Interpreting Aave V4: A Transition from Product to "Bank"

Original Title:：《Decoding Aave V4: A Shift from Product to “Banking”》

Original Author: Eric, Foresight News

On the evening of March 30 Beijing time, the Aave V4 version—officially greenlit in 2024—was launched on the mainnet, bringing the first good news since the Aave DAO governance controversies.

V4 can be described as a full teardown and rebuild of Aave. The most core change is that the previously independent lending markets have been integrated into a single liquidity pool architecture: Hub and Spoke (hub and spokes).

In the V4 version, each chain or L2 has a unified liquidity center (i.e., the Hub). All assets users deposit for lending will be stored in a single liquidity pool. The Hub is responsible for global coordination, credit limit control, system-level constraints (such as “total borrowing amount ≤ total supply amount”), and emergency pauses. The Hub does not directly face users; instead, it manages liquidity behind the scenes in a unified way.

It is worth noting that it is not that there is only one Hub on each chain—instead, different Hubs are designed based on different needs, which is essentially a form of risk isolation. For example, V4 has currently launched the Core Hub, Prime Hub, and Plus Hub. The Core Hub includes mainstream assets and is designed for all users. The Prime Hub is designed for suppliers seeking more “controllable” collateral. The Plus Hub is designed for strategy-based stablecoins, and its parameter design needs to take the project’s scale into account.

As for the Spoke, you can think of it as an independent market. Each market has independent lending/borrowing functions, risk parameters, and collateral rules. Within one Hub, users’ assets exist in the same liquidity pool, and borrowers need to select different Spokes based on their needs. For example, as shown in the figure above, users can deposit WETH as the borrowable asset. Borrowers can borrow WETH from the first four Spokes, but only the EtherFi Spoke can collateralize weETH.

Although the official claim is that fragmented liquidity can be integrated, in real operation, for users who borrow against high-quality collateral assets, the difference is actually not that big. For example, if you want to collateralize ETH to borrow assets, there is no difference between V3 and V4 in terms of operation—as long as you can ensure the health factor does not drop too low.

So, in terms of liquidity integration, V4 is indeed more granular to manage than independent markets, but it’s not a qualitative leap. What truly brings the difference is Spoke-specific parameter customization and a new liquidation engine.

In V4, borrowers’ interest rates depend on the base rate and the risk premium. The base rate is still the same as in V3: it uses a utilization curve, rising slowly below the optimal utilization and then steeply after that point. The risk premium depends on the nature of the collateral asset. If the collateral is more stable assets such as USDT, ETH, or WBTC, the risk premium will be small or even 0. But if it is a high-risk meme coin, the risk premium will be much higher—avoiding the situation of “subsidizing bad assets with good assets.”

Take a simple example: in V3, interest rates depend entirely on supply and demand. If you borrow USDT in the same way, although there may be differences in the loan-to-value (LTV) limit and liquidation threshold, the interest rates for collateralized ETH and LINK under the same supply-and-demand conditions are the same. But obviously, LINK’s volatility is higher than ETH’s. If the interest rate is the same, borrowers collateralizing LINK will raise utilization, which creates a problem where borrowers using ETH collateral see their borrowing cost not go down but actually rise.

V4 fixes this flaw. Borrowers who collateralize high-risk assets must pay higher costs, and providers of funds can also earn higher returns. At the same time, higher interest rates curb borrowing demand, making the cost advantage for users borrowing against high-quality collateral assets even more obvious.

In the liquidation mechanism, liquidators will only restore the health factor to the target value preset for the Spoke. And the lower the health factor, the higher the liquidation bonus. This design not only gives borrowers more room for action, but also reduces bad-debt risk across the entire platform. In addition, the new liquidation engine adds a “dust prevention mechanism”: when the remaining debt or collateral falls below a threshold (such as $1000), liquidators must liquidate the entire position, preventing small residuals from accumulating and reducing capital efficiency.

Finally, idle liquidity in the Hub can be automatically deployed into low-risk yield strategies approved by governance (such as short-term treasuries, stablecoin LPs, money market instruments, etc.). While increasing income for liquidity providers, it also increases the DAO’s revenue, which might be one of the few advantages under the “unified liquidity” model.

Overall, the advantages brought by Aave V4’s unified liquidity in lending/borrowing are not very significant. And the so-called composability—i.e., borrowing users can manage positions across different Spokes in a unified way—doesn’t make it much more convenient than V3. But as the author says in the title, V4 turns Aave from a product into a financial infrastructure similar to a “bank.”

Strip away all kinds of complex business, the most core business of a bank is to accept deposits, leave part of them as reserves for users’ daily needs such as payments and transfers, and then earn the spread by making loans. As for idle funds, banks can also allocate them to different investments within the limits of their risk tolerance.

St. George Bank Headquarters, St. George’s Palace

Established in 1407 in Genoa, Italy, St. George’s Bank is usually considered the earliest bank in the world. The bank didn’t just provide deposit and lending services; it also handled government debt management, currency exchange, and fund transfers—meeting the commercial needs of Genoa at the time as a major European trading hub.

From 2017’s launch of ETHLend to Aave V4’s rollout in 2026—less than 10 years—Aave has managed to resemble the original model of a bank. Of course, Aave and banks are not identical, and this is only a comparison. Compared with P2P, a bank model that has been beaten and refined for hundreds of years by countless black swans is naturally a better choice—just as V4 is to V3.

If you look closely, you’ll find that many “innovations” in the DeFi space have almost turned into historical dust—for example, the hot DeFi 2.0 in the second half of 2021. Instead, it’s projects like Aave—where the business is simple and the logic has matured in traditional finance over hundreds of years—that have survived, and the more they operate, the better they thrive. After years of exploration, many DeFi projects likely discovered the same issue: DeFi has a very high ceiling, but not a single step on the path traditional finance has taken can be left out.

Aave V4 concentrates liquidity. There’s a lot it can do in the future. For example, it could invest assets whose idle time exceeds a certain number (such as one year) into relatively higher-risk investments—such as providing ETH/USDT LP on Uniswap—running entirely in a business bank model, and gradually expanding into other businesses of commercial banks, such as credit cards (refer to Ethfi’s model of stablecoin consumption via collateralized borrowing), and so on.

Going one step further, Aave could also expand into “investment banking.” For example, by launching an ICO platform, users who deposit assets to earn interest could borrow USDT and USDC to participate in investments, without needing to withdraw the assets and sell them to obtain the stablecoins required to join an ICO. In this way, it can charge fees to projects on one side, while earning interest on the other.

Although there isn’t anything particularly groundbreaking in the Hub&Spoke mechanism itself for lending/borrowing, it does set up the most important foundation for the next step.