The restaking sector — $15.3 billion in TVL on EigenLayer alone, plus $3.6 billion on Babylon and $897 million on Symbiotic — entered April 2026 as the fastest-growing primitive in decentralized finance. It exited the month's third week with $13 billion in aggregate DeFi TVL erased, a $292 millio...
"rsETH was configured with a single verifier: the LayerZero Labs DVN. No second DVN had to agree." — Chainalysis, Inside the KelpDAO Bridge Exploit (April 2026)
The restaking sector — $15.3 billion in TVL on EigenLayer alone, plus $3.6 billion on Babylon and $897 million on Symbiotic — entered April 2026 as the fastest-growing primitive in decentralized finance. It exited the month's third week with $13 billion in aggregate DeFi TVL erased, a $292 million exploit at Kelp DAO, up to $230 million in bad debt at Aave, and an emergency industry bailout fund scrambling to raise 100,000 ETH.
The Kelp DAO bridge exploit on April 18 did not reveal a new class of vulnerability. It confirmed one that researchers had been modeling for over a year: liquid restaking tokens (LRTs) create recursive collateral loops across lending protocols, and a single point of failure in bridge infrastructure can trigger cascading liquidations across the entire DeFi stack. An arxiv paper published weeks before the exploit — "Financial Dynamics and Interconnected Risk of Liquid Restaking" (arXiv:2604.03274) — had mapped precisely this contagion pathway.
This report compares the security architectures, risk profiles, and economic models of EigenLayer, Babylon, Symbiotic, and the liquid restaking token layer (Ether.fi, Kelp, Renzo, Puffer) to assess where systemic risk concentrates and what structural changes are now underway.
The restaking market as of early April 2026, prior to the Kelp exploit, was dominated by three base-layer protocols and a liquid restaking layer built on top of them.
Base-Layer Restaking Protocols:
| Protocol | TVL | Market Share | Asset | Operators | |---|---|---|---|---| | EigenLayer | $15.3B (4.36M ETH) | 93.9% | ETH | 1,900+ | | Symbiotic | $897M (256K ETH) | 5.5% | ETH | ~200 | | Karak | $102M (29K ETH) | 0.6% | Multi-asset | ~50 |
Bitcoin Restaking:
| Protocol | TVL | Asset | Status | |---|---|---|---| | Babylon | $3.6B | BTC | Operational, 91,332 BTC staked |
Liquid Restaking Token (LRT) Layer:
| Protocol | TVL | Token | LRT Market Share | |---|---|---|---| | Ether.fi | $5.6B–$9.2B | eETH | 50–74% | | Kelp DAO | ~$1.6B (pre-exploit) | rsETH | ~16% | | Renzo | ~$800M | ezETH | ~8% | | Puffer | ~$400M | pufETH | ~4% |
EigenLayer's TVL trajectory tells its own story. After slashing went live on April 17, 2025, TVL slid from peak levels toward $7 billion as yield farmers exited. By March 2026, it had recovered to $8.9–$18.6 billion (sources vary based on measurement methodology), reflecting what Fensory described as "the honest number — users who actually want the product." The protocol now secures over 20 Actively Validated Services (AVSs), with EigenDA as the largest consumer of restaked security.
Babylon operates in a structurally different market. Its $3.6 billion in TVL represents 91,332 BTC (0.46% of total Bitcoin supply), though this figure dropped approximately 74% from its early 2026 peak according to BingX data. Babylon's model — providing cryptoeconomic security to proof-of-stake chains using Bitcoin as collateral — avoids Ethereum's LRT composability risks but introduces its own set of bridge dependencies.
On April 18, 2026, at approximately 14:30 UTC, attackers drained 116,500 rsETH ($292 million) from Kelp DAO's LayerZero-powered bridge. The attack was not a smart contract exploit. It was an off-chain infrastructure compromise.
Attack Sequence:
Immediate Contagion:
Attribution: LayerZero attributed the attack to North Korea's Lazarus Group, specifically the TraderTraitor subunit. Evidence cited includes pre-funding through Tornado Cash ten hours before the attack, self-destructing binaries on compromised infrastructure, and post-drain consolidation patterns consistent with documented DPRK operations. If the attribution holds, Lazarus Group drained more than $575 million from DeFi in 18 days during April 2026.
Recovery: Arbitrum's Network Security Council froze 30,766 ETH ($71 million) of attacker funds, recovering approximately 25% of stolen assets.
The exploit exposed structural differences in how restaking protocols handle security. Not all models carry the same risk profile.
EigenLayer (Ethereum): EigenLayer's core security model relies on economic slashing — validators who misbehave on AVSs lose staked ETH. Slashing went live in April 2025. The protocol itself does not operate bridges; LRT providers like Kelp, Ether.fi, and Renzo build their own cross-chain infrastructure on top. EigenLayer's systemic risk is concentrated in two areas: correlated slashing events (multiple AVSs failing simultaneously, a scenario Vitalik Buterin has warned about as "cascading slashing") and the downstream composability of LRTs built on its deposits.
Babylon (Bitcoin): Babylon's model keeps Bitcoin on the Bitcoin network using cryptographic time-locks rather than bridges. Validators post BTC bonds that can be slashed if they equivocate on connected PoS chains. This avoids Ethereum-style LRT composability risks — there is no widely used liquid restaking token for Babylon BTC that circulates through DeFi lending markets. The trade-off: lower capital efficiency and limited DeFi integration. Babylon's primary risk is bridge-related if users move derivative tokens cross-chain, but the base protocol does not require it.
Symbiotic (Ethereum): Symbiotic differentiates through permissionless vault design — any ERC-20 token can be used as restaking collateral, and security parameters are set per-vault rather than protocol-wide. This distributes risk across isolated vaults rather than pooling it. Symbiotic's $897 million TVL is too small to create systemic DeFi exposure at present, though that changes if it scales.
Karak (Multi-chain): Karak accepts multiple asset types and operates across chains. At $102 million TVL, it is not systemically relevant. Its differentiation is asset diversity — accepting stablecoins and other tokens as restaking collateral — which introduces different risk profiles than ETH-only systems.
The core finding: the vulnerability that caused the Kelp exploit was not in the restaking layer itself. It was in the bridge infrastructure connecting LRTs to multiple chains. EigenLayer's base protocol functioned as designed. The failure was one layer up, where Kelp's bridge relied on a single verification point to secure $292 million in cross-chain assets.
The deeper systemic issue is not any single exploit. It is the recursive collateral structure that LRTs create across DeFi.
The loop works as follows:
This "leverage loop" amplifies yield but creates a chain of dependencies. According to a Bank of Canada staff analytical paper published in April 2026, "leveraged staking can amplify the risk of cascading liquidations by introducing heightened selling pressure from liquidations, posing systemic risks to the broader ecosystem."
The arxiv paper "Financial Dynamics and Interconnected Risk of Liquid Restaking" (arXiv:2604.03274, March 2026) found that revenue dynamics of liquid restaking protocols are "primarily predicted by the value locked in the underlying EigenLayer ecosystem." The paper concluded that "multi-blockchain expansion of the liquid restaking token presents a double-edged sword: bridging to other networks is crucial for user adoption, but it adds the bridge risks to the existing risks of restaking."
In the Kelp case, the 116,500 unbacked rsETH minted by the attacker represented approximately 18% of all circulating rsETH. When these tokens were deposited into Aave as collateral, the protocol's automated systems treated them as legitimate — because on-chain, they were indistinguishable from legitimately minted rsETH. The result: $190 million in borrowing against worthless collateral, and potential bad debt of $123–$230 million depending on how the shortfall is allocated.
Aave's incident report found the protocol "functioned as designed." The problem was not in Aave's code. It was that Aave had no mechanism to distinguish between legitimate and fraudulently minted LRTs. The protocol's risk parameters — LTV ratios, liquidation thresholds — were calibrated for market risk, not for the risk that the underlying collateral might be entirely fabricated.
On April 23, Aave service providers launched "DeFi United," a cross-protocol fund targeting 100,000 ETH to restore rsETH backing. As of April 25, the fund had gathered 69,642 ETH (~$161 million) from 14 contributors.
Major contributors:
| Contributor | Amount | Structure | |---|---|---| | Mantle | 30,000 ETH | Low-interest loan | | Aave DAO | 25,000 ETH | Treasury (Snapshot vote pending) | | Stani Kulechov (Aave founder) | 5,000 ETH | Personal pledge | | Ether.fi | 5,000 ETH | Direct contribution | | Lido | 2,500 stETH | Direct contribution | | Other contributors (9) | ~2,142 ETH | Various |
The structure of contributions is notable. Mantle's 30,000 ETH is a loan, not a donation — the largest contributor has a repayment expectation. Aave DAO's 25,000 ETH requires governance approval. Kulechov's personal 5,000 ETH pledge — he called Aave his "life's work" — signals the severity of the situation from the protocol's own founder.
This is, effectively, the first coordinated industry-wide bailout in DeFi history at this scale. It raises questions about moral hazard: if restaking protocols know that a sufficiently large failure will trigger an industry rescue, it may reduce incentives for conservative bridge architecture.
The academic literature predicted the structural vulnerabilities that materialized in April 2026 with notable precision.
Pre-exploit warnings:
Market response to warnings: TVL continued growing. EigenLayer's TVL recovered from the post-slashing-launch dip of $7 billion to $15.3 billion. Ether.fi's cumulative staking revenue exceeded $400 million. LRT adoption in lending protocols expanded. The yield premium of 8–12% APY on restaked positions versus 3–4% on standard staking continued to attract capital.
The gap between academic risk assessment and market behavior is not unique to restaking — it is a recurring pattern in financial markets. What is notable is the speed at which theoretical risk became realized loss: weeks, not years, separated the published research from the $292 million exploit.
Restaking TVL concentration creates systemic exposure. EigenLayer holds 93.9% of base restaking market share. A failure at the EigenLayer level — distinct from the Kelp bridge failure — would affect $15.3 billion in assets with downstream exposure through LRTs in lending protocols.
LRT bridge infrastructure is the weakest link. The Kelp exploit targeted a 1-of-1 DVN bridge configuration, not the restaking protocol itself. Bridge security standards across the LRT sector remain inconsistent.
Recursive collateral loops amplify contagion. The LRT-to-lending-to-restaking loop creates leverage that accelerates losses. A single exploit generated $13 billion in DeFi TVL drawdowns — a 44x multiplier from the initial $292 million theft.
Babylon's architecture avoids Ethereum LRT risks. By keeping Bitcoin on-chain via cryptographic time-locks and avoiding deep DeFi composability, Babylon's $3.6 billion in restaked BTC does not create the same recursive collateral exposure.
The "DeFi United" bailout sets a precedent. The 100,000 ETH recovery fund, led by Mantle (loan) and Aave DAO (treasury), establishes that large DeFi failures trigger coordinated industry responses — with potential moral hazard implications.
Academic literature anticipated the failure mode. Research published weeks before the exploit mapped the contagion pathway from LRT bridge failure to lending protocol bad debt to ecosystem-wide TVL drawdown.
The restaking sector's value proposition — higher yields through security-as-a-service — remains economically rational. EigenLayer's 20+ AVSs represent real demand for shared security. Babylon's model proves Bitcoin can participate in proof-of-stake security without the composability risks of Ethereum's LRT layer.
The structural problem is not restaking itself. It is the infrastructure built on top of it: bridges with single-verifier configurations, lending protocols that cannot distinguish legitimate from fraudulent LRTs, and collateral loops that amplify a $292 million exploit into $13 billion in TVL losses.
The immediate response — DeFi United's $161 million in commitments, Arbitrum's $71 million asset freeze, Aave's market freezes — demonstrates that DeFi can coordinate crisis response. Whether that coordination produces lasting structural reform — multi-DVN bridge requirements, LRT provenance verification in lending protocols, leverage limits on restaked collateral — or merely resets the conditions for the next exploit remains the open question.
The data indicates that the restaking market is repricing risk in real time. Whether it is repricing enough is not yet clear.