Aave (AAVE)

Aave is a decentralized, non-custodial liquidity protocol that allows users to supply cryptocurrency assets to earn interest or to borrow against deposited collateral, without relying on a bank or other centralized intermediary. Built primarily on the Ethereum blockchain and subsequently deployed across a number of additional blockchain networks, Aave operates through smart contracts that automatically manage lending pools, interest-rate calculations, and collateral-backed borrowing, and is governed by holders of its native token, AAVE, through a decentralized autonomous organization (DAO) structure. By the mid-2020s, Aave had grown into one of the largest protocols in decentralized finance ("DeFi") by total value locked, and had expanded from its original crypto-native lending markets into institutional real-world-asset (RWA) lending and a native, decentralized stablecoin called GHO.¹²

Aave originated as ETHLend, a peer-to-peer cryptocurrency lending platform founded in November 2017 by Finnish entrepreneur Stani Kulechov, which raised approximately $16.2 million through an initial coin offering.¹ The project was rebranded to Aave in September 2018 — "aave" being the Finnish word for "ghost" — and transitioned from its original peer-to-peer lending model to a pooled-liquidity model. The AAVE governance token launched in October 2020 as a successor to the earlier LEND token, with existing LEND holders able to migrate their holdings to the new token at a fixed 100:1 ratio.¹ Aave has a fixed maximum token supply of 16 million AAVE, nearly all of which had entered circulation by the mid-2020s.³

By the end of 2025, Aave reported that it had surpassed $3 trillion in cumulative, all-time assets supplied and roughly $950 billion in all-time loans originated, and that it accounted for a majority of the DeFi lending sector's active loans, total value locked, and revenue.⁴ Net deposits reached roughly $70 billion during 2025, while total value locked as measured by the analytics site DefiLlama peaked in the range of $30–40 billion, making Aave at various points the single largest DeFi protocol by that measure.⁵⁶ The protocol's governance token, AAVE, has traded with substantial volatility, reaching an all-time high of roughly $661–667 during the 2021 cryptocurrency bull market before a prolonged decline and partial recovery.³⁷

Ticker	AAVE
Category	Decentralized Finance (DeFi)
Website	https://app.aave.com/
Twitter	@aave
Telegram	Aavesome
Reddit	https://www.reddit.com/r/Aave_Official
Contract Addresses
ethereum	0x7f...e9Copied!
optimistic-ethereum	0x76...78Copied!
near-protocol	7f...arCopied!
huobi-token	0x20...25Copied!
polygon-pos	0xd6...0bCopied!
fantom	0x6a...3bCopied!
harmony-shard-0	0xcf...0fCopied!
arbitrum-one	0xba...96Copied!
avalanche	0x63...d9Copied!
energi	0xa7...9aCopied!
sora	0x00...36Copied!
binance-smart-chain	0xfb...02Copied!

In April 2026, Aave was affected by the largest DeFi exploit of that year, stemming from a vulnerability in the cross-chain bridge infrastructure of a third-party protocol, Kelp DAO. The incident triggered billions of dollars in withdrawals from Aave's markets and led to an industry-wide recovery effort known as "DeFi United."⁸⁹¹⁰

History

Origins as ETHLend

Aave's predecessor, ETHLend, was founded in November 2017 by Stani Kulechov, a Finnish law student and entrepreneur who had previously experimented with blockchain-based applications. ETHLend was designed as a peer-to-peer lending marketplace built on Ethereum, allowing borrowers and lenders to be matched directly with one another for collateralized loans, and raised roughly $16.2 million through an initial coin offering of its own token, LEND.¹ While technically functional, ETHLend's peer-to-peer matching model proved to have significant liquidity limitations, since loans depended on individually matching a specific borrower with a specific lender rather than drawing from a shared pool of available capital.

Rebrand to Aave and the pooled-liquidity model

In September 2018, the project rebranded from ETHLend to Aave, adopting a new name — the Finnish word for "ghost" — intended to reflect the protocol's emphasis on transparency, in the sense of a system whose operations were fully visible and verifiable on-chain despite operating without a centralized intermediary. Alongside the rebrand, Aave shifted from ETHLend's original peer-to-peer lending design to a pooled-liquidity model, in which suppliers deposit assets into shared smart-contract-based liquidity pools that borrowers can draw against, with interest rates determined algorithmically based on the utilization rate of each pool rather than through individual loan negotiation.¹

Aave V1 and V2

Aave's V1 protocol, launched in January 2020, introduced the pooled-liquidity architecture along with several features that would become central to the protocol's identity, including flash loans — a mechanism allowing users to borrow assets without posting any collateral, provided the loan is borrowed and repaid within a single blockchain transaction, primarily intended for use cases such as arbitrage, collateral swapping, and self-liquidation. V1 also offered borrowers a choice between variable and stable interest rates. Aave V2, deployed in December 2020, improved the protocol's user experience and gas efficiency and refined its risk-management and rate-switching features.¹

AAVE token launch and governance

In October 2020, Aave launched its AAVE governance token as a successor to the original LEND token, with a migration mechanism allowing LEND holders to convert their tokens to AAVE at a fixed rate of 100 LEND per 1 AAVE.¹³ The new token, with a fixed maximum supply of 16 million, was designed to serve several functions within the protocol: enabling holders to vote on Aave Improvement Proposals (AIPs) governing protocol upgrades and parameter changes; allowing holders to stake AAVE within the protocol's Safety Module, a mechanism intended to provide a backstop of last-resort capital in the event of a shortfall, in exchange for staking rewards; and, later, providing a mechanism for revenue distribution to token holders.³² Staked AAVE within the Safety Module can be "slashed" by up to 30% in the event the protocol experiences a deficit that staked funds are called upon to help cover, giving stakers a direct economic stake in the protocol's ongoing solvency.³ Aave subsequently introduced Umbrella, an automated bad-debt backstop that uses asset- and network-specific staking vaults to absorb deficits without requiring slower governance intervention.¹¹

Multi-chain expansion and Aave V3

Beginning in 2021, Aave expanded beyond its original Ethereum deployment to additional blockchain networks, including Polygon and Avalanche, broadening its accessibility to users seeking lower transaction costs than those typically associated with the Ethereum mainnet.¹ Aave V3, released in 2022, introduced additional capital-efficiency and risk-isolation features, including "efficiency mode" (e-mode), which allows for higher loan-to-value ratios between correlated assets, such as different forms of staked ether, and isolation-mode markets that allow newer or riskier assets to be listed with contained risk exposure to the rest of the protocol. Stable-rate borrowing, a holdover from earlier versions, was progressively deprecated across V3 markets from 2023 onward, leaving variable rates as the standard borrowing mode.

By the end of 2024, Aave's net deposits had reached roughly $35 billion, up from about $10 billion a year earlier, and the protocol added markets on networks including BNB Chain, Scroll, and ZKsync Era.¹² Growth continued through 2025: Aave's share of total DeFi total value locked climbed from about 17% to 29% over the year, and it became the first protocol to reach $1 billion in TVL on six separate networks — Ethereum, Arbitrum, Avalanche, Base, Plasma, and Linea.⁴

GHO stablecoin and institutional expansion

Aave introduced GHO, a decentralized, overcollateralized stablecoin native to the Aave protocol, in July 2023, as part of its broader effort to diversify beyond simple asset lending and borrowing. GHO is minted by users who deposit collateral into Aave and is designed to maintain a peg to the U.S. dollar through a combination of overcollateralization and protocol-governed interest-rate facilities. Its circulating supply grew from roughly $35 million to more than $500 million over the course of 2025.⁷

In August 2025, Aave Labs launched Horizon, a permissioned, institution-focused market built on the Aave V3 codebase, allowing qualified institutional participants to borrow stablecoins such as USDC against tokenized real-world assets, including short-term U.S. Treasury instruments, money-market funds, and private credit, with collateral values kept updated using Chainlink's net-asset-value oracle feeds.¹³⁵ Horizon grew quickly, passing $500 million in total assets within roughly three months of launch, and by mid-2026 held approximately $540 million in total assets against about $163 million borrowed; its operators described it as targeting an institutional credit market they estimated could exceed $1 trillion in addressable size.⁵¹⁴

Separately, in December 2025 the U.S. Securities and Exchange Commission closed a long-running investigation into Aave without action, removing a significant regulatory overhang ahead of the protocol's institutional expansion.¹⁵

Aavenomics reforms and the "Aave Will Win" proposal

Aave's token economics — collectively referred to within the community as "Aavenomics" — underwent several significant revisions in the mid-2020s. In March 2025, the protocol introduced a mechanism known as "anti-GHO," a non-transferable token minted from a share of GHO-related protocol fees and distributed to AAVE and Balancer-pool-token stakers, intended to give stakers a more direct and efficient claim on protocol revenue while incentivizing continued GHO adoption.¹⁶ Around the same period, Aave's DAO approved a governance-controlled token-buyback framework and established an Aave Finance Committee to execute purchases and manage treasury assets under an explicit governance mandate. A pilot program running from May to November 2025 repurchased more than 94,000 AAVE from the open market at a cost of roughly $22 million.⁷

In October 2025, governance made the buyback permanent at approximately $50 million per year. That figure was subsequently reduced in March 2026 to about $30 million annually, with governance citing a roughly 25% decline in borrow-fee revenue from its peak and an ambitious 2026 operating budget of $190 million set against 2025 revenue of about $142 million; the adjustment also shifted primary buyback funding from stablecoins toward ETH-denominated treasury assets.¹⁷

The most consequential of these reforms was a governance proposal known as "Aave Will Win" (AWW), submitted by Aave Labs in February 2026 and approved by the Aave DAO by a wide margin on 12 April 2026. The proposal directed 100% of revenue generated by Aave-branded products — including swap fees collected through the aave.com interface, Aave Pro, the Horizon institutional market, Aave Kit, and Aave's mobile application — to the DAO treasury, structurally linking AAVE's value to the protocol's commercial performance rather than to governance rights alone. It resolved a months-long dispute that began in late 2025, when Aave Labs redirected aave.com swap fees away from the DAO treasury into a company-controlled wallet. In exchange, Aave Labs received DAO funding — $25 million in stablecoins plus 75,000 AAVE, with a further $17.5 million in grants tied to specific product milestones — and committed in writing to work exclusively on Aave-related products. The framework also contemplated a new Aave Foundation to hold the protocol's trademarks and intellectual property.¹⁸¹⁹

Technology

Lending pools and aTokens

When a user supplies an asset to Aave, the protocol issues that user an equivalent quantity of a corresponding "aToken" (for example, aUSDC for supplied USDC), an ERC-20 token that automatically accrues interest in real time and can be redeemed at any point for the underlying deposited asset plus accrued yield.²⁰ Borrowers on Aave must post collateral exceeding the value of the amount they wish to borrow — a design known as overcollateralization — and pay a variable interest rate that fluctuates algorithmically according to the utilization rate of the relevant lending pool.²⁰ Interest rates adjust automatically, without any centralized rate-setting authority, based on the proportion of a given pool's deposited liquidity that is currently borrowed: as utilization rises, borrowing costs increase to encourage repayment and attract additional deposits, and as utilization falls, rates decline correspondingly.²⁰

Flash loans

Among Aave's most distinctive technical features are flash loans, which allow a user to borrow any amount of a supported asset without posting collateral, provided that the borrowed funds are used and fully repaid, together with a fee, within the same blockchain transaction in which they were borrowed. Because a blockchain transaction either fully executes or is entirely reverted if any of its conditions are not met, this design allows Aave to offer uncollateralized borrowing while remaining fully protected against default, since a transaction that fails to repay the loan simply never takes effect. Flash loans are used by developers primarily for purposes such as arbitrage across decentralized exchanges, collateral swapping, and self-liquidation of an otherwise-unsafe borrowing position.¹

Aave V4 and the modular hub-and-spoke architecture

Aave V4, which launched on Ethereum mainnet on 30 March 2026, represented a substantial architectural overhaul, introducing a modular "hub-and-spoke" design in which a central hub acts as a unified liquidity layer serving specialized, customizable lending markets ("spokes"). Its developers described the goals as improved capital efficiency, a simpler process for listing new assets and markets, reduced governance overhead, and "health-targeted" liquidations intended to reduce unnecessary losses for borrowers whose positions become undercollateralized.⁷² Proponents described the upgrade as consolidating liquidity that had previously been fragmented across Aave's isolated V3 markets into a more unified but still risk-segmented structure.²

Multi-chain and non-EVM deployment

Beyond its original Ethereum deployment and subsequent expansion to EVM-compatible networks such as Polygon and Avalanche, Aave took its first step outside the Ethereum Virtual Machine ecosystem on 20 August 2025 with a V3 deployment on Aptos mainnet, a blockchain built using the Move programming language. The deployment required a complete rewrite of the protocol in Move, a rebuilt front end and SDK, and adaptation of Aave's business logic to the Aptos virtual machine; it followed a phased rollout incorporating audits from multiple firms, a mainnet capture-the-flag competition, and a $500,000 bug bounty payable in GHO. The initial market supported APT, sUSDe, USDC, and USDT.²¹²²

Governance

Aave is governed by holders of the AAVE token through a system of on-chain and off-chain proposals known as Aave Improvement Proposals (AIPs). Token holders can vote directly on proposals or delegate their voting power to other addresses, and approved proposals are typically executed automatically through Aave's on-chain governance smart contracts. Several third-party organizations have historically played specialized roles within this governance ecosystem, including risk-management providers responsible for setting and monitoring parameters such as collateral factors and liquidation thresholds across Aave's various markets, and community-coordination bodies responsible for stewarding specific initiatives, such as the growth of the GHO stablecoin.⁹

Aave's governance structure came under particular strain around the April 2026 Kelp DAO-related exploit (discussed below). Three independent organizations responsible for risk management, technical oversight, and governance coordination departed the ecosystem in the weeks immediately preceding the incident, citing disagreements over budget allocation, the protocol's architectural roadmap, and risk-management priorities; the last of them, Chaos Labs — which had set Aave's risk parameters since November 2022 — departed on 6 April 2026, twelve days before the exploit.⁹ The episode prompted subsequent governance reforms, including an overhaul of the criteria used to evaluate new assets proposed for listing, expanding the assessment framework beyond financial risk and volatility to encompass bridge dependencies, oracle design, rehypothecation depth, and the underlying technical architecture of candidate collateral assets, alongside proposals for tier-gated loan-to-value caps that would be harder to override for structurally risky assets.¹⁰²³

The April 2026 Kelp DAO exploit

On 18 April 2026, at approximately 17:35 UTC, an attacker exploited Kelp DAO's LayerZero V2 bridge route between Unichain and Ethereum, which had been configured with a single verifier (a "1-of-1 DVN" setup). A forged inbound message was accepted on the Ethereum side without a corresponding burn on Unichain, releasing approximately 116,500 rsETH — a liquid-restaking token representing staked and restaked ether — from the Ethereum-side adapter. The tokens, worth roughly $292 million and representing about 18% of rsETH's circulating supply, appeared fully backed to downstream protocols but were not.²⁴²⁵ Rather than immediately selling them, the attacker deposited 89,567 rsETH across seven wallets on Aave as collateral and borrowed 82,650 WETH and 821 wstETH — roughly $190 million — across Aave's Ethereum and Arbitrum deployments.²⁶⁸

The exposure was amplified by a governance decision three months earlier. On 29 January 2026, Aave activated e-mode for rsETH, adding WETH as a borrowable asset against rsETH collateral for the first time at a 93% loan-to-value limit and a 95% liquidation threshold, with the stated aim of attracting roughly $1 billion in new rsETH inflows; no assessment of the underlying bridge risk was conducted. Competing lenders had set materially lower limits — 72% at SparkLend and 75% at Fluid — and SparkLend halted new rsETH supply entirely on the same day Aave raised its limits.⁹²⁷

Within hours of detecting the exploit, Aave's Protocol Guardian froze rsETH and wrsETH markets across all deployments, set the asset's loan-to-value ratio to zero, and halted further borrowing against it; WETH was subsequently frozen on Core, Prime, Arbitrum, Base, Mantle, and Linea to prevent new borrows while the fallout was assessed.²⁶²⁸ Aave's own smart contracts, oracles, and liquidation systems were not compromised and continued to operate as designed throughout.¹¹

The resulting impairment triggered a rapid, large-scale wave of withdrawals. Aave's total value locked fell from $26.4 billion on 18 April to roughly $20 billion by the following morning, to about $16 billion within four days, and to approximately $14.6 billion a month later — a decline of some 45%, which cost Aave its position as the largest DeFi protocol by TVL. Total DeFi TVL fell by more than $13 billion over the same initial two-day span, with some capital rotating into competing lenders such as SparkLend. The AAVE token fell roughly 16% in the immediate aftermath.²⁹³⁰³¹

An incident report published on 20 April by LlamaRisk and other Aave service providers modeled two outcomes depending on how Kelp allocated the shortfall. Under uniform socialization across all rsETH holders on all chains — a roughly 15.1% haircut — Aave's bad debt was estimated at $123.7 million, with Ethereum Core absorbing $91.8 million. Under an alternative in which losses were isolated to rsETH on layer-2 networks, remote-chain tokens would be repriced to a 26.5% backing ratio and bad debt would rise to $230.1 million, concentrated on Mantle (a 71.5% WETH reserve shortfall) and Arbitrum (26.7%), leaving Ethereum mainnet unaffected. The total backing shortfall was calculated at 112,204 rsETH.²⁶³²

DeFi United and recovery

On 23 April 2026, Aave's founder Stani Kulechov and a coalition of other DeFi protocols and service providers — including Lido Finance, Ether.fi, Ethena, Mantle, and later Compound — launched an industry-wide recovery initiative known as "DeFi United," aimed at recapitalizing rsETH's backing directly through voluntary contributions rather than relying solely on Aave's insurance mechanisms. The initiative targeted a shortfall of roughly 68,900 ETH, about $160 million at the time; initial pledges totaled approximately 13,500 ETH from Kulechov personally, Ether.fi, Lido, and the Golem Foundation, plus a proposed 30,000 ETH loan facility from Mantle.⁸³³

Commitments accelerated rapidly. Consensys and Ethereum co-founder Joseph Lubin pledged up to 30,000 ETH; LayerZero contributed 5,000 ETH; and the Avalanche Foundation, Solana Foundation, TRON, and Circle Ventures joined in various capacities. By 27 April, Kulechov said the fund had secured enough to fully restore rsETH's backing, subject to pending governance votes and definitive agreements, with total commitments reported at roughly $300–320 million.³⁴³⁵

Separately, on 20 April the Arbitrum network's Security Council froze approximately 30,766 ETH — worth roughly $71 million at the time — that had been identified as linked to the exploit, placing the funds in a wallet that could not move without a governance vote. Much of the remainder of the stolen proceeds was bridged and swapped into bitcoin via Thorchain, complicating recovery.⁸

By mid-May 2026 the recovery was substantially complete. Kelp and Aave burned the exploiter's remaining rsETH on Arbitrum and began progressively refilling the LayerZero adapter on mainnet from the Aave Recovery Guardian and Kelp Recovery Safe; Kelp migrated rsETH's cross-chain messaging to Chainlink's CCIP; and on 17 May Aave restored pre-incident WETH loan-to-value ratios across all six affected V3 deployments, with more than 95% of the unbacked rsETH recovered.³⁶²⁸

Litigation over the frozen Arbitrum funds

The episode became the subject of a U.S. federal court dispute. On 1 May 2026, the law firm Gerstein Harrow LLP, representing plaintiffs holding three consolidated default judgments against North Korea with a combined face value exceeding $877 million, served Arbitrum DAO with a restraining notice under New York's CPLR §5222(b), seeking to reach the frozen 30,766 ETH. The firm relied on attribution of the Kelp exploit to North Korea's Lazarus Group and APT-38 by LayerZero and the blockchain analytics firm TRM Labs.³⁷³⁸

On 4 May, Aave LLC filed a 29-page emergency motion in the U.S. District Court for the Southern District of New York, through Morrison Cohen LLP, asking Judge Margaret M. Garnett to vacate the restraining notice outright or, failing that, to schedule an expedited hearing and require the plaintiffs to post a bond of at least $300 million. Aave argued that a thief acquires no lawful title to stolen property, that the frozen ether therefore represented recovered user funds belonging to Aave depositors with no connection to the hack, and that a contrary ruling would expose every future DeFi recovery effort to seizure claims whenever a state-backed actor was implicated.³⁷³⁹

On 8 May, Judge Garnett issued a two-page order that, invoking CPLR §5240, modified rather than vacated the restraining notice: it permitted an on-chain Arbitrum governance vote to transfer the ether to a wallet controlled by Aave LLC and shielded participants in that vote from violating the freeze, while preserving the judgment creditors' claims, which travel with the assets. Arbitrum delegates approved the release with about 91% of participating voting power. The underlying question of whether the recovered funds can ultimately be reached by the terrorism judgment creditors was left unresolved, and commentators noted that the case raised broad, largely untested legal questions about the interaction between conventional U.S. judicial enforcement mechanisms and decentralized, on-chain asset recovery.⁴⁰⁴¹³⁹

Market history

AAVE has traded on cryptocurrency markets since the token's October 2020 launch, with its price closely tracking the broader cycles of the DeFi sector. The token reached an all-time high in the range of $661–667 in May 2021, before entering a prolonged decline through the 2022–2023 bear market. By 2026, analysts noted that AAVE traded at what they characterized as a comparatively low valuation relative to the scale of the protocol's underlying business, with net protocol revenue of roughly $142 million in 2025 — up about 57% year-over-year — against a market capitalization representing a modest multiple of the protocol's total value locked.⁷¹⁷ Commentators attributed this apparent undervaluation in part to lingering investor caution stemming from the 2022–2023 DeFi bear market, a broader tendency for capital in early bull-market phases to favor layer-1 blockchain tokens over DeFi infrastructure tokens, and the residual effects of the April 2026 Kelp DAO exploit on market sentiment.⁷

Following the April 2026 exploit, AAVE's price fell sharply, and on-chain analytics reported that large holders ("whales") were net buyers of the token during the subsequent price weakness, in a pattern some analysts interpreted as a sign of longer-term conviction in the protocol's fundamentals despite the near-term disruption.⁵

Criticism and risks

Aave, like other DeFi lending protocols, faces a combination of smart-contract risk, market risk, and risks arising from its exposure to other, external protocols whose tokens it accepts as collateral. The April 2026 Kelp DAO exploit illustrated this last category directly: although Aave's own smart contracts functioned exactly as designed throughout the incident, the protocol nonetheless faced substantial losses because it had accepted, as collateral, a token whose backing was compromised by a vulnerability in an entirely separate protocol's cross-chain infrastructure — and had done so at an aggressive 93% loan-to-value ratio set three months earlier without a corresponding bridge-risk assessment.²⁶⁹ The incident also demonstrated a second-order risk: a crisis of confidence prompted depositors across unrelated Aave pools to exit regardless of their specific exposure, driving utilization toward 100% and spiking borrowing rates for every remaining borrower.⁹

Critics have also questioned aspects of Aave's evolving token-economics mechanisms — such as the "anti-GHO" incentive system introduced in 2025 — on the grounds that features tied closely to expectations of profit distribution and centralized treasury-management authority could attract heightened regulatory scrutiny in jurisdictions applying tests such as the U.S. securities law's Howey Test to determine whether a digital asset constitutes an unregistered security.¹⁶ Separately, the "Aave Will Win" proposal drew objections from some delegates on governance grounds, with critics noting that its $50 million total funding request represented a substantial share of the DAO's treasury and non-AAVE reserves, granted to a single service provider in a single vote with limited recourse if deliverables were not met.¹⁹ Aave's founders and governance participants have generally expressed optimism that evolving U.S. regulatory policy toward decentralized finance — including the SEC's December 2025 decision to close its investigation into the protocol — would provide greater clarity and reduced regulatory risk going forward.¹⁶¹⁵

References