---
title: "What one USDC actually equals"
author: Mariusz Szyma
date: 2026-09-15
lang: en
canonical: https://szyma.co/en/blog/stablecoins/what-one-usdc-means/
series: "Stablecoins: Past, Present and Future"
series_part: 1/11
series_url: https://szyma.co/en/blog/stablecoins/
data_as_of: 4–10 Sep 2026
---

# What one USDC actually equals

The sentence "1 USDC = 1 USD" looks like a definition. It is three independent claims, and only one of them explains the price. This chapter splits the sentence, shows what held and what failed when Circle's bank was seized in March 2023, and leaves three questions that work on any dollar token.

## What you'll learn

- Split "1 USDC = 1 USD" into a technological claim, a legal claim, and an economic claim.
- Name the loop that holds the price, and say what a reserve is worth when that loop is closed.
- Ask three questions about any peg: who can use the loop, what it costs, and when it is open.
- Tell a payment stablecoin from a bank deposit, a tokenized deposit, a CBDC, and a tokenized Treasury.

## Three layers of one sentence

**A stablecoin** is a token built to hold a fixed price against a reference currency. USDC, issued by Circle, is the usual classroom example because its contract is public, its issuer publishes a reserve fund, and the weekend that tested it is documented. Tether's USDT is the larger sibling by supply and runs on the same three-layer structure.

Most confusion about these tokens comes from mixing the layers. The contract, the claim, and the price can each be true while the other two are in trouble.

**The technological claim** is that one token on the ledger stands for one dollar of denomination. USDC is an ERC-20 contract: a mapping from addresses to integers, six decimal places, so a holding of 100 USDC is the integer `100000000` at that address. The contract has no price oracle and no bank connection. If USDC traded at 0.60 USD tomorrow, the contract would behave exactly as it does today. What the chain does give is a public liabilities counter. Anyone can read `totalSupply()`, and only Circle can mint or burn. That number is how many tokens exist. It is not proof that anything sits behind them.

**The legal claim** is that a named party owes the holder a dollar. Possession of USDC does not, by itself, create a right to walk up to Circle and be paid. **Circle Mint** is the issuer's primary-market desk: a verified client, after KYC and a signed agreement, can create tokens against dollars and hand tokens back for dollars at par. The current minimum ticket for that desk is not established in the sources used here. A holder who bought USDC on an exchange is a secondary-market buyer. That holder's exit is to sell to someone else at the market price. Tether Limited runs the same structure for USDT: direct redemption requires an account with the issuer, verification, and a minimum.

A stablecoin is not a bank deposit. There is no FDIC-style insurance on the token. If the issuer fails, the holder is a creditor of a private company, and the ranking of that claim depends on whether the reserves were legally separated from the issuer's estate. The sources describe that separation as bankruptcy-remote structuring. They do not, in the material used for this chapter, walk through a court that has tested it.

**The economic claim** is that the market price stays near one dollar. That claim is a statement about a trade that someone must want to do, not about a number in a reserve report.

> **Watch out:** A page that says "fully reserved" has answered the technological question about supply and the legal question about assets. It has not answered the economic one. The March 2023 weekend is the documented case where the first two stayed intact and the third failed for about sixty hours.

![1 USDC = 1 USD holds on three layers at once; only the economic layer explains the price, which is why the SVB weekend of 10–13 Mar 2023 moved it.](https://szyma.co/blog/img/three-layers-of-one-usdc.0ef426b0.svg)

## The arbitrage loop that holds the price

**A peg** is the target price a token is built to hold, here one US dollar. **Redemption** is the primary-market trade in which a verified client hands the token back to the issuer and is paid that target in cash.

The trade that actually keeps the screen price near the target is not the reserve. **An arbitrage loop** is the round-trip that is profitable whenever the market price leaves the target, and that, by being done, pushes the price back.

When USDC trades at 0.995 USD, a Circle Mint client buys the token in the market and redeems it at Circle for 1.00 USD, keeping the half-cent. The buying itself bids the price up. When USDC trades at 1.005 USD, the same client mints new tokens at 1.00 USD and sells them, and the selling offers the price down. Both directions have to remain open, cheap, and fast enough that someone will take the trade before the gap is interesting.

A reserve without that loop holds nothing that a secondary-market seller can use. A reserve reachable through an open, cheap, and fast loop is what holds the peg. The March 2023 weekend is the empirical case, not a thought experiment.

![Below $1, a Circle Mint client buys cheap tokens and redeems at par, which bids the price up; above $1, the same client mints at par and sells, which offers it down. The SVB weekend of 10–13 Mar 2023 is the documented close of that loop.](https://szyma.co/blog/img/arbitrage-loop.4c6fba5b.svg)

Three questions decide whether a given token's loop is a mechanism or a slogan.

1. Who can pass through it. A desk limited to verified institutional clients corrects more slowly than an open window.
2. What it costs to pass. The redemption fee is the band inside which the price can drift before anyone bothers.
3. How long it takes, and when it is open. Settlement on bank days, and a weekend when the bank is closed, are the same gap Circle hit in March 2023.

The marketing page never answers the third question in those words.

## The weekend Circle's loop closed

On 10 March 2023 the Federal Deposit Insurance Corporation seized Silicon Valley Bank. Circle held 3.3 bn USD of USDC reserves there, about 8% of the token's reserve at the time. It was a weekend. US banks were closed, wires were idle, and Circle halted mint and redeem. The reserve still existed. Circle did not, on the record used here, lose the deposits. The loop was closed operationally for about sixty hours.

On 11 March 2023 USDC traded down to about 0.87 USD. In the Friday session, before the scale of the problem was fully public, holders redeemed more than 1 bn USD of USDC at the primary desk. That figure is the precision in the sources; a tighter number is not established. The primary-market drain and the secondary-market gap arrived on different clocks: redemption moved while the bank was still open, and the exchange price kept moving after the desk had shut.

On Monday, full access to SVB deposits was announced. The loop opened. The peg returned to 1.00 USD within hours. The same mechanism, open or closed, is the one that explains a fiat-backed token in a crisis. The cause was not fraud and not insolvency. Access to the reserve stopped for a few days, and the secondary market marked that stop in minutes.

> **Example:** Apply the three questions to that weekend. Who could pass: Circle Mint clients, already a narrow set. What it cost: the usual desk, until the desk froze. When it was open: not from Friday evening until Monday, because the bank behind the reserve was under FDIC control and US wires do not run on Saturday. The price that printed on 11 March 2023 was the answer to question three.

That is the mental model this series starts from. A later chapter walks the other constructions through the same three questions. The label on the token is the last thing to read, not the first.

## Three surviving pegs and one extinct design

Industry copy sorts tokens as regulated, institutional, or decentralized. None of those labels says what happens when a tenth of the holders want out in an hour. The split that does is the split by what stands behind the token and who controls access to it.

Three designs still have a market in 2026.

**Fiat-backed** tokens hold cash and short government paper against the supply, and they defend the peg by redemption at the issuer. USDT and USDC are this design. On 4 September 2026 DefiLlama put USDT at 183.34 bn USD and USDC at 74.46 bn USD. The failure mode is the bank or custodian that holds the cash, which is the SVB case.

**Crypto-backed** tokens hold crypto, posted in excess of the amount issued, and they defend the peg by liquidation of that collateral rather than by a cash window at an issuer. DAI, now issued as USDS by Sky, is this design. The failure mode is an oracle that is late, or a liquidation engine that cannot keep up with the collateral's price. Both showed up on 12 March 2020, the day DeFi still calls Black Thursday, when ether halved in under thirty-six hours and MakerDAO's auctions filled at zero. That episode belongs in a later chapter as a case; the point here is the mechanism: there is no Circle Mint equivalent.

**Synthetic** tokens hold a hedged position rather than a pile of cash or a vault of ether. Ethena's USDe, launched 19 February 2024, keeps a long spot position against a short perpetual futures position and calls the result delta-neutral. The peg lives in the hedge and in the secondary market, not in a legal right to redeem one token for one dollar at the issuer. The failure mode is a funding rate that stays negative, so the hedge costs money instead of paying it.

A fourth design is in the past tense. **An algorithmic stablecoin** has no collateral. It holds its peg by minting and burning a sister token that is supposed to absorb volatility. Terra's UST did that against LUNA. UST's supply was above 18 bn USD in early May 2022. Between 7 and 8 May a large holder sold about 285 mn USD of UST and the peg printed 0.98 USD. By 09:00 on 16 May 2022 UST was at 0.12 USD and LUNA was at zero. Sources record the combined loss as about 40 bn USD or about 60 bn USD, depending on whether they count UST alone or UST and LUNA together. The mechanism ran in reverse of its design: a falling UST minted more LUNA, cheaper LUNA weakened the defence, and the pair went to zero in a week. It is the one category in this material with a documented supply collapse to nothing.

US law does not treat the four designs as one object. **A payment stablecoin**, under the GENIUS Act signed 18 July 2025, is a digital asset meant for payment or settlement whose issuer is obliged to redeem it at a stated monetary value. The definition has no word for blockchain and no word for reserve. It is an issuer's redemption duty. USDe and DAI are not payment stablecoins under that definition, because nobody is formally obliged to take the token back at a stated dollar amount. Calling them that in a US legal sense is a category error, even when their market prices sit near a dollar.

## Six instruments all called a dollar

"A dollar" in this market is not one instrument. It is at least six claims, and the column that decides the rest is who owes the holder.

| Instrument | Who owes the holder | Public guarantee | Price | Settlement | Holder yield |
|---|---|---|---|---|---|
| Cash (banknote) | Central bank | Base money | Always par | Immediate, physical | No |
| Bank deposit | Commercial bank | Yes, to a limit (FDIC 250,000 USD in the US) | Par while the bank lives | Bank hours, working days | Yes, the bank sets it |
| Tokenized deposit | Commercial bank | Yes: the same deposit, new rails | Par | 24/7 inside the bank's network | Yes |
| CBDC | Central bank | Base money | Par | 24/7 by design | Usually no |
| Fiat-backed stablecoin | Private issuer | No | Near par, market, can drift | 24/7, final on-chain | No: GENIUS Act bars it |
| Tokenized Treasury / MMF | Fund, then the US Treasury | No (credit risk near zero) | NAV, floats | Depends on the product | Yes: that is the product |

**A tokenized deposit** is a bank deposit represented on new rails. JPMorgan's JPMD, live on Base since 12 November 2025, is this instrument: a claim on the bank, inside the banking system, and (where the law applies) inside deposit insurance. The bank that tokenizes a deposit is not issuing a new money. A stablecoin is a new private liability, outside that insurance. From the holder's seat those are different assets, not different pipes for the same one.

Banks prefer the deposit token for a reason that has nothing to do with blockchains. In the fractional-reserve model JPMorgan held about 2.5 tn USD of customer deposits against about 400 bn USD of central-bank reserves, in figures reported in November 2025. A dollar that leaves the deposit base for a 1:1 stablecoin has to be matched by reserve assets rather than lent onward. The same reporting puts the lost deposit capacity at more than four dollars for each dollar that migrates. That is why a large bank tokenizes the deposit it already has, and does not issue a stablecoin against a new pile of bills.

The other pair that collapses in conversation is the stablecoin and the tokenized Treasury. A payment stablecoin is built to print 1.00 USD. A tokenized Treasury or money-market fund is built so that the unit price rises with interest, which is why it is a poor unit for a coffee and a stablecoin is a poor savings asset. On 4 September 2026 DefiLlama put BlackRock's BUIDL at 2.78 bn USD, Hashnote's USYC at 2.69 bn USD and 1.136 USD per unit, and Ondo's USDY at 2.18 bn USD and 1.140 USD per unit. Those products share the series' RWA definition in [What an RWA actually is](/en/blog/rwa/what-is-an-rwa/): an enforceable claim against an identified party that controls an off-chain asset. They do not share a peg.

A fiat-backed stablecoin is, economically, close to a money-market fund: similar assets (bills, repo, cash), similar liquidity management. Three things split them. The fund pays the holder; the stablecoin does not. The fund is an investment product with that reporting regime; the US stablecoin is a separate payment category. The fund unit does not move peer-to-peer in seconds; the token does. The stablecoin is the fund that handed the holder transferability and kept the interest.

![Six kinds of "a dollar" differ first in whose liability they are, then in guarantee, price, and yield, with the GENIUS Act yield ban as of the OCC's 25 Feb 2026 proposal and JPMD live on Base since 12 Nov 2025.](https://szyma.co/blog/img/stablecoin-vs-deposit-vs-cbdc.da209d80.svg)

## The yield ban and the wrapper around it

The GENIUS Act bars a payment-stablecoin issuer from paying yield directly to holders. The OCC's Bulletin 2026-3, a notice of proposed rulemaking dated 25 February 2026, is the US banking agency's write-up of that reserve, redemption, and yield package. Final rules were not out by September 2026. The statute's backstop effective date is 18 January 2027.

The market's workaround is a second token. **A yield-bearing wrapper** is a separate token that holds the payment stablecoin, or sits next to it, and pays the interest that the issuer of the payment token may not. Ethena's sUSDe is not USDe. Sky's sUSDS is not USDS. The wrapper is, on its face, not the payment stablecoin, so it sits outside the letter of the ban. Whether a supervisor later reads that as inside the spirit of the ban is not established. It is the open regulatory question hanging over the whole yield-bearing segment, and a later chapter takes the economics of that spread apart.

USDT's market share is the other uncomfortable fact this chapter has to carry. On 4 September 2026 DefiLlama put USDT at 58.91% of USD-pegged supply. That share sat there after years in which Tether did not publish a full reserve audit, after an 18.5 mn USD settlement with the New York Attorney General in 2021 for misleading statements about USDT's backing, and after a 41 mn USD CFTC penalty the same year for untrue statements about dollar reserves. Transparency changes tail risk in a crisis. It has not, on this record, been what set the day-to-day price. Liquidity and the network that accepts the token have.

> **My take:** The token is an interface to a financial construction, not the construction. The chain does not know what a dollar is. What holds the price sits outside the code: in a contract, in a bank, and in an arbitrage loop that has to stay open. A copy of the ERC-20 can be deployed in a weekend. The loop cannot.

## Key takeaways

1. "1 USDC = 1 USD" is three claims at once (technological, legal, economic), and mixing them is how both a panic and a false sense of safety get started.
2. The USDC contract is a balance mapping with no oracle and no bank link; `totalSupply()` counts liabilities, not backing, and only Circle mints or burns.
3. Direct par redemption is limited to verified Circle Mint clients; a secondary-market holder can only sell at the market, and the current desk minimum is not established in these sources.
4. The arbitrage loop holds the peg. A reserve without an open, cheap, fast loop held nothing that a seller could use on 11 March 2023, when USDC printed about 0.87 USD with the reserve still intact.
5. Three designs still defend a dollar peg in 2026: redemption at the issuer (USDT, USDC), over-collateralization and liquidation (DAI/USDS), and a delta-neutral hedge (USDe). The algorithmic design without collateral, UST/LUNA, went to zero in May 2022.
6. A tokenized deposit is the same bank claim on new rails (JPMD on Base since 12 November 2025). A stablecoin is a new private liability. A tokenized Treasury does not target 1.00 USD.
7. The GENIUS Act, signed 18 July 2025, defines a payment stablecoin by the issuer's duty to redeem, not by a chain or a reserve, and it bars the issuer from paying yield directly; sUSDe and sUSDS are the wrappers that currently sit beside that ban.

## Glossary

- **Stablecoin**: a token built to hold a fixed price against a reference currency. USDC and USDT are the two largest dollar examples by supply on 4 September 2026.
- **Peg**: the target price. For the tokens in this chapter the target is one US dollar.
- **Redemption**: the primary-market trade in which a verified client returns the token to the issuer and is paid the target in cash.
- **Arbitrage loop**: the profitable round-trip (buy-and-redeem below the peg, mint-and-sell above it) that, by being done, pushes the market price back to the target.
- **Circle Mint**: Circle's gated primary-market desk. Only its clients redeem USDC at par with the issuer.
- **Payment stablecoin**: the US statutory category (GENIUS Act) of a payment token whose issuer must redeem it at a stated monetary value. USDe and DAI are not this category.
- **Tokenized deposit**: a commercial-bank deposit represented on new rails, still a bank claim. JPMD on Base, from 12 November 2025, is the example used here.
- **CBDC**: a digital claim on a central bank, base money by design, usually without holder yield in the designs discussed in these sources.
- **Yield-bearing wrapper**: a second token that pays interest the payment-stablecoin issuer may not pay directly. sUSDe sits next to USDe; sUSDS sits next to USDS.
- **Fiat-backed / crypto-backed / synthetic**: three surviving ways to hold a peg (cash window at the issuer; over-collateralized crypto plus liquidation; a hedged derivative book). Case studies of each sit later in the series.

## Go deeper

- [What an RWA actually is](/en/blog/rwa/what-is-an-rwa/): the claim test that tokenized Treasuries pass and a perpetual on a tokenized share fails, which is why BUIDL, USYC and USDY are in the table above and are not pegs.
- [How ten years of failures selected this market](/en/blog/stablecoins/history/): BitUSD, Tether, DAI, Terra and the SVB weekend as a selection process, not a progress story.
- [The business: a spread on T-bills, paid out to distributors](/en/blog/stablecoins/the-business/): who keeps the coupon the holder of a payment stablecoin does not receive.
- [How much tokenized value actually works in DeFi](/en/blog/rwa/rwa-in-defi/): the same split between a stablecoin and a tokenized deposit, read from the turnover gap rather than from the legal column.
- 🟢 U.S. Congress, GENIUS Act (P.L. 119-27), signed 18 July 2025: the statutory definition of a payment stablecoin as an issuer's redemption duty.
- 🟢 OCC, Bulletin 2026-3 (NPRM), 25 February 2026: proposed reserve, redemption, and yield-ban rules for permitted payment stablecoin issuers.
- 🟢 DefiLlama stablecoin dashboard, read 4 September 2026: supply by token, including USDT 183.34 bn USD, USDC 74.46 bn USD, and the tokenized-Treasury unit prices cited above.

## Sources

- 🟢 Author's knowledge base, module 1 (three layers of 1 USDC = 1 USD, the USDC contract as a balance mapping, Circle Mint as the gated par window, the arbitrage loop, the three diagnostic questions, the three surviving pegs and the extinct algorithmic design) and module 16 (tokenized deposit versus stablecoin, JPMD, the JPMorgan deposit-versus-reserve arithmetic, tokenized Treasuries as floating-NAV products).
- 🟢 Polish research edition, `content/blog/stablecoins/maszyna.md`: SVB weekend chronology, the more-than-1 bn USD Friday redemption, the six-instrument comparison, GENIUS Act definition and yield ban, wrapper tokens sUSDe and sUSDS, DefiLlama supplies of 4 September 2026.
- 🟢 U.S. Congress / GENIUS Act (P.L. 119-27), signed 18 July 2025; statutory backstop effective date 18 January 2027.
- 🟢 Federal Register, GENIUS Act implementation materials, 19 September 2025: legal definition of a payment stablecoin as used in the Polish edition.
- 🟢 OCC, Bulletin 2026-3 (NPRM), 25 February 2026: reserve requirements, redemption, yield ban.
- 🟢 DefiLlama API, read 4 September 2026: USDT 183.34 bn USD, USDC 74.46 bn USD, USDT share 58.91%, USDC share 23.93%, top-two 82.84%; BUIDL 2.78 bn USD; USYC 2.69 bn USD at 1.136 USD/unit; USDY 2.18 bn USD at 1.140 USD/unit.
- 🟢 Circle transparency page, circle.com/transparency (2026), and Circle Reserve Fund documentation, as cited in the Polish edition.
- 🟡 Eco Support, "What Is USDC? Circle's Regulated Digital Dollar in 2026", 2026: USDC/SVB depeg chronology used alongside the Polish edition.
- 🟡 Yahoo Finance / Bloomberg, November 2025: JPMorgan deposit base about 2.5 tn USD against about 400 bn USD of Fed reserves, and the more-than-four-dollars-per-dollar migration arithmetic.
- 🟡 MakerDAO community post-mortem, March 2020: Black Thursday, zero-bid ether auctions.
- 🟡 Bloomberg, "Why Did Terra's (UST) Algorithmic Stablecoin and Luna Crypto Coin Fail?", May 2022; CNBC, 2 June 2022, LUNA to zero; Ecos.am retrospective on the Terra Luna crash.
- 🟡 NYAG settlement with Tether and Bitfinex, 18.5 mn USD, 2021; CFTC penalty against Tether, 41 mn USD, 2021, both as cited in the Polish edition.
- 🟡 Ethena launch 19 February 2024, as dated in the history module of the knowledge base and the Polish edition.
- 🟡 JPMD live on Base since 12 November 2025, knowledge base module 16.

## Caveats

Circle's and Tether's current minimum redemption amounts are not in the sources used for this chapter. The desks exist, they are gated, and the dollar floor is unlocated.

The Friday pre-weekend USDC redemption is "more than 1 bn USD" in the Polish edition and the knowledge base. A more precise figure is not established, and none is used here.

The Terra collapse is recorded as about 40 bn USD or about 60 bn USD depending on scope (UST alone versus UST and LUNA together). Both figures are kept; they are not averaged.

The JPMorgan deposit and reserve figures, and the more-than-four-dollars-per-dollar claim, come from November 2025 secondary reporting (Yahoo Finance / Bloomberg as cited in the Polish edition), not from a JPMorgan filing read for this chapter.

JPMD's "live on Base since 12 November 2025" date is a knowledge-base item marked as secondary quality in the chapter brief.

The comparison matrix uses six categories (cash, deposit, tokenized deposit, CBDC, fiat stablecoin, tokenized Treasury), which is the framing in the series design. The knowledge-base table has eight rows; the two extra instruments are not dropped as facts, they are left for later chapters rather than forced onto a six-column figure.

GENIUS Act implementing rules were still proposals in September 2026. The yield ban is statutory; the OCC bulletin cited is an NPRM dated 25 February 2026, not a final rule.

Whether a court would treat USDC holders as unsecured creditors, and how far bankruptcy-remote structuring would move them, is not established by a case in this material. The ranking is described, not litigated.

USDT's 58.91% share is a DefiLlama reading of 4 September 2026. A later chapter records that aggregator taxonomies also disagree about how to classify synthetic and hybrid tokens; the share figure here is supply, not a verdict on construction.