CryptoQuant Reports Diverging Lending Trends: DeFi Borrowing Plummets 69% While CeFi Activity Rebounds 155%

yesterday / 14:16 3 sources neutral

New research from on-chain analytics firm CryptoQuant reveals a significant divergence in crypto lending behavior during the recent market correction. While decentralized finance (DeFi) borrowing has contracted sharply, centralized finance (CeFi) platforms are showing early signs of a recovery in credit demand.

DeFi borrowing has experienced a dramatic decline, with weekly volumes on major protocols falling in line with decreasing crypto prices since August. On Aave, one of the largest DeFi lending platforms, weekly borrowing of stablecoins USDT and USDC plummeted by 69%, dropping from a peak of $6.2 billion to just $1.9 billion by the end of November. This contraction mirrors the broader market downturn and indicates users are actively unwinding leverage rather than deploying fresh capital. Despite this sharp pullback, Aave still maintains $16.3 billion in outstanding loans, demonstrating the scale of DeFi credit markets even during periods of stress.

In contrast, CeFi activity is beginning to rebound. While centralized platforms initially followed a similar downward trajectory, recent data suggests a divergence. On Nexo, weekly retail credit withdrawals dropped sharply from $34 million in mid-July to $8.8 million by mid-November. However, the following week saw a strong rebound to $23 million—a 155% week-on-week increase. This behavior indicates users are increasingly opting to borrow against their crypto holdings rather than selling assets at depressed prices.

CryptoQuant's analysis highlights the structural importance of centralized lenders during market stress. While DeFi borrowing contracts rapidly as leverage is reduced, CeFi platforms absorb liquidity demand when investors seek flexibility and capital preservation. Nexo's cumulative credit withdrawals reached $817 million in 2025, positioning it as one of the most active venues for crypto-backed lending. The data suggests both models play complementary roles within the ecosystem, with clearly differentiated risk profiles and user behavior.