet route has no effect on the peg.

The peg deserves particular care as the facility moves toward implementation. GHO’s principal redemption anchor is the Stability Module inventory that holders can exchange for dollar stablecoins at near-par, subject to the redemption fee. That inventory supports the arbitrage that keeps the price stable. Since GHO can move across its deployment through CCIP, the relevant resource is the aggregate redemption inventory. New GHO entering circulation can increase the claim on that resource, while conversions through the Stability Module reduce what remains available to defend the peg.

GHO’s peg has been weak in recent weeks. The discount widened through the first half of September before recovering, showing how quickly strong borrow-side demand for GHO can pressure the peg. This demand is positive for GHO, but it requires careful liquidity management.

This facility adds a distinct form of pressure because its collateral sits outside the protocol’s incentive levers. Once GHO is minted and converted into the lending currency, any Stability Module execution draws on the same inventory that guarantees redemptions. The custodian monitors the collateral, issues margin calls and liquidates automatically through its regulated trading desk, which keeps each loan over-collateralized and the GHO behind it safe. Its rate is fixed by contract, so it has to be actively managed, and it can stay stale for up to the contractual notice period, typically 90 days. We agree with the ARFC that the effect on the peg is the material question for this proposal.

Matched sGHO inflows are the right first route for funding. They source the lending currency without consuming Stability Module inventory, so the proposed order of execution is strong. However, they alone are not a complete answer. The matched inflow must respect an equal or longer duration than the borrower draws in order to solve, and not only delay, the liquidity stress.

We strongly support agreeing on a maximum acceptable peg deviation in advance. With the peg only recently recovered and redemption inventory modest relative to the size of the facility, there is limited headroom. TokenLogic will approach each conversion from that starting point when agreeing the route, timing, and size with Aave Labs.

The ARFC also assigns the deployment of the facilitator contract to TokenLogic, and we are supportive of conducting that work.

TokenLogic has spent the past months building an allocation framework for GHO backing, which this allocation asset will be part of. Its purpose is to support safe allocation rules across the assets and channels that back GHO. Viewed against that work, we are comfortable with the ARFC’s initial parameters, including the initial bucket capacity of 25M GHO, and with the GHO Stewards managing expansion of the GHO Borrow Cap within their existing mandate. Any later increase should be assessed further against the peg and the backing available at the time.

We will publish a broader post in the following weeks that develops the next steps for GHO and its considerations. In the meantime, we support progressing this ARFC and look forward to working with Aave Labs and LlamaRisk on its implementation.

1 Like

The ARFC to approve Aave Institutional has been raised to snapshot. Voting will begin in less than 12 hours. You may vote here

2 Likes

And you keep pushing without giving any clarity for holders, shameful behavior.

What are @AaveLabs and @TokenLogic views on this model vs allocating DAO capital through an external credit manager, along the lines of Spark–Maple?