Two and a half months after the Kelp DAO shock, Aave no longer looks like a protocol in crisis. But it does not look fully repaired either. On 18 Apr, an attacker drained roughly $292mn of unbacked rsETH from Kelp DAO's cross-chain bridge and used that collateral on Aave to borrow real assets, mainly wrapped Ether (WETH). The result was a sharp liquidity squeeze. Utilization across major pools hit 100%, meaning supplied assets had been fully borrowed and lenders could not withdraw on demand unless borrowers repaid or fresh liquidity entered the system.
Aave Rebuilds Function After Kelp Shock, but Not Its Old Depth
1 July 2026 - 12:30 CEST
Aave's emergency controls did what they were designed to do. The Guardian froze rsETH markets on V3 and V4 and cut the loan-to-value ratio to zero, limiting additional collateral risk. But the episode exposed a more uncomfortable issue: even when the protocol works as intended, lenders can still rush for the exit once confidence in immediate liquidity breaks down.
Panic faded but balance sheet still impaired
The picture since then is mixed. The panic phase has clearly passed. Ethereum V3 is no longer trapped at full utilization, borrowing rates have normalised, and the largest pools are functioning again. But Aave's balance sheet has not recovered. Total value locked has fallen from $41.7bn before the shock to $21.9bn today, a decline of 47.6%. Active loans have dropped almost as much, from $17.4bn to $9.36bn, down 46.2%. This was not simply a case of lenders pulling deposits while borrowers remained active. Both sides of the market shrank at roughly the same pace.
The loans-to-TVL ratio, however, is largely unchanged, moving from 41.7% to 42.8%. That suggests Aave has not become structurally weaker as a credit venue. Rather, it has become a smaller one. Activity has stabilised, but inside a much reduced balance sheet. While the system is working again, the depth that existed before the Kelp incident has not returned.
Chart
(Source: Token Terminal)
Core Ethereum V3 pools are functioning again
The Ethereum V3 market, where Aave's largest pools sit, now sits at $16.94bn supplied and $7.21bn borrowed. Core markets are active again, but liquidity remains uneven.
WETH, the largest asset pool, has $3.22bn supplied and $2.60bn borrowed, leaving utilization at 80.6%. That is no longer a bank-run state, but it is still tight for Aave's most important Ether market.
Stablecoins look functional, though not uniformly comfortable. USDT has $2.73bn supplied and $2.06bn borrowed, putting utilization at 75.4% and suggesting a meaningful buffer has returned. USDC is more fragile: $2.10bn supplied and $1.90bn borrowed, or 90.2% utilization. That is well below the April stress point, but still high enough that another wave of withdrawals would quickly become relevant.
The more telling weakness is in Ether derivatives and restaked collateral. weETH has $1.70bn supplied but only $88,300 borrowed, equal to roughly 0.01% utilization. rsETH has $703.2mn supplied and just $1,800 borrowed, effectively zero utilization. These are not functioning credit markets in any meaningful sense. They are pools where collateral remains, but borrowing demand has largely disappeared. Trust has not returned to these tokens, as the April event was specifically tied to restaked Ether collateral risk. The market appears willing to use Aave again for core assets such as WETH, USDT and USDC. It is far less willing to extend the same trust to every form of Ether-linked collateral.
Conclusion: functional, not fully trusted
Aave survived the Kelp shock. Its core pools are no longer frozen, borrow markets are operating, and the protocol did not suffer a direct smart-contract failure. But the recovery remains incomplete. The clearest reading is that Aave has regained functionality, not its former depth. TVL and active loans are still almost half below pre-shock levels, while restaked and Ether-derivative pools show little real borrowing activity.
April repriced liquidity risk. In DeFi lending, trust is not only about whether the code works, it is also about whether lenders believe they can get out when stress returns.
