The Engine of DeFi: Inside the Aave Crypto Lending Protocol

This article, by S&P Global and S&P Global Ratings, is a thought leadership report that neither addresses views about individual entities nor is a rating action. S&P Global Ratings is a separate and independent division of S&P Global.

Decentralized finance (DeFi) lending leverages blockchain technology to power protocols that enable users to borrow and lend digital assets. These DeFi lending platforms allow permissionless, overcollateralized transactions that enable users to earn interest on their digital assets. 

Unlike traditional finance, where overcollateralized lending typically requires intermediaries and permissions, DeFi lending operates via smart contracts, enabling users to interact directly. Interest rates are determined by the supply and demand of capital through predefined algorithms. The key innovation is that lending, borrowing, collateral management, liquidations and risk monitoring occur automatically through smart contracts rather than through financial intermediaries. DeFi lending brings credit markets on-chain. This will contribute to global integration of credit markets, market acceleration and lower costs.

The most significant use case of the DeFi lending platforms is to create leveraged exposure, which compounds both yield and risk. By allowing users to borrow against posted collateral, these platforms expand investors' portfolios, increasing both potential returns and losses. Consequently, growth in TVL is linked to demand for leveraged exposure rather than traditional consumption or commercial credit. 

Using Kaiko data and the Dune platform, this report examines lending, borrowing, withdrawal, liquidation, and interest rate trends across major digital assets, providing a unique perspective on the evolution of liquidity, leverage, and risk in DeFi lending markets. 

DeFi protocols' total value locked (TVL) has fluctuated significantly with crypto market conditions. In September 2026, the TVL in DeFi lending markets was about $52 billion (out of a total DeFi TVL of almost $90 billion). This was a decline from the October 2025 DeFi lending market TVL, which was about $92 billion (out of a total DeFi TVL of almost $171 billion). The broader DeFi ecosystem also encompasses decentralized exchanges, perpetual futures, and derivatives, among other blockchain-driven investment vehicles.

Aave is a prominent DeFi lending protocol operating on Ethereum and other blockchains. As of September 2026, Aave had about $19.2 billion in TVL, most of which was held in its most popular version, Aave v3. This large TVL demonstrates that on-chain credit intermediation has achieved scale and has the potential to increase the attractiveness of DeFi to institutional investors whose primary focus is liquidity and operational capacity. Aave is the largest DeFi lending protocol by a significant margin, followed by Morpho, the second largest with a TVL of $11.0 billion, and Sky with a TVL of $5.9 billion, according to DeFiLlama. Aave depositors can earn interest by supplying liquidity to the protocol. For each digital asset, all deposits are pooled together and can be borrowed by anyone on the platform. To take out a loan, borrowers must provide collateral, and both the principal and interest must be repaid to unlock their collateral. The onboarding and offboarding of collateral is set through governance votes by AAVE token holders. In addition to borrowers and depositors, liquidators play a crucial role in Aave. If a borrower fails to maintain adequate collateral, liquidators act as third parties that sell collateral to repay a portion (or the entirety) of the loan and bring the position back into compliance. They earn liquidation bonuses as an incentive to help maintain the protocol's stability. Most of these liquidators are bots operating through smart contracts.

The collateral requirements for Aave v3 are determined through a multi-layered governance process. While AAVE token holders cast the deciding votes, the process is underpinned by technical expertise from risk advisers who propose parameters based on market volatility and liquidity profiles. Central to Aave's risk architecture is the Umbrella Module, which functions as an asset-specific, first-loss buffer capital that absorbs protocol bad debt. Lenders to the module receive higher interest rates in exchange for absorbing the first losses in a bad debt scenario for each market. 

Aave also has a Real-World Asset (RWA) platform, Aave Horizon, targeted at permissioned and know-your-customer (KYC)'d institutional lenders. This platform allows for collateral such as tokenized Treasuries and collateralized loan obligation (CLO) funds to be posted against stablecoin debt on Ethereum.

Figure 1 summarizes a typical lending/borrowing transaction. 

Three key factors set Aave v3 apart in the DeFi lending landscape:

• Early adoption and liquidity stickiness: Aave v3 was among the first DeFi lending protocols to achieve widespread usage, enabling it to amass significant liquidity that persists today. Liquidity remains notably sticky across different versions, with most activity concentrated on v3 due to its strong connection between liquidity pools and market rates — even as v4 is live and brings isolated markets into the protocol.

• Dual thresholds for borrowing and liquidation: Aave v3 employs two distinct thresholds, a maximum loan-to-value (LTV) ratio for borrowing and a separate LTV for liquidation. This optimizes capital efficiency while preventing liquidations due to cliff effects.

• Universal lending model: Each lending market on Aave is built around a single lending asset, in which a wide range of collateral assets can be used to borrow this lending asset. Users can also swap collateral or debt to take advantage of better yields or interest rates. From the perspective of the lender deploying a given lending asset into the protocol, this model diversifies the collateral asset types to which the lender is exposed. 

In contrast, isolated lending markets provide lenders with the ability to select a single specific collateral asset, which simplifies the lender's risk exposure, but results in liquidity fragmentation risks. Vaults are an innovation designed to balance the trade-off between market liquidity fragmentation and a lender's ability to select risk exposures. They are on‑chain investment vehicles that pool deposits and deploy them according to a defined strategy. (Digital Assets Primer: How Vaults Can Shape The Capital Markets of Tomorrow | S&P Global Ratings). While Aave operates a pooled lending model in which depositors are exposed to a diversified portfolio of borrowers and collateral assets, newer competitors such as Morpho have introduced vaults that allow capital to be allocated to curated sets of isolated lending markets, offering more customized risk exposures and potentially higher yields.

Aave features several participant roles, including depositors, borrowers, and liquidators. It allows five activities: borrowing, repaying, depositing, withdrawing, and liquidating. Since all loans are overcollateralized, borrowers are also depositors by default. If a borrower's debt becomes undercollateralized, a liquidator can partially repay the loan and acquire the collateral assets at a discounted rate. Lending protocols like Aave were instituted and adapted to extend globally permissionless lending and borrowing in DeFi.

1. The health factor measures the safety of a borrowing position. 

The health factor in Aave v3 is calculated as the value of collateral, weighted by each asset's specific liquidation threshold, divided by the value of outstanding borrowings. A health factor above 1 signals that the position is sufficiently overcollateralized, meeting loan-to-value (LTV) requirements and preventing liquidation.

A health factor of less than 1 allows for partial or full liquidation. A liquidator may liquidate the user's collateral and pay up to 50% of the total position debt in a single call (see example in Figure 20).

One of the most significant DeFi lending strategies is looping, which creates leveraged exposure. For example, looping would allow a user to deposit ether (ETH); borrow stablecoins, such as USDC; use the borrowed USDC to buy more ETH; redeposit the additional ETH; and then repeat the process, as shown in Figure 2. (Please see glossary at the end of this paper for more detailed definitions of various digital assets.)

Looping becomes less attractive when liquidity is low and utilization rates surge, causing borrowing costs to rise sharply. Looping makes a position's health factor highly sensitive by compounding leverage, while enabling liquidity access and capital efficiency.

The Oct. 10, 2025, flash crash in crypto markets triggered a wave of liquidations that noticeably altered the distribution of health factors. On Oct. 9, 2025, over 98% of the aggregated TVL in positions had a health factor greater than 1. Following the crash, the market saw an overall 10% decline in health factor, as illustrated in Figure 3. Leverage, not technology or smart-contract failure, was the primary risk. The losses arose because market participants borrowed aggressively against volatile collateral and when prices fell, liquidation followed mechanically. The event also highlighted that the future challenge for DeFi is to manage leverage and concentration risk, as the technology worked as it was designed.

3. Flash loans on Aave are unique to blockchain technology and DeFi.

Flash loans, which were pioneered by Aave, are an innovation unique to DeFi, as they solve a core problem of traditional finance: capital efficiency and access. While in traditional systems, one needs to provide collateral or credit scores to borrow, flash loans eliminate these barriers by relying on smart contracts. Flash loans enable users to instantly borrow substantial funds from a protocol without collateral beyond what is needed to cover transaction costs. Borrowers must complete their intended transactions and repay the principal plus fees within the same operation in one block; if they fail to do so, the entire transaction is canceled with the borrower paying only transaction costs. Lenders suffer no default risk. In other words, a flash loan is temporary liquidity used for a specific strategy.

Figure 9 shows a typical flash loan transaction flow.

To start the process, a user requests a flash loan from the Aave liquidity pool. The flash loan smart contract then checks that the funds do exist and that the logic will return the loan amount plus a fee within the same transaction.

One of the primary use cases for flash loans is to expand the pool of liquidators. Flash loans enable any party to liquidate a position below the liquidation LTV without needing large sums of capital in the debt asset beforehand. In traditional finance, only large, established players with access to capital act as liquidators. Flash loans have supported the resilience of Aave's liquidation mechanism through periods of stress and have proven economically beneficial to the protocol as they mitigate the accrual of bad debt.

Another popular use case for flash loans is for collateral restructuring to optimize collateral and capital efficiency. In DeFi lending protocols, users must overcollateralize assets. When market conditions or interest rates change, users may want to adjust collateral types or reduce borrowing costs. Flash loans allow users to repay debt, release collateral, swap assets, and repledge in one single transaction. 

Lastly, flash loans are used for arbitrage opportunities allowing for efficient price discovery of decentralized exchanges (DEXs). 

Flash loan activity may increase during liquidity shocks, which cause market inefficiencies in the form of temporary asset mispricing as prices move faster than reporting can capture.

WETH dominated cumulative daily flash loan volumes with $4.5 billion between Jan. 1, 2025, and July 31, 2026, followed by USDC, with $3.1 billion. Figure 10 shows the borrowed amount of flash loans within Aave since January 2025, including a July 21, 2025, spike in WETH traced to one, automated Instadapp vault transaction. The vault transaction used flash loans to rotate its Aave v3 collateral from wrapped Lido staked ETH (wstETH) to wrapped ether.fi staked ETH (weETH) while keeping roughly the same WETH debt.

Because Dune reports flash loan volume as gross notional, the vault's repeated flash loan loops (four rounds) added up to $912.5 million, or about 99% of that day's WETH flash loan volume of $918.6 million, even though net borrowing demand was essentially unchanged. 

4. Aave's interest rate model is not tied to any central bank's base rate and is determined by AAVE governance entirely.

Parameters for the Aave interest rate model are determined by AAVE token holders, based on the advice of risk advisers and a highly centralized governance participation. Rates are derived from utilization-driven curves that apply different slope parameters depending on whether utilization is low or high. The utilization rate, defined as the ratio of borrowed funds to available liquidity, is the primary factor determining interest rates. These rates are set algorithmically, reflecting the balance between borrowing demand and the supply of liquidity provided by lenders. 

When utilization is low, borrowing costs remain relatively cheap to encourage more borrowing. As utilization approaches a certain threshold, typically around 80% to 95%, liquidity becomes constrained and rates increase sharply. Beyond this point, interest rates rise rapidly to incentivize additional liquidity supply from lenders or prompt borrowers to deleverage. Additionally, highly leveraged positions will become eligible for liquidation once their accrued interest pushes their health factor below 1. It is important to note that the system corrects itself as high rates reduce borrowings and attract deposits, thus reducing the utilization rate.

The borrow rate is higher than the supply (lending) rate to cover the protocol's reserve factor, or the percentage of borrow interest that the protocol keeps, rather than paying it out to suppliers. The supply rate is calculated as the product of the borrow rate, utilization rate and (1 minus the reserve factor). The borrow and supply rates increase sharply once the utilization rate exceeds a certain threshold.

Figure 11 shows the borrow interest rate curves for USDT, USDC, USDe, WETH and wstETH, as functions of the utilization rate on June 18, 2026, and the optimal utilization rate for each. Note that the interest rate curve for wstETH has a low optimal utilization threshold of 80%, and the interest rate increases more rapidly after this point compared to the other four coins. This aligns with the fact that wstETH has counterparty and operational risk. Out of USDT, USDC and USDe, the latter has the lowest optimal utilization point.  WETH exhibits a higher optimal point combined with a smaller slope. This reinforces that the optimal utilization rates reflect the Aave governance perception of the relative risk of these assets.

Figure 12 illustrates four real-world cases when the interest rates spiked. Across all events, the trigger is a liquidity shock caused by withdrawals (either voluntarily or by an exploit), demand or liquidations. The transmission mechanism is through a surge in utilization, which prompts a rate spike, for example, from a range of 2% to 4% up to a range of 10% to 15%. 

In smaller markets, such as PayPal USD (PYUSD), with a TVL of $9.3 million as of Aug. 22, 2026, a $1.8 million single withdrawal on the same day triggered utilization rates up to 100% and interest rate spikes of 50% and above.

We also studied the trend in interest rates. Figures 13 and 14 show interest rates and utilization rates from mid-2024 to mid-2026 for stablecoin USDT for the first block after midnight UTC. Between November and December 2024, stablecoin borrow rates on Aave v3 rose sharply amid a broader surge in stablecoin borrowing demand across the market. During this period, utilization on major stablecoin markets rose above 95%, which mechanically pushed variable borrow rates above 40% under Aave's utilization-based interest rate model. These spikes coincide with the looping strategies that took place in late 2024 with staked USDe (sUSDe) increasing demand for stablecoin borrowing. Borrowing demand is fundamentally tied to the ebbs and flows of leverage in crypto markets. For USDT, Aave's high interest rate environment in 2025 is different than the one in December 2024 for comparable high utilization rates, as the 2025 environment experienced lower spikes. While often treated as a reference rate for DeFi lending, the pronounced volatility in interest rates highlights how large-scale withdrawals can rapidly tighten liquidity and underscores the significance of concentration risk within Aave.

5. What are historical trends for deposits, withdrawals, borrowings, and repayments on the Aave lending protocol?

Figure 15 shows a baseline upward trend in 2025 for Aave v3 deposits with matching increases in borrowing and withdrawals, as users actively rotated positions (this analysis did not distinguish between new and repeat depositors).

The July 2025 liquidity outflow of over $1.7 billion in WETH caused borrowing rates to surge above 10%. This event highlights how large players can rapidly impact key protocol metrics. Following this liquidity shock, rates gradually returned to normal, attracting liquidity providers back to the platform. During this period, TVL increased, and both WETH collateral and utilization rates rose. This trend was further supported by the launch of Aave Horizon in mid-2025.  

Figure 16 illustrates several liquidation events on Aave v3 that exceeded $200 million. These liquidations unfolded in a stepwise manner: as collateral values fell, borrowers' health factors dropped below 1, triggering liquidation bots to close positions. The resulting collateral sales further depressed prices, causing even more accounts to become eligible for liquidation. While liquidations are often viewed negatively, they represent the protocol functioning as designed, particularly during periods of stress. 

Among three of the most notable liquidations events:

• On Aug. 5, 2024, ether saw notable price declines, dropping to its lowest value of the year, below $2,200, as the broader cryptocurrency markets and risk assets in general suffered widespread losses. This accounted for the largest liquidation spike in Aave v3 history, with $462 million in crypto assets liquidated.

• On Feb. 2, 2025, a steep fall in the crypto market led to $360 million in liquidated assets, triggered by fears of rising inflation related to geopolitical uncertainty regarding tariffs.

• On Oct. 10, 2025, trade tensions triggered a fast crash in the highly leveraged crypto market, which caused a cascading liquidation spiral amplified by low liquidity and infrastructure stress. During this event, which led to global derivatives contract liquidations, Aave experienced $250 million in liquidated assets. 

Figures 17 and 18 show the daily trends in deposits, withdrawals, borrowings and repayments on Aave v3 for WETH, USDT, and USDC since Jan. 1, 2025.

Figure 17 shows that in mid-2025, when WETH hit peak levels, deposits also soared, and investors aggressively supplied assets into the protocol for use as collateral. After the price decline towards the end of 2025 and in 2026, withdrawals increased and deposits remained low.

Figures 18a and 18b show high volatiltity for USDT and USDC deposits and withdrawals, with USDT exhibiting more frequent spikes in both directions. Notably, deposits and withdrawals expanded in July 2025, tracking the on-chain reaction to the US GENIUS Act, when massive amounts of supply were rotated into decentralized lending pools. 

The spike in deposit and withdrawal activity following the Oct. 10, 2025, crash saw more than $20 billion in leveraged positions liquidated within 24 hours. Investors moved funds out of volatile assets like bitcoin (BTC) and ether (ETH) and into stablecoins for safety, driving the sharp rise in deposits. Once prices stabilized at lower levels, much of this same supplied capital was withdrawn.

Figure 19 shows the daily liquidation profile for WETH, USDC and USDT on Aave v3. The largest liquidation peaks are during the February 2025, market sell-off; the Oct 10, 2025, flash crash; and the February 2026, price decline of WETH.

Since 2020, the value of DeFi lending protocols has expanded exponentially from a TVL of a few billion to more than $52 billion following the growth of blockchain technology adoption. Aave v3 remains the largest DeFi lending protocol and a key player in the universal lending markets. 

Aave's key advancement is that it has transformed lending and borrowing into a decentralized, transparent, and global financial service. It enables users to access liquidity, earn yield, and leverage digital assets without traditional financial intermediaries, demonstrating that large-scale financial markets can operate through smart contracts. Aave made DeFi more open, programmable, and capital efficient. However, achieving greater adoption will require addressing security vulnerabilities, liquidity and concentration risks and ecosystem interconnectedness.

How does liquidation impact the health factor of a borrower?

In figure 20, we show a simple, idealized health factor calculation incorporating no interest and assuming the loan's principal stays the same. If a user borrows $80 against $100 in ether (ETH) collateral at time t_0, assuming a liquidation loan-to-value (LTV) of 85, the health factor is 1.06. When the ETH value drops and the health factor falls below 1, a partial liquidation is triggered in which a liquidator steps in and repays 50% of the loan by selling collateral at a discount and charging a fee (80 cents), thus bringing the health factor back above 1.

How does looping create a leveraged position?

Figure 21 shows a simple, idealized example (assuming a flat price and no accrued interest) of a loop on a lending protocol where the user borrows (at max LTV) $80 USDC against $100 ETH collateral, then swaps the USDC on more ETH to increase the deposit and price exposure to $180 ETH collateral, thus increasing leverage to 1.8. The loop is repeated, and leverage increases further to 2.44.