At the end of March, Aave launched its V4, a new version of its lending protocol with a simple approach, open small, then expand over time. Six months later, adoption seems to be there. Indeed, V4 of the protocol reached one billion dollars in deposits.
On a protocol like Aave, you do not deposit just any quantity of any asset. Each market has a ceiling and beyond that, deposits are refused. These ceilings start low on purpose and raising them allows for growth. Indeed, Aave V4 introduces a new approach to managing the maximum capacities of its pools.
Key Points
V4 has been live since March 30, 2026. According to the counters, it is around the billion dollars deposited; the exact figure varies from source to source.
This rise followed a series of ceiling raises, not a spontaneous rush.
Most of these increases were not voted on one by one by AAVE holders.
They were carried out by a small group that controls a multi-key vault, after a public recommendation of risk.
A typical vote takes around two weeks. This committee can act in a few hours.
Announced in 2024 by Stani Kulechov, then postponed, Aave V4 was deployed on Ethereum on March 30, 2026. Its architecture is called Hub and Spoke. This consists of a central liquidity reserve (the hub) which feeds distinct borrowing markets (the spokes), each with its own collateral rules. The idea is to pool money instead of letting it sit in as many silos as there are markets. An approach focused on capital efficiency.
In mid-September, several counts brought V4 above the billion dollars in deposited assets. These are the calculations presented by Sumcap.xyz, which has closely followed the evolution of V4 since its deployment. V3 still exceeds the 30 billion provided.
This growth has a well-identified lever. On V4, the old “supply caps” and “borrow caps” become more or less add caps (what a spoke has the right to deposit in the hub) and draw caps (what it has the right to borrow). A saturated cap refuses new deposits. It also limits Aave’s exposure to one asset. Thus, if its price drops or its liquidity evaporates, the volume of possible bad debts remains limited.
Not a vote for each increase
Until V4, cap increases, specifically allowing the protocol to accept more of an asset, were decided via a governance vote. From now on, it is LlamaRisk, a risk service provider mandated by the DAO, which publishes the recovery “rounds” on the forum. The changes are now applied via the Security Council, a multisig, which only executes a transaction if several keys sign. AAVE holders did not vote “yes” or “no” each time a milestone of 10,000 to 15,000 ETH passed. They vote for major acts: turning on V4, listing a market, deploying a channel, granting a mandate. On a daily basis, the ceilings are the responsibility of LlamaRisk and the Security Council.
Ainsei, the majority of the increases that made this growth possible were signed by a multisig rather than adopted by an on-chain vote. In other words, it was not the assembly of tokenholders that opened the floodgates one by one.
Risk Stewards
The regime that the DAO voted for in V3 is called Risk Stewards. It authorizes a risk committee to adjust certain parameters without convening all AAVE holders. The perimeter is locked by the code: maximum amplitude per update, delay before retouching the same asset, prohibition of listing a new asset or going outside the limits. This is the “the DAO sets the corridor, the committee leads within” model.
On V4, it is not yet the tool that has come most of the way. An ARFC of August 2026, then a Snapshot and an AIP (proposition 523), aim to connect Stewards to Ethereum and Avalanche. At the end of September, LlamaRisk still indicated that Ethereum and Avalanche adjustments were going through the Security Council, except on Arc, already processed via the Stewards.
The recommendations today come mainly from LlamaRisk, with Aave Labs. Chaos Labs left the system in spring 2026.
The gain is mainly temporal. Forum, Snapshot of about five days, on-chain voting, turnaround time, the whole procedure took about two weeks. The Council or Steward reduce this to a few hours.
The tension remains the same, the contract is not the same
A handful of keys which arbitrate the exposure of a market which aims for a billion does not fit well with the image of a protocol governed in a decentralized manner. Historically, the Aave Chan Initiative, long one of the main delegates, founded by Marc Zeller, defended this division of labor. ACI announced in March 2026 that it was not renewing and withdrew. The argument did not go away with the team. Supporters of systematic voting respond that a mandate that is too broad ends up emptying the ballot box of its substance.
Delegation also has a security cost. A multisig is private keys. A Steward limited by the code limits the damage, even compromised, he can neither list a phantom asset nor multiply a ceiling by ten with a single gesture. The Security Council of the launch phase has a broader mandate, because V4 opened its doors empty and we had to be able to open quickly. This is another risk profile, even if each increase remains a public transaction, preceded by a dated acknowledgment on the forum.
In a bank, this type of committee raises its limits behind closed doors. At Aave, you can reread the transaction. AAVE holders can revoke a mandate by voting. Since V3, they have rather expanded it. On V4, they are only starting to transpose it into the contract provided for it.
