Aave is a pooled crypto lending protocol for supplying assets and borrowing against collateral

Borrow stablecoins without selling an ETH position

Collateralized stablecoin borrowing lets an ETH holder access USDC or GHO while retaining exposure to the supplied asset. The user deposits ETH or WETH into an eligible market, enables that position as collateral and borrows less than the collateral's permitted capacity. Borrowed tokens arrive in the connected wallet, while the debt begins accruing variable interest immediately.

The arrangement is overcollateralized: borrowing power stays below the recognized value of the collateral. That gap absorbs some price movement, although it does not eliminate liquidation risk. A falling collateral price, a rising debt balance or appreciation of the borrowed asset against the collateral reduces the position's health factor. Supplying without borrowing avoids that personal liquidation path, yet the supplier still depends on reserve liquidity and protocol accounting.

Which charges shape supply returns and borrowing costs?

Supply APY and variable borrow APY are market-set rates driven mainly by reserve utilization, which is the borrowed amount relative to supplied liquidity. A rate curve raises borrowing costs as utilization approaches its configured optimal point. Suppliers receive the portion allocated to liquidity providers after the reserve factor, while V4 Spokes also support collateral-sensitive risk premiums that add to a borrower's base interest cost.

Network gas is separate from protocol interest. One basis point equals 0.01%, a useful unit when reading fee parameters, but ordinary supplying and borrowing have no universal fixed APY to quote. An ERC-20 supply requiring an onchain allowance uses two transactions - approval and supply - whereas a supported permit turns approval into a signature and leaves one onchain supply transaction. Withdrawals, collateral changes, borrows and repayments each require their own network execution.

How pooled liquidity becomes a withdrawable position

Pooled liquidity converts a supplier's deposit into an accounting claim rather than matching it with one named borrower. In V3, supplying an asset mints aTokens whose value tracks the underlying asset at a 1:1 unit relationship while accrued interest increases the claim. A holder of aUSDC therefore has a USDC-denominated supply position, not exposure to a separate floating-price lending token.

V4 moves core liquidity accounting into Hubs and lets user-facing Spokes draw from that shared capital under defined limits. Hub balances use shares internally, while Tokenization Spokes issue transferable ERC-4626 vault shares for a single underlying asset. This division separates liquidity storage from the collateral rules and borrowing logic attached to each Spoke.

Withdrawal capacity still depends on unborrowed liquidity. When borrowers have drawn most of a reserve, a full withdrawal does not execute until repayments or new supplies restore enough liquidity; an amount within the available balance remains withdrawable. Interest compensates suppliers for lending demand, not for a guaranteed redemption schedule.

Where do V3 and V4 operate, and why does the market matter?

Market and network selection determine the contracts, supported assets, risk parameters and liquidity available to a position. Aave V4 launched on Ethereum on March 30, 2026 with three Liquidity Hubs - Core, Prime and Plus - and later expanded to Avalanche, where one Core Hub serves three initial markets: Main, AVAX Correlated and Forex. V3 deployments include Ethereum, Polygon, Avalanche, Arbitrum One, Optimism and Base, among other networks. Positions remain local to their deployment and version; a USDC supply on Base does not fund an Ethereum reserve or secure debt on Arbitrum One. Even identical tickers require the correct network-specific token contract.

How the health factor converts price movement into liquidation risk

The health factor measures the collateral-adjusted value supporting a borrow relative to its debt. The calculation multiplies each collateral value by its liquidation threshold, combines those adjusted values and divides the total by outstanding debt. A health factor below 1 makes a borrow position eligible for liquidation. Values closer to 1 leave less room for collateral declines, debt interest or adverse movement in the borrowed asset.

V3 applies a default 50% close factor while the health factor remains above 0.95, then permits up to 100% under the lower-health-factor condition. V4 instead calculates a repayment amount designed to restore a governance-defined target health factor, and its liquidation bonus rises as the position deteriorates. Those version-specific rules matter; a V3 close-factor assumption does not describe a V4 Spoke.

Hypothetical worked example. Use a hypothetical collateral value of 10,000 accounting units, a hypothetical 80% liquidation threshold and hypothetical debt of 6,000 units. These three inputs are illustrative: collateral value and debt change over time, while governance controls the threshold. The starting health factor is (10,000 × 0.80) ÷ 6,000 = 1.333. If the hypothetical collateral value falls to 7,500 while the hypothetical debt stays at 6,000 and the hypothetical threshold stays at 80%, the health factor becomes exactly 1.000. A 25% collateral decline has consumed the entire 2,000-unit threshold-adjusted buffer.

Which controls limit collateral and reserve exposure?

Isolation Mode, E-mode, supply caps and borrow caps constrain how liquidity and collateral interact. V3 Isolation Mode allows an isolated asset to serve as the account's sole enabled collateral and restricts borrowing to governance-approved stablecoins. E-mode selects one category of correlated assets with category-specific loan-to-value, liquidation and borrowing parameters. Higher efficiency narrows the buffer when the assumed correlation weakens.

V4 expresses comparable separation through Spokes, credit lines and per-Spoke add or draw caps, allowing several risk environments to use Hub liquidity without sharing identical collateral rules. Governance votes set these limits and approve deployments. The governance token AAVE follows ERC-20 and supports EIP-2612 permits; its 2020 migration converted 100 LEND into 1 AAVE, creating 13 million redeemable tokens plus a 3 million ecosystem reserve for a 16 million initial supply. Aave Labs builds interfaces and contributes software, while the DAO controls protocol decisions through governance.

Why flash loans behave differently from ordinary debt

A flash loan borrows and repays liquidity within one atomic blockchain transaction, so it creates no lasting collateralized position. If the transaction cannot return principal plus the required premium, every operation in that transaction reverts. The contracts expose two principal entry patterns: a simple call for one reserve and a multi-reserve call for more complex operations. The Ethereum V4 launch configuration matched the V3 premium structure at 5 basis points, or 0.05%, with 4 basis points allocated to the protocol and 1 basis point allocated to liquidity providers. Developers use this mechanism for atomic refinancing, collateral changes and arbitrage involving venues such as Uniswap; transaction complexity and gas determine whether the operation remains economical.

From wallet connection to a first supply

A first supply requires a self-custodial wallet, a supported token on the selected deployment and enough of that network's native asset for gas. MetaMask and Rabby connect directly or through WalletConnect-compatible sessions. The network identifier provides a durable confirmation of context: Ethereum Mainnet uses chain ID 1, Polygon PoS uses 137, Arbitrum One uses 42161, Base uses 8453 and Avalanche C-Chain uses 43114.

After choosing a market, the wallet grants the relevant ERC-20 allowance or signs a supported permit, then authorizes the supply transaction. Aave Pro presents V4 Hubs and Spokes, whereas V3 interfaces organize assets by deployment and market. Before borrowing, the account needs collateral enabled, enough borrowing capacity and an understood health-factor change. Supplying alone is simpler because it creates no debt, although asset, contract and withdrawal-liquidity exposure remain.

Smart-contract, oracle, liquidity and governance trade-offs

Protocol risk spans contract execution, price inputs, listed assets, reserve liquidity and changeable governance parameters. Chainlink feeds and specialized price adapters supply values used for borrowing limits and liquidations; a delayed or constrained feed changes how quickly the system reflects market movement. Asset depegging, thin secondary liquidity or a frozen reserve also affects withdrawals and collateral quality even when the lending contracts operate as designed.

V4 underwent approximately 345 cumulative days of security review involving four audit firms, four independent researchers and a six-week Sherlock contest with more than 900 verified participants. Trail of Bits, Blackthorn, ChainSecurity and Certora contributed review or formal-verification work. That depth reduces identifiable implementation risk without removing it. Upgrades, cap changes and new asset listings add operational flexibility, yet they also require users to distinguish immutable contract rules from parameters that governance updates.

Compound III, Morpho Blue and SparkLend separate the alternatives

Protocol choice turns on market structure, collateral breadth, rate formation and who controls risk parameters. Compound III organizes each market around one borrowable base asset with supported collateral assets. Morpho Blue defines an isolated market through five immutable parameters: loan token, collateral token, oracle, interest-rate model and liquidation loan-to-value. SparkLend is a pooled money market whose stablecoin liquidity and governance-set rates connect closely with the Sky ecosystem.

Aave fits users seeking pooled multi-asset markets, established health-factor accounting and deployments spanning several chains, while V4 adds shared Hubs with specialized Spokes. Morpho Blue offers finer isolated-market selection at the cost of evaluating each configuration separately. Compound III provides a narrower base-asset structure, and SparkLend emphasizes Sky-linked stablecoin liquidity. None supplies unsecured, fixed-term consumer credit; every onchain option carries its own contract, oracle, collateral and liquidity assumptions.