Aave TVL lost 15.1 billion in just three and a half days, while Spark bucked the trend and grew into the biggest beneficiary

On April 22, on-chain analysts Yuyan showed that, affected by the rsETH/KelpDAO incident, Aave funds continued to flow out, with total deposits falling from $48.5 billion to $30.7 billion in just three and a half days. Morpho saw a modest outflow of $1.5 billion over the same period. The most notable contrast came from Spark: its SparkLend business TVL went against the trend, rising from $1.9 billion to $3.2 billion.

A split among three major platforms: Aave bleeds while Spark benefits in the opposite direction

（Source：Aave）

During this incident, three major mainstream DeFi lending platforms showed sharply different trends:

Aave: down from $48.5 billion to $30.7 billion, with net outflows of $15.1 billion in three and a half days (a drop of about 31%).

Morpho: down from $11.7 billion to $10.2 billion, net outflows of about $1.5 billion, with relatively limited impact.

Spark (SparkLend): up from $1.9 billion to $3.2 billion, net inflows of about $1.3 billion, making it the biggest beneficiary of the market turbulence this time.

Some institutional funds withdrawing from Aave redeployed into Spark, indicating that after the rsETH incident, some market participants chose a comparatively safer alternative platform.

Aave’s bad-debt outlook: Defillama’s founder presents two scenario analyses

Defillama founder @0xngmi provided two scenario analyses on how the recovery of $71 million worth of ETH on Arbitrum would affect Aave’s bad-debt size:

Scenario one (loss socialized): If Arbitrum prioritizes support for the Aave market and socializes the loss, then each rsETH holder’s position would be reduced evenly by 18.5%, Aave’s bad debt would be about $216 million, and the Treasury and Umbrella Fund would need to be used.

Scenario two (loss borne by L2 holders): If the loss is borne only by L2 rsETH holders, then Aave’s bad debt on Arbitrum would fall from about $88 million to about $17 million, cutting it by roughly 80%, but it could trigger dissatisfaction and potential lawsuits.

Market sentiment stabilizes: AAVE rebounds from the low, the DeFi industry calls for facing the recovery

After the Arbitrum Security Council froze about 307,776 ETH (about $71 million) in KelpDAO-related addresses, Santiment data showed that Aave market sentiment was no longer falling to new lows. The AAVE token rebounded from the $80 low on April 20 to around $93.59, and the TVL decline trend also began to level off.

Dragonfly managing partner Haseeb Qureshi said, “AAVE may take on some bad debt, but it has enough net worth to repay it.” Defillama’s founder emphasized, “DeFi will be hit, but it won’t disappear,” and believes that through protocol treasuries and adjustments, affected protocols such as Aave can “fully recover.”

Frequently asked questions

Why can Spark grow against the trend while Aave bleeds?

Spark benefited from this incident mainly because some market participants view it as a relatively safe alternative platform, especially amid the backdrop of Aave facing bad-debt risk. The transfer of institutional funds also further attracted other watch-and-wait funds to follow, creating a positive inflow effect.

How much could Aave’s bad debt ultimately reach?

Based on Defillama’s founder’s two scenario analyses, Aave’s bad debt is expected to range from $17 million to $341 million, depending on how Arbitrum disposes of the seized ETH and how the rsETH loss is allocated. Aave DAO currently holds $181 million in assets and about $54 million in the WETH Umbrella Fund as a buffer.

How does the DeFi industry assess the impact of this incident on the broader ecosystem?

Mainstream views tend to see the incident as a process of DeFi “learning from failure.” Executives from Dragonfly and Defillama both said the protocols have enough capital and mechanisms to absorb losses. Arbitrum’s coordinated freezing actions are viewed as a positive governance precedent, and overall optimism has begun to emerge in the community.