For about two years, Terra had one of the most confident stories in crypto. It ran a stablecoin called UST that held its dollar peg not with dollars, but with math. It paid near-20% yields on UST deposits through a protocol called Anchor. It had a charismatic founder, billion-dollar treasury reserves, and a fast-growing ecosystem of apps built on top of it. By early 2022 the combined market capitalization of UST and its sister token LUNA had cleared $60 billion.

In one week in May 2022, nearly all of it was gone.

The Terra collapse was not the result of a hack or a rug pull. It was the result of a design that worked until confidence broke, and then did not work at all. For anyone trying to understand systemic risk in crypto — or the kind of traps that structured financial products can quietly create — it is one of the most instructive case studies available.

The Design: What UST Actually Was

A stablecoin is a token that is supposed to trade at a fixed value, usually $1. Most of them do it the boring way. Tether, USDC, and similar tokens hold actual dollar-equivalent reserves — cash, short-term Treasuries, bank deposits — and issue one token per dollar held.

UST did not work that way. It was an algorithmic stablecoin, and its peg was maintained through a mint-and-burn mechanic with its sister token, LUNA.

The rule was simple on paper:

• You could always burn 1 UST to mint $1 worth of LUNA.
• You could always burn $1 worth of LUNA to mint 1 UST.

In theory, arbitrageurs kept the peg in place. If UST fell to $0.98, a trader could buy UST cheap, burn it for a dollar of LUNA, and pocket 2 cents. That buying pressure should pull UST back to $1. If UST went above $1, the same loop worked in reverse.

That is the sales pitch. What made this design actually fragile was the second half of the sentence: $1 worth of LUNA. When LUNA fell, the amount of LUNA you received for each UST burned went up. In a calm market that is fine. In a panic, it is a demolition charge.

Anchor: The Yield That Funded the Ecosystem

There was a second piece to the story, and it was the reason most UST holders were there at all.

Anchor Protocol was a lending market on the Terra blockchain that offered depositors a roughly 19.5% annual yield on UST. That yield was not funded by market-rate borrowing. It was funded by a subsidy — mostly from Terraform Labs and the foundations behind Terra — that topped up returns to keep the headline number high.

For most of 2021 and early 2022, Anchor pulled in deposits at a rate that no traditional financial product could match. By the peak, about $14 billion of UST was sitting in Anchor, earning a return that a Treasury yield could not come close to. Retail investors used it as a savings account. Crypto-native funds parked idle capital there. Entire DeFi strategies were built on top of the 19.5% rate.

The problem was not that Anchor was paying too much. The problem was that the yield was a marketing expense. Its burn rate was unsustainable. The subsidy reserves were projected to run dry sometime in 2022 unless something changed. Everyone involved knew this. The question was what would happen when the rate finally had to come down.

In the end, something else happened first.

The Run: May 7 to May 13, 2022

Several things went wrong in the same week, and each of them amplified the next.

Saturday, May 7 — The Curve Pool Imbalance

Terra’s treasury had been migrating UST liquidity from an older Curve pool to a newer 4pool on the same platform. During the transition, a pair of very large UST sell orders — $84 million and $108 million — hit the thinner pool within minutes.

Those trades by themselves were not catastrophic. But they tilted the pool meaningfully, and they set off the first whisper in crypto chat rooms that UST might be losing its peg.

To this day, there is debate about whether the sells were a deliberate attack, a risk-management decision by a large holder, or ordinary trading. The identity of the counterparty has never been definitively established. What matters is not who did it. What matters is how a system that everyone insisted was stable reacted when it was prodded.

Sunday, May 8 — The Peg Slips

UST dropped to $0.985. Under normal conditions this would be a minor wobble — UST had briefly depegged before and recovered. But Anchor deposits started flowing out. By the end of the day, withdrawals from Anchor had hit a pace of roughly $1 billion per day.

Depositors were not waiting to see what happened. They had noticed the peg was moving and they wanted out before the exit got crowded.

Monday, May 9 — The Reflexive Loop Begins

UST sellers started using the mint-and-burn mechanic. They burned UST to mint LUNA, then sold LUNA on open markets. As more UST was burned, more LUNA was created. As more LUNA was created, the price of LUNA dropped.

Here is where the design’s fragility showed up in the open. Every UST holder exiting through the mechanism added selling pressure to LUNA. Every drop in LUNA meant the next holder exiting would create even more LUNA per UST burned. The system was rewarding people for running faster.

By end of day, LUNA was down from about $60 to around $30.

Tuesday, May 10 — The Death Spiral Accelerates

UST traded below $0.70. LUNA lost another 40%. The Luna Foundation Guard, a reserve entity set up specifically to defend the peg, deployed most of its Bitcoin reserves — reportedly around 80,000 BTC — in an attempt to buy UST and restore confidence.

It did not work. The buying was absorbed. Sellers kept coming.

Wednesday, May 11 — Binance and Other Exchanges React

Binance, Coinbase, FTX, and others began suspending trading on UST and LUNA as the mechanics of the mint-and-burn dynamic generated massive LUNA supply inflation. LUNA, which had started the week near $60, was now trading below $1. Its circulating supply had expanded from around 340 million tokens to billions.

Thursday, May 12 — LUNA Goes to Effectively Zero

LUNA traded below $0.01. The mint-and-burn rule was still technically functioning, but it was functioning as the mathematical identity that tells you a broken system will keep breaking. Terraform Labs halted the Terra blockchain to prevent governance attacks in what had become a nearly-free token supply.

Friday, May 13 — The Count

By the end of the week, the combined market cap of UST and LUNA had fallen from roughly $60 billion to under $1 billion. Anchor Protocol’s deposits were gone. An entire ecosystem of DeFi projects that had been built on Terra was effectively unwound.

Retail holders — many of whom had used UST as a savings vehicle and never understood the mechanism underneath it — lost in days what they had accumulated in years.

The Contagion

Terra was not a self-contained failure. The reach of its balance sheet, and of the institutions that had been heavily exposed to UST or LUNA, turned the collapse into a broader event.

Three Arrows Capital (3AC), one of the largest crypto hedge funds in the world, had taken major exposure to Terra through LUNA and related positions. The fund was reportedly deeply underwater within days. Its subsequent default on loans from other crypto lenders in June 2022 helped trigger a wave of failures across the centralized crypto lending industry.

Celsius Network, a major crypto lending platform, halted withdrawals in June 2022 and filed for bankruptcy the following month. A part of its insolvency traced back to Anchor deposits and related exposures.

Voyager Digital filed for bankruptcy in July 2022, directly citing its loan exposure to 3AC.

BlockFi followed later in 2022, with its solvency hit further by its FTX-related credit line.

None of these firms would have failed in the shape that they did without the Terra collapse. The crypto bear market of 2022 had several causes — rising interest rates, generalized risk-off flows, and the FTX collapse later in the year — but Terra was the shock that started the chain.

Do Kwon and the Criminal Case

Do Kwon, the co-founder and public face of Terraform Labs, had been defiant throughout the collapse. He announced a revival plan — a new LUNA token issued to the original holders, while the old chain would continue as “Luna Classic.” The revival version traded for pennies and continued trending downward.

By fall 2022, South Korean authorities had opened a criminal investigation. Investigators alleged that Kwon had made knowingly false statements about UST’s stability and about the stablecoin’s use cases in real-world payment processors. A warrant was issued in Seoul. Interpol issued a red notice.

In March 2023, Kwon was arrested in Montenegro while attempting to board a flight using a forged Costa Rican passport. He was initially held in Montenegrin custody, convicted there of document fraud, and then became the subject of a long extradition fight between the US and South Korea.

He was eventually extradited to the United States, where he faced federal fraud charges brought by the Department of Justice. In 2025, he pleaded guilty to conspiracy and wire fraud charges. Sentencing proceedings continued into 2026.

The SEC brought a separate civil case against Terraform Labs and Kwon for securities fraud. In 2024, a jury found Terraform Labs and Kwon liable. The court approved a settlement that required Terraform Labs to pay roughly $4.47 billion in disgorgement, penalties, and prejudgment interest — an amount that, in practical terms, will never fully reach creditors, but which served as a formal finding of responsibility.

What Terra Taught the Industry

Algorithmic Stablecoins Were a Category Error

After Terra, the market quietly retired the idea that an algorithmic peg backed by a floating sister token was a stablecoin at all. The category did not recover. Newer stablecoin designs that use overcollateralization, real-world reserves, or mixed collateral pools became the dominant model.

Regulators treated it as an object lesson as well. The EU’s MiCA regulation, which took effect in 2024, imposes reserve and disclosure requirements on stablecoin issuers that effectively ban unbacked algorithmic designs from being marketed as stablecoins in the EU. Similar provisions made it into stablecoin bills in the UK, Singapore, Japan, and — in a slower form — the US.

Yield Is a Risk Spread, Not a Reward

Anchor’s 19.5% rate was the single most effective marketing tool Terra ever had. It was also the single most important warning signal that almost nobody acted on.

There is no financial product in the world that produces a risk-free return significantly above the risk-free rate. When a yield is well above Treasury yields or short-term investment-grade corporate yields, the difference is not free money. It is compensation for some kind of risk — counterparty, structural, liquidity, or all of the above. With Anchor, the risk was that the yield was subsidized from a reserve that would not last, sitting on top of a peg that depended on a token that could fall to zero.

The discipline to ask where does this yield come from? is not a sophisticated analytical skill. It is a survival skill. In the aftermath of Terra, a long list of retail investors who had never heard of algorithmic stablecoins turned out to be Anchor depositors.

Reflexive Designs Fail Reflexively

The specific thing that made Terra terminal — rather than merely distressed — was that the mechanism designed to stabilize the peg did more damage the more it was used. Under stress, the stabilization loop became a destabilization loop.

Reflexive designs appear throughout crypto. Perpetual futures with cascade-liquidation risk. Leveraged staking where discounts widen during stress. Liquidity pools where impermanent loss compounds as informed traders front-run withdrawals. None of these is automatically catastrophic. All of them deserve particular scrutiny, because the normal operating mode and the crisis operating mode can look nothing alike.

Systemic Exposure Is Not Theoretical

3AC, Celsius, Voyager, BlockFi — the post-Terra insolvency cluster — showed something that the crypto industry had previously preferred to wave away: that positions in supposedly uncorrelated venues can become tightly correlated during a single bad week. Institutional crypto lenders had extended credit against collateral that included LUNA, against counterparties that were themselves long UST-related positions, with yield obligations that depended on the entire system remaining solvent.

The lesson for individual traders is that “diversification across counterparties” is not the same as “diversification across risks.” If four exchanges, three lenders, and one hedge fund all have the same correlated exposure, holding assets across all of them does not reduce the risk. It spreads it.

A Short Comparison Table

Event	Year	Peak Exposure	Core Issue
Terra / UST	2022	~$60B	Algorithmic peg broke under stress
Iron Finance (TITAN)	2021	~$2B	Similar reflexive mechanism, smaller scale
Basis Cash	2021	~$100M	Earlier algorithmic attempt, faded
TerraUSD Classic (USTC/LUNC)	2022+	Residual	Revival never restored confidence
MakerDAO DAI (for contrast)	2014+	Still active	Overcollateralized, survives because of design

DAI is on that list as the contrast case, not because it is risk-free, but because its design choice — overcollateralization in visible, on-chain collateral — is structurally different. When DAI has come under stress, it has adjusted collateral ratios and fees. It has never had the reflexive failure mode that Terra built in.

What a Trader Should Take Away

Look at the mechanism, not the yield. If you cannot explain in one paragraph how a protocol keeps its promised peg, return, or liquidity, do not treat it as risk-free. “Algorithmic” should be a prompt for questions, not a closing argument.

Peg stability in good weather is not peg stability. Every broken peg in history traded at par until the moment it did not. Historical stability is a prerequisite for a stablecoin, not a guarantee about the future.

Read the subsidy carefully. When a yield is paid by the protocol itself out of foundation reserves, treat that as a temporary marketing expense. Ask when the reserve runs out. Ask what is supposed to replace it. If the answer is “new user adoption,” what you actually have is a growth-dependent promise, not a yield.

Understand the downside of reflexivity. Structures that work by encouraging arbitrageurs to defend a price can work the other direction too. In a panic, the arbitrage becomes an exit tool.

Price systemic risk even when the system looks stable. Terra had $60 billion of market cap and a treasury with tens of thousands of Bitcoin. It also had a single mechanism that, once it started failing, was nearly impossible to stop.

Final Takeaway

Terra and LUNA were not a cautionary tale about crypto as a whole. They were a cautionary tale about a specific kind of financial engineering — one that dresses a highly reflexive design in the language of stability, funds it with a subsidy that has an expiration date, and markets it on a yield number that traditional finance could not match.

The mechanism worked until it did not. When it stopped working, it collapsed in a straight line in roughly a week, took an ecosystem with it, and seeded a cascade of insolvencies that reshaped the institutional side of crypto for years.

Any product that offers a stable value and an outsized yield deserves the same hard question: what is really paying for this, and what happens to it under stress? If the honest answer is new inflows, probably, you are not looking at a stablecoin. You are looking at a confidence vehicle. Those end, and they usually end fast.