Stable by Design

What stability means

"Pegged to $1" doesn't mean every exchange quotes exactly $1.000000 at every moment. It means the price is kept close to $1 by a mechanism that makes deviations self-correcting. The mechanism is what matters — not the price at any given second.

The peg works through arbitrage. If USDC trades at $0.99 on a DEX, arbitrageurs buy it there and redeem it directly with Circle for $1.00 — a risk-free $0.01 profit per token. Demand for cheap USDC pushes the price back up. If USDC trades at $1.01, arbitrageurs mint new USDC from Circle for $1.00 and sell it on the DEX for $1.01. Supply increases until the premium disappears. As long as redemption is reliable and unrestricted, price stays within a tight band around $1. This is why the redemption mechanism is the actual product — everything else is implementation detail.

When the peg breaks

USDT, 2018. Tether's peg slipped to $0.94 amid concerns about whether reserves actually existed. Arbitrage eventually restored it — but only because enough traders were willing to bet on Tether's survival. Confidence in the redemption mechanism, not the mechanism itself, drove recovery.

USDC, March 11, 2023. Circle disclosed it held $3.3 billion at Silicon Valley Bank as regulators took the bank over. USDC fell to $0.88 within hours — the largest de-peg for a regulated stablecoin in history. The FDIC's announcement that all SVB deposits would be made whole brought USDC back to $1.00 over the weekend. The peg recovered not because of USDC's mechanism, but because the US government backstopped the bank.

TerraUSD (UST), May 2022. The defining failure. UST was an algorithmic stablecoin with no real reserve — its $1 peg depended entirely on confidence in the mint/burn mechanism. When that confidence broke, the mechanism accelerated the collapse rather than containing it. UST went from $1.00 to $0.01 in five days. $40 billion in combined UST and LUNA value was destroyed.

Why it matters

Stablecoins are DeFi's money market. Lending protocols denominate debt in them. AMMs use them as the reference leg of every pair. Cross-border payments depend on them as settlement layer. When the dollar-denominated unit drifts by 5%, every protocol assuming stability takes unexpected losses. When it collapses by 90%, the damage is existential — for users, for protocols, and increasingly for regulatory frameworks trying to contain the fallout.

Three architectures

Fiat-backed (reserve model). Every token is backed by a dollar (or dollar-equivalent) held off-chain by a custodian. Peg holds as long as the custodian is solvent and the redemption channel stays open.

Crypto-backed (overcollateralized). Tokens are minted against on-chain collateral worth more than the tokens issued. No off-chain custodian. Peg holds as long as collateral stays above the liquidation threshold.

Algorithmic (seigniorage). No reserve. The protocol mints or burns a second token to balance supply and demand. Peg holds as long as enough market participants believe it will hold. When belief collapses, the mechanism fails in the worst possible way.

Fiat-backed (USDC, USDT)

The simplest model: for every stablecoin in circulation, the issuer holds $1 of cash or cash equivalents in bank accounts and money market funds. Send 1 USDC to Circle, get $1 wired to your bank. The cryptographic part handles settlement speed and composability; the reserve handles solvency.

USDT: the first mover

Tether launched in 2014, and with $100B+ in circulation it remains the largest stablecoin by market cap. It's also the most controversial. For years Tether claimed 1:1 cash backing. A 2021 CFTC settlement revealed the reality: Tether's reserves had included commercial paper, secured loans, and other assets — not pure cash. Tether paid an $41M fine without admitting wrongdoing.

Current Tether attestations show reserves consisting primarily of US Treasuries, which is a meaningful improvement. But Tether is incorporated in the Cayman Islands, has never completed a full independent audit (only quarterly attestations from a smaller accounting firm), and has historically been opaque about its banking relationships. Traders use it because the liquidity is unmatched. Institutional holders tend to prefer USDC.

USDC: regulated, transparent, and still vulnerable

Circle is licensed by the New York Department of Financial Services and publishes monthly reserve attestations from Grant Thornton. Holdings are restricted to cash and short-duration US Treasuries. This is as close to "fully audited fiat-backed stablecoin" as currently exists in the market.

The SVB event exposed the limits of even this model. On March 10, 2023, Circle disclosed that $3.3 billion of USDC's approximately $40 billion in reserves was held at Silicon Valley Bank — which regulators had just seized. USDC fell to $0.88 as holders panicked about whether those funds would be recoverable. The FDIC's Sunday announcement that all SVB depositors would be fully backstopped restored confidence, and USDC returned to $1.00 before markets opened Monday. The peg held — but only because of a government intervention that wasn't guaranteed when the crisis began.

The trust model

Using a fiat-backed stablecoin means trusting the issuer, the custodian banks, and the regulator. This is deliberately centralized — and that centralization enables censorship. Circle maintains a blocklist of wallet addresses that cannot send or receive USDC; as of 2024, roughly 75,000 addresses had been frozen. This is explicitly intentional: regulatory compliance requires the ability to freeze funds associated with sanctions violations, court orders, and law enforcement requests. It works exactly as designed. It also fundamentally contradicts the permissionless premise of crypto.

Overcollateralized (DAI)

MakerDAO's DAI has been running continuously since 2017 — longer than any other decentralized stablecoin. It works by requiring users to post more collateral than they borrow, making undercollateralization structurally impossible under normal conditions. The on-chain mechanism, not issuer trust, enforces solvency.

How Vaults work

To mint DAI, you open a Vault and lock collateral — originally only ETH, now a range of assets. An ETH vault might require a 150% collateralization ratio: you lock $150 of ETH to borrow 100 DAI. If ETH's price falls and your collateral drops below the minimum ratio, the protocol automatically liquidates your position — selling collateral at a discount to repay the DAI debt and restore system solvency. No bank, no counterparty. The math enforces it.

The Stability Fee is the interest rate on DAI borrowing, set by MKR governance. A high stability fee makes borrowing DAI expensive, reducing supply and pushing the price up toward $1. A low stability fee makes borrowing cheap, expanding supply and pushing price down. The DAI Savings Rate (DSR) works on the demand side: the protocol pays interest to DAI holders who lock into the DSR contract, increasing demand and supporting the peg from below.

Multi-collateral DAI and the centralization tradeoff

MakerDAO expanded from ETH-only collateral to include WBTC, real-world assets (US Treasuries held via Centrifuge and Monetalis), and USDC via the Peg Stability Module. The PSM allows direct 1:1 swaps between DAI and USDC, functioning as a hard peg anchor. It works — but it means roughly half of DAI's collateral is now USDC, introducing the same issuer and regulatory risks that DAI was designed to avoid. Decentralization and peg stability pulled in opposite directions, and peg stability won.

Spark Protocol, MakerDAO's lending front-end, lets users borrow at DSR-linked rates using the same collateral system. In 2024, MakerDAO rebranded to Sky and introduced USDS as a soft successor to DAI — same mechanics, new name, with old DAI gradually being soft-deprecated.

Battle-tested, including the hardest test

March 12, 2020 — "Black Thursday" — ETH dropped roughly 50% in hours. DAI's collateral auctions broke down under network congestion: liquidation bots couldn't get transactions included, gas spiked, and some bots bid $0 on collateral auctions and won. The protocol accumulated $8.3M in bad debt. MakerDAO covered it with a dilutive MKR token auction. Critically, DAI held — it briefly traded at $1.06 as ETH collateral collapsed, but never broke significantly. The protocol survived and patched its auction mechanism. That event, and seven subsequent years of operation, is why DAI is the benchmark for decentralized stablecoins: not the most capital-efficient, but the most proven.

Algorithmic — and why it failed

The appeal of algorithmic stablecoins was obvious: a $1 token with no bank account, no custodian, no reserve to audit or trust. Just code and incentives. The problem is that the mechanism designed to restore the peg becomes the mechanism of destruction when confidence fails.

Terra/LUNA: the defining collapse

TerraUSD (UST) maintained its $1 peg through a mint/burn relationship with LUNA, the native token of the Terra blockchain. To mint 1 UST, you burned $1 worth of LUNA. To redeem 1 UST, you received $1 worth of LUNA. No reserves, no collateral — just the expectation that LUNA would always be worth something.

The ecosystem had $18 billion of UST in circulation by May 2022. Anchor Protocol was offering 20% APY on UST deposits — an unsustainable rate funded by ecosystem reserves that were visibly depleting. When large UST positions began unwinding (possibly coordinated), the peg slipped to $0.98. This was enough to trigger the mechanism: holders redeemed UST for LUNA at the protocol rate. But LUNA's price was falling as more of it was minted into existence, which meant each UST redemption required minting even more LUNA to reach $1 of value — which pushed LUNA's price lower — which required more LUNA to be minted — which pushed it lower still.

UST hit $0.10, then $0.01. LUNA went from $80 to fractions of a penny. $40 billion in combined value was destroyed in 72 hours. The mechanism worked exactly as designed. The design was fatal.

The same pattern, repeatedly

Terra was not an isolated failure. Empty Set Dollar (ESD), Basis Cash (BAC), and Dynamic Set Dollar all attempted seigniorage-based stability mechanisms in 2020 and 2021. All collapsed to near-zero following the same structural logic: in a crisis, the governance token inflates precisely when demand for it has collapsed. The death spiral is not a bug in any particular implementation — it is intrinsic to the architecture. Minting the token that is already crashing to stabilize the stablecoin that is already de-pegging cannot work.

What survived

FRAX began as a partially algorithmic stablecoin — partially collateralized, partially stabilized by its FXS governance token. Post-Terra, FRAX progressively increased its collateral ratio toward 100%. The "algorithmic" component has been largely abandoned. The lesson the market drew from the failure of every seigniorage stablecoin is that stability without reserves is stability in name only. Serious projects have either added real collateral or shut down.

Yield-bearing stablecoins

A new category emerged from the convergence of DeFi and traditional finance: stablecoins that pay yield to holders without requiring any action. The yield is embedded — either through token price appreciation (wrapped model) or rebasing. The catch is that yield always comes from somewhere, and where it comes from determines the risk.

sDAI: the conservative baseline

Spark Protocol (MakerDAO's lending front-end) wraps DAI at the DAI Savings Rate. If the DSR is 5%, sDAI appreciates 5% annually against DAI — no staking, no claiming, no protocol interaction required beyond the initial wrap. The yield source is the stability fees paid by DAI borrowers. Risk is limited to MakerDAO's smart contract risk, which has been production-tested for over seven years. sDAI is the conservative end of the yield-bearing stablecoin spectrum.

USDe: delta-neutral yield

Ethena's USDe launched in early 2024 and reached $3 billion in circulating supply within months — the fastest growth in stablecoin history. The mechanism: Ethena holds long spot cryptocurrency (ETH, BTC) and simultaneously holds short perpetual futures positions of equal size, creating a delta-neutral portfolio. The funding rate paid on the short position — typically positive in bull markets, as longs pay shorts — generates yield. During the 2024 bull market, USDe yields ranged from 15% to 30%.

The risks are specific and real. Ethena's futures positions are held at centralized exchanges — Binance, OKX, Bybit — creating exchange counterparty risk. If an exchange fails or freezes assets, the hedge breaks. Funding rates can and do go negative in bear markets, at which point holding USDe earns a negative yield and the protocol must draw on reserve funds to maintain the peg. USDe is not a reserve stablecoin and not a risk-free dollar. It is a sophisticated structured product packaged as a stablecoin.

RWA stablecoins: tokenized T-bills

USDY (Ondo Finance) tokenizes short-term US Treasuries and bank demand deposits. Yield equals approximately the T-bill rate — around 5% through 2024. Issued by a regulated entity, requires KYC, and is not available to US retail investors. This is the institutional RWA model: TradFi yield accessed via DeFi rails, with TradFi compliance requirements still attached.

Mountain Protocol's USDM follows the same model — fully backed by short-term Treasuries, yield passed directly to holders, regulated issuer. The distinction from USDY is structural detail and jurisdiction rather than fundamental approach.

The risk spectrum

sDAI carries smart contract risk only — the simplest risk profile in this category. USDM and USDY add issuer and regulatory risk: the entity holding the Treasuries can be sanctioned, regulated out of existence, or freeze your tokens. USDe adds exchange counterparty risk (three major CEXs hold the hedge) and funding rate risk (yield can go negative). Higher yield reflects higher and more varied risk, not free money.