Stablecoins, demystified: what makes them stable (and what makes them not)
A few hundred billion dollars of crypto pretends to be the US dollar. Some of it is genuinely dollar-backed; some of it is a clever computer-science magic trick. The difference matters a lot.
Stablecoins are the most boring and most important thing in crypto. Boring because their entire pitch is "we will be worth exactly one dollar, forever." Important because the global stablecoin float is now north of $200 billion and growing, and on any given day they settle more transaction volume than Visa. They're how crypto actually moves.
But "stablecoin" is a marketing word, not a category. Under the hood, there are at least three different mechanisms doing very different things. Mixing them up is how people lose money in spectacular ways.
The plain-English bit
A stablecoin is a token that claims to be worth $1. Different stablecoins make that claim in different ways, and the way matters. Some hold actual dollars in a bank. Some hold other crypto and pinky-promise the math works. Some use code and crossed fingers. Only one of those three has a track record of holding up under stress.
Type 1: Fiat-backed (USDC, USDT)
The simplest design. Issuer takes a real US dollar, puts it in a bank or a short-term Treasury bill, and mints one token. Burn the token, get the dollar back. The peg holds because there's an arbitrage: if the token trades at $0.99, you buy it, redeem it, get $1.00, pocket a penny, repeat until the price recovers.
Circle's USDC and Tether's USDT are the two giants here. How they hold those reserves is wildly different.
| Coin | Issuer | Audit cadence | Reserve quality | Past wobbles |
|---|---|---|---|---|
| USDC | Circle (US) | Monthly attestations, Big-4 auditor | T-bills, repo, cash at large US banks | SVB weekend 2023: briefly traded at $0.88 before redemptions reopened |
| USDT | Tether (offshore) | Quarterly attestations, less detailed | T-bills majority, plus secured loans, BTC, gold | Multiple short de-pegs over the years; survived all of them |
USDC is the one regulators are comfortable with. USDT is the one everyone actually uses outside the US. Both of these have, so far, paid out at par when it mattered. Both also depend entirely on the issuer's ability to liquidate reserves and on the trustworthiness of the bank holding them, which is not nothing.
Type 2: Crypto-collateralized (DAI, sDAI)
MakerDAO's DAI is the OG of this category. The mechanic is roughly: lock up $150 of ETH (or other crypto) in a smart contract, mint $100 of DAI against it. If your collateral falls in value, the contract auctions it off automatically to keep the system solvent. Over time, DAI has quietly become a hybrid — a big chunk of its backing is now actually USDC and Treasury bills, which is a fascinating philosophical retreat from "purely crypto-native money."
The strength: it doesn't depend on a bank or an issuer's honesty. The weakness: if crypto crashes faster than the system can liquidate, the peg breaks. So far, the math has held.
Type 3: Algorithmic (RIP, mostly)
These ones have no real backing. The peg is supposed to be maintained by a clever feedback loop between two tokens: print more of the unstable one when the "stable" one trades above $1, vice versa when below. In theory, it's beautiful financial engineering. In practice, the design has a name in the industry: "the death spiral."
Terra/Luna in May 2022 was the loudest example. The "UST" stablecoin lost its peg, the algorithm minted huge amounts of LUNA to defend it, LUNA's price collapsed to fractions of a cent, and roughly $40 billion of paper value evaporated in 72 hours. Several smaller projects have run the same experiment since, with the same result.
Cult-warning bit
If a stablecoin claims to maintain its peg "algorithmically" without significant real collateral behind it, treat it the way you'd treat a bridge that's only held up by a really compelling math proof. It might be fine. The downside, if it isn't, is total.
Why the boring kind keeps getting bigger
The genuinely useful thing about USDC and USDT is that they're dollars that work in places dollars don't easily work. Argentina, Turkey, Nigeria, parts of Southeast Asia — anywhere local currency is unstable or moving money cross-border is painful — stablecoins are doing what the US dollar already does informally, just faster and on phones.
That use case is the actual product. Speculators and traders use them as a parking spot between trades. Builders use them as the unit of account inside DeFi. Treasurers at crypto firms use them like a checking account. None of that has much to do with the crypto-bro mythology. It's just rails.
What we'd watch
- Reserve transparency. Monthly, third-party attestations of holdings, with composition breakdowns. If you can't see what's behind the token, treat the token as a piece of paper that says "trust me."
- Yield share. Some new entrants — Ondo, Mountain, sDAI — pay the Treasury-bill yield through to holders. The big two (USDC, USDT) currently don't, which is partly why they print money. Watch whether competition forces that to change.
- Regulation. The US is slowly inching toward a stablecoin framework. The clearer the rules, the more boring the category gets — and "boring" is the goal.
A stablecoin is only as boring as its boringest reserve asset. Audit the reserve, not the marketing.
Disclosure: No member of the Money Market desk has a position in MakerDAO, Circle, Tether, or any specific stablecoin issuer at publication. Token sizes and reserve breakdowns reflect publicly disclosed information as of early 2026; figures change frequently and should be verified against current issuer attestations.