Skip to content
Decision guide // updated July 2026

Fiat-Backed vs Crypto-Backed vs Algorithmic vs Synthetic: Choosing a Stablecoin Model

A builder's comparison of fiat-backed, crypto-backed, algorithmic, and synthetic dollars in 2026, from a team that ships stablecoin and yield-bearing-dollar infrastructure.

TL;DR

Fiat-backed stablecoins (USDC, USDT) hold cash and short Treasuries 1:1, dominate the ~$300B market, and fit new regulation, but concentrate trust in one issuer and pay holders nothing. Crypto-backed CDP dollars (USDS, ex-DAI) are over-collateralized and fully on-chain, at the cost of capital efficiency. Pure algorithmic stablecoins are effectively dead after Terra. Synthetic, delta-neutral dollars (Ethena's USDe) are a distinct fourth model that earns yield by hedging, and must not be confused with algorithmic. The right model depends on your regulatory footing, capital efficiency, and whether holders should earn yield.

Protofire builds stablecoin infrastructure, native issuance, yield-bearing dollars, and cross-border rails, so we have a commercial interest in the category. The comparison is model-agnostic, and the supply and peg figures come mostly from DefiLlama. See our other decision guides, including RWA token standards and cross-chain messaging.

Scorecard

At a glance

01Fiat-backed02Crypto-backed03Algorithmic04Synthetic
BackingCash + short Treasuries, 1:1Over-collateralized cryptoLittle/no collateralLong spot + short perp (hedged)
Peg mechanismRedemption + arbitrageOver-collateralization, liquidations, and a peg-stability moduleSupply expand/contract (seigniorage)Basis-trade hedge + arbitrage
Capital efficiencyHigh (~100%)Low (110-175%+ collateral)Very high, but fragileHigh (~100%, off public chain)
Main riskReserve, bank, regulatoryLiquidation, collateral volatilityReflexive depeg / death spiralFunding rate, exchange counterparty
TransparencyAttestations (monthly)Fully on-chain, real-timeOn-chain but reflexivePartial (positions on CEX/custody)
Holder yieldNone (banned for payment stablecoins)Via savings rateHistorically noneVia staked token (sUSDe)
ExampleUSDC, USDTUSDS (ex-DAI)Terra UST (defunct)Ethena USDe
In detail

The four models

01

Fiat-backed

Cash and Treasuries, 1:1
Strengths
  • +Simple, deeply liquid, and the regulatory default (GENIUS Act and MiCA are built for it)
  • +Capital-efficient: every token is a dollar of reserves, held 1:1
  • +Dominant by far: USDT and USDC are ~83% of all stablecoin supply
Trade-offs
  • Trust concentrates in one issuer and the banking system (USDC depegged to $0.87 during the SVB failure)
  • Holders capture no yield; once the GENIUS Act is in force, payment stablecoins cannot pay interest to holders
  • Reserves are shown via monthly attestations, not full audits (Tether has never completed one), and issuer jurisdiction is an ongoing risk

A fiat-backed stablecoin holds cash and short-dated US Treasuries roughly 1:1 and holds its peg through issuer redemption and arbitrage. It is the simplest and most liquid model, and the one regulation is written for: USDC (~$73B) and USDT (~$184B) together are about 83% of the ~$310B market, a share that is slowly declining, perDefiLlama.

The trade-offs are trust and yield. Reserves sit with a centralized issuer and its banks, which is why USDC briefly depegged to $0.87 during the 2023 SVB failure, and the GENIUS Act (signed July 2025) both codifies the reserve rules and, once it takes effect, prohibits payment stablecoins from paying interest to holders, which is already why yield routes through separate products. In the EU, MiCA has treated these as e-money tokens since June 2024, and after its transition period ended in July 2026, USDT was delisted for EEA retail while USDC stayed.

02

Crypto-backed (CDP)

Over-collateralized and on-chain
Strengths
  • +Fully on-chain and verifiable collateral in real time, the transparency advantage
  • +Censorship-resistant and DeFi-native, no bank dependency for the core mechanism
  • +Recovered its peg across multiple cycles, though DAI briefly fell to ~$0.90 in the 2023 SVB event via its USDC exposure
Trade-offs
  • Capital-inefficient: you lock $1.10 to $1.75+ of collateral per $1 minted
  • Exposed to liquidation cascades when collateral falls sharply
  • Decentralization is eroding as peg-stability modules lean on USDC and RWAs

A crypto-backed, collateralized-debt-position (CDP) stablecoin lets users lock volatile crypto in a vault and mint a dollar against it, with over-collateralization, automated liquidations, and a peg-stability module holding the peg. The canonical example rebranded in this cycle: MakerDAO became Sky in 2024 and launched USDS as DAI's upgrade; in April 2026 major exchanges force-converted DAI balances to USDS, though DAI itself remains live and immutable. Others include Liquity (LUSD and V2 BOLD), Curve's crvUSD (whose LLAMMA soft-liquidation is designed to avoid hard-liquidation cascades), and Aave's GHO.

Its advantage is transparency: the collateral is on-chain and verifiable in real time, and the mechanism does not depend on a bank. The costs are capital efficiency (you post well over a dollar of collateral per dollar minted) and liquidation risk if collateral drops fast. A subtle 2026 caveat: to defend the peg, these systems increasingly hold USDC and real-world assets in their stability modules, which dilutes the pure-decentralization claim.

03

Algorithmic

Seigniorage, and effectively dead
Strengths
  • +In theory, capital-efficient with little or no collateral to post
  • +Decentralized by design, with no reserve or custodian
  • +Included as a cautionary category; there is no healthy example at scale
Trade-offs
  • Empirically fragile: reflexive depeg and death spirals with no collateral floor
  • Terra's UST (~$18B) collapsed in 2022; the combined UST and LUNA wipeout exceeded $40B
  • No pure algorithmic dollar has survived at scale since

A pure algorithmic stablecoin holds its peg by expanding and contracting supply against a paired token, with little or no hard collateral. In theory that is capital-efficient and decentralized; in practice it is reflexive and prone to death spirals, because there is no exogenous collateral floor to catch a loss of confidence.

The model is effectively dead at scale. Terra's UST collapsed in May 2022, and the combined UST and LUNA wipeout exceeded $40B, the defining seigniorage failure. Survivors reframed around real collateral (Frax, for one, moved from a fractional-algorithmic model toward full backing). We include it because a guide should say plainly why not to build a pure algorithmic dollar, not because it is a live option.

04

Synthetic (delta-neutral)

Hedged, yield-bearing, and often mislabeled
Strengths
  • +Native yield: the hedged position earns funding, paid through a separate staked token (sUSDe)
  • +Capital-efficient (~1:1), with a real hedged portfolio behind it
  • +Largest synthetic dollar, Ethena's USDe, is a real, sizable product
Trade-offs
  • Funding-rate risk: negative perp funding erodes the backing
  • Exchange and custody counterparty risk (reserves sit on CEXs and with custodians)
  • A demonstrated flash-depeg history (USDe printed ~$0.65 on one venue in October 2025)

A synthetic, delta-neutral dollar is backed by a hedged position, long spot crypto and an equal short perpetual future, so price exposure nets to about zero and the peg rests on the hedged portfolio plus mint-and-redeem arbitrage. This is not seigniorage, so calling it algorithmic (as some trackers do) is wrong. Ethena's USDe (~$4B, down from a ~$14B peak) is the largest example, and its staked form, sUSDe, pays a variable yield, in the mid-teens through 2024 to 2025 but compressed to roughly 4% by mid-2026 as perpetual funding cooled.

The risks are specific. Funding rates can turn negative and erode the backing, so Ethena runs a reserve fund to absorb that and rotates part of the backing into stablecoins and tokenized Treasuries when funding is thin (meaning the backing is not purely spot-plus-perp). The collateral also lives on exchanges and with custodians rather than in a public vault, a real counterparty risk, the same bucket that sank Elixir's deUSD in November 2025 when its backing was parked with a failing external manager and fell to about $0.02. USDe itself printed as low as ~$0.65 on one venue during the October 2025 selloff, an order-book dislocation rather than a reserve failure, and held near par elsewhere. One legal nuance: once the GENIUS Act's ban on payment-stablecoin interest is in force, designs like this must separate yield into a staked token, which is already why USDe itself is not yield-bearing but sUSDe is.

Verdict

Which should you use?

If your priority is
Regulatory fit, liquidity, and simplicity

you want the model regulation is written for (GENIUS, MiCA), maximum liquidity, and simple 1:1 backing, and holders earning yield is not the point.

Fiat-backed
If your priority is
On-chain transparency and censorship-resistance

you want a fully on-chain, verifiable dollar and can accept capital inefficiency and liquidation risk.

Crypto-backed (CDP)
If your priority is
Native yield with a different risk profile

you want a capital-efficient dollar that earns yield through hedging, and you understand the funding-rate and exchange-counterparty risks; never market it as cash-like.

Synthetic
If your priority is
A pure algorithmic design

there is no healthy example at scale; if capital efficiency is the goal, use over-collateralization or a hedged model, not seigniorage.

Avoid
Also consider

Adjacent instruments and context

  • Tokenized money-market funds (BlackRock BUIDL, Circle USYC): not stablecoins. They are tokenized securities with a stable NAV that DO pay yield, restricted to whitelisted holders. They are the institutional-cash complement to a payment stablecoin, not a substitute, and they exist partly because payment stablecoins legally cannot pay yield.
  • RWA-backed / yield-bearing dollars (Ondo USDY, Ethena USDtb): tokens that look like a stablecoin but pass through Treasury yield, usually structured as securities to stay legal under the GENIUS no-yield rule.
  • Regulatory currency: the US GENIUS Act (2025) sets reserve, disclosure, and no-yield rules for payment stablecoins, with rulemaking through 2026 and effect no later than January 2027; MiCA's e-money-token rules applied from June 2024, and its CASP transition period ended on 1 July 2026, after which EEA venues delisted USDT while USDC stayed. Confirm the framework with counsel before issuing.

FAQ

Is USDT banned in Europe under MiCA?
Not banned outright, but restricted. Under the EU's MiCA regime, fiat-backed stablecoins are regulated as e-money tokens and must meet reserve, authorization, and disclosure rules. After MiCA's transition period ended in July 2026, USDT was delisted for EEA retail users on regulated European venues because it lacked that authorization, while USDC, which secured it, stayed listed; PayPal's PYUSD is another authorized entrant. So in the EU the real question is whether a stablecoin holds MiCA e-money-token authorization, not whether stablecoins are allowed at all.
What is the most common type of stablecoin?
Fiat-backed, by a wide margin. USDT (~$184B) and USDC (~$73B) together are roughly 83% of the ~$310B stablecoin market, and slowly declining, per DefiLlama. They hold cash and short-dated Treasuries 1:1 and hold their peg through issuer redemption. Crypto-backed dollars like USDS (formerly DAI) are a much smaller share, pure algorithmic stablecoins are effectively gone after Terra, and synthetic dollars like Ethena's USDe are a newer, smaller category. Over 99% of stablecoin value is US-dollar-denominated.
Are algorithmic stablecoins dead?
Pure algorithmic stablecoins, the seigniorage model with little or no collateral, are effectively dead at scale. Terra's UST collapsed in May 2022, and the combined UST and LUNA wipeout exceeded $40B, which shows why: without an exogenous collateral floor, a loss of confidence becomes a death spiral. What often gets mislabeled as algorithmic is the synthetic, delta-neutral model (Ethena's USDe), which is different because it is backed by a hedged long-spot, short-perp position rather than a paired governance token. If capital efficiency is the goal, over-collateralization or a hedged design is the answer, not seigniorage.
Is Ethena's USDe an algorithmic stablecoin?
No, and the distinction matters. USDe is a synthetic, delta-neutral dollar: it is backed by long spot crypto hedged with an equal short perpetual future, so its value comes from a real, hedged portfolio, not from expanding and contracting a paired token. Its risks are funding-rate risk (perp funding can turn negative) and exchange and custody counterparty risk, not the reflexive death-spiral risk of algorithmic designs, though it did print as low as ~$0.65 on one venue during the October 2025 selloff. Note that USDe itself does not pay yield; its staked form, sUSDe, does, a structure driven partly by the GENIUS Act's ban on payment stablecoins paying interest.
How do the GENIUS Act and MiCA affect which model to build?
Both are written primarily for fiat-backed payment stablecoins, which is part of why that model dominates. The US GENIUS Act (2025) sets 1:1 reserve requirements, monthly disclosure and attestation, and, once it takes effect, prohibits paying interest to holders, which is already why yield-bearing designs route yield through a separate staked token or structure as a security. In the EU, MiCA has classified fiat-pegged stablecoins as e-money tokens (requiring a licensed issuer and 1:1 segregated reserves) since June 2024, and after its transition period ended in July 2026, USDT was pushed off EEA retail venues while USDC stayed. If you are issuing, the regulatory model often decides the design as much as the technology does.

Reviewed by Luis Medeiros, Field CTO at Protofire. Last updated: July 2026.

Build it

We build stablecoin systems: the collateral and peg mechanics, oracle and liquidation design, reserve attestation, and the governance and monitoring that keep them solvent.

native stablecoin development

Book a call with Alejandro Losa

Issuing a stablecoin or a yield-bearing dollar? Talk to us about the right model and the reserve, peg, and compliance stack.

Protofire 2026. All rights reserved

Message us on Telegram