On-chain private credit grew to $2.29 billion by March 31, 2026, up from roughly $0.40 billion at the start of 2025 — and almost all of that expansion landed on a single protocol. Maple Finance's active loan book climbed from $0.21 billion to $2.13 billion over the same period, giving it about 93.1% of the market, according to data CoinGecko compiled from DeFiLlama's on-chain metrics.
Where the growth came from
The surge has been driven by loans to crypto-native market makers and trading firms. Those borrowers' balance sheets are tightly coupled to crypto market volatility, which helps explain why the growth concentrated in Maple rather than spreading across the sector. It also means the book is only as stable as the trading desks underneath it.
How Maple ran the book
Maple's High Yield Secured product delivered a 16.83% net APY during 2024. Secured pools stayed overcollateralized through the year, and the protocol executed a partial liquidation in August 2024. Maple issued 61 margin calls in 2024, cured on average in roughly three hours.
TVL reached as high as $600 million during Q4 2024, with loans outstanding growing 23% quarter over quarter. Syrup USDC, a product listed on DeFiLlama's RWA dashboards, shows an active market cap above $1.3 billion.
The concentration problem
The growth is concentrated in Maple, and within Maple, in crypto-native credit exposures. That introduces correlation risk: multiple borrowers could face simultaneous pressure during a market drawdown, and the protocol's margin-call system would be tested all at once rather than one loan at a time.
The oracle risk
Academic work on Protocols for Loanable Funds flags oracle manipulation, often via flash loans, as a distinct risk channel for lending pools. It's a known vulnerability class for the model Maple runs, even if no such attack has materialized in the data here.
The open question is whether Maple can diversify its borrower base beyond crypto-native firms before the next drawdown tests the concentration. The numbers so far suggest it hasn't needed to.
