Kelp DAO's rsETH LayerZero bridge was attacked on April 18, resulting in the unauthorized minting of 116,500 rsETH (approximately $292 million). The attacker subsequently borrowed approximately $190 million worth of assets on Aave V3, and the Arbitrum Security Council froze 30,766 ETH involved at the chain level.
The DeFi United mechanism, led by Aave and centered on cross-protocol joint rescue efforts, was immediately established with the goal of filling the total gap of 112,200 rsETH (approximately $258 million).
As of April 24, multiple protocols have publicly stated their participation in DeFi United, with intentional commitments totaling approximately $100 million. However, most are still in the DAO voting stage, forum discussion, or earlier pre-discussion phases.
Disclosed Funding Parties and Terms
Golem Foundation and Golem Factory have collectively allocated 1,000 ETH (approximately $2.3 million) from their treasury to restore rsETH collateral support and the orderly disposal of affected users. Golem stated that they have been working closely with the Aave team, and this funding is part of DeFi United's cross-protocol coordinated response to risks.
Aave founder Stani Kulechov personally committed 5,000 ETH (approximately $11.5 million) to support DeFi United. Stani stated in a tweet that Aave is his life's work, the team is fully committed to finding the most beneficial solution for users, and is working with partners to secure more funding commitments. Personal funding does not go through the DAO voting process and is a direct commitment.
EtherFi Foundation authorized the deployment of up to 5,000 ETH (approximately $11.5 million) from the DAO treasury as EtherFi's share in the cross-protocol joint rescue, to absorb user losses and prevent bad debt in lending markets like Aave.
The terms are constrained to automatically stop the program after the rescue is completed, the 5,000 ETH upper limit is reached, or a subsequent governance vote terminates it; unused portions will be returned to the DAO treasury; if compensation is later obtained through legal proceedings, insurance payouts, or on-chain recovery, returned funds will be repaid to the DAO within the funding amount. The foundation committed to releasing the rescue framework and allocation mechanism details within 7 days after the proposal is passed.
Lido DAO proposed funding up to 2,500 stETH (approximately $5.75 million) to a dedicated rescue vehicle, bridged and transferred through a multi-signature wallet operated by Lido Labs Foundation. This allocation will only be released if the rescue fund is fully subscribed, sufficient to completely cover the rsETH gap.
Lido argued that if only part of the gap is covered, its EarnETH treasury depositors would still face a loss exposure of up to approximately 9,000 ETH, therefore refusing to unilaterally fund a partial rescue structure. The rescue vehicle's authorization is limited to filling the rsETH gap itself and not for supporting position health factors, secondary loss capital reset, and other second-order effects. This allocation is independent of the EGG 2026 budget, and unused or recovered funds will be returned to the Lido Aragon Agent. The proposal has not yet entered the Aragon on-chain voting stage.
Mantle drafted a proposal for Mantle Treasury to provide a loan of up to 30,000 ETH (approximately $69 million) to Aave DAO, with interest calculated at LIDO yield +1% APR, a maximum term of 36 months, and no penalty for early repayment.
The terms require Aave DAO to provide collateral of no less than $11 million worth of AAVE tokens plus 5% of future protocol revenue, with funds entering a multi-signature wallet where Mantle holds first-priority rights, while delegating voting rights for 130,000 AAVE tokens to Mantle. Mantle positioned this in the draft as a fixed-income tool priced at a crisis-period premium and hopes to promote Aave's native deployment on the Mantle network.
Several other protocols have publicly expressed participation but have not disclosed amounts, including Ethena, LayerZero, INK Foundation, etc.
Aave's Own Funds May Cover the Bad Debt Upper Limit
Based on current intentional commitments and the total gap, DeFi United's remaining gap is approximately $50 million. Even if external commitments stop at the current level, the Aave protocol itself has sufficient buffers. The Aave treasury holds approximately $180 million in assets, and the Umbrella security module holds approximately $56 million in insurance funds, totaling about $236 million, which already covers the upper limit of the bad debt range of $123.7 million to $230.1 million calculated by LlamaRisk.
Aave partially unfroze the previously emergency-frozen WETH market. The approximately $300 million borrowing squeeze that occurred in the early stages of the attack has largely eased, and the AAVE token has begun to stabilize after an initial decline of over 25%. From the perspectives of protocol self-rescue and market sentiment, the systemic risk of the event has moved away from the most critical stage.
However, Spark Strategy Lead monetsupply.eth criticized Aave's decision to unfreeze the core Ethereum WETH market in a tweet, calling it ill-considered and primarily beneficial to loop borrowers. He pointed out that since Aave's core Ethereum borrowing rate is capped at 5.15%, while LST/LRT offers higher returns under discounts and staking yields, arbitrageurs can achieve approximately 45% annualized returns through loop borrowing with up to 14x leverage.
This strategy would keep the WETH market utilization rate consistently at 100%, preventing ordinary aWETH holders from withdrawing collateral or refinancing high-cost debt positions to other markets. He further questioned whether the decision was more about public relations than sound risk management, actually worsening the situation for ordinary users trapped in the market over the past few days.
0xngmi's Three Paths: Vast Differences in Aave's Bad Debt Scale
The actual bad debt that Aave needs to absorb ultimately depends on KelpDAO's loss-sharing plan. DefiLlama founder 0xngmi conducted a complete numerical simulation of three possible paths, with each path having a huge impact difference on Aave.
Path One: All rsETH holders share the loss proportionally. KelpDAO imposes a uniform 18.5% reduction on all rsETH holders, a proportion corresponding to the share of the unauthorized minted 116,500 rsETH in the total circulation. Aave currently has approximately 666,000 rsETH as collateral across the network, with positions on the mainnet and L2 mostly in a state接近 maximum loop leverage (95% liquidation LTV).
Once socialized losses are implemented, the 18.5% reduction far exceeds the approximately 5% equity buffer of loop positions, and the equity of all positions on the mainnet and L2 will be completely wiped out, leading to large-scale liquidations. According to 0xngmi's calculations, the final bad debt沉淀 in Aave would be approximately $216 million, or 13.5% of the total collateral value.
The coverage structure for this bad debt is: the Umbrella insurance module covers approximately $55 million, the Aave treasury subsidizes approximately $85 million, leaving a gap of approximately $76 million. KelpDAO could cover the shortfall through borrowing or selling AAVE tokens held by the Aave treasury (current market value approximately $51 million). This plan shares the loss among all users, has a controllable impact on the Aave single market, but directly reduces the equity of rsETH holders.
Path Two: Abandon L2 holders, only protect the mainnet. KelpDAO chooses to repay only mainnet rsETH holders at the original value, treating rsETH on L2 as worthless assets. The scale of rsETH collateral on Aave's L2 is approximately $359 million (priced at the parity used by the oracle). If all positions are at maximum loop leverage, the bad debt would directly膨胀 to $341 million.
In this scenario, the Umbrella insurance module cannot cover any L2 bad debt. Aave can only use treasury or external borrowed funds to rescue部分 markets. The rsETH markets on the chains with the largest losses, Arbitrum, Mantle, and Base, are most likely to be abandoned and collapse. This plan has a relatively small direct impact on Aave's mainnet but would severely damage L2 ecosystem credibility and could trigger a trust contraction across the entire L2 restaking track. For Mantle, its own L2 rsETH market would directly崩盘 in this scenario, and the rationale for its 30,000 ETH loan proposal would also be反噬.
Path Three: Repay only original holders based on pre-attack snapshot. KelpDAO attempts to fully repay only rsETH holders before the attack based on a snapshot, with subsequent buyers or transferees bearing the losses themselves. Theoretically, this can compress the loss to the smallest range, but actual execution is extremely complex: funds have flowed extensively between DeFi protocols after the attack, and the nature of lending and liquidity pools is a mixed fund pool, making it impossible to truly distinguish between different batches of depositors.
In terms of specific bad debt calculation, the hacker borrowed $124 million on the Aave mainnet and $18 million on Arbitrum, totaling $142 million; after deducting Umbrella coverage, there is still a net loss of approximately $91 million. This plan has the smallest theoretical loss but is almost impossible to execute technically, and is also prone to long-term controversy legally and community-wise. 0xngmi judged its probability of implementation to be extremely low.
Among the three paths, Aave's net loss ranges from $76 million, $91 million, to $341 million, showing orders of magnitude differences. KelpDAO's choice of plan determines whether Aave ultimately needs to use all its treasury resources.
Rescue funds have gathered around Aave, but the key switch for the plan's success lies in KelpDAO's hands.
On the afternoon of April 23, KelpDAO issued a statement reiterating the core principle of always putting users first, stating that the team and partners have made substantial progress on multiple paths over the past four days, and publicly thanking the Arbitrum Security Council for the on-chain freeze and SEAL 911 for early intervention in the investigation. KelpDAO did not disclose specific plans, timelines, or loss-sharing details, only承诺 to continue sharing specific updates through official channels.
The current DeFi United rescue plan is gathering around Aave, and the structure of intentional commitments from various protocols implies Path One as the baseline scenario. If KelpDAO moves towards other paths, the narrative of Mantle's 30,000 ETH loan, the sharing比例 of EtherFi and Lido, and even the entire rescue framework would need to be recalculated.
