After successfully rescuing 100k ETH

On April 26th, Arbitrum DAO approved a proposal to release 30,765 frozen ETH. Coupled with Mantle’s pledged credit line of 30k ETH, Aave DAO’s commitment of 25k ETH, and other contributions of various sizes, DeFi United has collectively raised over 100k ETH for the KelpDAO incident (worth approximately $230 million at market prices).

Industry media cheered: this is the largest coordinated rescue in DeFi history, a testament to the self-healing ability of decentralized finance.

Seeing this news, the blockchain community’s mind was actually on the sentence written on April 19th — the true endpoint still returns to that simplest starting point: holding your own Bitcoin.

The rescue was successful, so what’s next?

What follows is not a review of the rescue operation, but a reflection on what comes after.

1. What did they rescue?

First, let’s look at what DeFi United actually did.

On April 18th, KelpDAO’s rsETH cross-chain bridge was attacked, with about 116,500 rsETH stolen, worth roughly $292 million. The attacker deposited the stolen rsETH into Aave as collateral, borrowed WETH, and triggered a bad debt panic. The utilization rate of WETH on Aave skyrocketed to 100%, locking the funds of countless innocent depositors.

DeFi United is a community-initiated emergency rescue mechanism, without regulatory directives or central bank backing.

As of April 26th, the rescue results are as follows:

• Mantle pledged a credit line of 30k ETH, three-year term, with an interest rate of Lido staking yield +1%, plus Aave 5% income and AAVE token collateral

• Aave DAO committed 25k ETH, pending governance vote

• Arbitrum Security Council 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, amounts to be disclosed

Adding these up, they cover approximately 86% of the funding gap. The remaining gap is about 5,600 ETH (roughly $13 million).

Listing these figures isn’t for statistical purposes. The blockchain community wants to emphasize the details.

Careful reading of Mantle’s terms: 30k ETH is a credit line, not a donation. It includes Aave 5% income and AAVE token collateral. This is a business negotiation, not charity.

Aave DAO’s 25k ETH, titled “No Ghost Left Behind,” cites the precedent of DAO bearing bad debt in the 2022 CRV incident. This is using historical legitimacy to endorse this rescue effort.

LayerZero’s contribution amount is still marked as Confirmed, TBD.

The rescue is rational, conditional, and driven by business negotiations. It’s not idealistic decentralized mutual aid. The industry’s self-rescue is about saving its own balance sheet.

2. What if the rescue hadn’t come?

The above describes a scenario where the rescue arrived. Now, consider: what if DeFi United hadn’t appeared?

Aave has a risk absorption mechanism called Umbrella. Its design logic is: when bad debt occurs, it is absorbed in the following order:

First layer: Umbrella staking pool. WaWETH stakers (users who stake WETH to insure the protocol) are automatically reduced. This pool is about $55 million.

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

Third layer: WETH depositors. This is socialized loss — all users who deposit WETH on Aave share the remaining bad debt proportionally.

According to LlamaRisk’s calculations, in the worst case, bad debt could reach about $230 million. The first two layers combined cover roughly $140 million, leaving a shortfall of $90 million, borne by WETH depositors.

With the WETH pool’s deposits at about $5 billion at that time, each ETH depositor’s wallet’s aETH would be forcibly reduced by about 1.8%.

You’ve never touched rsETH. You don’t know how KelpDAO’s cross-chain bridge was configured. You did nothing wrong. But your money would be less.

And what about those holding USDT and USDC? They won’t have their principal reduced this time — because bad debt only occurred in the WETH pool, and Umbrella has asset isolation. But they will experience another issue: utilization hitting 100%, unable to withdraw, funds locked in the protocol, watching interest rates soar to 15%.

The blockchain community experienced a lock-up during the Compound incident. Back then, they didn’t touch deUSD, nor did they understand xUSD. But xUSD de-pegged, dragging deUSD with it. Compound accepted deUSD as collateral, and in the early morning, paused withdrawals urgently. The community’s funds were locked.

That lesson stuck: you don’t need to directly interact with the bad apples; just being in the same pool means you’re affected.

So, when the blockchain community sees the successful rescue of 100k ETH by DeFi United, what they see is that someone saved the day this time. But what they don’t see is: who will come next?

3. Three structural flaws exposed by the rescue

What’s being discussed here points not just to this incident, but to deeper, structural flaws in DeFi.

Flaw 1: Rescue relies on temporary coordination, not a permanent mechanism.

DeFi United was a temporary team assembled on the spot. No pre-planned procedures, no permanent fund pools, no automatic trigger rules. Mantle’s 30k ETH credit line comes with commercial terms, LayerZero’s contribution is still TBD. When the next crisis hits, who will lead? If major players are unwilling to act, what then?

This isn’t blaming anyone for not doing enough. It’s already remarkable that such a rescue mechanism could be assembled at all. But the question is: is this model sustainable? Can it be relied upon?

Flaw 2: The first line of bad debt absorption is the staked funds of others, not the protocol’s profits.

Umbrella’s design logic is: let stakers bear the risk in exchange for rewards. It sounds fair. But the problem is, waWETH stakers earn about 6% annualized, but risk being cut by up to 100%. Have they calculated this risk-reward ratio?

More critically, the so-called safety module essentially shifts risk onto ordinary users chasing yields. It’s not the protocol setting aside bad debt reserves from profits, nor using income to cover losses. It’s one user paying for another.

Flaw 3: Innocent participants in shared pools can never truly protect themselves.

This is the most concerning to the blockchain community. Even if Umbrella functions perfectly, WETH depositors still face a 1.8% socialized loss. They didn’t touch rsETH, don’t know what KelpDAO is, just want to deposit ETH to earn interest. But when losses come, they can’t escape.

This isn’t a bug in Umbrella; it’s inherent in the design of shared pools and lending. To achieve composability and liquidity efficiency, this cost must be borne.

Worse, these three flaws can’t be truly fixed. Flaw 1 might be partially improved (by establishing a permanent fund), but that requires industry-wide coordination. Flaw 2 is fundamental to Umbrella’s design — not a bug, but a feature. Flaw 3 is a core contradiction of shared pools: fixing it would mean abandoning composability, which is the core value of DeFi.

4. The industry’s dilemma: funds vote with their feet

Having discussed these three flaws, someone might ask: what should DeFi do?

As of April 2026, the combined market cap of USDT and USDC surpassed $30k. The total issuance of tokenized US Treasuries exceeded $25k.

These figures reflect the community’s choice.

Funds are flowing into predictable, regulated, low-risk products. Stablecoins are backed by issuers, US Treasuries by the US government. Yields are low but certainty is high. In DeFi, earning 5-6% annualized involves risks like cross-chain bridge misconfigurations, contagion in shared pools, security module reductions, and uncertain DAO votes.

Not everyone can accept this risk-reward ratio.

This isn’t to say DeFi is doomed. It won’t disappear; it will still exist and innovate. But perhaps, unfortunately, DeFi is shifting from a narrative of alternative finance to a supplementary financial layer.

Scenes requiring complex composability are where DeFi is most effective. For large, simple, and certainty-seeking funds, stablecoins and US Treasuries are more practical.

This isn’t speculation. The $100k versus a few hundred billion is already a reality.

So, what should DeFi do? That depends on industry builders finding an answer.

Perhaps DeFi’s task isn’t to add more features or design more complex yield strategies, but to do less. Isolate risks so users only bear the risks they choose. Increase transparency so ordinary users can understand what a protocol does within five minutes. Make rescue mechanisms permanent, rather than assembling ad hoc during crises.

Maybe a fundamental solution doesn’t exist. Maybe the core logic of DeFi — permissionless, composable, shared liquidity — is inherently at odds with protecting innocent users. Who knows?

5. Two paths, one choice

Long-term crypto holders probably see two paths over the next five to ten years.

Path A: Fully self-custody, no yield. Hold native BTC in wallets you control. Don’t rely on any protocol, don’t earn interest. The cost is giving up liquidity and composability. Risk shifts from protocols to private key management.

Path B: Moderate participation in mainstream protocols, accepting controllable risks. Use wrapped assets to participate in top protocols like Aave and Compound, earning 3-8% yields. Risks include another rsETH-like event, socialized losses deducting a few points, or funds being locked for days or nights.

The more risk-averse, the more likely they are to choose Path A. Those more willing to accept risk and seek higher yields tend toward Path B.

For most, it’s probably a mix between the two. They want some returns and believe mainstream protocols are sufficiently safe.

Perhaps, a hybrid approach is the most pragmatic middle ground:

• 60-70% core holdings: Native BTC, stored in cold wallets, not participating in any DeFi. This is the baseline, ensuring assets are safe even in worst-case scenarios.

• 30-40% satellite holdings: Engage with top DeFi protocols like Spark, Aave, and Compound, using ETH, USDC as collateral. These funds accept risks for higher yields and liquidity. The returns from this portion help offset the opportunity cost of the core holdings earning nothing.

This approach offers more flexibility than Path A and more security than Path B.

In the long run, after accounting for risks, the returns of Path B might not actually outperform Path A. Who knows?