Stablecoins look simple on a trading screen: one token, one dollar. Under the hood, however, the path to that dollar differs radically across fiat-backed, crypto-backed, and algorithmic designs. That distinction matters because stablecoins are now core DeFi collateral, the quote currency for perpetual futures, the settlement layer for cross-border crypto flows, and the base asset for a large share of on-chain yield strategies. A wallet holding USDT, USDC, DAI, LUSD, or an algorithmic dollar is not holding the same risk; it is choosing a balance sheet, a liquidation engine, or a game-theoretic promise.
The market has already voted on the first-order question. Fiat-backed stablecoins dominate because they are capital-efficient, liquid, and easy for centralized exchanges and payment firms to integrate. But DeFi users should not confuse dominance with purity. Fiat-backed tokens import banking, custody, and regulatory risk; crypto-backed tokens import collateral volatility and oracle risk; algorithmic stablecoins import reflexivity, where confidence itself is the collateral. The correct framework is not which model is best in the abstract, but which failure mode you are being paid to underwrite.
Fiat-backed stablecoins: the money-market fund in token form
Fiat-backed stablecoins such as Tether's USDT and Circle's USDC are structurally closest to narrow banks or tokenized money-market funds. Users deposit dollars or dollar-equivalent assets with an issuer, and the issuer mints tokens that trade on public blockchains. Redemptions run in reverse: authorized participants return tokens and receive fiat. The peg is maintained primarily by arbitrage. If USDC trades at $0.995, a qualified redeemer can buy it in the market, redeem at $1, and capture the spread, assuming the issuer honors redemption quickly.
The key source of strength is capital efficiency. A fully fiat-backed issuer can theoretically mint $1 of stablecoins for every $1 of reserves, unlike overcollateralized crypto systems that may require $1.50 or $2.00 of collateral for each dollar issued. That efficiency explains why fiat-backed stablecoins tend to have the deepest exchange liquidity, the tightest spreads, and the broadest merchant and institutional acceptance. USDT has historically been especially dominant in offshore trading pairs, while USDC has had stronger penetration in U.S.-regulated venues and DeFi integrations.
The reserve composition is the analytical center of the model. The safest version is cash and short-dated U.S. Treasury bills held with segregated custodians, marked transparently, with frequent attestations or audits. Tether reported tens of billions of dollars in Treasury exposure in 2024 and generated multi-billion-dollar quarterly profits during the high-rate cycle because reserve assets yielded more than it paid token holders: usually zero. Circle similarly benefited from higher short-term rates through its reserve structure, including cash and Treasury-backed arrangements with large asset managers and custodial banks.
That spread is good for issuer solvency but raises a tokenomics point often missed by retail users: stablecoin holders supply cheap funding while issuers capture the interest income. In DeFi, this creates a yield hierarchy. Passive USDC in a wallet earns nothing, but USDC lent through Aave, supplied to a Curve pool, or converted into a yield-bearing wrapper can share some market yield. The additional return is not free; it layers smart contract risk, liquidity risk, and sometimes rehypothecation risk on top of issuer risk.
The main weakness is off-chain dependency. When Silicon Valley Bank failed in March 2023, USDC briefly depegged to roughly $0.88 on some venues after Circle disclosed $3.3 billion of exposure. The peg recovered after U.S. authorities protected deposits, but the episode proved that a transparent, regulated stablecoin can still suffer a bank-run discount if reserve access is uncertain over a weekend. USDT has faced the opposite critique: deep liquidity and strong profitability, but persistent market scrutiny over reserve transparency and jurisdictional complexity. In both cases, the token is only as strong as redemption plumbing, custodian quality, and legal enforceability.
Crypto-backed stablecoins: decentralization bought with overcollateralization
Crypto-backed stablecoins replace bank deposits with on-chain collateral. MakerDAO's DAI is the most important example, although its design has evolved from ETH-heavy collateral to a mixed portfolio that includes centralized stablecoins and real-world assets. Liquity's LUSD represents a more purist version: ETH-backed borrowing with algorithmic redemptions and liquidation mechanisms. The basic principle is straightforward: users lock volatile crypto assets in a smart contract and mint a smaller amount of stablecoin against them.
The peg is protected by collateral ratios and liquidations. If a user deposits $10,000 of ETH and mints $5,000 of DAI, the vault starts at 200% collateralization. If ETH falls, the system liquidates before the debt becomes undercollateralized. In Maker, each collateral type has its own liquidation ratio, debt ceiling, stability fee, and oracle configuration. In Liquity, Troves historically required a minimum collateral ratio near 110%, with system-level safeguards and redemptions designed to push LUSD back toward par.
This model is more transparent than fiat backing because liabilities and collateral can be monitored on-chain in real time. It is also composable: DAI can be supplied to Aave, paired in Curve pools, routed through Uniswap, or locked into Maker's savings products. But the price of decentralization is inefficient collateral usage. To mint $1 billion of a crypto-backed stablecoin safely, the system may need $1.5 billion to $3 billion of volatile collateral, depending on asset quality, liquidation depth, and governance risk tolerance.
Live market context makes this concrete. With ETH trading around $1,725.66 in the provided snapshot and up 1.52% over 24 hours, a DAI borrower using ETH collateral is not just making a dollar liquidity decision; they are running a leveraged ETH position. If ETH drops 25%, a vault that looked conservative at 180% collateralization can move close to liquidation thresholds depending on parameters. The stablecoin may remain near $1, but the borrower absorbs volatility through liquidation penalties, auction slippage, and gas costs.
The deeper risk is not merely price volatility but liquidation capacity. During the March 2020 market crash, Maker suffered stressed auctions and bad debt when Ethereum congestion and falling collateral prices overwhelmed keepers. The system survived, but only after emergency governance actions and recapitalization. Since then, DeFi protocols have invested heavily in oracle design, keeper incentives, circuit breakers, and collateral diversification. Even so, crypto-backed stablecoins remain vulnerable to correlated market drawdowns, especially when collateral includes assets whose liquidity disappears precisely when liquidations are needed.
Maker's evolution also shows a philosophical trade-off. DAI began as a decentralized, ETH-backed stablecoin, but the Peg Stability Module allowed large amounts of USDC to enter the system to stabilize the peg. Later, Maker expanded into Treasury bills and real-world assets, making DAI more robust from a yield and scale perspective but less purely crypto-native. Spark Protocol and the DAI Savings Rate helped route real-world yield back to users, at times making sDAI one of the cleanest on-chain expressions of Treasury-linked yield. The result is neither purely fiat-backed nor purely crypto-backed; it is a hybrid balance sheet governed on-chain.
Algorithmic stablecoins: when confidence becomes collateral
Algorithmic stablecoins attempt to maintain a peg without full external collateral. Designs vary, but most rely on supply expansion and contraction, seigniorage shares, bonding mechanisms, or a volatile companion token. When the stablecoin trades below $1, users are incentivized to burn it for another asset or future claim; when it trades above $1, new supply enters circulation. In theory, the market arbitrages the peg. In practice, the mechanism works only while participants believe the future claims are valuable.
TerraUSD, or UST, is the canonical failure. UST could be swapped for $1 worth of LUNA, and LUNA absorbed peg pressure. The design scaled quickly because Anchor Protocol paid roughly 20% yields on UST deposits, creating massive demand for the stablecoin. But when withdrawals accelerated in May 2022, UST redemptions minted increasing amounts of LUNA, LUNA's price collapsed, and the collateral illusion vanished. The death spiral destroyed tens of billions of dollars in market value and turned algorithmic stablecoins from a growth narrative into a systemic-risk case study.
The lesson is not that all dynamic-supply mechanisms are useless. It is that undercollateralized pegs need a credible buyer of last resort. If the stabilizing asset is the same ecosystem token whose value depends on the stablecoin's success, the design is circular. Reflexivity works upward during growth and violently downward during stress. This is why Frax, once known for a partially algorithmic model, moved toward full collateralization after 2022. The market learned to price hard collateral above elegant incentive design.
Algorithmic systems also struggle with liquidity timing. A stablecoin can appear stable in shallow markets when redemptions are small, but the relevant test is exit capacity under fear. If $500 million attempts to leave a protocol whose backstop liquidity is $50 million of immediately sellable assets plus a volatile governance token, the peg is not a dollar; it is an option on market confidence. For DeFi risk managers, the correct metric is not the average peg deviation in calm periods but the size and quality of collateral available during a one-day or one-week redemption shock.
Peg defense is a stack, not a slogan
Across all three models, the peg is defended by a stack of mechanisms: redemption rights, collateral quality, arbitrage incentives, market liquidity, oracle reliability, and governance response. Fiat-backed stablecoins have the cleanest redemption path but the most off-chain opacity. Crypto-backed stablecoins have the most transparent collateral but require robust liquidation infrastructure. Algorithmic stablecoins can be capital-efficient but are fragile when confidence breaks.
Investors should separate price stability from solvency. A stablecoin can trade at $1 because liquidity is strong, even if its reserve quality is uncertain. Conversely, a solvent stablecoin can temporarily trade below $1 if redemption rails are closed or market makers are balance-sheet constrained. USDC's 2023 depeg was a liquidity and banking-access shock more than an asset-quality collapse. UST's 2022 depeg was a solvency and reflexivity collapse. Those are different risk categories, and they deserve different position sizing.
Three metrics are particularly useful for due diligence:
- Redemption quality: who can redeem, in what size, on what schedule, through which legal entity, and into which bank account or wallet?
- Collateral liquidity: how much of the backing can be converted into dollars or ETH within 24 hours without large slippage?
- Stress incentive alignment: do arbitrageurs make money restoring the peg when volatility spikes, or does the mechanism require them to catch a falling knife?
Regulation is adding another layer. The European Union's MiCA framework imposes reserve, disclosure, and authorization requirements for stablecoin issuers, while U.S. legislation remains fragmented. For fiat-backed issuers, clearer rules could deepen institutional adoption but compress margins and limit offshore flexibility. For DeFi-native stablecoins, regulatory pressure on centralized collateral may increase demand for ETH-backed alternatives, but also raise scrutiny of governance, front ends, and real-world asset exposures.
Yield strategies: know what pays you and what can break
Stablecoin yield is often marketed as low risk because the unit of account is stable. That is misleading. The yield source determines the risk. Lending USDC on Aave earns borrower interest and sometimes token incentives. Supplying liquidity to Curve's 3pool-like structures earns trading fees and governance-directed rewards but exposes users to pool imbalance if one asset depegs. Holding sDAI earns yield routed from Maker's asset base, including real-world assets and protocol revenues. Providing liquidity to a smaller algorithmic stablecoin pool may offer high APR precisely because the market demands compensation for peg and exit risk.
A practical portfolio approach is to diversify by failure mode rather than by ticker. Holding USDT and USDC diversifies issuer-specific risk but not the broader fiat-custody model. Combining USDC, DAI or sDAI, and a smaller allocation to ETH-backed LUSD-style assets spreads exposure across bank reserves, on-chain collateral, and governance mechanisms. For treasuries, the safest operational setup also includes venue diversification: do not rely on a single bridge, lending market, or stablecoin pool for liquidity during stress.
Bridges deserve special caution. A bridged stablecoin is not the same as the native asset. USDC on Ethereum issued by Circle differs from a wrapped representation on a smaller chain secured by a bridge contract or multisig. Many historical DeFi losses, from bridge exploits to oracle manipulation, occurred not because the stablecoin issuer failed but because the transport layer failed. In a crisis, liquidity concentrates on mainnet and the deepest centralized exchanges first; peripheral chains can trade at discounts even when the underlying issuer remains solvent.
The stablecoin endgame is hybrid, regulated, and yield-aware
The next phase of stablecoins will not be a single winning model. Fiat-backed tokens will likely remain the dominant settlement asset because they scale efficiently and integrate with banks, exchanges, and payment companies. Crypto-backed stablecoins will remain essential for users who value censorship resistance, transparent collateral, and composability. Pure algorithmic stablecoins will struggle to regain institutional trust unless they carry far more exogenous collateral than earlier designs allowed.
The most important trend is convergence. Fiat-backed issuers are becoming more like regulated money-market funds. Crypto-backed systems are using real-world assets to improve yield and peg depth. Algorithmic projects are quietly rebranding into overcollateralized or hybrid models. For DeFi users, the opportunity is not to memorize labels but to underwrite mechanisms: reserves, redemption, liquidations, governance, and incentives.
The stablecoin question is no longer whether a token usually trades at $1. It is what has to be sold, frozen, liquidated, governed, or believed for that $1 to remain real under stress.
That is the analytical edge. Stablecoins are the base layer of DeFi risk, not the risk-free asset. The protocols and investors that treat them as engineered financial products, rather than digital cash equivalents, will be better positioned when the next peg test arrives.