Disclosure: This newsletter is for informational purposes only and does not constitute financial advice. Always DYOR before making any investment.

Each week in Edge, we share data-driven insights, highlight risk ratings, and showcase new product updates.

Let’s dive in 👇

The Context

Recursive lending has become one of the dominant strategies in DeFi over the past two years and most participants know the mechanic but not so many think carefully about what it does to the rates they are earning (and what happens when it unwinds). A recent analysis posted on TechFlow about DeFi's $8 billion on-chain yields in 2025 by researcher Vadym says that roughly half of all borrowing demand across major protocols is recursive.

With that said, knowing how DeFi can sometimes hide a lot risks, especially on complex strategies, we are here to breakdown how recursive leverage works.

How It Works

A user deposits collateral into a lending protocol, borrows against it, and deploys the borrowed funds back into a yield source, often the same protocol or a correlated one. That position becomes new collateral, enabling another borrow, and so on until the position approaches the liquidation threshold.

A concrete example: deposit 1,000 USDC into Aave at 75% LTV, borrow 800 USDC, redeploy it as supply, borrow 640 USDC, repeat. After a few iterations, a user builds roughly 3,500 USDC in notional exposure from their original 1,000. If supply APY is 4% and the borrow rate is 3%, the net spread looks like yield. Utilization looks elevated. TVL looks healthy. The problem is that this loop creates supply and demand simultaneously -- with the same capital doing both jobs.

Vadym's data makes the scale concrete: on Aave's Ethereum deployment, ~39% of borrowing demand serves leveraged ETH staking strategies, and another 11.6% recycles into Ethena's sUSDe. That is roughly half of Aave's borrow-side activity driven by internal recycling, not external demand.

What It Does to Rates

Recursive borrowers compress the spread between supply and borrow rates structurally. Because the looper is both lender and borrower, they have a built-in incentive to keep borrow rates low. This is why stablecoin supply yields have converged to 2-3% despite billions in TVL. High utilization stops being a reliable signal of real demand when much of it is synthetic, and synthetic demand unwinds fast under stress, pulling rates down with it.

Protocols whose yield originates outside the loop behave differently. Sky's USDS pool, for example, draws ~70% of its revenue from off-chain sources (Coinbase PSM rewards, RWA exposure via BlackRock's BUIDL and Janus Henderson), which is why its 3.75% savings rate has held steady while on-chain rates compressed. That structure carries its own risks, off-chain revenue concentration adds counterparty and regulatory exposure, but it does not unwind the same way a loop does.

The Risks

Now we know how it works, and lets face it, recursive lending will stay in DeFi for a long time and you can make good returns using this strategy. However, it is pretty important to know the risks, especially

• Unwind risk. Looped positions are correlated. A rate spike, collateral reprice, or parameter change causes them to unwind simultaneously collapsing utilization and supply yields overnight.

• Health factor sensitivity. 3x leverage on a 75% LTV market leaves little buffer. Correlated collateral moves can trigger cascading liquidations across positions running the same strategy.

• Yield sustainability misjudgment. A market paying 6% USDC supply APY in a 2% rate environment is almost certainly sustained by looping, emissions, or both. Neither lasts. Before deploying: ask who is borrowing and why. If that is not legible from the market structure, the yield is not either.

Recursive lending is rational it creates real liquidity depth and tightens spreads in normal conditions. But it means that a significant portion of the yield you see on any given dashboard is the market lending to itself. Knowing when you are on the right side of that loop, and when you are not, is one of the more useful edges in DeFi credit allocation.

Winners for the YO Hackathon are out!

PistachioFi x YO and Exponential Partnerships 🔥