Most institutional DeFi yield strategies either sit inside a lending market or depend on one. Stablecoin vault depositors are the lenders, borrowers pay interest that flows back to the vault, and the protocol enforces the rules through smart contracts and liquidation mechanics. Understanding how DeFi lending markets work is therefore understanding the mechanism behind a large share of onchain yield.

By April 2026, DeFi lending protocols held over $54 billion in deposits, with aggregate TVL in the lending category sitting between $75 and $80 billion up from roughly $50 billion at the start of 2025. Aave, Morpho, Spark, Compound, and Fluid dominate that field, but they do not all work the same way, and the architectural differences between them translate directly into different risk, yield, and liquidity profiles for the vaults that sit on top.

This guide explains the core mechanics of DeFi lending, the interest-rate model that drives yield, the liquidation engine that keeps it solvent, the architectural designs and their tradeoffs, and what a vault allocator needs to understand about the lending market underneath their position. Our complete guide to DeFi vaults covers the vault layer, and the ERC-4626 standard companion covers how vaults integrate with these lending markets.

How Does Overcollateralized Lending Work in DeFi?

The fundamental design is simple. A borrower deposits collateral, the protocol evaluates its current value using an oracle, and the borrower can draw a loan worth less than that collateral. A borrower who posts $1,500 of ETH might be able to borrow up to $1,000 in USDC, depending on the loan-to-value limit the protocol sets for that collateral type.

The lender on the other side, which for a vault strategy is the vault itself or its depositors, supplies the asset being borrowed and earns the interest the borrower pays. That interest is calculated continuously as a function of how much of the pool is currently borrowed, the utilization rate. High utilization means high borrowing demand relative to available supply, which drives rates up. Low utilization means idle capital, which pushes rates down. Rates on USDC across major protocols in early 2026 ranged from 3% to 8% depending on chain, protocol, and curator strategy, with the spread reflecting differences in utilization, risk selection, and how efficiently the protocol routes capital to its highest-yielding opportunities.

The overcollateralization requirement is what makes the system function without credit checks or legal enforcement. If the borrower simply disappears, the collateral can be sold to repay the loan. The protocol does not need to trust the borrower; it trusts the value of the collateral and its ability to liquidate it before that value falls below the loan amount.

How Does the Interest Rate Model Work?

The interest rate a borrower pays and a lender earns is not fixed. It is an algorithmic function of utilization, and understanding that function helps explain why DeFi yields can spike dramatically during stress and collapse during quiet markets.

The basic model has a target utilization point, sometimes called a kink, below which rates rise slowly as utilization increases, and above which they rise steeply. At 80% utilization, the rate might be 6%. At 95% utilization, it might be 50%. This design means that as the pool becomes nearly fully lent out, borrowers pay sharply more, which is designed to attract new supply and deter additional borrowing. Early DeFi interest-rate models were deliberately simple, but they created predictable problems during stress: rates could spike to hundreds of percent within hours when demand surged, and collapse to near zero in quiet markets regardless of external conditions. Protocols have progressively refined these models to include steeper rate kinks at high utilization, base rates that track macro interest rate environments, and more granular parameters per asset and per chain.

For a vault that earns by lending into these markets, the interest-rate model is the mechanism that translates market conditions into yield. A vault that earned 7% in a high-demand environment might earn 3% when the same asset is less sought as borrowers de-lever. That variability is why the beyond-APY evaluation of a lending-strategy vault requires understanding when and why its yield is high, not just that it has been.

How Does Liquidation Work, and Why Does It Matter?

Liquidation is the enforcement mechanism that keeps overcollateralized lending solvent, and its design determines how well the system handles stress.

When a borrower's collateral value falls and the loan-to-value ratio breaches a liquidation threshold, the protocol activates liquidation. A liquidator, which is typically a bot, repays part of the borrower's debt and receives the borrower's collateral at a discount, usually 5-15% below market price. The liquidator profits from that discount, and the protocol restores the position's health.

The design works well in slow-moving, liquid markets. It can fail in fast crashes. If collateral prices drop faster than liquidation bots can act, positions become undercollateralized before they can be closed, leaving bad debt that the protocol and its depositors absorb. Aave's Safety Module, which stakes AAVE tokens as a backstop that can be slashed to cover bad debt, is one institutional response to this. During fast collateral price drops, liquidation cascades can pile bad debt onto the protocol even when individual mechanisms fire correctly, because multiple positions becoming undercollateralized simultaneously creates more selling pressure than the market can absorb at normal discounts.

For a vault allocator, the key question is not whether a protocol has ever had bad debt but how its liquidation design handles tail scenarios. The oracle that prices collateral, the liquidation threshold relative to market depth, and the backstop available when cascades occur are the variables that separate protocols that survived 2022's crashes from those that did not.

What Are the Main Architectural Models?

The DeFi lending field has split into four designs, and each produces a different risk and yield profile.

A monolithic pool, the original Aave and Compound design, lets a governance-approved set of assets share a single liquidity pool. Depositors in the USDC pool collectively underwrite all the collateral types that borrowers can post against USDC loans. This delivers deep liquidity and a simple experience but means that adding a risky collateral type exposes every depositor in the pool. Aave holds around $20-26 billion in TVL across 22 chains and remains the benchmark for this design.

An isolated market pairs one loan asset with one collateral type in a separate contract. Compound V3 and Morpho Blue use this approach. The advantage is failure containment: a problem in one market does not drain another. The cost is liquidity fragmentation, since capital for each pair is separate.

A modular vault layer sits above isolated markets. Morpho's architecture is the leading example: Morpho Blue provides the immutable primitive with isolated markets, and curator vaults above it allocate depositor funds across multiple markets and rebalance as yields shift. Curators cannot withdraw user funds, they can only reallocate, and curator quality is now the primary risk management layer. Institutions including SG FORGE and Bitwise have deployed into Morpho vaults through this model, and Apollo's commitment to Morpho governance tokens signaled traditional finance's recognition of curated vaults as legitimate infrastructure.

A hybrid lending-DEX, like Fluid, combines lending and liquidity provision so the same liquidity serves both purposes simultaneously, improving capital efficiency at the cost of a more complex risk surface.

A fifth development worth noting is the emergence of RWA collateral. Protocols like Morpho and Euler have built isolated market structures specifically to accommodate tokenized Treasuries and other real-world assets as collateral, creating a direct transmission mechanism between Federal Reserve rate policy and DeFi borrowing costs. A borrower can post a tokenized T-bill vault token as collateral and borrow stablecoins against it, effectively creating exposure to the yield spread between T-bill rates and DeFi borrow costs. This intersection of RWA vaults and DeFi lending is where institutional credit markets and onchain infrastructure are converging most rapidly.

The emergence of undercollateralized lending for specific institutional borrowers is a separate evolution. Protocols like Maple Finance rebuilt their underwriting model after early defaults to require first-loss capital from pool managers, creating a structure closer to a traditional credit fund with blockchain settlement. These products carry real credit risk, not just collateral risk, which is a fundamentally different exposure than overcollateralized lending and should sit in a different part of an institutional allocation than the lending-market vaults this guide primarily covers.

What Should a Vault Allocator Understand About the Lending Layer?

For a vault allocator, the lending architecture underneath the vault is not a technical detail; it is part of the investment thesis.

Which architecture does the vault use, and what does that mean for risk isolation? A monolithic-pool vault exposes depositors to all the collateral types the pool accepts. A curated-vault approach gives depositors exposure to the curator's specific market selections. The distinction matters most in stress, when a bad collateral type in a shared pool can cause losses to USDC depositors who never intended to underwrite it.

What is the protocol's oracle setup, and how has it behaved in stress? Oracle manipulation remains the most acute tail risk in DeFi lending, and a protocol's oracle design is its first line of defense. Chainlink feeds, TWAP constructions, and circuit breakers all reduce the risk, but the exposure is not zero for any system.

How does the protocol's interest-rate model behave when utilization spikes? A vault locked in a high-utilization market in stress faces both rising rates, which is good for lenders, and the risk that a cascade of forced borrower de-levering creates a rush for exits that tests the redemption queue. The vault liquidity companion covers this exit dynamic in full.

Conclusion

DeFi lending markets are the engine underneath most institutional onchain yield strategies, and their mechanics, the utilization-rate model, the liquidation system, the architectural design, and the oracle setup, define both the yield and the risk a vault allocator actually takes on. Treating the APY as the output without understanding the input is the most common analytical gap in institutional DeFi diligence.

The sector has matured considerably: monolithic pools coexist with isolated markets and curator vaults, interest-rate models have grown more sophisticated, and $54 billion in deposits has survived multiple stress cycles with functional liquidation systems. That track record is evidence of resilience, not a guarantee against future risk, and an allocator who understands the mechanics is better placed to judge which lending architecture fits their mandate. Platforms like Lucidly Finance curate vault access with the lending architecture as a primary evaluation dimension, because the yield is only as durable as the mechanism producing it.