The Terra/LUNA and UST Collapse of May 2022

Key Takeaways

• UST was never backed by dollars. Its peg depended on LUNA's market value staying well above UST's outstanding supply so that arbitrage could always absorb sellers; once that inequality flipped on the evening of May 9, 2022, there was no reserve to fall back on.
• The Luna Foundation Guard's much-publicized bitcoin reserve, about 80,300 BTC as of May 6, 2022 per a later forensic audit, was built to defend the peg but was spent down by roughly $2.5 billion in three days and still failed, while its liquidation added selling pressure to bitcoin's own price.
• The yield that pulled money into UST was itself unsustainable: Anchor Protocol paid a fixed 19.5 percent that its own lending revenue never covered, requiring repeated cash injections from Terraform Labs and the Luna Foundation Guard, and the Terra community had voted to start cutting that rate six days before the run began.
• The mechanism that was supposed to stabilize UST instead hyperinflated LUNA. Its circulating supply grew from under 1 billion tokens to more than 5.89 trillion between May 11 and May 13, 2022, while its price fell from roughly $50 to a fraction of a cent over the same three days.
• The SEC's fraud case ended in a unanimous jury verdict in under two hours of deliberation, a settlement exceeding $4.5 billion, and, separately, a 15-year federal prison sentence for Do Kwon, handed down in December 2025 for conduct a federal judge called "a fraud on an epic, generational scale."

What Was TerraUSD, and How Was the Peg Supposed to Hold?

TerraUSD was built on the Terra blockchain, a network Do Kwon and Terraform Labs had raised money from investors for since April 2018, according to the SEC's complaint. UST carried no dollar reserves at all. Instead, its dollar peg depended entirely on an arbitrage contract linking it to a second Terra-native token, LUNA. The rule was symmetric and simple to state: 1 UST could always be swapped for exactly $1 worth of LUNA, and $1 worth of LUNA could always be swapped for 1 UST. If UST traded below $1 on the open market, a holder could burn 1 UST for $1 of LUNA and sell the LUNA for a profit, which contracted UST's supply and pushed its price back toward $1. If UST traded above $1, the reverse trade expanded UST's supply and pushed the price back down. Every LUNA burned in this process permanently reduced LUNA's supply, which was also how LUNA's price appreciated during Terra's growth phase.

The mechanism has a structural requirement that is easy to state and easy to forget while it is working: LUNA's total market value has to stay significantly larger than UST's outstanding supply, because a UST holder redeeming into LUNA needs enough LUNA in the system to actually receive. Researchers at the Federal Reserve Board later described the relationship precisely: the market capitalization of LUNA had to exceed the market capitalization of UST by a significant margin to guarantee redemption while limiting relative-supply volatility. The weaker LUNA's own market, the less the system could tolerate a swing in demand for UST. That is not a defect that shows up gradually. It is a threshold, and on May 9, 2022, Terra crossed it.

UST's supply had grown enormously on the strength of this arrangement, reaching $18.77 billion in circulation by May 7, 2022, on data from DefiLlama. LUNA's own peak came a month earlier: its price hit an all-time high of $119.18 on April 5, 2022, before drifting down alongside a broader decline in cryptocurrency prices through April, a decline that also thinned the very cushion the UST peg depended on. See how stablecoins are supposed to work for the collateralized alternative UST deliberately rejected.

What Was Anchor Protocol, and Why Did It Hold $12 Billion in UST?

A stablecoin only matters if people hold it, and the single biggest reason to hold UST was Anchor Protocol, a Terra lending market that advertised a fixed deposit rate as high as 19.5 to 20 percent. The SEC's complaint states plainly that Terraform and Kwon "touted and marketed UST as a 'yield-bearing' stablecoin, which they advertised as paying as much as 20 percent interest through the Anchor Protocol." It worked: researchers reconstructing Terra's on-chain data found that more than 70 percent of all UST ever minted ended up deposited in Anchor, and that right before the crash Anchor alone held around $12 billion of it, with a further $3 billion sitting in a related community account.

The rate was never something Anchor's own business could support. Interest paid to depositors consistently outpaced the interest Anchor collected from borrowers, and the shortfall was covered by a separate "yield reserve" that Terraform Labs and the Luna Foundation Guard had to top up by hand: $71 million from Terraform in July 2021, and $510 million worth of UST from the Luna Foundation Guard in February 2022, both funded by swapping LUNA. By April 2022 the daily subsidy needed to keep the 19.5 percent rate running had reached $6 million, and the Terra community passed a governance proposal to begin cutting the rate to something more sustainable, effective May 1, 2022, six days before the run began. Holdings were also concentrated: on the eve of the collapse the top 10 addresses on Anchor controlled 15 to 20 percent of all deposits, and the top 1,000 controlled roughly 60 percent, meaning a relatively small number of large, sophisticated holders could move first and move the market.

None of this made UST fraudulent by itself; a subsidized yield is a choice, not automatically a lie. What it did was manufacture demand for a token that had no independent reason to be worth $1 beyond the arbitrage contract, and it meant that when the subsidy became unsustainable, the exit door and the front door were the same door. Compare this to how yield farming and liquidity mining actually generate (or fail to generate) return elsewhere in decentralized finance.

Who Was Do Kwon, and What Was the Luna Foundation Guard For?

Do Kwon was the public face and, per the SEC's later complaint, the controlling mind of Terraform Labs, the Singapore-based company behind Terra, UST, LUNA, and a family of related tokens including MIR "mirror" tokens designed to track the price of U.S. equities. In January 2022, Kwon announced the formation of the Luna Foundation Guard, a non-profit whose stated purpose was to build reserves supporting the UST peg against exactly the kind of coordinated selling that had already caused smaller price dislocations on Curve and other exchanges. LFG was overseen by a governing council that included several Terraform Labs co-founders, and it raised its first $1 billion by selling LUNA tokens, with the trading firms Jump Trading and Three Arrows Capital as lead investors, according to research published by MIT and London School of Economics economists.

The most consequential decision LFG made was to acquire bitcoin. "The LFG purchased $3 billion worth of BTC to 'support the stability and adoption of the UST stablecoin,'" Kwon said, as recorded in Federal Reserve Board research on the collapse. That decision changed what UST actually was. A currency board backed by an external, independently valued asset is a materially different design from a purely endogenous mint-and-burn arbitrage, and layering one on top of the other created a hybrid nobody had fully reasoned through: bitcoin's price now mattered to UST's stability, and UST's stability now mattered to bitcoin's price, through the reserve that connected them. A forensic audit conducted after the collapse by JS Held found that, as of May 6, 2022, the day before the run began, LFG held approximately 80,300 BTC along with about $26 million in USDT and $24 million in USDC.

What Happened the First Time UST Broke Its Peg, in May 2021?

UST had already depegged once before, a full year before the collapse, and how that episode was handled is central to the fraud case that followed. According to the U.S. Attorney's Office for the Southern District of New York, UST slipped below its $1 peg in May 2021, and Kwon told investors publicly that a computer algorithm known as the "Terra Protocol" had automatically restored the value. Prosecutors alleged that this was not what actually happened: instead, Kwon arranged for a high-frequency trading firm to secretly buy up large quantities of UST on the open market to push the price back to $1, and then let the public believe the algorithm alone had done the work.

Kwon eventually admitted to exactly this in his 2025 guilty plea. "I made false and misleading statements about why it regained its peg by failing to disclose a trading firm's role in restoring that peg," he told the court. "What I did was wrong." The significance of the episode is not that a stablecoin wobbled once and recovered; plenty of pegged assets have survived a scare. It is that the recovery investors were shown, and that gave the market confidence to keep pouring money into Anchor for another year, was not the mechanism it was represented to be. When the same mechanism was tested for real in May 2022, at a scale no trading firm's balance sheet could quietly absorb, there was no version of that fix available.

What Triggered the Run on May 7, 2022?

UST's deepest pool of external liquidity was not on the Terra blockchain at all. It sat in a Curve Finance "metapool" on Ethereum called UST-3pool, paired against USDC, USDT and DAI, which by 2022 had become the primary hub of stablecoin liquidity in decentralized finance. In April 2022, Terra and the stablecoin project FRAX had proposed and won a governance vote to migrate that liquidity into a new "4pool" that would replace DAI with FRAX and UST, integrating UST even more deeply into Curve.

On the morning of May 7, 2022, the Terra Foundation began withdrawing funds from the old UST-3pool to prepare for that migration. According to the blockchain analytics firm Chainalysis, the initial withdrawal was $150 million, and pulling that much liquidity out made the remaining pool "shallower" and more sensitive to large trades. Within the following hour, two traders swapped a combined 185 million UST for USDC in that thinned pool, and the pool's own pricing formula registered UST trading below its peg against the other three stablecoins for the first time. Independent on-chain research narrows the timing further: two large wallets, since alleged on social media to belong to Jane Street and Celsius, withdrew a combined 400 million UST from Anchor that same day, moving in batches between roughly 5 a.m. and 9:48 p.m. UTC to Binance, to Ethereum via the Wormhole bridge, and to Curve itself.

What followed was less a single trigger and more a coordination event happening in public, on a ledger everyone could watch in real time. As the withdrawals accumulated on-chain, other large Anchor depositors could see them and began exiting too, well before the broader public understood anything unusual was happening. Whether the initial $150 million withdrawal was itself an attack, an accident of bad timing around a routine migration, or simply the first mover in an already-fragile system remains debated; the Federal Reserve's own research notes a competing theory, that a trader who had built a bitcoin short worth potentially $4 billion alongside a $1 billion UST position may have anticipated LFG's coming bitcoin sales and profited from them, without concluding the theory is proven.

How Did the Peg Actually Break, Hour by Hour?

The clearest way to see the collapse is in UST's own market price, tracked day by day through DefiLlama's price feed for TerraClassicUSD (as UST is now tracked, sourced originally from CoinGecko).

Prices below are single daily snapshots, each taken close to the time shown; because UST moved sharply within single days during the run, a snapshot understates the true intraday range. Anchor and swap figures are from Liu, Makarov and Schoar (2023); the withdrawal and Curve sequence is from Badev and Watsky (2023), citing Chainalysis.

Total TFL and LFG spending trying to defend the peg between May 7 and May 9 came to roughly $2.5 billion, made up of 80,071 BTC, 26,281,671 USDT and 23,555,590 USDC, according to the post-collapse JS Held audit. Across the full week of the run, from May 7 through May 13, holders swapped 7.42 billion UST worth $4.65 billion through the native UST-LUNA exchange contract alone, on top of everything sold directly on exchanges. Terra's own on-chain total value locked, tracked by DefiLlama, tells the same story from a different angle: it peaked at $20.38 billion on May 5, 2022, had fallen to $848.7 million by May 13, and stood at $35.4 million by the end of the month, a decline of more than 99.8 percent from its peak in twenty-six days.

What Was the Death Spiral, and How Did LUNA's Supply Reach 6 Trillion?

Once UST traded meaningfully below $1, the mint-and-burn arbitrage gave holders two ways out: sell UST directly on the open market, or burn UST for $1 worth of freshly minted LUNA and sell that instead. The second route was, in theory, always more profitable, because it captured the full discount rather than whatever price a direct market sale could get. As the UST discount widened during the week of May 7, more holders took that route, and the Terra protocol did exactly what it was designed to do: it minted new LUNA to honor every redemption, without limit.

That is the mechanism, and it is also the flaw. Selling pressure that would normally have been absorbed by a large, liquid LUNA market instead diluted a LUNA supply that was shrinking in relative value at the same time sellers needed more of it. Every redemption pushed LUNA's price down, which meant the next redemption of the same dollar amount of UST required minting even more LUNA, which pushed the price down further still. Researchers who reconstructed the sequence from on-chain data found that LUNA's circulating supply grew from under 1 billion tokens on May 11, 2022, to 5.89 trillion by May 13, while its price fell from roughly $50 to about one-millionth of a dollar over the same three days. The Federal Reserve's independent analysis puts the multiple at "almost 20,000-fold." A token that traded above $100 five weeks earlier was, by design, worth a fraction of a cent, not because anyone chose to destroy it, but because the protocol's own rules required unlimited minting to keep honoring a promise it could no longer keep.

One further detail matters for anyone trying to draw a general lesson from this: this was not the first time LUNA's supply had moved by billions in a single event at Terraform Labs' own discretion. In November 2021, governance proposals introduced by Kwon directed that 88.675 million LUNA sitting in a community pool be swapped for UST, which pushed UST's own supply from 2.5 billion to 6.5 billion almost overnight and generated swap fees of 30 percent that were used to temporarily boost LUNA's staking yield. The network's supply mechanics were never purely mechanical market outcomes; they were also periodically steered by decisions made inside Terraform Labs, which is one reason the SEC would later argue the system was not the decentralized, autonomous mechanism it was marketed as.

Could the Luna Foundation Guard's Bitcoin Reserve Have Saved the Peg?

On paper, the Luna Foundation Guard's roughly 80,300 BTC reserve, worth several billion dollars at May 2022 prices, looked like exactly the kind of independent backstop UST's algorithmic design lacked. In practice, it was consumed almost immediately and still failed. Alongside its BTC, LFG and Terraform Labs spent essentially all of the $2.5 billion identified in the post-collapse audit within roughly 72 hours, buying UST directly on the market and attempting to slow the outflow from Anchor and the exchanges. The UST price barely responded; it went from roughly $1.00 on May 8 to about $0.80 by the following evening despite the spending, then swung as low as $0.43 intraday before continuing to fall over the following days.

Two structural problems explain why. First, the reserve was sized against ordinary volatility, not against a coordinated exit from a token with $18.77 billion in circulation; a few billion dollars of buying power cannot outlast selling pressure an order of magnitude larger once panic becomes self-reinforcing. Second, liquidating the reserve created a second casualty. Selling a meaningful share of 80,300 BTC on the open market in a matter of days put direct downward pressure on bitcoin's own price at a moment when broader crypto markets were already softening, which is part of why the Terra collapse is remembered as a market-wide event rather than a contained failure of one token. A reserve that is large relative to a calm market and small relative to a panicked one is not really a reserve in the sense depositors assumed; it is a delaying action, and delaying actions that fail still cost real assets on the way down. The parallel worth holding in mind is the broader 2022 rate-shock backdrop: the Federal Reserve had just delivered a 50-basis-point rate increase on May 4, 2022, three days before the run began, pushing the effective federal funds rate to 0.83 percent from near zero in March, part of a general tightening in risk appetite that LUNA's price had already been sliding against since its April 5 peak.

Who Made Money During the Collapse?

Losses were not evenly distributed, and neither were the exits. On-chain research that classified Anchor depositors by sophistication and speed found a consistent pattern: large, financially sophisticated wallets ran first and ran hardest, using multiple channels simultaneously, while smaller and less experienced holders were slower to react and in some cases actually bought UST on May 9 or 10, when the price briefly appeared to be stabilizing below $1, trying to "buy the dip" into what turned out to be the middle of a collapse. Those addresses fared particularly badly.

The single largest identified beneficiary of the native UST-to-LUNA swap mechanism during the run was Alameda Research, the trading firm co-founded by Sam Bankman-Fried and closely tied to the FTX exchange. Researchers found Alameda conducted the third-largest volume of UST-LUNA swaps of any address during the crisis, exploiting a structural advantage: swap fees and execution uncertainty on other exchanges discouraged most Anchor depositors from using the native swap contract as an exit, but Alameda's privileged access to FTX let it execute the arbitrage at a scale few other participants could match. The connection is not incidental. Alameda and FTX's own collapse followed just six months later, and the same names, and often the same balance-sheet fragility, run through both stories; see the FTX collapse of November 2022 for what happened when Alameda's own risk-taking eventually caught up with it.

How Did Terra's Collapse Spread Through the Rest of Crypto?

Terra's collapse did not stay contained to Terra. Three Arrows Capital, the Singapore-based crypto hedge fund, had been a lead investor in the Luna Foundation Guard's original $1 billion fundraise and held significant direct exposure to LUNA; the fund's own subsequent collapse in June 2022, one of the largest hedge fund failures in crypto history, followed within weeks of Terra's implosion and left a trail of counterparty losses across crypto lenders that had extended it credit. Federal Reserve researchers studying the episode found a broader mechanical channel too: "wrapped" versions of UST and LUNA had been bridged onto other blockchains for use in their own decentralized finance applications, and once the underlying assets became worthless, every application relying on those wrapped tokens lost real value simultaneously. Blockchains that shared a bridge with Terra at the time of its collapse were measurably more likely to lose market share afterward, and the researchers estimated the odds of a meaningful, time-bound loss of share rose by roughly 40 percent for each additional shared bridge, evidence that the risk did not stay confined to a single chain the way isolated failures sometimes do.

The wider effect is now generally described as the start of the 2022 "crypto winter": a cascade in which Terra's failure damaged Three Arrows Capital, whose failure damaged lenders including Celsius and Voyager Digital that had extended it credit, compounding a decline that had already begun with the Federal Reserve's tightening cycle earlier that year. It is worth being precise about what can and cannot be said here: Terra's collapse was a major accelerant and a direct source of large, identifiable losses for specific counterparties, but crypto asset prices broadly were already declining before May 2022, and attributing the entire 2022 downturn to Terra alone would overstate a single event's role in a much larger repricing.

What Did the SEC Charge Terraform Labs and Do Kwon With?

The SEC filed its complaint in the U.S. District Court for the Southern District of New York on February 16, 2023, charging Terraform Labs PTE Ltd and Do Hyeong Kwon with what it called "a multi-billion dollar crypto asset securities fraud." The complaint alleges that from April 2018 until the May 2022 collapse, Terraform and Kwon raised billions of dollars by offering an interconnected suite of crypto asset securities, many sold without registration: UST itself, LUNA, MIR "mirror" tokens designed to track U.S. equity prices, and "mAssets" structured as security-based swaps. Beyond the unregistered-offering claims, the SEC alleged straightforward misrepresentation on two fronts. It said Terraform and Kwon repeatedly and falsely claimed that a popular Korean mobile payment application used the Terra blockchain to settle transactions in a way that would accrue value to LUNA, and separately that they misled investors about the underlying stability of UST itself, most notably by concealing the trading firm's role in the May 2021 depeg recovery described above.

Gurbir S. Grewal, then Director of the SEC's Division of Enforcement, was unambiguous in the same press release about what the agency believed the case represented: "the Terraform ecosystem was neither decentralized, nor finance," he said. "It was simply a fraud propped up by a so-called algorithmic 'stablecoin': the price of which was controlled by the defendants, not any code." That framing, that a system marketed as autonomous code was in fact steered by discretionary human decisions, tracks directly onto the November 2021 supply interventions and the concealed May 2021 peg defense described earlier in this article.

How Did the SEC's Case Against Terraform and Kwon End?

The litigation moved in stages over roughly sixteen months. On December 28, 2023, the district court found Terraform and Kwon liable for offering and selling crypto asset securities in unregistered transactions, resolving that portion of the case before trial. Terraform Labs then filed a voluntary Chapter 11 bankruptcy petition in the U.S. Bankruptcy Court for the District of Delaware on January 21, 2024. The fraud claims went to a jury, and on April 5, 2024, after a nine-day trial, the jury unanimously found Terraform and Kwon liable for securities fraud following less than two hours of deliberation.

The settlement followed on June 13, 2024. Terraform agreed to pay $3,586,875,883 in disgorgement, $466,952,423 in prejudgment interest, and a $420,000,000 civil penalty, an obligation the SEC's own case-tracking page puts at $4,473,828,306 in total. Kwon separately agreed to pay $110,000,000 in disgorgement and $14,320,196 in prejudgment interest, on a joint-and-several basis with Terraform, plus his own $80,000,000 civil penalty. Two structural details matter more than the headline number. First, Terraform's payment obligation is explicitly "deemed satisfied" only through actual payments to harmed investors and creditors inside the bankruptcy case, and the SEC receives nothing until those investors and creditors are paid in full, meaning the settlement figure is a ceiling on what the agency can extract, not a guarantee of what victims will receive. Second, Terraform agreed to wind down its operations entirely, replace two of its board directors, and distribute its remaining assets through a court-supervised liquidation, ending the company as an operating entity rather than merely fining it.

What Happened to Do Kwon Personally?

Kwon left South Korea after the collapse and was arrested at Podgorica Airport in Montenegro on March 23, 2023, attempting to board a flight while carrying a forged Costa Rican passport; a search reportedly turned up additional forged Belgian identity documents as well. A Montenegrin court separately convicted him on the passport charge and sentenced him to four months in prison there, entirely apart from the fraud case. What followed was a lengthy, three-way extradition dispute among Montenegro, South Korea and the United States, both of which wanted Kwon for different sets of charges; Montenegro's Supreme Court nullified an earlier appellate ruling favoring South Korea, and the country's Minister of Justice ultimately signed off on extradition to the United States, a transfer that concluded on December 31, 2024.

U.S. prosecutors indicted Kwon on nine criminal counts covering securities fraud, wire fraud, commodities fraud and money laundering conspiracy. In August 2025 he pleaded guilty to two of them, conspiracy to commit commodities fraud, securities fraud and wire fraud, and a substantive wire fraud count, and agreed to forfeit more than $19 million in proceeds from his schemes. At his sentencing on December 11, 2025, U.S. District Judge Paul A. Engelmayer imposed 15 years in prison, above the roughly five years Kwon's own lawyers had requested (so that he could return to South Korea to face separate charges there) and below the at-least-12-years federal prosecutors had sought. "This was a fraud on an epic, generational scale," Engelmayer told him. "In the history of federal prosecutions, there are few frauds that have caused as much harm as you have, Mr. Kwon." Kwon apologized in court to the hundreds of victims who had submitted impact statements. As part of the plea agreement, prosecutors agreed not to oppose a future application by Kwon to be transferred abroad to serve the remainder of his sentence after he completes half of it.

Can Investors Who Held UST or LUNA Recover Any Money?

Whatever recovery exists runs through Terraform Labs' bankruptcy case, not directly through the SEC. On September 20, 2024, the U.S. Bankruptcy Court for the District of Delaware approved a liquidating Chapter 11 plan, establishing a trust to liquidate Terraform's remaining assets and pursue further claims for the benefit of creditors and harmed investors. Kwon's own required contribution under the settlement's Final Judgment is at least $204,320,196, made up of $7 million in cash, all of the Luna Foundation Guard's remaining crypto assets, and Kwon's personal holdings of PYTH tokens.

A dedicated claims portal, at claims.terra.money, opened to victims around March 31, 2025, with an original bar date that was later extended to May 16, 2025, for filing "Crypto Loss Claims." The SEC's own investor-distribution page warns explicitly about phishing attempts impersonating the claims administrator, a reminder that a fraud this large and this public draws a second wave of opportunistic scams targeting the same victims. As of this article's last review, distributions were still being administered through the bankruptcy process, and no verified, aggregate recovery percentage for UST or LUNA holders as a class has been published; unlike a bank failure with an insured deposit floor, there is no minimum guaranteed recovery here, and the ultimate percentage will depend on how much value the liquidating trust can extract from Terraform's remaining and pursued assets.

What Regulatory Changes Followed the Collapse?

Terra's failure became a central exhibit in the case for federal stablecoin regulation, and that case eventually became law. On July 18, 2025, President Trump signed S.1582, the Guiding and Establishing National Innovation for U.S. Stablecoins Act, known as the GENIUS Act, into law as Public Law No. 119-27, after the Senate passed it 68 to 30 on June 17, 2025. The act creates a federal regulatory framework specifically for "payment stablecoins," and its central requirement directly targets the design UST used: a permitted issuer must maintain reserves backing its stablecoin on a strict one-to-one basis in U.S. currency or similarly liquid assets, must be a subsidiary of an insured depository institution or a federally or state-qualified nonbank issuer, and must publicly disclose its redemption policy and publish its reserve composition monthly. Three years after enactment, digital asset service providers will be prohibited from offering stablecoins that are not issued by a permitted issuer under the act, a provision explicitly aimed at algorithmic, endogenously collateralized designs like UST's.

This is worth flagging plainly as a rule that can change over time: the GENIUS Act's implementing regulations are still being written by the Treasury Department and federal banking regulators, its permitted-issuer framework and $10 billion state-versus-federal threshold could be adjusted by rulemaking, and Congress could amend the statute itself. A reader relying on any specific compliance date or numeric threshold from this section for a present-day decision should verify it against Treasury's current rules rather than this summary. What will not change is the historical fact the law responds to: a stablecoin backed by nothing but a second, freely mintable token failed catastrophically, and the first major piece of comprehensive federal crypto legislation in the United States was written with that specific failure mode in mind.

Common Myths About the Terra/LUNA Collapse

"UST was backed by bitcoin, so the reserve should have covered the losses." The Luna Foundation Guard's roughly 80,300 BTC reserve was a late addition meant to supplement the arbitrage mechanism, not replace it, and it was never sized to be a one-to-one backing for an $18.77 billion stablecoin. It was built to smooth ordinary volatility, was spent down by about $2.5 billion in roughly 72 hours, and still failed to hold the peg.

"It was a hack or an external attack." No credible forensic account, including the Federal Reserve's and the academic reconstruction from on-chain data, identifies a smart contract exploit or unauthorized access. The mechanism did exactly what it was coded to do: it minted LUNA without limit to honor UST redemptions. The failure was a design flaw and a run dynamic, not an intrusion, though researchers have not ruled out that a specific trader anticipated and profited from the predictable sequence of events.

"Nobody could have seen this coming." UST had already depegged once, in May 2021, and Do Kwon later admitted in his own guilty plea that the recovery investors were shown then was concealed and misrepresented. Anchor's 19.5 percent yield required a growing subsidy that had reached $6 million a day by April 2022, and the Terra community itself had voted to start cutting the rate six days before the run, a clear signal that insiders understood the model was not sustainable at that level.

"Do Kwon got away with it because he fled the country." He evaded arrest for roughly ten months, but the outcome was a four-month Montenegrin sentence for the passport charge, extradition to the United States, a guilty plea, and a 15-year federal prison sentence, on top of the $4.5 billion-plus civil settlement against him and Terraform Labs. Flight delayed accountability; it did not prevent it.

"UST and LUNA holders will be made whole through the SEC settlement." The settlement money is explicitly structured so the SEC collects nothing until investors and creditors are paid in full through Terraform's separate bankruptcy case, and no verified aggregate recovery percentage for those investors has been published as this is written. The $4.5 billion figure describes what Terraform and Kwon owe, not what any individual UST or LUNA holder will actually receive.

What a Reader Can Actually Carry Forward

It is tempting to file Terra away as a crypto-specific story: an algorithmic stablecoin, a token that got printed into worthlessness, a founder who fled and was eventually caught. That framing discards the part of the mechanism that shows up, in less exotic clothing, well outside crypto.

What generalizes

• A backing mechanism that depends on the backed asset's own market confidence is not really a reserve. UST's peg required LUNA's market value to exceed UST's supply; once selling pressure shrank LUNA's value, the very mechanism meant to defend UST accelerated LUNA's decline instead. Ask of any pegged or collateralized instrument: is the collateral's value independent of the thing it is backing, or can the two fail together?
• An unsustainable yield is a warning with a specific, checkable number attached. Anchor's 19.5 percent required a growing daily subsidy, visible on-chain, long before the collapse. When a yield cannot be traced to underlying revenue, the honest question is not "is this too good to be true" as a vague feeling, but "who is paying the difference, and for how much longer."
• Reserves sized for calm markets fail in panicked ones. LFG's multi-billion-dollar bitcoin reserve looked substantial against ordinary volatility and was exhausted within three days against a coordinated exit. Stress-test a backstop against the tail scenario, not the average one.
• "Decentralized" and "automated" are marketing claims that need independent verification. The SEC's central allegation, later validated by a unanimous jury, was that a system marketed as autonomous code was in fact steered by discretionary decisions inside one company, including a concealed intervention in 2021 that built the confidence later exploited in 2022.

What does not generalize

• The specific mint-and-burn arithmetic. Most financial instruments cannot be diluted by 20,000-fold in three days; that is a property of a small number of algorithmic stablecoin designs, and the GENIUS Act's reserve requirements are aimed specifically at foreclosing this exact structure going forward.
• The speed of the criminal and civil resolution relative to the size of the fraud. A unanimous jury verdict, a multi-billion-dollar settlement, and a 15-year sentence inside three and a half years of the collapse is unusually fast for a case of this scale, aided by extensive public, immutable blockchain evidence that does not exist for most financial fraud.

The question worth asking now

Not "could a stablecoin fail again," which is really a question about a specific design category the GENIUS Act now regulates directly. The more durable question is narrower and applies to any yield-bearing product, crypto or otherwise: if this rate stopped being subsidized tomorrow, what would the honest, unsubsidized number be, and who is currently paying the gap? For most conventional savings and lending products the answer is visible and small. For a product where nobody can point to where the yield actually comes from, the gap is the risk, whether or not the word "algorithmic" ever appears in its marketing.

Related Reading

• The FTX collapse of November 2022, the next domino: Alameda Research's outsized role in the Terra run six months before its own exchange failed.
• The 2022 rate shock, for the macro backdrop of Fed tightening running alongside Terra's collapse.
• LTCM and the Russian default, for an earlier case of a reserve and a hedge both failing under the same stress at once.
• How stablecoins are supposed to work, for the collateralized designs UST's algorithmic approach deliberately avoided.
• All Swoopr market history case studies.

References

Every figure on this page was verified against the following sources, each retrieved on 26 August 2026:

• U.S. Securities and Exchange Commission: SEC Charges Terraform and CEO Do Kwon with Defrauding Investors in Crypto Schemes: the February 16, 2023 charge date, the securities named, the Chai payment-app allegation, and the Gensler and Grewal quotations.
• U.S. Securities and Exchange Commission: Terraform and Kwon to Pay $4.5 Billion Following Fraud Verdict: the $40 billion market-value figure, the December 28, 2023 and April 5, 2024 court findings, the January 21, 2024 bankruptcy filing, and every dollar figure in the June 13, 2024 settlement.
• U.S. Securities and Exchange Commission: SEC v. Terraform Labs PTE, Ltd. and Do Hyeong Kwon, Distributions to Harmed Investors: the $4,473,828,306 total obligation, the $204,320,196 Kwon contribution and its components, the September 20, 2024 liquidating-plan approval, and the claims-portal and bar-date dates.
• Federal Reserve Board Finance and Economics Discussion Series 2023-044: Badev and Watsky, Interconnected DeFi: Ripple Effects from the Terra Collapse: the arbitrage mechanism description, the $3 billion LFG bitcoin purchase and Kwon quotation, the Curve 3pool sequence citing Chainalysis, the $150 million and 185 million UST figures, the "almost 20,000-fold" LUNA supply increase, and the bridge-contagion findings.
• National Bureau of Economic Research Working Paper 31160: Liu, Makarov and Schoar, Anatomy of a Run: The Terra Luna Crash: the $50 billion abstract figure, LUNA's $119.18 April 5, 2022 peak, the Anchor 19.5 percent rate and $12 billion pre-crash balance, the $71 million and $510 million yield-reserve injections, the May 1, 2022 rate-cut vote, the May 7 wallet withdrawal timings, the paper's own narrative account that UST "plummeted to $0.75" the evening of May 9 and fell "below $0.2" by the end of May 13, the 80,300 BTC and $2.5 billion JS Held audit figures, the LUNA supply and price figures for May 11 to May 13, the 7.42 billion UST swap volume, the Alameda Research findings, and the November 2021 community-pool LUNA burn.
• Congress.gov: S.1582, Guiding and Establishing National Innovation for U.S. Stablecoins Act (GENIUS Act): the bill sponsor, the June 17, 2025 Senate passage, the July 18, 2025 enactment as Public Law No. 119-27, and the reserve and permitted-issuer requirements.
• Board of Governors of the Federal Reserve System: Federal Funds Effective Rate, Series DFF: the 0.83 percent effective rate in the week of the collapse and the rate's path from March 2022.
• DefiLlama: Terra Chain Total Value Locked, TerraClassicUSD Stablecoin Circulating Supply, and TerraClassicUSD Price Feed: the $18.77 billion UST peak supply, the daily UST prices used in the hour-by-hour table, and the Terra chain TVL figures from May 5 to May 31, 2022.
• U.S. Attorney's Office for the Southern District of New York: Do Kwon Pleads Guilty to Fraud: the May 2021 depeg allegation, the secret trading-firm purchase, and Kwon's guilty-plea admission quoted in this article.

Rules that can change, and when this page was checked. The GENIUS Act's permitted-issuer framework, its $10 billion state-versus-federal supervision threshold, and its three-year transition deadline are all subject to implementing regulations still being written by the Treasury Department and federal banking regulators as of this writing, and Congress can amend the statute itself. Terraform's bankruptcy distributions were still being administered as of this page's last review; any recovery percentage for UST or LUNA holders should be checked against the claims portal directly rather than this summary. Last checked on 26 August 2026.

Figures deliberately not stated. No aggregate recovery percentage for UST or LUNA holders, because the bankruptcy distribution was still in progress and no verified final figure existed as of this writing. No total dollar value of the Luna Foundation Guard's bitcoin reserve at the time of the collapse, because reconstructing it precisely requires an intraday bitcoin price this page did not independently verify; the reserve is instead stated in bitcoin units, as the source audit reported it. No total figure for crypto-wide losses attributed to the 2022 "crypto winter," because that period had multiple overlapping causes and assigning a single dollar figure to Terra's specific share would overstate what the cited sources actually establish.

Method note: UST's daily price in the hour-by-hour table is a single snapshot per date from DefiLlama's price feed for TerraClassicUSD, the symbol DefiLlama tracks the pre-collapse UST token under; each snapshot is timestamped in the table because UST's price moved sharply within single days during the run, and a daily snapshot cannot show the full intraday range. The $40 billion SEC figure and the $50 billion NBER figure both describe the same May 2022 loss from slightly different measurement points and are stated with their sources rather than reconciled into one number.

This is educational content about a historical episode. It is not investment advice, it is not a forecast, and nothing here should be read as a claim about the condition of any cryptocurrency, stablecoin, exchange or protocol today.

Frequently Asked Questions