Stablecoins Explained: How USDT and USDC Differ from Bitcoin

A stablecoin is a cryptocurrency built to do the opposite of what the rest of the market is known for: hold a steady price, almost always pegged one-to-one to the US dollar. That single design goal is what separates Tether (USDT) and USD Coin (USDC) from Bitcoin. Bitcoin fixes its supply and lets the price move. A stablecoin fixes its price and lets the supply move.

That distinction is the whole thing, and most explainers stop there. It is worth going further, because the interesting question is not how the three stablecoin designs work in theory. It is which one actually holds the money.

The three designs, and what each is actually worth

There are 328 dollar-pegged stablecoins with reported supply, totalling $306.6 billionDefiLlama asset register, all assets with pegType peggedUSD, summing self-reported circulating supply. as of 3 August 2026.¹ DefiLlama classifies them by peg mechanism, and the split is not close:

• Fiat-backed — 91.6% ($280.8 billion). The issuer holds cash and short-term Treasuries and promises to redeem. USDT and USDC are both here.
• Crypto-backed — 8.3% ($25.5 billion). Overcollateralised with on-chain assets. Sky Dollar (USDS), Dai and Ethena’s USDe sit in this group.
• Algorithmic — 0.09% ($273.9 million). No meaningful reserve; the peg is defended by supply mechanics.

The algorithmic argument is over. It lost.

Four years after Terra, algorithmic stablecoins still absorb a disproportionate share of the public argument about whether stablecoins are safe. Nine hundredths of one percent of the money is in them. Whatever you think of the design, it is no longer a systemic question, and treating it as one is a way of not discussing the actual exposure.

The actual exposure is concentration. Tether holds 59.7% of all dollar-pegged supply and USD Coin holds 23.5%DefiLlama asset register, 3 Aug 2026. Tether $183.1bn, USD Coin $72.0bn, against a total of $306.6bn.¹ — 83.2% between two companies. If you are worried about stablecoin risk, the mechanism debate is a distraction from a counterparty question with two names in it. That is our view, and the number is the reason for it.

How this differs from Bitcoin

Bitcoin’s supply is capped at 21 million and its price is whatever the market says. It has no issuer, no reserve and no redemption promise, which means it also has no counterparty to fail. A stablecoin inverts every one of those properties. Its price is supposed to be fixed, its supply expands and contracts with demand, and its stability depends entirely on an institution honouring redemptions. Bitcoin can fall 40% without anything breaking. A stablecoin falling 4% means something has already broken.

Neither is safer in the abstract. They fail in different directions, and the mistake is holding one while pricing the risk of the other.

What to check before you trust one

Attestation is not audit. Most large issuers publish periodic attestations, in which an accounting firm confirms that reserves existed on a stated date. That is a photograph, not a film, and it is a weaker assurance than a full audit. Check who signs it, how often, what sits in the reserve, and whether the issuer commits to redeeming at par for ordinary holders or only for large institutional accounts. The last of those is where the difference between a peg and a price usually shows up.

How a fiat-backed stablecoin actually holds its peg

The peg is not held by the reserve. It is held by the redemption right, and the two are not the same thing. What keeps a fiat-backed token at a dollar is a small number of approved counterparties who can mint new tokens by wiring dollars to the issuer and burn tokens by asking for dollars back. If the token trades at $0.995 on an exchange, those counterparties buy it there and redeem it at par for a dollar, and the buying closes the gap. If it trades at $1.005, they mint at par and sell. The reserve matters because it determines whether that redemption can be honoured at size and at speed. It is the collateral behind the mechanism, not the mechanism.

This is why the terms of access matter more than the headline reserve figure. If redemption is open only to institutional accounts above a minimum size, if it carries a fee, or if the issuer reserves the right to suspend it, then the arbitrage that defends the peg is available to a handful of firms and not to you. Most retail holders have never redeemed a stablecoin and never will. They sell it on an exchange at whatever the order book offers, which is a price, not a peg.

Where the reserve risk actually sits

“Backed by cash and cash equivalents” covers a wide range of things with different failure modes. Short-dated Treasury bills are about as close to money as an asset gets and can be sold in size in a bad week. Overnight reverse repo is similar. Longer-dated paper carries duration risk, which is the risk that you must sell before maturity into a market that has repriced. Commercial paper and secured loans carry credit risk, and both have been quietly removed from the largest reserves over the past few years precisely because that risk is hard to defend when questions start.

Then there is the part that has actually caused a depeg: the cash itself. Cash is a deposit at a commercial bank, and a deposit is an unsecured claim on that bank. A reserve can be entirely solvent and still fail to meet redemptions for a weekend because the institution holding the cash leg has been closed. That is a liquidity failure wearing the costume of a solvency failure, and it is the single most under-discussed line in any attestation.

So the questions worth asking, in order, are: what proportion is in bills rather than cash; which banks hold the cash and how many of them are there; what is the average maturity; and is any part of the reserve lent out, rehypothecated or invested in the issuer’s own affiliates. Those four answers tell you more than the total ever will.

Crypto-backed is a category, not a guarantee

The 8.3% of supply that DefiLlama classifies as crypto-backed does not share a single design. The original model, which Dai established, is overcollateralisation: you lock more value in volatile assets than you draw out in stablecoins, and if the collateral falls toward the debt an automated auction liquidates the position. It works, and it has kept working through several severe drawdowns, but it is capital-inefficient by construction and it depends on liquidations clearing in a falling market, which is exactly when they are hardest.

Two complications are worth naming. First, several nominally crypto-backed stablecoins have at times held large amounts of fiat-backed stablecoins as collateral, which does not remove counterparty risk from the system so much as move it one layer down and make it harder to see. A token described as decentralised that is substantially collateralised by another issuer’s dollars inherits that issuer’s bank relationships. Second, delta-neutral designs such as Ethena’s USDe are grouped here but are not overcollateralised in the Dai sense at all. They hold a long spot position hedged with a short perpetual futures position, and the yield comes from funding rates and staking. That is a trade, and its dependency is not a reserve but the continued availability of a counterparty willing to take the other side at a positive rate. It is a coherent design with a genuinely different risk, and it should not be read across from Dai just because a category label puts them together.

What has actually broken

The historical record is short enough to state plainly, and it is more instructive than any amount of theory.

Terra’s UST failed in May 2022 and took tens of billions with it. It was algorithmic, its peg was defended by an arbitrage loop with a sister token whose value depended on confidence in the loop, and once redemptions outpaced that confidence there was nothing underneath. Iron Finance had already demonstrated the same failure at smaller scale in June 2021. This is the category that now holds nine hundredths of one percent of the money.

USD Coin traded as low as roughly $0.88 in March 2023, not because its reserve was unsound but because a portion of the cash leg sat at Silicon Valley Bank, which had just been closed. The peg was restored within days once the deposits were made whole. Nothing about the reserve composition was wrong. The bank was wrong, and that was enough.

Tether has traded meaningfully below a dollar more than once, most visibly in October 2018, and has spent the years since materially changing what sits in its reserve, including removing commercial paper entirely. Whatever view you take of its disclosure, the direction of travel has been toward shorter and simpler assets.

The pattern across all of it: the algorithmic failures were failures of design, and they were terminal. The fiat-backed failures were failures of plumbing, and they were temporary. That is not an argument that fiat-backed tokens are safe. It is an argument that they fail differently, and that the thing to watch is the banking relationship rather than the whitepaper.

A price on a screen is not always a depeg

When a stablecoin prints $0.98 on an exchange, that figure is the clearing price in one order book at one moment. It may reflect genuine doubt about redemption. It may equally reflect a thin book on a weekend, a single large seller who needs liquidity now, or an exchange with a withdrawal problem of its own. The number that matters is whether the issuer is still redeeming at par, and that is not visible in a price feed. Where a venue is quoting a discount while primary redemption continues normally, what you are looking at is the cost of immediacy, not an impairment.

This cuts both ways, and it is the reason we do not treat a stable screen price as evidence of health either. A token can trade at exactly a dollar right up to the moment redemption is suspended.

What regulation actually changes

Setting aside the question of which bills are live in which jurisdiction, stablecoin regulation tends to do four structural things, and it is worth understanding them separately from the politics. It requires the issuer to be authorised, which turns an unregulated company into a supervised one. It prescribes what may sit in the reserve, usually pushing it toward cash and short government paper. It creates an enforceable redemption right at par, which converts the peg from a commercial promise into a legal obligation. And it requires reserve assets to be segregated from the issuer’s own balance sheet, so that a failure of the company is not automatically a failure of the token.

The European Union’s MiCA regime brought its stablecoin provisions into application from mid-2024 and is the most developed example of that template in force. The consequence worth noting is a structural one that most coverage misses: rules of this kind raise the fixed cost of being an issuer. That favours large incumbents, and in a market where two companies already hold 83.2% of supply, a compliance regime is as likely to entrench concentration as to reduce it. We have argued that case separately in The Weekly Take.

For the wider policy picture, see our guide to crypto regulation. For where these tokens are custodied and traded, see wallets and self-custody and our work on exchange market share, and for how they are used as collateral, DeFi explained.

What would change our conclusion

Our position is that the mechanism debate is a distraction and the real exposure is a two-name counterparty question. Four things would change that.

• The combined Tether and USD Coin share falling below roughly two thirds of supply, sustained across two quarterly re-runs rather than one. Concentration easing would move the argument back toward mechanism.
• A fiat-backed issuer at scale failing to honour redemptions for reasons of reserve quality rather than bank access. That would make composition, not counterparty, the primary risk.
• An algorithmic or delta-neutral design taking a materially larger share of supply. At 0.09% the category is not systemic; at ten percent it would be.
• Enforceable, audited reserve disclosure becoming standard rather than periodic attestation. That would narrow the gap between what we can count and what we would like to know.

How we counted

We took every asset in the DefiLlama stablecoin register carrying the pegType peggedUSD, used each asset’s self-reported circulating supply, and summed to a denominator of $306.6 billion across 328 tokens. Shares are that asset’s supply as a percentage of the total. Mechanism groupings are DefiLlama’s own classification, with one misspelled duplicate category folded into crypto-backed as noted below. Data pulled 3 August 2026. Supply figures are issuer-reported and are not independently verified by us; that limitation applies to every number on this page.

What we could not verify

Everything above is supply data. It says nothing about reserve quality, which is not observable on-chain and which we are not in a position to confirm independently. One data-quality note in the interest of showing our working: DefiLlama’s register carries a misspelled duplicate category holding $2.6 million, which we folded into crypto-backed. It is immaterial at this scale, but we would rather say so than quietly clean it up.

---

---

Next in this series

This measurement re-runs quarterly at this address. The next question we intend to count is which chains the 91.6% actually settles on, and whether stablecoin settlement is as concentrated by network as it is by issuer.

Revision history

• 6 August 2026 — expanded from the original measurement into the full Stablecoins pillar. Added peg mechanics, reserve composition, the historical failure record, the regulation section and a stated falsification test. Findings and figures unchanged; data pull date remains 3 August 2026.
• 5 August 2026 — first published. DefiLlama register, pulled 3 August 2026.