AssessmentsBorrowingLiquidation
Coinbase's crypto-backed borrowing, assessed as a custodial door onto an on-chain loan
The app is a familiar exchange account. The loan underneath it is a smart-contract position that a price feed can close without asking. Assessed on the gap between those two facts.
By Yusuf Karim
A crypto-backed loan does not end in a phone call. It ends when a number crosses a threshold and a contract sells the collateral, and the pledged assets can be gone in full before anyone has read an alert. That is the risk this product is built around, and where the assessment starts.
The commercial framing is simple: pledge assets, receive USDC, repay when you like. Underneath it sits a custodial account at the front and an over-collateralised on-chain position doing the actual lending.
This assessment grades disclosure and mechanics, not desirability. Figures appearing only in the company’s own documentation are not reproduced — this desk reports what a named source states.
What the published reporting sets out
The structure, and why it matters
Two counterparties, not one. The account, the interface and the USDC balance are the exchange’s. The loan is a position in a Morpho market on Base, with collateral held by contract code rather than by the firm the borrower signed up with.
That split is better than the alternative this sector spent a cycle failing at. In the deposit model, the user was frequently an unsecured creditor of an operator whose lending book they could not see. Here the credit leg is over-collateralised and on-chain — the structure that came through intact, and inspectable rather than asserted.
It also means exposure to both layers. A failure at the front end and a contract failure at the back are different events, and either can end badly.
Liquidation, read properly
The reporting describes the trigger as total loan value — including accrued interest — reaching a defined threshold relative to collateral. Two phrases there carry most of the risk.
Including accrued interest means the measured quantity grows on its own. A position can walk toward its threshold with the collateral price flat, purely because interest accumulated against it. Borrowers modelling this as a price question are modelling the wrong variable.
Defined threshold means the boundary is a published parameter of the underlying market, not a judgement call. That is a merit — nobody negotiates it, and nobody’s solvency decides whether it fires. It cuts both ways: no discretion is available to the borrower either.
Coinbase said borrowers are alerted ahead of liquidation by email and text, which is more than a bare protocol interaction offers. The reporting does not state how much notice that is, nor what happens if a message is delayed. An alert is a courtesy, not a brake.
The UK regulatory position, stated plainly
This product is not FCA-regulated, and nobody should read its UK availability as implying otherwise.
The FCA’s cryptoassets regime policy statements, published 30 June 2026, place lending and borrowing inside PS26/11 on regulated cryptoasset activities, with retail protections the regulator says it is maintaining — enhanced disclosures, consent, appropriateness testing, and safeguards including over-collateralisation and negative balance protection. The FCA also states that the full scope of regulated activities expands from 25 October 2027, with the savings-provisions application window scheduled to open on 30 September 2026 and close on 28 February 2027.
Nothing on offer today carries that authorisation, because the perimeter has not yet arrived. A borrower using this product now sits outside it.
What the published material does not settle
The price input. Liquidation depends entirely on a feed, and the reporting relied on here names neither the oracle, its update cadence, nor its failure behaviour. That is the most consequential unspecified item in the product.
Alongside it: the opening loan-to-value ratio, the liquidation penalty, and the notice period behind those alerts. None appear in the material cited here; a borrower should establish all three from the firm’s own documentation before pledging anything.
Pros and cons
Verdict
The score reflects a real structural advantage — an over-collateralised, inspectable, on-chain loan reached through an interface most users can already operate — marked down for what the published material omits at exactly the point where a borrower loses money. It grades how well the product explains itself, not whether anyone should use it. Collateral pledged here can be liquidated in full, and none of this is financial advice.