Aave V4 is the largest architectural revision the protocol has shipped since launch in 2020. The Aave V3 vs V4 comparison matters because Aave is the largest DeFi lending protocol — $19.4B TVL across 15+ chains as of April 2026, per DefiLlama's Aave V3 page — and the V4 design changes the unit of liquidity, the cross-chain model, and the risk-isolation primitive. This guide walks the differences, what they enable, and the migration path for users and integrators.
​

The short version: V3 is one liquidity pool per chain with per-asset risk parameters; V4 is a "Liquidity Hub" with "Spokes" — borrow markets that draw from a shared liquidity layer per network, with cross-chain liquidity fungibility scheduled for later phases. Rates, listings, and risk all change.
​

Aave V3 launched in March 2022. Its core design is a monolithic pool per chain with isolation mode, e-mode (efficiency mode for correlated assets), and a portal feature for cross-chain transfers via supported bridges. The aave-v3-core GitHub repo contains the canonical contracts; Aave's developer docs detail the integration surface.
​

V3 deployed to 15+ EVM chains: Ethereum, Polygon, Avalanche, Arbitrum, Optimism, Base, BNB Chain, Gnosis, Scroll, Linea, Metis, Fantom (deprecated), Harmony (deprecated), zkSync Era, Polygon zkEVM, and others. Each chain runs an independent pool with its own asset listings, risk parameters, and governance-approved configurations.
​

E-mode allowed correlated assets (e.g., stETH and ETH, or USDC and USDT) to be borrowed against each other at LTVs up to 97%. This unlocked levered staking strategies and stablecoin yield arbitrage. Isolation mode allowed riskier assets to be listed with a debt ceiling, capping protocol exposure to any single asset.
​

Aave V4, outlined in the May 2024 Aave Companies temp-check governance proposal and shipping in phases through 2025–2026, restructures the protocol around two layers: a Liquidity Hub and Borrow Spokes.
​

The Liquidity Hub is a shared liquidity pool per network. Suppliers deposit into the Hub. Spokes — independent borrow markets — draw from the Hub when they need liquidity. Each Spoke can have its own risk parameters, collateral list, and rate curve, but does not maintain its own depositor pool. This decouples deposit from borrow risk: a depositor in the Hub does not have direct exposure to the failure of any single Spoke (the Hub holds the bad-debt buffer).
​

Other notable V4 features:

The Hub-Spoke split solves a structural problem with V3's monolithic design. In V3, listing a new asset means exposing every depositor to that asset's failure mode (mitigated but not eliminated by isolation mode). The trade-off was: keep the asset list short and conservative (Compound's approach) or list aggressively and accept correlated risk (Aave's approach). Neither is satisfying when the goal is to support hundreds of long-tail assets without endangering the core pool.
​

V4's Hub-Spoke architecture lets the protocol list a long-tail Spoke (e.g., a vault for an emerging LST) without depositors in the main USDC Hub being affected. The Spoke pays a premium for liquidity, holds its own risk parameters, and absorbs first-loss bad debt before tapping the Hub buffer.
​

The model rhymes with Morpho Blue's vault-and-curator design, but with one structural difference: Aave's Hub is governed by AAVE-token holders, while Morpho Blue's markets are permissionless. Aave V4 trades flexibility for the AAVE governance umbrella, which is what large institutional depositors prefer.
​

Aave's governance plan is to deploy V4 alongside V3 rather than force-migrate. V3 markets continue running, and depositors can choose when to move. The migration interface is expected to be a one-click swap of aTokens between versions, similar to the V2-to-V3 migration tool that ran 2022–2023.
​

Integrators face a larger lift. The V3 Pool contract surface — supply, borrow, repay, withdraw, liquidationCall — maps to a different surface in V4 (Hub and Spoke split). Anyone integrating Aave at the smart-contract level (vault aggregators, levered yield apps, treasury automation) will need to rewrite the interaction code. The aave-v3-core repo remains the V3 source of truth; V4 contracts are tracked in a separate repo per the governance proposal.
​

The Hub-Spoke split is expected to compress rate volatility. V3 utilization spikes — common during stablecoin yield rotations — caused brief rate spikes when borrow demand on a single asset hit the kink point. V4's unified borrow rate model averages utilization across Spokes, smoothing the curve.
​

TVL impact is harder to forecast. The structural advantage of Hub-Spoke (lower listing friction, broader long-tail support) should grow TVL; the partial migration period (Q3 2025 onward) may temporarily fragment liquidity between V3 and V4. Token Terminal's Aave dashboard tracks the rollout in real time.
​

Aave V3 is multi-chain but liquidity does not commute across chains: a USDC depositor on Polygon Aave can't withdraw on Arbitrum without bridging. V4's planned cross-chain liquidity fungibility will eventually solve this for Aave-internal flows, but it doesn't help users moving between protocols (e.g., from Aave on Base to Morpho on Ethereum, or from Spark to Compound).
​

Eco handles this case as a first-class primitive. A user or treasury team submits a stablecoin movement intent across 15 supported chains, and the network routes liquidity in seconds. For lending users chasing the best supply APY across protocols and chains — Aave on one network, Morpho on another — Eco Routes (CLI + API) eliminates the bridging round-trip. The pattern is described in the pillar Best DeFi Lending Protocols 2026.
​

The Hub is conceptually a single shared deposit pool per network, not a single deposit pool across all networks. On launch, V4 will run independent Hubs on Ethereum, Polygon, Arbitrum, and so on. Cross-network liquidity fungibility — the longer-term feature — relies on Chainlink CCIP message passing and is structurally separate from the Hub-Spoke split itself.
​

Within a network's Hub, depositors supply assets and receive aTokens (the same yield-bearing receipt token used in V3, with adjusted accounting). The Hub holds the global liquidity pool; each Spoke is a separate contract that requests liquidity from the Hub on demand. When a borrower in Spoke A wants to draw USDC, the Spoke pulls USDC from the Hub, lends it to the borrower, and pays the Hub a liquidity premium per second.
​

Spokes can be created with different risk postures. A blue-chip Spoke (ETH, wBTC, stables) might have an LTV of 80%; a long-tail Spoke (specific LRT or RWA) might have 50% with a tight debt cap. The Spoke's first-loss buffer absorbs bad debt before the Hub becomes exposed — meaning Hub depositors have layered protection that V3 depositors did not.
​

One subtle V4 change: risk parameters move closer to the market. In V3, every parameter (LTV, liquidation threshold, debt ceiling, reserve factor) is set by AAVE governance through proposal-and-vote. In V4, Spoke-level parameters can be delegated to a "Risk Steward" — a smart contract empowered to adjust within governance-set bounds. Gauntlet has run a Risk Steward model for Aave V3 risk recommendations since 2023; V4 formalizes the role on-chain.
​

The result: faster risk responses to market moves (a stablecoin depeg, an LST de-syncing from ETH) without waiting for a governance proposal cycle. Whether this is a feature or a risk depends on perspective — institutional depositors value the speed; risk-conservative depositors prefer governance friction.
​

Aave V2-to-V3 migration ran 14 months from V3 launch to <50% of V2 TVL remaining. V3-to-V4 is expected to be slower for two reasons. First, the architecture change is more disruptive — V2-to-V3 was a contract upgrade with similar surfaces; V3-to-V4 is a structural rearchitecture. Second, V3 is widely integrated into yield aggregators (Yearn, Beefy, Aera), wallets (MetaMask, Rabby), and treasury automation (Aera, Karpatkey, Llama Risk). Each integration needs to update before the V3 user can move smoothly to V4.
​

Practical takeaway for users: V3 will remain the dominant Aave deployment through 2026 and likely most of 2027. V4 adoption will start with greenfield deposits and large institutional repositories that justify the migration cost.
​

The largest practical impact of V4 will fall on yield aggregators that integrate Aave at the smart-contract level. Yearn V3 vaults, Aera, Llama Risk, Karpatkey, Pendle, and Aave's own GHO Stability Module all maintain integration code against the V3 Pool contract surface. Each will need to update — separately, with separate audit cycles — before users can move smoothly to V4.
​

The migration burden is uneven. A vault that simply supplies USDC to Aave for yield needs minimal updates. A vault running active leverage strategies (e.g., Yearn's stETH levered vaults, or Pendle's PT-aUSDC strategies) needs more substantial reworking because the Hub's accounting differs from V3's per-asset state.
​

For users, the practical timeline: greenfield deposits will move to V4 quickly. Existing positions in integrated vaults will lag by 6–12 months as each aggregator ships V4 support. During that window, V3 will remain the dominant Aave deployment in TVL terms even as V4 ships strategically important features.
​

One V4 feature worth dwelling on: GHO becomes natively mintable from any Spoke. In V3, GHO is mintable only from the Ethereum deployment, with cross-chain GHO transfers handled via CCIP wrapping. In V4, a borrower with collateral in a Polygon Spoke can mint GHO directly into the Polygon Hub, and the GHO is fungible with Ethereum-issued GHO via the CCIP-managed cross-chain supply.
​

This matters for GHO adoption because the V3 limitation has constrained GHO usage outside Ethereum. Treasury teams running multi-chain operations have largely picked USDC or USDS instead because the operational overhead of managing GHO across chains is high. V4's native cross-Spoke GHO removes that overhead and is expected to materially expand GHO TVL.
​

For users tracking V4 readiness, the operational checkpoints worth watching:

The first three items are protocol milestones; the last two are ecosystem milestones. V4 is ready for power users at the protocol milestones; it's ready for everyone at the ecosystem milestones.
​

Aave V4 is shipping in phases through 2025–2026. The temp-check passed governance in May 2024; initial Hub deployments began on testnet in Q4 2024 with mainnet rollout phased through 2025. Track the Aave governance forum for stage-by-stage updates.
​

No — V4 deploys alongside V3. Depositors choose when to migrate. A migration tool is expected to ship at V4 mainnet, similar to the V2-to-V3 tool from 2022–2023.
​

V4 launches initially on Ethereum, then on the same major networks where V3 lives. The longer-term cross-chain liquidity fungibility roadmap requires CCIP integration on each network and will roll out incrementally.
​

E-mode is replaced by Spoke-level configuration. A correlated-asset Spoke can set its own LTV cap (potentially higher than V3's 97%) without affecting the Hub. The mechanism is the same; the location moves.
​

GHO becomes natively mintable from any V4 Spoke, not just Ethereum. This is the most material practical change for GHO users. See the DeFi Lending pillar for the broader stablecoin context.
​