On this page
Three stablecoins dominate crypto: USDT (Tether), USDC (Circle), and DAI(MakerDAO/Sky). All three are pegged to $1, all three are used as trading pairs and settlement assets, and all three are completely different under the hood. Understanding the differences — in reserves, in governance, in transparency, and in failure modes — is essential for anyone holding stablecoins for more than a few minutes.
The three contenders
USDT — Tether (US Dollar Tether)
Issuer: Tether Limited (private company, British Virgin Islands).
Reserves: US Treasury bills, money market funds, Bitcoin, gold, secured loans. Partially disclosed via quarterly attestations (not full audits).
Transparency: Low. Attestations from a Cayman Islands accounting firm; no full audit. Reserve composition is disclosed in aggregate but not itemized.
Market cap: ~$110B+ as of mid-2024 (largest stablecoin).
Chains: Ethereum, Tron, Solana, Omni, and many others.
Failure mode: Reserve illiquidity in an extreme run. Survived the 2022 stress tests ($10B+ redeemed after Terra, ~$3B after FTX; peg recovered).
The dominant stablecoin globally, especially outside the US. Its ubiquity is a network effect — it's where the liquidity is. The trade-off is opacity: you trust Tether Limited's attestations, not a regulator's oversight or a full audit. Full Tether deep dive →
USDC — Circle (USD Coin)
Issuer: Circle Internet Financial (US-regulated money transmitter).
Reserves: US Treasury bills and cash deposits at regulated US banks. Disclosed via monthly attestations (historically by Grant Thornton LLP; since 2023 by Deloitte, a Big Four firm), with reserve holdings disclosed weekly.
Transparency: High. Monthly reserve reports with specific T-bill maturities and bank counterparties. Circle is publicly pursuing a full regulatory framework.
Market cap: ~$35B as of mid-2024 (second-largest).
Chains: Ethereum, Solana, Avalanche, and many others.
Failure mode: Banking partner failure. In March 2023, USDC de-pegged to $0.87 when $3.3B of its reserves were stranded at Silicon Valley Bank (SVB). The peg recovered after the FDIC guaranteed SVB deposits. This was a banking failure, not a stablecoin design failure — the reserves were real, they were just temporarily inaccessible.
The “regulated” stablecoin. Circle publishes monthly attestations from a Big Four firm, operates under US money transmitter licenses, and follows OFAC sanctions. The trade-off is that USDC is more cautious — Circle freezes tokens when required by law, and has delisted certain jurisdictions. The SVB de-peg was instructive: it showed that even a well-backed stablecoin can wobble if its banking partner fails, but also that the peg recovered within days because the reserves were real.
DAI — MakerDAO / Sky
Issuer: MakerDAO (decentralized protocol, rebranded to “Sky” in 2024 — a change the community quickly fought over, with repeated pushes to revert toward the Maker brand). Governance is by MKR/SKY token holders voting on-chain.
Reserves: Overcollateralized assets — today largely USDC and tokenized T-bills (RWA), plus crypto assets (ETH, wBTC, and others). Each DAI is backed by collateral locked in a smart contract worth more than $1 — typically 130–150%+.
Transparency: Maximum. Every collateral position is visible on-chain in real time. You can verify the exact collateralization ratio at any block. No trust in a company required — the reserves are the smart contracts.
Market cap: ~$5B as of mid-2024 (third-largest).
Chains: Ethereum-native, available on L2s.
Failure mode: Smart-contract risk + collateral liquidation risk. If collateral crashes faster than liquidation auctions can process, DAI could become undercollateralized. Has survived multiple crypto crashes (2020, 2022) with minor de-pegs that recovered quickly.
The decentralized stablecoin. DAI is minted by locking up crypto collateral in a smart contract — no company holds your dollars. The trade-off is complexity (you're trusting smart-contract code, not a regulated entity) and crypto-native collateral risk (if ETH crashes 50% in an hour, liquidation auctions might not clear fast enough). MakerDAO has increasingly diversified into real-world assets (tokenized T-bills) to reduce crypto-collateral volatility.
The market caps above are as of mid-2024. Since then the three have kept growing: USDT passed $150 billion by mid-2025, and USDC grew past $70 billion by 2026, while DAI held roughly steady near $5 billion.
Head-to-head comparison
| Property | USDT | USDC | DAI |
|---|---|---|---|
| Issuer type | Private company (BVI) | US-regulated company | Decentralized protocol |
| Reserves | T-bills, BTC, gold, loans | T-bills + cash at US banks | Overcollateralized crypto + RWA |
| Transparency | Low (attestations, no audit) | High (monthly Big Four attestations) | Maximum (on-chain, real-time) |
| Primary risk | Reserve opacity / illiquidity | Banking partner failure | Smart-contract + collateral crash |
| Censorship resistance | Low (Tether can freeze) | Low (Circle can freeze) | Medium (governance can freeze) |
| De-peg history | Minor, brief (survived FTX) | $0.87 (SVB, recovered in 2 days) | Minor, recovered quickly |
| Market cap | ~$110B+ (mid-2024) | ~$35B (mid-2024) | ~$5B (mid-2024) |
The three failure modes
Each stablecoin has a different way it can break. Understanding these is more important than knowing the current market caps.
USDT: the “reserves might not be real enough” risk
The concern with USDT is that, in an extreme run, the reserves might not be liquid enough at par to honor all redemptions. Tether's pivot to T-bills has reduced this risk significantly — T-bills are the most liquid dollar instrument in the world. But the lack of a full audit means you are ultimately trusting Tether Limited's word (plus a Cayman attestation) that the reserves are what they say. For more, see the Tether deep dive.
USDC: the “banking partner fails” risk
USDC's reserves are held at regulated US banks and in T-bills. The reserves are real and attested monthly (historically by Grant Thornton LLP, since 2023 by Deloitte). But as SVB showed, if a banking partner fails, the money can be temporarily stranded — and during that window, the peg can wobble. This is a traditional banking risk, not a crypto-specific one. The FDIC backstop for bank deposits (up to $250K per depositor, per ownership category) doesn't fully cover a stablecoin whose reserves passed $70 billion by 2026. Circle has since diversified its banking partners to reduce concentration risk.
DAI: the “collateral crash + smart-contract bug” risk
DAI's reserves are on-chain and overcollateralized, but the collateral is no longer mostly crypto: USDC and tokenized T-bills now back the largest share of DAI, with ETH and wBTC as the volatile remainder. If crypto crashes 50% in hours — as it has multiple times — the protocol must liquidate undercollateralized positions via auction. If the auction system can't keep up (as happened briefly in March 2020, “Black Thursday”), DAI can become undercollateralized. MakerDAO has mitigated this by diversifying into real-world assets, reducing reliance on volatile crypto collateral — at the cost of importing centralized-asset risk. There is also smart-contract risk: if a bug in the Maker protocol is exploited, collateral could be drained. This risk is mitigated by extensive audits and battle-testing, but it is non-zero.
There is no single “best” stablecoin — the right choice depends on your priorities:
- Maximize liquidity / global access: USDT (but accept the opacity risk).
- Maximize transparency / regulatory compliance: USDC (but accept that it can freeze funds and follows US sanctions).
- Maximize decentralization / censorship resistance: DAI (but accept smart-contract and crypto-collateral risk).
For most educational and trading purposes, any of the three is fine in the short term. For long-term holding of significant value, diversification across more than one is prudent — and nothing in crypto is risk-free.
What about algorithmic stablecoins?
The fourth “type” of stablecoin is the — backed not by reserves but by an algorithm and a sister token. (UST) was the largest and most spectacular example, and its $40 billion collapse in May 2022 demonstrated the fundamental flaw: the backing asset is inherently volatile, and the design has a death-spiral failure mode. No algorithmic stablecoin has survived a serious stress test. The market has largely concluded that algorithmic stablecoins are not stablecoins — they are structured products with a peg that works only in benign conditions. For the full story, see the Terra/Luna deep dive.
Key takeaways
- The three major stablecoins — USDT, USDC, DAI — represent three completely different trust models: a private company's word, a regulated company's attestations, and a decentralized protocol's code.
- USDT dominates by network effect and global reach, but is the least transparent. USDC is the most transparent and regulated, but can freeze funds. DAI is the most censorship-resistant, but carries smart-contract and crypto-collateral risk.
- Each has a different failure mode: USDT (reserve illiquidity), USDC (banking partner failure), DAI (collateral crash + smart-contract bug). All three have survived real-world stress tests, though USDC's SVB de-peg was a reminder that “backed by real reserves” still depends on the banking system.
- Algorithmic stablecoins (Terra/UST) are a fundamentally different and more fragile design. The market has largely rejected them after the Terra collapse.
- Diversification is prudent. No stablecoin is risk-free, and holding significant value in a single stablecoin concentrates that stablecoin's specific failure risk.