The short version: a stablecoin is a digital token designed to hold a fixed value — almost always one US dollar — so you can move it like cash without crypto’s price swings. It does that by being backed by reserves (cash and government debt), by other crypto, or, in riskier designs, by an algorithm. The catch is that the peg is only as good as the mechanism behind it, and the rest of this guide shows what happens — across five real cases — when that mechanism gets tested.
In March 2023, USDC, the second-largest stablecoin in the world, traded as low as $0.87. Circle, the company behind it, held $3.3 billion of USDC reserves at Silicon Valley Bank when the bank collapsed. The peg recovered within about 72 hours once the US government guaranteed SVB depositors, according to Circle's own statement at the time.
Just over two years later, in April 2025, a smaller stablecoin called FDUSD dropped to the same $0.87 level. This time there was no bank failure at all. Tron founder Justin Sun had publicly accused FDUSD's custodian, First Digital Trust, of being "effectively insolvent." The claim was never proven, but traders sold first and asked questions later. FDUSD was back near $0.99 by the next morning.
Then in June 2026, a third episode showed the pattern breaking somewhere new entirely. EURI, a euro-pegged token issued by Banking Circle, a licensed Luxembourg credit institution, fell as far as €0.956, a discount of 438 basis points. The issuer's reserves were never in doubt, and every redemption request that came in during the episode settled at the full one-euro rate.
This time the price was moved by Binance, EURI's dominant trading venue. They failed to secure a permanent license under the EU's MiCA framework and had to exit the European market. Holders who wanted euros before the exit deadline couldn't wait for the redemption queue, so they sold into the order book instead.
A Bundesbank research paper documenting the episode found the discount followed the token onto its one alternative venue too, while every other euro stablecoin and every dollar pair on Binance itself held par.
That's the thing about stablecoins: the peg isn't a property of the token itself. It's a claim about a mechanism sitting underneath it, and the only way to find out whether that claim is true is to watch what happens when the mechanism gets tested.
What a Stablecoin Actually Is
A stablecoin is a digital token designed to hold a steady value, almost always one US dollar (or, for a growing number of euro-denominated tokens, one euro), so it can be used and moved like cash without the price swings of Bitcoin or Ether.
It’s linked to something of equivalent value: cash in a bank account, short-term government debt, other crypto assets, or (in some cases) nothing but an algorithm and a promise. Broadly, there are three backing models, and each one fails in a different way when it's tested.
Fiat-Collateralized: USDC and USDT
This is the model behind the two largest stablecoins, USDC (issued by Circle) and USDT (issued by Tether). Each token in circulation is meant to be backed 1:1 by cash and short-dated US Treasuries held in reserve, checked periodically by an outside accounting firm.
Circle publishes its reserve composition weekly and has its reserves attested monthly by Deloitte & Touche LLP, which took over as Circle's auditor in fiscal 2022 (Grant Thornton had the role from 2015). The reserves themselves sit in the Circle Reserve Fund, a registered government money-market fund managed by BlackRock, split roughly 80% Treasuries and 20% cash.
Tether's most recent attestation, covering the first quarter of 2026 and prepared by BDO, put total assets at $191.7 billion against $183.5 billion in liabilities tied to issued tokens, an excess reserve buffer of $8.23 billion, a record high for the company. About 80% of that sits in US Treasuries, with the rest spread across cash, gold (roughly $8 billion) and bitcoin (roughly $7 billion).
An attestation, whether from Deloitte or BDO, confirms that the reserves matched the stated criteria on a specific date. It isn't a full financial audit of the underlying systems and controls, and neither Circle nor Tether currently publishes one of those for all reserves.
Crypto-Collateralized: DAI
DAI, issued by the MakerDAO protocol, is backed by other crypto assets rather than dollars, over-collateralized so that a drop in the value of the backing assets doesn't immediately break the peg. It held up through most stress events, but its March 2020 "Black Thursday" episode showed the model's real weak point: when Ethereum crashed 43% in a single day and gas fees spiked past 200 Gwei, the network itself got too congested for the liquidation bots meant to keep the system solvent to actually execute.
Algorithmic: The Cautionary Case of UST
TerraUSD (UST) tried a third approach. Operating with no hard collateral at all, just a sister token (LUNA) and market incentives designed to keep supply and demand in balance. In May 2022, confidence cracked, the incentive loop reversed, and UST and LUNA together lost tens of billions of dollars in value within days. An MIT Sloan analysis afterward described it as a run on a structurally unsustainable system, not evidence of any targeted attack.
Four Depegs, Four Different Causes
Put the events side by side and a pattern emerges: it's rarely the same failure twice.
- USDT, October 2018. Rumors about Tether's reserves and withdrawal issues at Bitfinex sent USDT down 10% to about $0.90.
- USDT, June 2023. A liquidity imbalance in Curve's 3pool, where Tether's share of the pool swelled past 70% instead of the expected third, pushed USDT down to $0.977. Not a reserve problem; a market-structure one.
- USDC, March 2023. Circle's exposure to Silicon Valley Bank, a real-world banking failure rather than anything crypto-native, dropped USDC to $0.87 before it recovered within about three days.
- FDUSD, April 2025. A public, unproven solvency accusation against the custodian triggered a same-day drop to $0.87 and a next-morning recovery, pure reputational contagion with no evidence the reserves were ever actually short.
A Fifth Kind of Depeg: When the Exchange, Not the Issuer, Fails
The EURI episode from June 2026 belongs on this list too, and it's worth spending more time on because it's a different species of failure than the four above. Daniel Fricke and Björn Imbierowicz of the Deutsche Bundesbank documented it in detail, and the mechanism is instructive for any business holding stablecoins.
Every prior depeg on this list traces back to something wrong, or feared to be wrong, with the issuer or its reserves. EURI's did not. Banking Circle, the regulated bank behind EURI, never stopped honoring redemptions at the full one-euro rate throughout the episode; the researchers' on-chain analysis shows zero mints and 17.76 million EURI redeemed at par during the four-day window. The problem was that redemption required know-your-customer onboarding and settled on a roughly five-business-day cycle, while the price moves minute by minute in the open market. When Binance, the exchange where nearly all EURI trading happened, lost its EU license and had to wind down, holders who needed euros before the cutoff couldn't wait a week for the official channel to clear. They sold at whatever the order book would pay, and the price fell to €0.956 before recovering over the following days as buyers willing to wait for the redemption caught up stepped back in.
Strikingly, the researchers found the first material on-chain redemption didn't settle until roughly 20 hours after the price had already bottomed out, and 99.9% of all redemptions came from a single wallet: Binance's own, unwinding its inventory as it exited.
When the New York regulator ordered Paxos to stop minting new BUSD in February 2023, the coin traded at a modest discount for weeks even though redemptions stayed open the entire time. That episode struck the issuer's ability to create new supply, not its reserves or its ability to pay out. EURI struck the exchange instead. The point of impact may be different, but the underlying lesson is the same: when a regulatory action hits one part of the system without touching the backing itself, the market still takes time to work out how much that actually threatens the peg, and in the meantime, the price moves anyway.
How to Actually Read a Reserve Disclosure
There are two things worth checking before taking any stablecoin's backing at face value:
- The cadence. Circle discloses USDC reserve holdings weekly, with monthly Deloitte attestations. Tether's BDO attestations run quarterly. Neither cadence is wrong, but a weekly disclosure catches a problem faster than a quarterly one.
- Attestation versus audit. An attestation checks a specific claim on a specific date against agreed criteria. An audit examines the systems and controls behind that claim over a period of time. Every major stablecoin issuer currently uses attestations. You have to know which is which.
There's a third thing worth checking that has nothing to do with reserves at all, and the EURI episode is the clearest illustration of why it matters: where the token actually trades.
EURI's own whitepaper named Binance as its initial venue, and almost all of its trading volume lived on that single exchange throughout its history. When the exchange lost its license, there was nowhere else for sellers to go in size, and the arbitrage that's supposed to hold a peg in place, buying the discount and redeeming at par, couldn't work fast enough to stop the price from moving, because the arbitrageurs faced the same days-long onboarding queue as everyone else.
The Bundesbank researchers priced that arbitrage directly: buying the discount and redeeming at par cost roughly 10 to 20 basis points in fees, against a discount that ran to 438.
A token spread across several exchanges doesn't have this problem in the same way. If one venue runs into regulatory trouble, trading and redemption pressure can shift elsewhere. A token concentrated on one exchange inherits that exchange's regulatory risk on top of its own.
If you’re trying to decide which stablecoin you want to hold or accept, make sure exactly how much red tape is involved when it comes to getting your money in or out of it.
Comparing the Major Stablecoins
| Stablecoin | Backing model | Disclosure / auditor | Worst historical depeg |
|---|---|---|---|
| USDC | Fiat-collateralized (cash + US Treasuries) | Weekly disclosure; monthly Deloitte attestation | ~$0.87, March 2023 (SVB exposure), recovered in ~72 hours |
| USDT | Fiat-collateralized (cash, Treasuries, gold, BTC) | Quarterly BDO attestation | ~$0.90, October 2018 (reserve rumors); ~$0.977, June 2023 (Curve pool imbalance) |
| DAI | Crypto-collateralized (over-collateralized) | On-chain, protocol-governed | Peg held during March 2020 "Black Thursday," but liquidation mechanism froze under network congestion |
| UST (discontinued) | Algorithmic (no hard collateral) | N/A | Collapsed to near zero, May 2022; tens of billions in value erased |
| EURI | Fiat-collateralized (euro e-money token, bank-issued) | Regulated under MiCA; par redemption via issuer | ~€0.956, June 2026 (dominant exchange lost its MiCA license, not a reserve issue), recovered over several days |
Each model has a specific trigger that breaks it, and knowing what that is for the stablecoin you're holding or accepting is the actual due diligence.
That now includes checking today's price, checking what backs the token, and checking how many places you could actually get your money out if you need to.
Related reading: The Real Cost of Crypto Payments: On-Ramp, Off-Ramp and the Hidden Spread · How to Accept Crypto Payments Without Holding Crypto (Instant Fiat Settlement) · How Pay-by-Crypto Works for B2B: Settlement, Chargebacks and Reconciliation