Lesson 3 — Depegs in depth: TerraUSD, USDC/SVB, and why one recovered

**TerraUSD (UST) — May 2022.** UST was the largest algorithmic stablecoin ever launched, peaking at roughly $18.7 billion of circulating supply in early May 2022. Its peg mechanism: 1 UST was always mintable by burning $1 worth of LUNA, and 1 UST was always redeemable for $1 of newly-minted LUNA. The Anchor Protocol on Terra offered ~19.5% APY on UST deposits, which had bootstrapped enormous demand. The system had no external collateral — the entire peg rested on LUNA's market value exceeding UST's market value and on the arbitrage incentive working in both directions.

The collapse mechanism: a series of large UST sales on May 7, 2022, dropped UST briefly to $0.985. The intended arbitrage response — buy cheap UST, redeem for $1 of LUNA, sell LUNA — was working, but the LUNA being minted was being dumped onto the market because the arbitrageurs had no reason to hold it. This selling pressure dropped LUNA's price. As LUNA dropped, it took more LUNA to absorb the same dollar of redeemed UST, which dropped LUNA further. By May 10 the death spiral was fully self-reinforcing: UST kept falling because more LUNA had to be minted to absorb redemptions; LUNA kept falling because the minted supply kept doubling. Anchor depositors trying to withdraw accelerated the run. By May 13, UST was below $0.10 and LUNA was effectively zero — a market cap collapse of approximately $40 billion across both tokens in less than a week. Nothing recovered. The Luna Foundation Guard's $3 billion of BTC reserves, deployed to defend the peg, were exhausted within hours and made no difference.

**The key insight on UST.** The death spiral wasn't a market accident — it was the system functioning exactly as designed under the specific conditions of stress. The mint/burn mechanism that maintained the peg in calm markets was the same mechanism that destroyed both tokens during the run. No amount of external defense could fix the structural feature, because the defense (buying UST to support the peg) was simply funding the LUNA inflation that was killing the protocol. The lesson generalises: every no-collateral algorithmic stablecoin launched at scale has eventually failed for the same structural reason, and the mechanism is visible in the design before any failure occurs.

**USDC and the SVB weekend — March 2023.** USDC is a fiat-backed stablecoin issued by Circle. Reserves are held in a mix of cash and short-term US Treasuries. On Friday March 10, 2023, Silicon Valley Bank — a US regional bank — was placed into FDIC receivership. Circle disclosed late that evening that approximately $3.3 billion of USDC reserves (about 8 percent of the total) had been held at SVB. Markets opened Saturday-morning trading on crypto exchanges with USDC selling off; by Saturday afternoon USDC was trading around $0.87 on most centralized venues.

The mechanism: secondary-market participants didn't know how much of the $3.3 billion was recoverable (FDIC insurance covers $250,000 per depositor; the rest was uninsured), and even if it was eventually recovered, the timeline could have been weeks or months. Holders who needed liquidity sold at a discount. Arbitrageurs, normally the peg-maintenance mechanism, were unwilling to buy at $0.95 because the worst-case outcome wasn't $0.92 — it was that Circle might survive but with a permanently impaired reserve, with knock-on effects on the issuer's other operations.

**The resolution.** Late Sunday March 12, the US Treasury, Federal Reserve, and FDIC announced that all SVB depositors would be made whole — insured and uninsured — beginning Monday morning. Circle confirmed that USDC was fully backed and that redemptions would process normally as banking resumed. By Monday afternoon March 13, USDC had returned to within a few basis points of $1.00. The depeg event lasted approximately 72 hours peak-to-trough.

**Why USDC recovered and UST didn't.** Different family, different mechanism, different external dependencies. USDC's failure mode was bank-side counterparty risk on a small fraction of the reserve, and the broader US banking system intervened to remove that risk within 72 hours. The asset backing was real; the temporary liquidity gap was the entire problem. UST's failure mode was structural mint/burn dynamics with no external collateral, and no possible external intervention could have addressed it. Defending the peg by buying UST simply funded LUNA dilution; defending LUNA by burning UST didn't address the underlying market cap mismatch. The same kind of policy response that saved USDC would have done nothing for UST.

**Other depeg events worth knowing.** Iron Finance / IRON (June 2021) — partially-collateralized algorithmic design, classic death spiral, around $2 billion of value destroyed in 24 hours. USDR (October 2023) — Real-Estate-collateralized 'stablecoin' that depegged when liquidity dried up because the real-estate collateral was illiquid and couldn't be sold to defend redemptions; tokens still trade around $0.50 today. NUSD, FEI (multiple events), DAI's near-miss on Black Thursday 2020 — each has its own mechanism worth understanding. The pattern is always: the specific failure mode for the design family was foreseeable from the structure.