Aave v4 at a billion dollars

Published 2026-09-26 · data as of

tl;dr

Aave v4 deposits are >$1.2B as of 26th September, with recent step changes from EtherFi Cash and Arc deployments.
Borrowing demand varies across markets, with $378M of debt against $723M of borrowable supplied assets (52%).
EtherFi Cash accounts for ~92% of active addresses, but only 9% of debt. The other deployments have far fewer users, with much larger borrowing positions.
v4 deposits are still only 4% of v3's. Current and former v3 users account for a majority of analysed v4 deposits (excluding Cash), but it is early days.
There are plenty of new developments in the works, with a Base v4 tokenized stock deployment in the last couple of days, and multiple active proposals.

Aave v4 is the latest protocol version from the Aave ecosystem, introducing a new Hub and Spoke architecture. Lending liquidity is pooled in Hubs, which authorized Spokes can draw upon (with limits). Spokes are where risk profile diversity and innovation can exist, with a simple way to seed liquidity for new Spokes. This contrasts with Aave v3, where each market has its own liquidity pool and risk configuration. Shared liquidity remains within a Hub on one chain; v4 does not combine liquidity across chains.

Aave v4 went live on 30 March 2026, and recently hit a billion dollars in deposits across instances. We will take a closer look at the current state and trajectory of Aave v4.

The headlines

As of , Aave v4 has $B of deposits across Ethereum, Avalanche and Arc, alongside a separately operated, licensed EtherFi Cash deployment on Optimism. There are ~k addresses with open balances across all deployments (of which k are on the EtherFi Cash instance), with in outstanding debt, of which is on Ethereum.

Table as of , including EtherFi Cash. The newly launched Base deployment is excluded throughout the analysis and discussed in next steps. An active address has a positive deposit or debt balance. Addresses are not distinct people or Cash cardholders.

Deposits are distributed across assets: stablecoins are roughly , BTC is , while ETH and staked ETH are . EtherFi Cash also holds a substantial book of vault and wrapped receipts. On the demand side, the majority of borrowing is stablecoins (), with most of the remainder borrowed ETH.

As of , including EtherFi Cash. Categories use token metadata with explicit overrides in the pipeline. Yield-bearing dollars include sUSDe, syrupUSDG and principal tokens.

Growth over time

Aave v4 has seen steady growth in deposits over the past few months, with two spikes, one for EtherFi Cash on Optimism, and the other with the deployment on Arc, while the Avalanche deployment has remained relatively small.

Weekly observations from to , plus the latest snapshot date, for seven Hubs stacked to total gross supply. The four blue series are Ethereum: Core, Global Dollar, Prime and Plus. Arc Core, Avalanche Core and EtherFi Cash on Optimism are shown separately.

Borrowing has generally grown alongside deposits. is borrowed against of supplied assets with borrowing enabled (). The remaining supply consists of collateral-only assets, alongside a small restricted balance of rsETH.

Weekly observations through , plus the latest snapshot date. Borrowable supply is held in Hub assets with borrowing enabled through at least one linked Spoke at that time. Collateral / other supply cannot be borrowed under those settings; it includes restricted assets. Caps, pauses and available liquidity can further limit new borrowing.

Ethereum has the highest aggregate debt-to-borrowable-supply ratio at , compared with on Avalanche. The newest deployment on Arc currently has of debt against of supply.

The ratios describe balances, not a target of full utilization. Caps, pauses and other constraints can still limit borrowing.

The Ethereum Etherfi Spoke is the largest by debt at $91M, followed by Ethereum Main, which have both seen steady growth over the past few months. Meanwhile Arc Main's debt jumped up in the last couple of days, with one large cirBTC holder taking out a $49.0M USDC loan. Maple SyrupUSDG on Ethereum and EtherFi Cash on Optimism round out the five largest markets below.

The five largest current Spokes by debt across all four deployments, including EtherFi Cash, shown through .

Incentives and supply growth

Supply incentives have accompanied growth in Core USDG, Core frxUSD and Global Dollar USDG, adding rewards on top of the base lending APY. The first recorded rewards for Core frxUSD and USDG are in May, with Core USDG seeing continued growth while frxUSD's supply has plateaued after its initial growth. Global Dollar USDG's rewards began in August.

At the snapshot, these three markets account for of gross Aave v4 supply across all deployments. The test of durable liquidity will come as and when these incentives are reduced or removed.

Blue: supplied tokens. White: base supply APY. Purple: APY including rewards; shading shows the uplift. Dashed vertical lines mark the first recorded reward day. Each market includes up to seven days before rewards; growth compares the preceding daily balance with the latest observation.

Variety in user behaviour

Not all active addresses are doing the same thing on Aave v4, and we can do some rough categorisation. "Lender" addresses supply borrowable assets without debt ( in total across deployments). addresses hold only non-borrowable deposits and have in debt. Another have borrowable deposits as well as in debt. A final have non-borrowable deposits with no debt (the purpose of these deposits is uncertain). These users are all getting different things out of Aave v4 today.

Excluding EtherFi Cash reduces the depositor count from to , but those depositors account for of the debt. Among those remaining addresses, have debt, compared with when Cash is included.

As of , each depositor address is classified once across the deployments selected in the table. Counts including Cash are approximate. The toggle affects this table only. Addresses are not necessarily distinct people.

There is significant variety across Spokes in terms of borrower count and average debt. borrower addresses on EtherFi Cash have a mean of in debt (and a median of , including small residual balances), while Maple SyrupUSDG on Ethereum has addresses with a mean of . Lots of Spokes have a relatively small number of borrowing addresses today, though of course not all addresses are equivalent, and the differences may reflect the range of distribution and maturity across Spokes.

As of , including EtherFi Cash on Optimism. Each dot is one Spoke. Both axes are logarithmic. Hover for the full name, total debt, address count and median debt per address.

Migration versus new activity

Tracking addresses is an imperfect method, but can provide an indication of how v3 users are using v4. Currently, only 4.7% of the v3 deposits are by addresses with deposits on v4, but those addresses account for 29% of v4 deposits. 28% of v4 deposits are from addresses which have made deposits on v3 (but no longer have more than $100 supplied there), while 43% is from addresses with no v3 history detected.

As of . The dotted line links the same addresses' current deposits on each version; it does not show transfers. v4 here covers Ethereum, Avalanche and Arc, excluding EtherFi Cash and addresses with less than $100 supplied and borrowed.

This implies that v4's growth is coming from some new sources, but also a large proportion of deposits is coming from users who are or were active on v3. On the flip side, there is still a lot of value deposited in v3 which feels pretty sticky - only 3 of the 50 largest current deposit addresses on Ethereum's v3 have funds on v4.

Visualising the Hubs and Spokes

It is possible to visualise the structure of Aave v4 and its Hubs and Spokes as a graph, with separate deployments on different chains. Below is a snapshot: explore the latest data at aave-v4.azfuller.com/graph.

All networks appear in one snapshot as separate clusters. Click the graphic to open the live site.

What is next for v4?

Aave v4 is roughly six months old. What might the next 6 months hold?

Continued growth of v4 certainly appears to be a priority of the Aave ecosystem, and there is still lots of headroom for growth, given that deposits are ~4% of the v3 total. v3 deposits will likely remain pretty sticky, and the Aave DAO have rowed back plans about strict migration timelines, but it is reasonable to expect more migration (aided by dedicated tooling), and further growth of the existing markets.

New markets will also drive growth, with or without supporting incentive campaigns (for example there are open proposals for a Custodied Collateral Lending hub and spoke, USDe on Avalanche, and BTC.b and EURCV on the Core Instance on Ethereum).

Arc's launch brought significant deposit growth on a fresh v4 deployment, and debt is picking up, with $56.6M of debt against $239.0M of deposits (though that is dependent on a couple of large debtors). We can expect more v4 deployments across chains - even in the last few days, Coinbase tokenized stocks have gone live on v4 on Base, introducing a new primitive on a new network. And the EtherFi Cash playbook could be repeated, for an existing userbase (implying another nice bump), or for a new product. The exact economics and addressable market for these future license deals remain to be seen.

Meanwhile there are active efforts to improve the robustness of v4, proposing Umbrella coverage for certain markets and activating Risk Stewards.

So plenty of avenues for growth and development, and if the market trend continues to be bullish, we can expect that to generally act as a tailwind for Aave v4 as well as the broader Aave ecosystem.

Method

Methodology & reference data

Source and scope. The data is frozen on . The headline table covers Ethereum, Avalanche, Arc and the separately operated EtherFi Cash instance on Optimism. The newly launched Base deployment is deliberately excluded from all analytical totals, charts and address cohorts. It is discussed in next steps. This is not a census of every deployment using v4 code. Current balances and holders come from Aave's public GraphQL API. Cash requires its own instance header. Historical Hub balances also use archive RPC reads. The downloads below retain source responses and retrieval times. Separate reads can differ as balances and prices change.

Deposits, debt and addresses. Headline balances count each Hub asset once. They are gross balances, including collateral and recursive lending. They do not measure external capital or net inflows. Active addresses have an open balance, not necessarily recent transactions. Holder lists are fully paginated and addresses are deduplicated across reserves and deployments. Repeated addresses use the last observed balance and mark the affected count as approximate. Live pagination can still miss addresses as balances change. Addresses can represent contracts or multiple users; there is no ownership or wrapper look-through.

Borrowable supply. A Hub asset counts as borrowable if at least one linked reserve permits borrowing. These deposits can also serve as collateral. Collateral / other supply cannot be borrowed under those settings and includes restricted assets. Caps, pauses and available liquidity can limit new loans further. Token categories use metadata and explicit pipeline overrides. Wrapped receipts are not decomposed into their underlying portfolios.

Growth over time. The two supply-history charts use weekly observations from launch, plus the latest snapshot date. Each historical block supplies Hub balances, linked Spoke borrowing flags and oracle prices. Both charts cover the same seven Hubs and reconcile to the same total at each date. The contract set is the one discovered for this snapshot. Dollar changes include price movements; they are not constant-price flows. The five-largest-Spokes chart combines API history with the historical Cash Hub book. Its selection uses current debt, not historical rank.

Incentives. Daily supply history is selected for each token's specific Hub. It is measured in tokens, not historical dollars. The APY lines compare daily average base lending APY with APY including rewards. The dashed line marks the first positive reward observation, not an independently verified campaign start. Growth compares the preceding daily balance with the latest observation. The latest daily bucket is provisional. Current rewards are not projected backwards. Growth alongside rewards does not establish that rewards caused the deposits. The share quoted in the prose uses current dollar balances for the three markets and all four deployments.

Lenders, borrowers, or both. Each address's positions are combined across the selected deployments before classification. Any positive deposit in a borrowable Hub asset counts as lending. Any positive debt counts as borrowing. The remaining depositors hold collateral / other assets without debt. This describes balances and permissions, not user intent. Debt-only addresses are outside this depositor table. Excluding Cash removes its positions and recomputes the categories; Ethereum's Etherfi Spoke remains included. Settings and balances are separate reads. The borrower-size chart combines each address's debts within each Spoke. Its mean is total debt divided by distinct borrowing addresses. Tooltip medians use those per-address totals, including small residual debts.

v3 and v4 overlap. The v4 migration sample covers Ethereum, Avalanche and Arc. It includes addresses with at least $100 supplied or borrowed. Cash is excluded. “v3 + v4” combines current overlap with v3 evidence whose timing is unresolved. “Prior v3” means earlier detected v3 activity and less than $100 supplied and borrowed on v3 now. “No v3 detected” is limited by the covered history and does not establish a new user. v3 deposits use the matching market set from the same day. “v3 only” is the market total less the shared cohort's v3 deposits, not a separate holder census. It can include addresses with Cash-only or small v4 positions. Each bar is normalized to its own total. The dotted line connects an address cohort, not traced transfers. Current v3 positions cover 24 markets on 21 chains. History is unavailable for Celo, Metis, Plasma, Ink, Mantle and MegaETH. Earlier activity is compared with the oldest currently returned v4 position, not a verified first-ever v4 interaction. The separate top-50 v3 check covers Ethereum Core, Lido, Etherfi and Horizon; it is not a ranking across all v3 chains. The downloads include cohort definitions, source observations and the reconciliation to the headline.

Hubs and Spokes. The graph uses registered reserve relationships on each chain. Node and link balances reconcile within their source books. A draw-only link has zero add caps and permission to draw liquidity. It does not accept deposits into that Hub through the Spoke. The Spoke can hold collateral through a different Hub on the same chain. The header is a decorative rendering of the same topology; node size and spacing do not encode balances.

Checks and limitations. Pipeline checks cover pagination, identities, currencies, finite balances, coverage and breakdown sums. History checks cover complete observations, positive prices and agreement across linked oracles. Current holder totals and the latest API history must each agree with Hub asset totals within one percent. Residual differences remain visible below. These are consistency checks, not an independent onchain audit of every current position. Historical RPC reads are cross-checked against the API.

Balance cross-check

This table covers Ethereum, Avalanche and Arc only. It compares API measurement bases and therefore has a smaller scope than the headline table, which includes Cash. Hub summary supply is excluded from composition charts.

Frozen data and reproduction

Rebuild instructions and QA commands are in pipelines/aave-v4/README.md.