Algorithmic Stablecoins: The Promise, the Failures, and What's Actually Left

Introduction

When algorithmic stablecoins promised to eliminate the idle capital trap of collateralized designs, investors believed they had found money's future. Anchor Protocol's 19.5–20% yield on UST drew $14–15 billion in deposits in under two years — proof that demand for a peg-maintained-by-code could scale rapidly. Yet on May 14, 2022, LUNA's supply ballooned from 350 million to 6.5 trillion tokens in five days, destroying approximately $40 billion in combined market value. This article explains why algorithmic stablecoins failed across multiple architectural generations, how the structural flaw makes recovery impossible, and which mechanisms survived by abandoning the "algorithmic" label. The reader will understand the difference between a mechanism that holds a peg in equilibrium and one that survives a coordinated exit — and why the difference matters more than any engineering improvement.

Why Did Algorithmic Stablecoins Promise So Much — and Attract So Much Capital?

Algorithmic stablecoins offered something collateralized models lacked: a peg maintained entirely by code, with no locked dollars diluting capital efficiency. That promise attracted billions before the structural failures became undeniable.

What Capital Efficiency Promised

Every dollar locked in a collateralized stablecoin — USDC, DAI, LUSD — is a dollar doing nothing else. At scale, that idle capital represents an enormous drag on DeFi's capital velocity. Algorithmic models proposed eliminating that drag entirely. A stablecoin backed by protocol-native governance tokens rather than external reserves required no upfront collateral deposit — in theory, a $1 stablecoin could exist without any underlying dollar locked anywhere. For protocols competing on yields and liquidity depth, that efficiency appeared transformative. Anchor Protocol demonstrated the appeal: by pairing UST with a 19.5–20% yield, it drew ~$14–15B in deposits in under two years, proving that demand for a well-incentivized algorithmic stablecoin could scale rapidly (Anchor Protocol / Bocconi research, 2022). The capital efficiency narrative wasn't irrational — it was incomplete.

The Central Bank in Code Vision

Behind the capital efficiency argument sat a broader ideological claim: monetary policy itself could be encoded in smart contracts. Traditional central banks stabilize currency value by expanding or contracting the money supply at human discretion, subject to political pressure and institutional lag. Algorithmic stablecoin designers proposed that smart contracts replicate that function without discretion or delay — burning supply when the peg fell below $1, minting supply when it exceeded $1, and distributing governance token seigniorage to absorb the imbalance mechanically. The vision attracted serious capital and serious talent. It also rested on a fatal assumption: that market participants would consistently cooperate with the arbitrage mechanism ↗ even during stress, because the protocol's long-run solvency appeared unquestionable.

Data current as of September 2026.

How Does a Seigniorage Model Work — and Where Does the Logic Break?

Seigniorage models appear coherent in equilibrium but contain a structural flaw that only becomes visible under stress — and stress is precisely when the flaw matters most.

How Seigniorage Models Work

In a seigniorage stablecoin system, a two-token architecture separates the stable asset from the absorbing asset. When the stablecoin trades above $1, the protocol mints new stablecoin supply and distributes the excess value to governance token holders — this profit is seigniorage. When the stablecoin trades below $1, the protocol incentivizes holders to burn stablecoin in exchange for newly minted governance tokens — supply contracts and the peg theoretically restores. The arbitrage logic is straightforward: if burning 1 UST yields $1 worth of LUNA at any price, rational actors will burn until supply falls enough to push UST back to $1. The mechanism depends on two conditions holding simultaneously: the governance token must maintain positive market value, and the market must trust that the arbitrage will close the gap before the governance token is diluted past the point of return (SNB research, 2022).

The Circular Dependency Flaw

The model's stability depends on the governance token retaining value. The governance token retains value only if the stablecoin retains demand. Demand collapses when the peg weakens — precisely when the governance token must absorb losses. This circularity is not an edge case. It is the load-bearing assumption of the entire architecture. Under normal conditions, small depegs trigger arbitrage ↗ that closes quickly — each successful close reinforces confidence in the mechanism. The governance token gains speculative value as the stablecoin grows, which in turn makes the absorption mechanism look more credible. The system can scale for months or years without the flaw surfacing. But the flaw never disappears — it compounds. The larger the stablecoin supply relative to the governance token's market cap, the more governance tokens must be minted per dollar of stablecoin burned during a contraction. At sufficient scale, no feasible governance token price can absorb a rapid stablecoin contraction without entering hyperinflationary dilution. The UST/LUNA system crossed that threshold in May 2022 (Bocconi research, 2022).

What Is a Rebase Stablecoin and Why Is That Label Misleading?

Rebase stablecoins occupy a distinct category from seigniorage models, but the stabilization claim they make is equally hollow — and the mechanism is, if anything, more opaque.

Rebase Mechanics Explained

A rebase stablecoin adjusts every holder's wallet balance in proportion to the gap between the current price and the $1 target. When Ampleforth's AMPL trades at $1.10, every holder's balance increases by approximately 10% at the next rebase epoch — new tokens appear in wallets automatically. When AMPL trades at $0.90, balances contract by 10%. The protocol's claim is that this mechanical supply adjustment pushes price back toward $1 by shifting the supply curve. If everyone holds 10% more tokens overnight, the price must fall to restore equilibrium — theoretically. No external collateral backs the system. No governance token absorbs contraction losses. The mechanism relies entirely on market participants interpreting the rebase as a stabilizing signal and adjusting their bids accordingly.

Why Rebase Isn't Really a Stablecoin

The rebase mechanism does not stabilize purchasing power — it redenominates units. A holder who enters with $100 at $1.10/AMPL, receives a positive rebase, and then watches the price fall back to $1.00 ends up with more tokens worth less each. The dollar value of the position depends entirely on price, which the rebase mechanism does not control. External demand determines AMPL's price, not supply adjustments. In practice, rebase stablecoins track their pegs only during periods of organic demand growth. When demand stalls or reverses, the rebase contracts balances, which reduces the nominal value of holdings, which reduces demand further. Ampleforth survived by abandoning the stablecoin identity and repositioning as a "non-correlated asset" — an implicit acknowledgment that the rebase mechanism fails to deliver monetary stability. No rebase stablecoin operates at meaningful scale in 2026.

How Did UST and LUNA Create a $40 Billion Death Spiral?

The UST/LUNA collapse between May 7–14, 2022 was not a black swan. It was the seigniorage model's circular dependency executing exactly as its logic predicted — at $18.7 billion in stablecoin supply and insufficient governance token liquidity to absorb the shock (SNB/Bocconi research, 2022).

The Mint-Burn Loop

Terra's protocol maintained UST's peg through a mint-burn arbitrage: burning $1 of UST always yielded $1 worth of LUNA at current market price, and burning $1 of LUNA always yielded $1 of UST. When UST traded below $1, arbitrageurs bought discounted UST, burned it for LUNA at face value, and sold the LUNA — pocketing the difference and reducing UST supply. When UST traded above $1, arbitrageurs minted new UST by burning LUNA — supply expanded and price fell back toward $1. This loop functioned through 2021 and into early 2022. LUNA's market cap rose above $40B at its April 2022 peak price of $119, lending apparent credibility to the mechanism's absorptive capacity (The Block/Finexus, 2022). At that market cap, even a $5B UST contraction looked manageable.

Anchor Protocol: The Demand Engine

Anchor Protocol held approximately 75–80% of all circulating UST supply at the time of collapse (UNC/Bocconi academic papers, 2022). It paid depositors 19.5–20% APY on UST — a yield funded not by lending revenue but by a protocol reserve that Luna Foundation Guard (LFG) periodically topped up. That yield was not organic. It was a subsidy designed to bootstrap UST demand at scale, on the assumption that growing adoption would eventually generate sustainable yield. The architecture created a structural concentration risk: nearly all UST demand was rate-sensitive institutional and retail capital seeking yield, not organic monetary demand for a stablecoin. When Anchor's yield reserve came under pressure in early May 2022, large depositors began withdrawing UST. A coordinated attack on Curve's UST-3pool deepened the depeg, and the mint-burn loop engaged at the worst possible scale (Bocconi research, 2022).

The Death Spiral Step by Step

LUNA's supply stood at approximately 350 million tokens on May 9, 2022. Over the following five days, the mint-burn mechanism minted approximately 6.5 trillion LUNA ↗ as the protocol attempted to absorb UST redemptions — an expansion of roughly 18,500x (Finexus/luncmetrics, 2022). Walk through the arithmetic: at $0.01/LUNA, 6.5 trillion tokens imply a $65 billion market cap. The actual LUNA market cap at that price was effectively zero, because no buyer existed at any price that made the arbitrage worth executing. The feedback loop accelerated: UST depegged → protocol minted LUNA to absorb redemptions → LUNA dilution crashed LUNA price → less LUNA value per burned UST → larger depeg required to incentivize burning → more LUNA minted. Total value destroyed across UST and LUNA reached approximately $40 billion by the time trading halted (SNB/Bocconi research, 2022). UST reached $0.01. LUNA, which had traded at $80 at the start of the crash, reached $0.0001 by May 14.

Data current as of September 2026.

The UST/LUNA collapse demonstrated that seigniorage models scale their risk faster than their resilience. A $1B seigniorage system can survive a stress event. An $18.7B one, concentrated in a single yield protocol, cannot.

How Did Frax Survive When Every Other Algorithmic Model Failed?

Frax survived not because its fractional-algorithmic design was inherently superior, but because it retained real protocol revenue as a floor ↗ — and then abandoned the algorithmic component entirely before that floor was tested at scale.

How Frax Built a Hybrid

Frax Finance launched in December 2020 as the first fractional-algorithmic stablecoin: FRAX was partially backed by USDC (external collateral) and partially backed by FXS (the protocol's governance token). The collateral ratio started at 100% and adjusted algorithmically based on market demand. When FRAX traded above $1, the protocol reduced the collateral ratio — less USDC required per FRAX minted, more seigniorage captured by FXS holders. When FRAX traded below $1, the ratio increased. At its peak, Frax ran a collateral ratio of approximately 85%, meaning 15% of every FRAX was backed by FXS rather than dollars. That 15% endogenous exposure looked small. During the UST collapse, contagion pressure tested Frax's peg, and FXS's price fell sharply — compressing the effective collateral buffer. Frax held its peg, but the margin was thinner than the model implied (Frax docs, 2026).

Why Frax Survived and Others Didn't

The critical distinction: FXS captured real protocol cash flows, not speculative value alone. Frax collected fees from AMM pools, lending markets, and protocol integrations — revenue streams that anchored FXS to a fundamental value floor independent of FRAX demand. Pure seigniorage models like Terra had no equivalent revenue stream. LUNA's value derived entirely from UST adoption; without UST demand, LUNA had zero backing. Because FXS captured protocol revenue, it retained residual value even if FRAX demand fell, which meant the collateral buffer never collapsed to zero under moderate stress. The governance lesson was not lost on Frax's team. In early 2023, FIP-188 committed Frax to a collateral ratio of at minimum 100%, phasing out the fractional-algorithmic model entirely (Frax docs / Portals blog, 2026). The 2025 North Star upgrade (FIP-428) completed the pivot: the stablecoin FRAX was renamed frxUSD — backed 1:1 by tokenized U.S. Treasuries such as BlackRock BUIDL and Superstate USTB — while the FXS governance token was renamed FRAX (Frax governance, 2025). As of August 2026, frxUSD's market cap stands at approximately $83M — a fraction of its 2022 peak, but the only surviving lineage of any model that once called itself algorithmic. The survival required abandoning the algorithm.

What Algo-Adjacent Stablecoins Are Still Operating in 2026?

Several stablecoins operating in 2026 use algorithmic mechanisms as components of their peg defense — but none relies on endogenous collateral as its primary stability layer.

crvUSD and LLAMMA

Curve Finance's crvUSD uses a mechanism called LLAMMA (Lending-Liquidating AMM Algorithm) — a continuous soft-liquidation system that rebalances collateral positions as price moves, converting ETH or other assets gradually into crvUSD and back, rather than executing hard liquidations at a single threshold (Curve docs, 2026). The design reduces liquidation cascade risk by spreading collateral conversion across a price range. The stability mechanism is algorithmic, but the collateral is entirely exogenous — ETH, wBTC, staked ETH derivatives. crvUSD carries no exposure to Curve's governance token (CRV) as collateral. Its market cap stood at approximately $120M as of August 2026 , constrained by the deliberate conservatism of the collateral parameters (CoinPaprika, Aug 2026).

GHO and USDe: What's Different

Aave's GHO, with a market cap of approximately $599M as of August 2026 , functions as a CDP stablecoin ↗ backed by the full range of Aave V3 collateral — predominantly ETH and staked ETH (CoinPaprika, Aug 2026). GHO's interest rate adjusts algorithmically to influence supply, and stkAAVE holders receive a discount on borrowing rates — a soft governance token integration, not a collateral exposure. But stkAAVE is not used as collateral; the connection is a rate incentive only. Ethena's USDe sits in a different category entirely ↗ — and, at approximately $4.1B as of August 2026 , it is now by far the largest "algo-adjacent" dollar, though it is not algorithmic in the seigniorage sense. It generates its dollar peg through delta-neutral derivatives positions — long spot ETH, short ETH perpetual futures — capturing the funding rate differential as yield. USDe's stability mechanism is market-structural rather than algorithmic; it breaks under funding and market-liquidity stress, not under token confidence loss. That distinction was tested on October 10, 2025, when USDe briefly printed ~$0.65 on Binance during a record ~$19B liquidation cascade — an isolated, single-venue oracle dislocation. Mint-and-redeem stayed operational, the protocol remained overcollateralized, and USDe held near its peg on Curve and other deep-liquidity venues, repegging within minutes (CoinDesk / Galaxy Research, Oct 2025). None of these models accepts endogenous collateral as their primary backing, which is the defining feature that separates surviving algo-adjacent designs from the failed pure algorithmic models.

Data current as of September 2026.

What Does the GENIUS Act Actually Say About Algorithmic Stablecoins?

The GENIUS Act bars endogenous-collateral stablecoins from payment stablecoin licensing ↗, which effectively excludes every functioning algorithmic model from the payment category — the segment with the most significant regulatory and commercial weight.

What the GENIUS Act Actually Says

The GENIUS Act, signed July 18, 2025, establishes a licensing framework for "payment stablecoins" — stablecoins intended for use in payments and commerce (Sullivan & Cromwell / Congress.gov, 2025). To qualify as a payment stablecoin, issuers must maintain 1:1 reserves in high-quality liquid assets: Treasury bills, central bank deposits, insured bank deposits, or similar instruments. Governance tokens, protocol-native assets, or any endogenous collateral category do not qualify. The Act directs the Treasury — in consultation with the Federal Reserve, OCC, FDIC, and SEC — to conduct a study of endogenously collateralized stablecoins, with the report due within one year of enactment, on July 18, 2026. As of August 2026 that deadline has passed; the published findings were not yet publicly available at the time of writing. That study will inform whether future rulemaking addresses the category separately or continues to exclude it. The effective date for the licensing framework is the earlier of January 18, 2027, or 120 days after regulators issue final rules (Sullivan & Cromwell, 2025).

Endogenous Collateral and Payment Status

The practical consequence of the GENIUS Act's collateral requirements: any stablecoin issuer seeking payment stablecoin status — and the ability to partner with banks, payment processors, and merchants at scale — cannot use endogenous collateral. A stablecoin backed partly by its own governance token, like pre-FIP-188 FRAX, fails to qualify. A stablecoin backed entirely by protocol-native tokens, like UST, cannot qualify. This does not prohibit DeFi protocols from issuing stablecoins for on-chain use; it blocks the path to regulated payment integration. For protocols targeting institutional adoption, that exclusion amounts to a functional ban — the payment channel is where the addressable market lies. The GENIUS Act's treatment of endogenous collateral reflects the policy lesson of May 2022: regulators concluded that any model whose collateral value correlates with the stablecoin's own demand creates an unacceptable reflexivity risk for users transacting in payments (Sullivan & Cromwell, 2025).

Is the Algorithmic Stablecoin Idea Dead or Just Incomplete?

The structural verdict on pure algorithmic stablecoins is not that the engineering was immature — it is that the mechanism is categorically incapable of delivering unconditional monetary stability, which is the only property a stablecoin must have.

The Structural Verdict

Seigniorage and rebase models share a foundational failure: their stability is conditional on the continuation of the same demand conditions that create the need for stability. A stablecoin that holds its peg only when demand is growing is not stable — it is pro-cyclical. The conditional nature of algorithmic stability is not a design defect that iteration can correct. It follows from the absence of a loss-absorbing layer that exists independently of the stablecoin's own demand. External collateral provides that layer precisely because its value is not determined by the stablecoin's success. Governance tokens are determined entirely by that success. Every proposed improvement to seigniorage models — bonding curves, multi-collateral buffers, protocol-owned liquidity — either adds exogenous collateral (at which point the model is no longer purely algorithmic) or patches the symptom without addressing the circular dependency. No team has demonstrated a pure algorithmic mechanism that survives a coordinated exit at scale. IRON/TITAN (2021), UST/LUNA (2022), and the Basis Protocol shutdown (2018) across multiple architectural generations constitute a definitive empirical record.

What Survives and Why

What survives from the algorithmic stablecoin era is the insight that smart contract mechanics can improve upon traditional CDP designs — not replace their collateral requirements. LLAMMA's soft liquidation mechanism, Frax's AMO (Algorithmic Market Operations) controllers, and GHO's rate-based governance integration all demonstrate that algorithms serve effectively as collateral management tools. They fail as collateral substitutes. The surviving models in 2026 — crvUSD, GHO, FRAX post-FIP-188 — use algorithms to optimize how exogenous collateral is deployed, not to replace it. That reorientation separates them from every model that collapsed. frxUSD's market cap of approximately $83M after Frax's pivot to full collateralization, compared to the $3B+ peak of its fractional-algorithmic phase, also illustrates the market's own verdict: capital prefers verified reserves to elegant mechanism design.

Data current as of September 2026.

The algorithmic stablecoin era produced real advances in DeFi mechanism design. It produced no viable unconditional stablecoin.

Algorithmic Stablecoins: Frequently Asked Questions

The UST/LUNA collapse and subsequent regulatory developments generate persistent questions across investor, developer, and researcher audiences. The following addresses the most frequent.

FAQ

What happened to UST and LUNA?

UST was an algorithmic stablecoin issued by Terraform Labs on the Terra blockchain. It maintained its $1 peg through a mint-burn mechanism with LUNA, Terra's governance token. In May 2022, coordinated selling and Anchor Protocol withdrawals triggered a depeg. The protocol minted approximately 6.5 trillion LUNA attempting to absorb UST redemptions — expanding supply from ~350 million tokens — which diluted LUNA to near-zero and broke the arbitrage mechanism entirely. Approximately $40B in combined UST and LUNA market value was destroyed in five days (SNB/Bocconi research, 2022).

Is Frax still an algorithmic stablecoin?

No. Since FIP-188 in early 2023, Frax Finance targets a minimum 100% collateral ratio backed by exogenous assets (Frax docs / Portals blog, 2026). The fractional-algorithmic model that defined Frax V1 and V2 — where a portion of each FRAX was backed by the FXS governance token — no longer applies. Frax retains algorithmic mechanisms for managing collateral deployment (AMO controllers), but the stablecoin — rebranded frxUSD in the 2025 North Star upgrade — is fully collateralized by tokenized U.S. Treasuries.

Are algorithmic stablecoins banned?

Not in DeFi generally. The GENIUS Act (signed July 18, 2025) bars stablecoins backed by endogenous collateral from receiving payment stablecoin licensing, which is required for regulated payment use cases (Sullivan & Cromwell, 2025). A Treasury study that was due July 18, 2026 may inform further policy. For on-chain protocols, the restriction blocks the path to regulated payment integration but does not prohibit stablecoin issuance itself.

What does endogenous collateral mean?

Endogenous collateral refers to assets whose value is determined by the protocol that issues the stablecoin — typically the protocol's own governance token. UST was backed by LUNA; LUNA's value derived from UST adoption. That circularity is what "endogenous" describes. Exogenous collateral — ETH, USDC, Treasury bills — holds value independently of the stablecoin it backs.

Does any pure algorithmic stablecoin operate at meaningful scale in 2026?

No. Every stablecoin operating at meaningful scale in 2026 uses either full external reserves (USDC, USDT, FRAX) or exogenous crypto-collateral with algorithmic liquidation management (crvUSD, GHO). Pure seigniorage models and rebase models with no external collateral have either collapsed or abandoned the stablecoin designation entirely.

FAQ

References / Sources