Who should skip this?
Skip this briefing if you wanted a click-path to loop Pendle principal tokens, a live liquidation bot, or a claim that Morpho was hacked. Nothing was exploited. Losses sat on borrowers who had looped PT-reUSD with health factors around 1.03. Use the Morpho guide and the incidents hub for the checklist, not this recap, if you needed how-to steps.
- You wanted a first Aave supply tutorial
- You needed a live price or health-factor calculator
- You assumed this was a smart-contract bug you could patch by switching apps
Skip this briefing if you wanted a how-to that tells you to loop Pendle PT-reUSD on Morpho, or a headline that Morpho was hacked. It was not. On the morning of 25 August 2026 a wallet spent on the order of $320,000 buying YT-reUSD on Pendle. That mechanically pushed the paired PT-reUSD price down about 3%. Morpho’s oracle for those markets takes the lower of a 15-minute time-weighted average and a fixed accretion curve toward $1 at the 10 December 2026 maturity. The lower number printed. Highly levered loops sitting at health factors around 1.03 — less than a 3% buffer — were liquidated in 14 minutes.
Primary narrative sources for the tape are The Defiant at thedefiant.io/news/defi/pendle-oracle-move-liquidates-usd36-million and CoinDesk at www.coindesk.com/tech/2026/08/25/a-3-token-move-just-triggered-usd36-million-in-ethereum-defi-liquidations Dollar figures below are those desks’ Morpho API / PeckShield counts, not a live ticker.
What actually printed on Morpho?
- 33 liquidation events between 04:37:47 and 04:51:23 UTC on 25 August 2026 (The Defiant, Morpho API)
- About $36.14 million of debt repaid; CoinDesk rounded the cascade near $36.4 million
- USDC market about $35.19 million; USDT market about $956,000
- About 38.6 million principal tokens seized as collateral
- Zero realised bad debt on either market; Steakhouse said vault lenders were not left short
- Largest liquidated positions on The Defiant’s count: about $13.01 million, $11.01 million and $6.83 million
- One liquidator contract handled about 96% of repayments in that window
Why a 3% move was enough
The Pendle reUSD pool maturing 10 December 2026 was the pricing surface. The Defiant put that pool near $8.97 million of liquidity against a Morpho market that, before the cascade, held on the order of $67.5 million of PT-reUSD collateral against $52.2 million of borrows at a 91.5% liquidation loan-to-value. A lending book several times the size of the pool that prices the collateral is the recurring failure mode in looped yield tokens — not a novel bug.
Borrowers were not holding spot reUSD. They had deposited PT-reUSD, borrowed USDC (or USDT), bought more PT-reUSD, and repeated. Each loop raises advertised carry and shrinks the buffer. At a health factor near 1.03, a 2.6–3% oracle print is a liquidation, not a dip you can ‘wait out’.
On 17 August a Morpho forum comment (user SrAugust, cited by The Defiant) already flagged the mismatch — tens of millions borrowed against a few million of apparent liquid depth. That is a design warning, not a prediction we claim as our own.
Oracle-as-designed versus ‘manipulation’
Pendle said the dual-reference oracle was configured correctly and functioned as intended. Steakhouse Financial, curator of the Morpho markets that take PT-reUSD, said the same: no smart-contract exploit, underlying reUSD did not depeg, lenders in its vaults were made whole. CoinDesk reported Steakhouse briefly pulled liquidity from the affected markets while it reviewed, then began restoring it.
Andrew Hong (Herd) publicly called the same tape oracle manipulation and estimated on the order of $1.3 million of liquidator profit, concentrated in one contract. That is a researcher’s characterisation of a thin-pool TWAP, not a court finding and not a Morpho bug report. This desk reports both frames: the contracts did what the parameters said; looped borrowers with a sub-3% buffer should not have been surprised.
Morpho’s liquidation incentive at ~91.5% LLTV is a few percent, not a 15% fire-sale. Isolation still did its job: the cascade stayed in those PT markets. Isolation does not protect the wallet that chose that market.
Who should skip PT collateral and looping?
Skip Pendle principal tokens as Morpho collateral if you cannot name the oracle (TWAP versus accretion), the LLTV, the curator, and the depth of the Pendle pool relative to the Morpho book. Skip looping if your health factor would sit under about 1.10 — a round number, not a promise. Skip this entire stack if you have never supplied on Aave without borrowing.
Spot reUSD, Curve Lend, and Resupply’s own CDP are adjacent products, not this Morpho market. Do not treat a PT loop as ‘the same as holding reUSD’. The June 2025 Resupply wstUSR incident was a different failure mode (new-market oracle / donation). This August event is leverage plus a short averaging window.
UK readers who looped and were liquidated may have a Capital Gains Tax disposal. That is a tax fact pattern, not advice — see our DeFi tax answer and Digital Assets UK’s lending-tax guide at digital-assets.co.uk/tax/crypto-lending-borrowing-uk/, not this news page.