DeFi is moving from a narrative of alternative finance back to a complementary financial position.

Written by: Liu Jiao Lian

On April 26, Arbitrum DAO voted to approve the proposal to release 30,765 frozen ETH. Along with the 30,000 ETH credit line promised by Mantle, the 25,000 ETH promised by Aave DAO, and other contributions large and small, DeFi United has raised over 100,000 ETH (approximately $230 million) for the KelpDAO incident.

Industry media is cheering: this is the largest coordinated rescue in DeFi history, proving the self-repairing capability of decentralized finance.[1]

Jiao Lian looked at this news but thought of the sentence written on April 19 - the real endpoint, after all, is returning to that most straightforward starting point: self-custody of one's own bitcoins.[2]

The rescue was successful, and then what?

This is not a review of the rescue operation; it is a reflection on what comes next.

1. What Are They Rescuing

First, let's take a look at what DeFi United actually did.

On April 18, KelpDAO's rsETH cross-chain bridge was attacked, resulting in approximately 116,500 rsETH being stolen, worth about $292 million. The attacker deposited the stolen rsETH into Aave as collateral to borrow WETH, triggering a bad debt panic. The utilization rate of WETH on Aave surged to 100%, locking the funds of countless innocent depositors.[1]

DeFi United is an emergency rescue mechanism spontaneously established by the community, without regulatory directives or central bank endorsement.

As of April 26, the rescue results were as follows:

• Mantle promised a credit line of 30,000 ETH for three years at a rate of Lido staking yield +1%, with Aave 5% income and AAVE token collateral attached.
• Aave DAO promised 25,000 ETH, pending governance vote.
• The Arbitrum security committee released 30,765 frozen ETH.
• Stani Kulechov (founder of Aave) personally pledged 5,000 ETH.
• Lido pledged 2,500 ETH, pending vote.
• LayerZero, Ethena, Ink, Frax, and others confirmed participation, with amounts to be disclosed.

Altogether, this has covered about 86% of the funding gap. The remaining gap is approximately 5,600 ETH (about $13 million).[1]

Listing these numbers is not for statistics. Jiao Lian wants to discuss the details.

Looking closely at Mantle's terms: the 30,000 ETH is a credit line, not a donation. It comes with Aave's 5% income and AAVE token collateral. This is a business negotiation, not charity.

Aave DAO's 25,000 ETH proposal is titled "No Ghost Left Behind," referencing the precedent of DAO bearing bad debts in the 2022 CRV incident. This is using historical legitimacy to endorse this rescue.[1]

The contribution amount from LayerZero still says Confirmed, TBD.

The rescue was rational, conditionally constrained, and driven by business negotiations. It was not an idealistic decentralized mutual aid. The industry can save itself; it is saving its own balance sheet.

2. If the Rescue Had Not Come

The above speaks to the situation where the rescue came. Conversely, think about this: what if DeFi United had not appeared?

Aave has a risk absorption mechanism called Umbrella. Its design logic is: when bad debts occur, they are absorbed in the following order.

First layer: Umbrella staking pool. Stakers of waWETH (users who stake WETH to provide insurance for the protocol) are automatically reduced. This pool is approximately $55 million.

Second layer: Aave DAO Treasury. The protocol treasury, approximately $85 million.

Third layer: WETH depositors. This is socialized loss - all users who deposit WETH on Aave share the remaining bad debts proportionally.[1]

According to calculations by LlamaRisk, in the worst case, the bad debt could be about $230 million. The first two layers together amount to about $140 million, leaving a gap of $90 million to be borne by WETH depositors.

Based on the then WETH pool with about $5 billion in deposits, each person storing ETH would face a forced deduction of about 1.8% from their aETH in the wallet.

You didn't touch rsETH. You don't know how KelpDAO's cross-chain bridge was configured. You did nothing wrong. But your money would be less.

What about those who deposited USDT and USDC? This time they would not face a principal deduction - because the bad debt only occurred in the WETH pool, Umbrella did asset isolation. But they would experience another thing: utilization reaching 100%, wanting to withdraw but unable to, with funds locked in the protocol, watching interest rates skyrocket to 15%.[1]

Jiao Lian was once locked during the Compound incident. At that time, Jiao Lian hadn't touched deUSD nor knew what xUSD was. But with xUSD de-pegged, deUSD also de-pegged, and Compound accepted deUSD as collateral, emergency halting withdrawals at five in the morning. Jiao Lian's funds were locked.[2]

That lesson is something Jiao Lian always remembers: you don’t need to directly touch bad apples; just being in the same pool will cause spillover effects.

So when Jiao Lian sees DeFi United's rescue of 100,000 ETH succeed, what Jiao Lian can see is that someone saved the day this time, but what Jiao Lian cannot see is, who will come next time?

3. Three Structural Defects Exposed by the Rescue

What has been discussed above points to not just the issue of this incident, but rather deeper structural flaws in DeFi.

Defect 1: The rescue relies on temporary coordination, not a permanent mechanism.

DeFi United is an ad-hoc team. There is no contingency plan, no permanent funding pool, no automatically triggered rules. Mantle's 30,000 ETH comes with commercial terms, and LayerZero's contribution still says TBD. Who will take the lead when the next crisis occurs? What if several major players are unwilling to act?

This is not to blame anyone for not doing enough. Under those circumstances, managing to pull together such a rescue mechanism is already commendable. But Jiao Lian is considering: is this model sustainable? Can it be relied upon?

Defect 2: The first line of defense for absorbing bad debts is money staked by others, not protocol profits.

The design logic of Umbrella is: let stakers bear the risk in exchange for returns. It sounds fair. But the problem is, waWETH stakers earn a yield of only around 6% annually while bearing a risk of being reduced by up to 100%. Have they calculated this risk-reward ratio themselves?

More critically, the so-called safety module essentially shifts the risk to ordinary users who chase returns. It is not that the protocol allocates bad debt reserves from profits, nor does it cover losses with income. It is one user footing the bill for another user.

Defect 3: Innocent participants in a shared pool can never self-insure.

This is what Jiao Lian is most concerned about. Even if Umbrella runs perfectly according to design, WETH depositors still face a 1.8% socialized loss. They have not touched rsETH, do not know what KelpDAO is, and just want to deposit ETH to earn interest. But when the loss arrives, they cannot escape.

This is not a bug in Umbrella; it is determined by the design of shared pool lending itself. To achieve composability and liquidity efficiency, one must bear this cost.

Worse still, these three defects cannot be truly repaired. Defect 1 may be partially improved (by establishing a permanent fund), but it requires coordination across the entire industry. Defect 2 is the essence of Umbrella's design and is not a bug that can be fixed. Defect 3 is the fundamental contradiction of shared pools; to fix it would mean giving up composability, which is the core value of DeFi.

4. The Industry's Dilemma: Capital Votes with Its Feet

The three defects have been discussed. Some may ask: what should DeFi do?

By April 2026, the total market value of USDT plus USDC exceeded $200 billion. The total scale of tokenized US Treasury securities surpassed $10 billion.[1]

When these two numbers are put together, Jiao Lian sees the market's choice.

Capital is flowing towards predictable, regulated, low-risk products. Stablecoins have the endorsement of issuers, and US Treasury bonds have the endorsement of the U.S. government. The returns are not high, but the certainty is high. Meanwhile, in DeFi, earning a 5-6% annual yield carries risks, including cross-chain bridge configuration errors, shared pool transmission, security module deductions, and uncertainty of DAO voting results.

Not everyone can accept this risk-reward ratio.

Jiao Lian is not saying DeFi should perish. DeFi will not die; it will still exist and continue to innovate. However, perhaps unfortunately, DeFi is moving from a narrative of alternative finance back to a complementary financial position.

In scenarios that require complex composability, DeFi is the optimal solution. For a large number of simple, certainty-seeking funds, stablecoins and US Treasury bonds are easier to use.

This is not a guess. Over $200 billion to several hundred billion, this has already happened.

So what should DeFi do? This question relies on industry builders finding an answer.

Perhaps what DeFi needs to do is not to add more functionalities or design more complex yield strategies, but to simplify. Isolate risks and allow users to only bear risks they choose. Increase transparency so that ordinary people can understand what a protocol is doing in five minutes. Make the rescue mechanism permanent, so that temporary arrangements and negotiations are not needed in every crisis.

Perhaps there are fundamentally no solutions. Perhaps DeFi's underlying logic - permissionless, composable, shared liquidity - is inherently contradictory to protecting innocent users. Who knows?

5. Two Paths, One Choice

For those who hold cryptocurrencies long term, there are likely two paths in the next five to ten years.

Path A: Complete self-custody, no yield generated. Holding native BTC in a wallet they control. Not relying on any protocol, not earning any interest. The cost is the sacrifice of liquidity and composability. The risk shifts from protocol to personal private key management.

Path B: Moderately participate in mainstream protocols and accept controllable risks. Using wrapped assets to participate in top protocols like Aave and Compound, earning yields of 3% to 8%. The risks include potentially facing the next rsETH event, possibly being deducted a point or two due to socialized losses, or having funds locked for days and nights.

The higher the aversion to risk losses, the more one tends toward Path A. Conversely, the higher the acceptance of risk and the stronger the pursuit of yields, the more one tends toward Path B.

For most people, perhaps they are distributed between A and B. They need yields and are willing to believe that mainstream protocols are safe enough.

Perhaps it need not be black or white. Perhaps a mixed approach could become a pragmatic middle ground:

60-70% core position: native BTC in cold storage, self-custodied, not participating in any DeFi. This is the bottom line, ensuring asset safety in the worst-case scenario.

30-40% satellite position: participating in mainstream DeFi, only engaging with top protocols like Spark, Aave, and Compound, only providing collateral like ETH or USDC. This portion of funds is willing to bear risk in pursuit of yields and liquidity. The yield from the satellite position compensates the opportunity cost of the core position not earning interest.

This scheme adds some flexibility compared to Path A and some sense of security compared to Path B.

In the long run, Path B's returns after deducting risk losses may not really outperform Path A in the end.

References:

[1] Cryptoslate, "DeFi lost $13B this month as the KelpDAO rescue shows both the best and worst of DeFi", Apr 26, 2026. [Link](https://cryptoslate.com/defi-lost-13b-this-month-as-the-kelpdao-rescue-shows-both-the-best-and-worst-of-defi/)

[2] Liu Jiao Lian, "The Lesson of $292 Million: Looking at DeFi Security from the rsETH Theft," April 19, 2026.