DeFi lending protocols hold $54B in deposits as of April 2026, according to DefiLlama's lending category, and the design space has fragmented into four distinct architectures: monolithic pool, isolated market, modular vault, and lending-plus-DEX hybrid. Choosing among them comes down to three measurable variables — total value locked (TVL), supply and borrow rates, and the risk surface a user is willing to underwrite. This guide compares Aave, Morpho, Spark, Compound, Fluid, Euler, and adjacent venues by those three axes, names the chains where each is live, and shows where stablecoin liquidity sits.
​

Rather than rank in the abstract, the comparison treats each protocol as a system: who can list a market, who sets risk parameters, what liquidation engine fires, and how interest is set. Rates and TVL move daily; the architectures do not. Readers leave with a model for evaluating any DeFi lending protocol — including ones that do not yet exist.
​

Lending is the second-largest DeFi category by TVL after liquid staking. As of mid-April 2026, DefiLlama tracks 380+ active lending protocols across 80+ chains, with the top ten capturing 78% of deposits. Aave V3 leads at $19.4B, followed by Spark ($6.8B), Morpho Blue ($4.9B), Compound V3 ($2.7B), and JustLend on Tron ($2.4B). Newer entrants — Fluid ($1.6B), Euler V2 ($890M), Kamino ($1.1B on Solana) — have grown 3–5x year over year per Token Terminal market data.
​

The composition of borrows has shifted decisively toward stablecoins. Dune dashboards on stablecoin borrowing show 84% of outstanding DeFi debt is denominated in USDC, USDT, USDS, DAI, FDUSD, or similar. The collateral side is split between ETH (39%), liquid staking tokens (28%), BTC wrappers (14%), and the remainder in major altcoins and stablecoin LP tokens. This means the lending market in practice is a stablecoin borrow market backed by crypto collateral — which is also why borrow rates anchor closely to the funding rate paid for stablecoin liquidity onchain.
​

Every onchain lending protocol implements the same three primitives: deposit, borrow, liquidate. What separates designs is how those primitives compose.
​

Monolithic pool. All deposits sit in a single contract. Borrowers post collateral and draw any supported asset against it. Risk parameters — loan-to-value (LTV) caps, liquidation thresholds, and reserve factors — are set per asset by governance. Aave V3 and Compound V2 are the canonical examples. The advantage is deep liquidity and a single risk surface; the trade-off is that adding an asset exposes every depositor to that asset's failure mode.
​

Isolated market. Each lending market is a single collateral / single loan pair, deployed as its own contract. Liquidity does not commingle across markets. Compound V3 (Comet) ships one base-asset pool per market — USDC, ETH, USDT — and isolates collateral risk to that pool. The advantage is failure containment; the cost is liquidity fragmentation.
​

Modular vault. Lending becomes a permissionless primitive: anyone can deploy a vault with an oracle, an interest-rate model, an LTV, and a liquidation incentive. Curators allocate depositor funds across vaults. Morpho Blue and Euler V2 represent this design. The advantage is that risk choices are unbundled from liquidity provision; the cost is curator dependency.
​

Lending plus DEX. The deposit serves both as borrow collateral and as DEX liquidity for the same asset. Fluid Protocol pioneered this hybrid: depositors earn lending interest plus swap fees from the same capital. The advantage is yield stacking; the cost is correlated risk between the two markets.
​

Interest rates in all four designs are set by a utilization curve. As borrow demand pushes utilization toward 100%, rates kink upward sharply to reserve liquidity for withdrawals. The exact curve — slope1, slope2, optimal utilization — is a per-asset parameter, and curve calibration is one of the more consequential governance decisions a lending protocol makes.
​

The following matrix summarizes the seven largest DeFi lending venues by TVL as of April 2026. TVL figures are pulled from DefiLlama; rate ranges are observed across the prior 30 days from each protocol's public dashboard.

A few patterns stand out. Aave V3's TVL lead does not translate into the highest yields — the protocol's deep liquidity floor caps utilization, which caps APY. Morpho and Fluid carry higher rates because their architectures concentrate borrow demand into specific markets. Spark's USDC rates track DAI Savings Rate (DSR) closely because both are governed by the Sky ecosystem.
​

For deeper architectural breakdowns of each, see the supporting cluster: Aave V3 vs V4, Morpho Blue explained, Spark vs Aave, Compound V3 architecture, Fluid Protocol, and Euler V2.
​

Onchain lending exposes depositors to four kinds of risk, in order of historical loss frequency.
​

Smart-contract risk. A bug in lending logic can drain deposits. The Euler exploit in March 2023 cost $197M; Cream Finance lost $130M in October 2021; and Mango Markets lost $114M to a price-oracle manipulation that same month. DefiLlama's hack tracker records $7.6B in cumulative DeFi losses, with lending protocols accounting for roughly 28% of that total. Audit count and time-since-deployment are the cheapest proxies; production track record matters more than either.
​

Oracle risk. Lending protocols read collateral prices from price feeds. A manipulated or stale feed can trigger bad-debt liquidations or under-collateralized borrows. Aave and Compound use Chainlink with multiple aggregators; newer protocols sometimes use Pyth or Redstone. Read the oracle architecture in each protocol's docs before depositing.
​

Liquidation risk. Borrowers who let LTV drift past the liquidation threshold lose collateral at a discount to liquidators. The discount is typically 5–15%. During fast collateral price drops (e.g., March 2020, May 2022, August 2024), liquidations cascade and pile bad debt onto the protocol. Aave's "Safety Module" stakes AAVE to cover bad debt; Compound and Morpho rely on overcollateralization buffers.
​

Governance risk. Risk parameters are upgradeable via governance. A captured DAO can raise LTVs or list a malicious asset. The November 2022 Mango Markets governance attack proved this is more than theoretical. Spark mitigates by inheriting Aave V3 contracts; Morpho Blue mitigates by making each market immutable post-deployment.
​

The general framework: count audits, read the oracle docs, check the liquidation parameters, and skim the governance forum for in-flight proposals before depositing more than a test amount.
​

The most common DeFi lending use case is borrowing a stablecoin against crypto collateral — the user wants liquidity without selling. Three patterns dominate.
​

Levered ETH staking. Deposit stETH or wstETH; borrow USDC or USDS against it; swap that USDC for more stETH; repeat. Looped 3–5x, this stacks Lido's ETH staking yield against the borrow rate. Profitable while ETH staking yield exceeds USDC borrow rate plus liquidation buffer.
​

BTC-backed stable borrow. Deposit cbBTC, WBTC, or tBTC; borrow USDC. Used to stay long BTC while accessing dollar liquidity for spending or other deployments. Spark and Aave V3 both list BTC variants with 70–75% LTV.
​

Stablecoin yield arbitrage. Deposit USDC where supply APY is highest (often Morpho Blue or Fluid); borrow stablecoin where rate is lowest (often Spark when DSR is below 5%); pocket the spread. Requires monitoring; spreads compress fast as more capital arrives.
​

For deeper breakdowns of each, see Stablecoin Borrowing Strategies.
​

Lending TVL is fragmented across chains. Aave V3 lives on 15+ EVM chains; Spark concentrates on Ethereum, Gnosis, and Base; Morpho is largely Ethereum and Base; Kamino is Solana-only. A user holding USDC on Polygon who wants to lend on Morpho Blue needs that USDC on Ethereum — and the round-trip cost (bridge, slippage, gas) often eats the rate advantage.
​

Eco is the stablecoin execution network that handles this round-trip as a single intent. A team or user submits a stablecoin movement intent — "USDC on Polygon to USDC on Ethereum" — and Eco's solver network routes the liquidity, settles in seconds, and delivers funds to the destination address. The same primitive works for treasury teams rebalancing stablecoin positions across lending venues, for vault curators sourcing depositor funds across chains, and for retail users chasing the best supply APY without manually bridging. Routes (CLI + API) is the developer surface; the network handles solver selection and finality across 15 supported chains.
​

The fastest way to read a lending protocol's risk posture is to look at its asset listing process. Every protocol falls into one of three buckets.
​

Governance-approved listings. Aave V3, Compound V3, Spark, and JustLend require an on-chain vote (AAVE, COMP, MKR/SKY, JST) to list a new asset. The risk-parameter discussion happens in public forum threads — Aave's governance forum typically runs 100+ comments per material listing. Listing turnaround is two to six weeks. The benefit: every market depositor implicitly endorses the listing through the vote. The cost: long-tail assets rarely list.
​

Permissionless listings. Morpho Blue, Euler V2 (via EVK), and Fluid (via vault deployments) let anyone create a market without governance approval. The list grows quickly — Morpho Blue has 200+ markets after 18 months — and curators or vault choosers handle the risk vetting downstream. The benefit: long-tail and exotic collateral lists. The cost: users must vet markets themselves or trust a curator.
​

Hybrid listings. Spark uses Sky governance for risk parameters but ships markets faster than Aave because the Sky DAO has a more aggressive risk appetite for USDS-aligned markets. Some protocols also operate "growth markets" with looser approval requirements, isolated from the main liquidity surface.
​

The listing model maps directly to how depositors should think about the protocol. Governance-approved venues require trust in the DAO. Permissionless venues require trust in a curator or in the depositor's own market-vetting capacity. Neither is universally better; the choice depends on the depositor's risk-evaluation appetite.
​

Every lending protocol must answer: when a position is underwater, who liquidates it, and who covers any leftover bad debt? The answers split the protocols.
​

Aave V3 runs an open liquidation market — anyone can call liquidationCall() and seize collateral at a discount (typically 5–10%). Bad debt that exceeds collateral coverage falls on Aave's Safety Module, where AAVE stakers underwrite the protocol in exchange for staking yield. The Safety Module has been tapped twice historically (the 2020 BLY incident and the 2022 CRV bad debt), per the Aave docs.
​

Compound V3 uses an absorb mechanism rather than open liquidation. The protocol itself absorbs underwater positions, paying liquidators in base asset for collateral seized. Bad debt sits on the protocol's reserve account; if reserves go negative, COMP holders are expected to vote a recapitalization.
​

Morpho Blue runs open liquidation with a fixed-formula liquidation incentive. Bad debt from underwater positions sits with depositors in the affected market — Morpho explicitly does not socialize bad debt across markets. This is why market-level isolation matters so much in the Morpho design.
​

Euler V2 runs a Dutch-auction-style liquidation: the discount starts at zero and rises over time until a liquidator takes the position. The mechanism reduces MEV extraction relative to open liquidation. Bad debt sits per-vault.
​

For depositors, the practical question is: if a market collapses, do my deposits survive? In Morpho Blue and Euler V2, market collapse means depositors in that market take the loss; depositors in other markets are unaffected. In Aave V3 and Compound V3, the protocol's reserve / Safety Module absorbs first; depositors are exposed only after those buffers are depleted. Different exposure profiles for different appetites.
​

Aave V3 has the highest TVL among DeFi lending protocols at $19.4B as of April 2026, deployed across 15+ EVM chains. Spark ($6.8B) and Morpho Blue ($4.9B) follow. See the full comparison matrix for chain coverage and rate ranges.
​

Aave V3 uses a monolithic pool — all assets share one liquidity contract. Compound V3 (Comet) uses isolated markets — one base asset per pool, with collateral siloed. Aave offers more borrowable assets per market; Compound contains failure modes per pool. See the Compound V3 deep dive.
​

Risk varies by protocol and asset. Audits, oracle architecture, liquidation parameters, and governance posture all matter. Cumulative DeFi lending losses exceed $2.1B per DefiLlama's hack tracker. Treat any deposit above 1–2% of net worth as a risk underwriting decision, not a yield decision.
​

Yes — most DeFi lending protocols accept ETH, stETH, wstETH, cbETH, and rETH as collateral. LTV caps typically range from 70–82.5%. Aave V3 and Spark both list multiple ETH-derivative collaterals. Check the stablecoin borrowing strategies guide for tactic-by-tactic mechanics.
​

Interest rates follow a utilization curve: as borrow demand approaches 100% of supplied liquidity, rates kink upward to reserve liquidity for withdrawals. Each protocol publishes its slope1, slope2, and optimal utilization parameters per asset. Curve calibration is a key governance lever.
​