Every published on-chain borrow rate carries an invisible asterisk. It is not a quote, it is not an offer, and it is not what you will pay. It is a reading, taken at a moment, of a formula that recalculates continuously.
Borrowers coming from custodial lending routinely miss this. They compare a 3.62% on-chain rate against a 9.25% custodial rate and conclude the first is cheaper. Sometimes it is. But one of those numbers is a contract term and the other is a snapshot of a variable that can multiply in a single block.
The mechanism: utilisation drives everything
A lending pool has capital supplied by lenders and capital borrowed by borrowers. Utilisation is the ratio between them, and it is the input to the rate.
Each market has an interest rate curve written into its smart contract. Feed utilisation in, get a borrow rate out. Nobody quotes it. Nobody negotiates it. When a large borrower draws from the pool, utilisation rises and every existing borrower's rate rises with it, immediately, without anyone being asked.
The curve is almost always two-sloped:
Below optimal utilisation (often 80 to 90 percent) the rate climbs gently. This is the normal operating band, and it is where most published rates are taken from.
Above it the slope steepens hard. The protocol needs to force liquidity back into the pool so suppliers can withdraw, and the only lever it has is price. Rates can go from single digits to something unrecognisable across a few percentage points of further utilisation.
That kink is why on-chain rates spike rather than drift. There is no gradual repricing, there is a threshold and a cliff behind it.
What this looks like across real markets
The deepest liquidity and the longest-audited contracts, at the highest transaction cost. Topping up collateral during a fast drawdown costs the most here.
An Ethereum layer 2. Near-identical contracts to mainnet at a fraction of the gas, which matters when you need to add collateral quickly.
A Bitcoin layer. Uses sBTC, which is redeemable for actual Bitcoin rather than issued by a company, so the wrapper risk is different in kind.
A separate chain with its own bridged Bitcoin representation, BTC.b.
An Ethereum layer 2 with mature lending markets and lower fees than mainnet.
High throughput and low fees, with a separate contract ecosystem and its own history of network outages to weigh.
High throughput and low fees, with a more concentrated validator set than Ethereum.
Lender facts on this page render live from our comparison database, last verified August 4, 2026. Figures refresh weekly; for the current set and your own loan size, see the comparison tool.
The spread across these markets is not mainly a judgement about Bitcoin's risk. It is mostly a snapshot of how busy each pool happens to be. A market at the low end may simply have plentiful supply and modest borrowing demand today, and a market at the high end may be running near its kink.
Two markets quoting similar rates can behave completely differently the moment utilisation moves, because their curves are not the same. The optimal point, the base rate and the steepness above the kink all vary by protocol and often by individual market.
The three things that actually move your rate
Someone else borrows. The most common cause. A single large position entering a shallow pool can move utilisation several points, and if that crosses the kink the rate multiplies for everybody in the pool including you.
Suppliers withdraw. Utilisation is a ratio. It rises when the denominator shrinks, so lenders pulling capital out raises borrow rates exactly as new borrowing does. This tends to happen in stressed markets, which is when you least want it.
Broader stablecoin demand shifts. These are USDC markets, and USDC has alternative uses. When yields rise elsewhere, supply leaves lending pools and rates rise across the board without anything happening in Bitcoin.
Notice what is absent from that list: your creditworthiness, the size of your loan relative to your collateral, and how long you have held the position. None of them affect your rate. The pool does not know who you are.
Why this compounds into a liquidation risk
A variable rate is not only a cost problem. Interest accrues into your debt, and debt is the numerator of your loan-to-value. A rate spike accelerates the drift of your position toward its liquidation threshold.
Combine that with the fact that on-chain liquidation has no margin call and no cure period, covered in what happens in a DeFi liquidation, and a rate you did not agree to can move you toward a liquidation you cannot contest. Borrowing near max LTV in a pool close to its kink is the combination to avoid.
What to check before borrowing on-chain
Current utilisation, not just the rate. The rate tells you today's cost. Utilisation tells you how much room exists before the curve turns steep. Both are on each market's page under DeFi Bitcoin loans.
Pool depth. Total supplied capital determines how far a given borrow moves the needle. A $500,000 borrow is noise in a large pool and an event in a small one.
Where the kink sits for that specific market. Protocols publish their curve parameters. A pool at 70% utilisation with an 80% optimal point has far less headroom than the raw number suggests.
Whether you need certainty at all. If the borrowed money funds something with a fixed obligation, a tax bill or a purchase, a rate that can triple is a genuine mismatch. That is an argument for a custodial lender rather than for a different protocol.
The honest comparison against custodial
Custodial lenders in our set charge a rate fixed at origination, and effective APRs run from about 7.25% upward:
| Lender | Effective APR (incl. origination) | Max LTV | Lender page |
|---|---|---|---|
| APX Lending | 9.99% to 11.49% | 60.00% | Learn More |
| Arch (Deferred) | 9.49% to 10.99% | 60.00% | Learn More |
| Arch (Standard) | 8.99% to 10.49% | 60.00% | Learn More |
| CoinRabbit | 11.95% to 16.80% | 90.00% | Learn More |
| Figure | 9.76% to 12.35% | 75.00% | Learn More |
| Ledn | 9.99% to 11.49% | 50.00% | Learn More |
| Nexo | 15.90% | 50.00% | Learn More |
| SALT | 7.49% to 10.50% | 70.00% | Learn More |
| Strike | 7.75% to 10.47% | 50.00% | Learn More |
| Unchained | 14.18% | 50.00% | Learn More |
Across our index history the gap between the two rails has averaged about six percentage points, measured daily and set out in the custody premium. Part of what that premium buys is precisely the thing this article is about: a number that does not change while you are not looking.
Whether that is worth six points depends on the size and duration of your loan, and on whether a rate spike would force you into a decision you cannot afford to make. The two models are compared in full in CeFi vs DeFi Bitcoin loans.
Where we track it
We snapshot every Bitcoin-collateral market daily and publish the TVL-weighted average as the DeFi half of the BoB Rate Index, with per-protocol history on the protocol index and the whole series downloadable from the data page. The Rate Index page also charts total tracked TVL and utilisation over time, which is the aggregate version of the mechanism described here.
This is not financial advice
borrow/on/bitcoin is a comparison publisher, not a lender, broker, or financial advisor. On-chain borrow rates are variable, uncapped, reset continuously, and the figures shown here are captured at our daily snapshot rather than being rates you can lock in. Nothing here endorses any protocol. Borrowing against Bitcoin can result in the total loss of your collateral, and a rising variable rate can move a position toward liquidation without any fall in the Bitcoin price.