Maple lends billions to people who already have the money, and the decision is made by a person, not a formula

What does it do, in one line?

The difference from Aave sits entirely in who decides. There a curve decides: if the pool is nearly empty the rate rises, if it is full the rate falls, and anyone with the collateral can borrow without asking permission. Here there is a pool delegate instead, usually a firm with full identity verification, who looks at the accounts of whoever asks, decides whether to lend and at what price, watches the collateral day by day and liquidates when needed.

That delegate doesn’t work for free and doesn’t work uncovered: they keep a slice of the interest, but they also have to put their own capital into the pool they run. If their assessments are wrong they lose alongside you. The loans are mostly overcollateralized — to take a hundred you have to deposit a hundred and fifty — and that makes them as safe as the collateral, not as safe as the borrower.

Put bluntly: an overcollateralized loan is like asking a bank for money while carrying the money you need — it works for whoever has capital tied up and wants liquidity without selling it, not for whoever hasn’t got the money.

How much adoption does it really have?

The number to hold alongside it, though, is the peak: $3.25 billion on 30 October last year. From there the capital fell about 16 percent and never climbed back. So the real story isn’t “it is growing”, it is “it grew for a year, hit a ceiling in the autumn, and has settled a quarter lower since”.

In a credit market that shape isn’t necessarily bad. Capital coming in has to find somebody to be lent to: if the applicants are a few dozen verified institutions, past a certain point the extra money would sit there earning nothing, and whoever put it in leaves. The ceiling here is the size of the niche — not a verdict on the product.

Who keeps the interest?

Almost everybody skips this step and tells you the turnover as though it were the take. Maple’s annual turnover is $109.5 million, the protocol’s take is $13.56 million: whoever quotes the first number is describing somebody else’s money. Against the capital deposited the take is worth about half a percentage point a year — little, but it is the right percentage to look at if you want to know what the structure weighs.

The same numbers also give the price of the credit: on $1.89 billion actually lent, $109.5 million of interest works out at about 5.8 percent a year paid by the borrower and about 3.5 percent collected by the lender. That is a secured-credit yield, not a speculative one: anyone expecting double digits is looking at the wrong product.

How easy is it to get out?

That money sitting there is the door. As long as whoever asks to leave stays within that figure, they leave; beyond it, they queue and wait for a loan to come back. It isn’t a hidden defect and it isn’t a broken exchange: it is how a credit fund works, where the capital is committed to a maturity and doesn’t come back before it.

The thing to watch is the ratio between the two parts. High utilization means the capital is working and earning; very high utilization means the door narrows exactly when it is needed, because the requests to leave all arrive together in the bad moments — and that is precisely when the loans don’t come back early.

How many things does it depend on?

The second is the delegates. The billions in the pools aren’t allocated by a public algorithm but by a handful of firms deciding who to lend to. A delegate who consistently gets the assessments wrong doesn’t damage one position: they damage every pool they run, and the selection of borrowers is the same across all of them.

The third is the borrowers. Institutional crypto credit is a club of a few dozen names, and they are all exposed to the same market: when it goes badly, it goes badly for all of them at once. Diversification here isn’t a portfolio choice, it is a limit of the sector — there aren’t enough different counterparties to spread the risk across.

Does the governance count for anything?

On what actually determines the yields and the losses, though, the vote doesn’t reach. Who to lend to, at what rate, against how much collateral and when to liquidate is decided by the pool delegate, one case at a time, on criteria that don’t sit onchain. The governance governs the frame; somebody else does the credit.

SYRUP governance votes on protocol parameters, strategies for idle liquidity, opening new pools and allocating funds. What stays outside the vote are the credit decisions: which applicants to approve, at what rate, against what collateral and when to liquidate — those are made by the pool delegate.

On the token itself there is one thing to know: it comes from converting the old MPL at one to a hundred — a cosmetic operation, the value didn’t change — and it carries a declared inflation of five percent a year. Whoever holds SYRUP is diluted by that much every year, and the fees the staking collects have to at least match it not to lose.

Does the team hold up?

That failure changed the product, and it is why the loans today are mostly overcollateralized instead of granted on somebody’s word. A team that has watched its own model break and rewritten it is worth more, as a judgment, than one that has never been tested — but what happened has to be kept in the account, not filed away.

Since then the execution has been steady: simple-deposit products for people who don’t want to pick a pool, an architecture of separate contracts using the common standard for deposits, the token conversion carried through, idle liquidity put to work on outside protocols. Nothing spectacular and nothing blown up.

In December 2022 a borrower defaulted on about thirty-six million dollars on a Maple pool, after the collapse of FTX: the loans were not sufficiently collateralized and the delegate who had approved them was removed. The model then moved to excess collateral, followed by simple-deposit products, the token conversion and putting idle liquidity to work on outside protocols.

Where does it break?

The weak link isn’t the first, it is the third. A borrower defaulting is normal and accounted for; collateral liquidating badly happens only on the days when everything falls together — which are, as it happens, the same days borrowers default. The two events aren’t independent, and a model that treats them as though they were underestimates the risk.

Then there are two risks that aren’t market risks. The first is regulatory: lending money to identity-verified institutions on negotiated contracts looks a great deal like banking, and if a regulator decided to treat it as such the model would have to change. The second is efficiency: asking a hundred and fifty to lend a hundred keeps out anyone with access to ordinary credit, and puts a ceiling on demand that no amount of execution can raise.

Comparison between maple and the direct competitors in the same category, on the measured criteria and all taken on 2026-08-20 from the same source (defillama). Capital: maple $2.46bn, centrifuge $1.63bn, clearpool $20m, truefi $0m. Revenue over 30 days: maple $1m, centrifuge $0.48m, clearpool $0m, truefi —. Interest over 30 days: maple $8.55m, centrifuge $5.95m, clearpool $0.04m, truefi —. On the first chain: maple 100%, centrifuge 77.8%, clearpool 99.3%, truefi 99.7%. 2 cells are empty: the source doesn’t publish that figure for that name.