A recent exploit of the KelpDAO protocol has triggered a substantial outflow of capital from the decentralized finance ecosystem. Over the past two days, the prominent DeFi lending platform Aave has seen a decline of $8.45 billion in deposits, contributing to a broader decrease of $13.21 billion in total value locked across DeFi.
The total value locked, which represents the combined dollar value of crypto assets deposited across DeFi protocols, dropped from $99.497 billion to $86.286 billion. Aave's total value locked decreased by $8.45 billion to $17.947 billion during the same period, according to data from DefiLlama.
Protocol-level data reveals double-digit percentage declines across various platforms, including Euler, Sentora, and Aave, with the majority of losses concentrated in lending, restaking, and yield strategies tied to the affected collateral. The incident began with a $292 million exploit of Kelp's bridge, allowing attackers to utilize stolen rsETH, a widely used liquid re-staking token in DeFi, as collateral to borrow funds on lending platforms.
As these stolen tokens lacked legitimate collateral backing, borrowing against them created potential shortfalls for lenders, similar to deceiving a traditional bank by depositing fake fiat and taking out loans against it. In response, protocols froze affected markets, while panicked users withdrew funds, resulting in a broad decline in total value locked. However, token prices have moved less sharply than deposits, with the AAVE token experiencing a 2.5% decline over 24 hours, while UNI and LINK have dropped less than 1% over the same period. According to Peter Chung, head of research at Presto Research, the incident highlights the risks associated with cross-chain infrastructure, particularly in verification systems used by bridges.
Initial analysis suggests that the issue may have originated in the verification layer rather than in the smart contracts themselves. Chung also noted that the episode demonstrates how interconnected DeFi protocols can transmit shocks beyond the initial point of failure, with withdrawal activity and market freezes extending to platforms without direct exposure to the exploit.