Crypto lending's rebound to $56 billion enlarges the pool exposed to the next Kelp DAO-style exploit

A $290 million hole in rsETH ^[3] was followed by a roughly $15 billion fall in Aave deposits within days ^[4]. That is about $52 of deposits gone for every dollar of unbacked collateral ^[15]. Aave's own contracts were never breached ^[4]. The Aave figure alone is larger than the $11.33 billion Galaxy counts leaving the entire sector over the second quarter ^[5][16], so either some of that money returned before June ended or the two numbers measure different pools.

The rebound figure has the same seam. Galaxy puts the quarter's outflow at $11.33 billion and the contraction at 16.78% ^[5][6]. If both describe one pool, lending started April near $67.5 billion and ended June near $56.2 billion ^[17], and a gain of more than 55% from there would put it near $87 billion today ^[18]. The $56 billion Cointelegraph reports comes from DeFiLlama ^[1]. Working back from it, the July starting point was about $36 billion at most ^[19]. The report does not say how much of the 55% is fresh deposits and how much is higher token prices.

That split decides how much risk came back with the money. "When a protocol accepts a token as collateral, it is also accepting that token's bridge, its verifier configuration, its oracle and its issuer's operational security," Stani Kulechov, chief executive of Aave Labs, told Cointelegraph's Magazine ^[7]. He said traditional reviews "missed the risk sitting in the bridges, verifier networks and other infrastructure an asset depends on" ^[14]. Aave now re-reviews every asset quarterly and "again after any material change" ^[8], and it has begun "an orderly wind-down" on six networks that missed its chain-level standards ^[9]. "No protocol can control the entire ecosystem, but it can control how much of that risk it takes on and how quickly it responds," Kulechov said ^[10].

Spark put a revenue test to the same question in January, months before the exploit. It began phasing rsETH out of SparkLend because "its low usage and revenue" did not justify the "additional risk" ^[11]. By its own account the fee stream was small, and giving it up meant Spark stopped carrying a bridge it did not run. "Every wrapper, bridge and oracle between the lender and the underlying asset is another place a loan can go wrong," said Thomas Wu, chief financial officer of Ledn ^[12].

If the 55% is mostly price, the same collateral is simply marked higher and each loan depends on the same bridges and oracles as before. Fresh deposits into wrapped and bridged tokens would mean the pool a single bridge failure can reach has grown, the "honeypot" Cointelegraph describes ^[2]. The third possibility is that the pruning holds and the next failure stays inside a capped market. Spark chief executive Sam MacPherson set that standard: "Preventing losses is only part of the challenge. Protocols also need to demonstrate how a loss would be contained if something does go wrong." ^[13]

In my view the problem Kulechov described is still there. Every change on the record either caps exposure or speeds the response ^{[8][9][10][11]}, and none takes a token's bridge or oracle out of the loan. The counter-case is that caps are enough, and that a capped rsETH market would not have set off a $15 billion run ^[4]. One figure would prove this view wrong: the share of the $56 billion posted as bridged or wrapped tokens. If that share falls as Aave's quarterly reviews proceed, exposure has shrunk along with the asset lists.