Was Anchor Protocol’s 20% UST yield a Ponzi? A look at what actually happened
Cryptowraith7 min read·Just now--
There’s a version of this story where Anchor was just another DeFi protocol that promised too much and blew up. Lots of those exist. Most are forgettable.
Anchor is not forgettable. Anchor was the engine that made $40 billion disappear in three days, and the wild part is that almost everyone who looked closely could see what was going to happen. The only real disagreement was about timing.
I’m going to walk through what 20% yield on UST actually was. Not the marketing version. The mechanical version. Because once you understand what the yield was made of, the collapse stops looking like a black swan and starts looking like a structural certainty.
The pitch was perfect, which should have been the first warning
Anchor launched in March 2021 on Terra, built by Terraform Labs, the same team that built UST. The pitch was almost obnoxiously simple: deposit UST, earn around 20% APY. Withdraw whenever you want.
That’s it. No lockups. No five-token reward structures. You held a dollar-pegged stablecoin and earned twenty percent. Pension-fund energy, in a space that mostly looked like a casino.
For about fourteen months, the pitch worked. By early 2022, Anchor held over $14 billion in UST deposits. Roughly 75% of all UST in existence was sitting inside Anchor. The protocol wasn’t a feature of the Terra ecosystem. It was the Terra ecosystem.
Which raises the question anyone with a pulse should have been asking: where was the 20% actually coming from?
How Anchor was actually supposed to work
The official model was savings-and-loans. Depositors put in UST and earned yield. Borrowers posted collateral: bonded LUNA or bonded ETH, and took out UST loans. Borrowers paid interest.
That interest funded the depositor yield.
The difference was that Anchor’s borrower side also paid out yield. The bonded collateral kept earning staking rewards, and a portion of those flowed back to subsidize depositor yield on top of the interest payments.
In theory, this meant Anchor could offer higher rates than a pure lending market because it tapped a second yield source: proof-of-stake rewards from the underlying collateral.
In practice, the math never closed.
Everyone could see the math wasn’t working
Borrower demand was weak. The amount of UST people wanted to borrow was always a small fraction of what depositors wanted to lend. Sometimes the loan-to-deposit ratio sat below 25%. Interest income from borrowers, plus staking yield from their collateral, came nowhere near covering 20% on the entire deposit base.
The shortfall was paid out of the Yield Reserve; a separate pool of UST topped up periodically by Terraform Labs and the Luna Foundation Guard (LFG), which subsidized the gap.
By late 2021, the Yield Reserve was draining at roughly $300,000 per day. By February 2022, closer to $1 million per day. By February, without a $450 million emergency top-up from LFG, it would have been depleted within weeks.
This was not a secret. Daily reports tracking the Yield Reserve were public. Researchers wrote about it. Twitter accounts dedicated themselves to monitoring it. Anyone who wanted to know exactly how subsidized that 20% was could have found out in about forty seconds.
The information was sitting in plain sight. The participants chose what to do with it.
Who knew, and what they did about it
Different groups understood the situation differently, and their behavior is the most revealing part.
The builders knew. Do Kwon was repeatedly asked about Anchor’s sustainability. His public position was a confident shrug; the rate would adjust, demand would catch up, the system was bootstrapping.
Internally, the picture was different. Later legal proceedings against Terraform Labs included allegations that the team was aware of structural problems beyond what was publicly acknowledged.
The institutions knew. Three Arrows Capital, Jump Crypto, Pantera, and Hashed held significant LUNA positions and had close relationships with Terraform Labs. Some helped fund LFG’s Bitcoin-backed reserve, meant to defend the peg in a crisis.
The fact that they were actively building emergency defenses tells you what they thought of the system’s stability. The fact that they didn’t reduce their LUNA exposure tells you what they thought of their ability to get out before everyone else.
Jump Crypto had been involved in defending UST’s peg during a smaller depeg in May 2021. The intervention worked at small scale. It established a pattern. It also created the assumption that someone would always step in.
The retail depositors mostly didn’t know. Or more accurately: they knew there was risk, but they didn’t know how much, and they didn’t know what would trigger it. The 20% was framed as a stable-asset savings rate. The word “stablecoin” was doing enormous work. People moved retirement savings into Anchor. The interface offered no warning that the yield was being subsidized from a reserve draining faster than it was being refilled.
The regulators were watching but moved slowly. The SEC had subpoenaed Terraform Labs in 2021. Scrutiny was increasing. No enforcement action hit before the collapse.
The collapse mechanics
What happened in May 2022 is well-documented now, but the speed still shocks me.
On May 7, a series of large UST sales hit Curve Finance, the main on-chain market for swapping between UST and other stablecoins. The pool became imbalanced. UST started trading slightly below a dollar.
Under normal conditions, Terra’s algorithmic mechanism was supposed to absorb this. UST holders could redeem 1 UST for $1 worth of newly-minted LUNA, which arbitrageurs would sell. This was supposed to push UST back to peg.
The mechanism worked exactly as designed. That was the problem.
As UST holders redeemed, the protocol minted enormous amounts of LUNA. LUNA’s price collapsed under supply pressure. Each redeemed UST minted even more LUNA, which collapsed the price further, which required minting more LUNA. The reflexive loop the system was supposed to prevent was the system.
By May 9, UST was around 70 cents. By May 11, around 30 cents. LUNA fell from $80 to under a dollar within days, eventually going essentially to zero.
In parallel, Anchor depositors withdrew fast. UST inside Anchor dropped from roughly $14 billion to under $2 billion in less than a week. Every withdrawal added selling pressure. Every drop in UST’s price made collateral inside Anchor less valuable, triggering more liquidations.
LFG deployed its Bitcoin reserve; over 80,000 BTC, roughly $3 billion at the time, trying to defend the peg. It drained the reserve. The peg kept falling.
The consensus number is around $40 billion erased between LUNA and UST. The figure runs higher when you include downstream effects: Celsius, Voyager, and Three Arrows Capital all had Terra exposure, and all collapsed in the months that followed.
The thing nobody wants to admit
The 20% was not a yield. It was a customer acquisition cost paid in subsidies.
Terraform Labs and LFG were buying TVL, paying users to deposit UST so that the appearance of demand for UST would attract more demand for UST. The money funding that 20% was coming from venture capital, token sales, and reserve draws. Not from any productive economic activity.
This is the same playbook every failed L1 has run. Avalanche Rush. Fantom DeFi summer. Berachain, three years later, with a sophisticated rebrand. Pay to attract liquidity. Hope the liquidity becomes organic before the budget runs out. It almost never does.
What made Terra different was scale and packaging. Most incentive programs are openly framed as such. Anchor’s 20% was packaged as a savings rate.
The structural similarity to a Ponzi was hidden behind the technical sophistication of the underlying chain. People who would have laughed at “deposit your dollars and we’ll pay you 20% from new investors’ deposits” were happy to deposit into Anchor because there was a smart contract involved and the yield was called staking.
The smart contract did not change the math. It just made the math harder to see.
What this should have taught us
The standard Terra retrospective treats this as a technical failure of algorithmic stablecoins. That’s not wrong, but it’s incomplete. The deeper failure was that almost every participant had partial information, and the partial information was enough to act rationally on a local level while collectively constructing something that couldn’t survive contact with reality.
Builders knew the yield was subsidized but believed organic demand would arrive. Institutions knew the system was fragile but believed they could exit first. Retail depositors didn’t know how the yield was generated but believed the word “stablecoin” implied a guarantee. Regulators saw the warning signs and moved on regulatory timelines while the system grew on internet timelines.
Each rationalization made sense from inside that participant’s view. None of them were correct.
What I keep coming back to is that Anchor wasn’t fooling anyone who looked. The depletion charts were public. The mechanism was, eventually, possible to understand if you wanted to.
What Anchor was doing was offering people a yield high enough to make them not want to look. That’s not a smart contract feature. That’s a human one.
And it’s the thing that keeps showing up in every cycle. The yields keep arriving in new technical clothes: proof of liquidity, restaking, points programs, real yield narratives, and most of them have, somewhere underneath, the same unanswered question Anchor never answered: where is the money actually coming from, and what happens when it stops?
If you can’t answer that in one sentence, you already have your answer.
Yours sincerely.
Cryptowraith.
