The foundational promise of on-chain lending was structural elegance: smart contracts would enforce collateral ratios automatically, removing the human discretion that blew up traditional credit cycles. That logic held — right up until the moment it didn’t. Between the Celsius implosion, the Euler Finance exploit, and the cascading liquidations that followed the Terra/LUNA collapse, the period from mid-2022 through 2024 produced a forensic record of where the design assumptions broke.

Collateral Quality Is Not Collateral Safety

The overcollateralization model assumes the collateral asset maintains sufficient liquidity to absorb a forced sale. What the 2022 cycles demonstrated is that highly correlated collateral — particularly tokens whose value was partially a function of the same ecosystem activity driving borrowing demand — compressed to near zero precisely when liquidation engines needed to execute.

LUNA was the sharpest example: borrowers using LUNA as collateral to borrow UST were caught in a reflexive loop where falling LUNA prices triggered liquidations, which increased LUNA supply, which accelerated the price decline. No collateral ratio, however conservative on paper, survives an asset losing 99 percent of its value in 72 hours. The structural lesson is that collateral quality must be assessed as a function of liquidity depth and correlation risk, not just current market value.

Liquidation Mechanisms Fail Under Stress Conditions

On-chain liquidation depends on solvent third-party liquidators and functioning gas markets. During peak volatility events in 2022, both conditions failed simultaneously. Ethereum network congestion during the LUNA collapse pushed gas prices high enough that small liquidation positions became economically irrational for bots to execute. Positions that should have been unwound at 120 percent collateralization instead held until they were deeply insolvent.

The Euler Finance exploit in March 2023 added a different dimension: flash loan mechanics allowed an attacker to manipulate collateral valuations within a single transaction, draining roughly $197 million before any liquidation logic could respond. The exploit didn’t defeat Euler’s collateral requirements — it circumvented the latency assumptions baked into how those requirements were priced. Protocols that survived this period generally had circuit breakers, oracle redundancy, or isolated lending pools that contained contagion.

Counterparty Risk Migrated, It Didn’t Disappear

A less-discussed structural observation: several on-chain protocols remained solvent while the centralized entities layered on top of them failed. Celsius, BlockFi, and Voyager were not pure on-chain lenders — they were centralized intermediaries who used on-chain infrastructure selectively while retaining discretionary control over customer funds. The counterparty risk that DeFi was designed to eliminate was reintroduced through the custody and yield-generation layer.

• Celsius used customer deposits in on-chain strategies, then gated withdrawals when liquidity stress hit — a classic bank-run dynamic with no lender of last resort.
• BlockFi’s exposure to FTX-related collateral illustrated how off-chain counterparty decisions can invalidate on-chain collateral structures entirely.
• The protocols themselves — Aave, Compound, MakerDAO — processed the volatility with losses that were large but bounded by their design constraints.

The distinction matters for anyone structuring exposure to this category. The risk profile of a permissioned on-chain lending protocol is structurally different from a centralized intermediary that markets itself using DeFi language.

The Operator Read

What the 2022–2024 period produced is not an indictment of on-chain credit mechanics — it is a map of where the design tolerances are. Protocols with oracle redundancy, isolated collateral pools, and no dependency on circular token ecosystems demonstrated measurable resilience. The operators and allocators paying attention to this space now are focused less on yield rates and more on liquidation architecture, collateral correlation matrices, and the precise boundary between on-chain enforcement and off-chain custody. Those structural distinctions — not marketing narratives — are where the due diligence sits.