Eighty-five percent is not a narrative. It is a measurement.
Tether and Circle now control roughly 85% of the stablecoin market by circulating supply — a concentration ratio sitting near the historical ceiling set during the 2021–2022 expansion. This is the highest sustained concentration since the post-2022 consolidation, when several competing issuers — BUSD, USDP, and a dozen smaller tokens — either wound down, were discontinued under regulatory pressure, or lost distribution entirely. The figure surfaced this week as a headline. It should surface as a structural warning.
Most readers skip the operative clause. The number describes issuance, not architecture. Stablecoins are not a decentralized technology layer; they are a centralized liability ledger rendered in token syntax. When I performed the forensic audit of the Parity multisig library in 2017, I learned that the most dangerous component of any system is rarely the one described in the whitepaper. It is the one the whitepaper assumes away. For stablecoins, that component has a name, a board of directors, and a banking partner.
—
To understand why 85% matters, you must first understand what a stablecoin is not. It is not a bearer asset. A holder of USDT or USDC does not own a dollar. The holder owns a contractual claim against an issuing entity, denominated in dollars, redeemable at that entity's discretion.
That distinction is the entire risk surface.
The supply divides into two issuers. Tether issues USDT. Circle issues USDC. Together they account for roughly eight and a half of every ten stablecoin units in existence. The remaining 15% is split across a long tail — DAI, FDUSD, PYUSD, TUSD, and a rotating cast of algorithmic and crypto-collateralized experiments, most of which have either failed, depegged, or remained structurally marginal.
Measure the concentration formally. The Herfindahl-Hirschman Index squares each participant's market share. If USDT holds 65% and USDC holds 20%, the index reads 4,625. United States antitrust guidelines define anything above 2,500 as "highly concentrated." This market is not merely concentrated; it is concentrated at a level that would trigger intervention in any comparable financial sector. And stablecoins are, functionally, a sector: payment rails, collateral, and settlement infrastructure compressed into a single asset class.
The bull market has a way of burying this. When prices rise, stablecoin supply expands, and concentration reads as adoption. Every new USDT minted looks like demand. It is not. It is a claim-creation event, and every claim created against a single issuer increases the correlated failure surface of the entire ecosystem. Hype builds the floor; logic clears the debris.
—
Begin with the redemption asymmetry. It is the mechanical root of everything that follows.
Minting is permissionless. Redemption is not. A user can acquire USDT on any exchange in seconds. Converting USDT back into a bank dollar requires the issuer's cooperation, its banking partner, its compliance desk, and its willingness to honor the claim at par. This asymmetry is invisible during normal operations. It becomes the only thing that matters during stress.
Consider the arithmetic. If 85% of supply sits with two issuers, then any event impairing one issuer's redemption capacity impairs at least 42.5% of the market — and likely far more once contagion pricing hits the second issuer. There is no diversification in a duopoly. There is only the illusion of a market.
Systemic risk in stablecoins is not a function of reserve quality; it is a function of claim correlation. Two issuers, one asset class, one redemption mechanism, one banking channel. The redundancy is cosmetic.
Now examine the reserve layer, where the public record is thinnest.
Circle publishes monthly attestations from a major accounting firm and discloses its reserve composition — largely short-duration Treasuries and cash held at regulated custodians. That is a verification posture. Tether publishes quarterly attestations rather than full audits and has historically held a broader reserve mix. That is a trust posture. The difference matters precisely because it is the difference between a variable and a constant.
Trust is a variable; verification is a constant.
Here I depart from the conventional critique. The problem is not that Tether's attestations are weaker. The problem is that the market prices both issuers as if the reserve question were settled. It is not settled. It is deferred. A deferred verification is not a passed verification; it is a liability carried off the visible balance sheet of market sentiment.
In 2022, I modeled the TerraUSD mechanism 72 hours before its collapse. The model was not sophisticated. It required only identification of a circular dependency: UST's peg depended on LUNA's market capitalization, and LUNA's market capitalization depended on the peg holding. A feedback loop with no external anchor. I hedged with inverse perpetuals and preserved capital while the market lost everything.
The stablecoin duopoly is not Terra. It has real reserves, real revenue, and real banking relationships. But it shares one structural property with Terra: the asset's value depends on the continued willingness of a small number of actors to honor an off-chain promise. The promise is better collateralized. The dependency is identical in kind.
Follow the dependency into DeFi, where it compounds.
Every major lending protocol — Aave, Compound, Morpho — accepts USDT and USDC as collateral and as the unit of account for debt. Every major DEX pairs against them. Every perpetual exchange settles in them. Stablecoins are not merely used within DeFi; they define its risk-free rate. The entire yield curve of on-chain credit is anchored to two issuers.
This is the composability trap. When a single asset is simultaneously the collateral, the quote currency, and the settlement medium, a shock to that asset is not a shock to one market; it is a shock to the coordinate system itself. Liquidations cascade not because borrowers are reckless, but because the denominator of every position moves at once.
I have argued before that the data availability layer is overhyped — that the overwhelming majority of rollups do not generate enough data to justify dedicated DA. The same logic applies here, inverted. Stablecoin issuance genuinely operates at the scale where concentration becomes systemically relevant. That is not hype. That is arithmetic.
The Regulatory Arbitrage Layer
Now the geographic dimension, which most analysts treat as background noise. It is not background. It is the mechanism.
Hong Kong's stablecoin ordinance and Singapore's MAS framework are not competing to "embrace innovation." They are competing for the same balance sheet: the settlement flows of Asia. When Hong Kong licenses stablecoin issuers, it is not endorsing decentralization. It is building a regulated corridor to capture the dollar-clearing volume that currently routes through Singapore and, historically, through offshore channels.
The concentration data explains the urgency. If 85% of stablecoin supply flows through two issuers subject to United States jurisdiction, then every other financial center is structurally subordinate to American regulatory reach. Hong Kong's licensing regime, Singapore's reserve requirements, the EU's MiCA — these are not harmonization efforts. They are jurisdictional bids for control over the same 85%.
Read the licensing frameworks as infrastructure, not ideology. A stablecoin regime is a claim on settlement sovereignty, and settlement sovereignty is what financial hubs actually sell.
This is why regulatory risk in stablecoins does not resolve. It redistributes. Each new framework adds a compliance layer without removing the underlying issuer concentration. The 85% constant survives every regulatory permutation, because regulation targets issuance behavior, not issuer identity.
The Algorithmic Alternative
The obvious counterargument: decentralized stablecoins will absorb the demand. The record says otherwise.
Crypto-collateralized models — DAI, LUSD, crvUSD — require overcollateralization and therefore capital inefficiency. They scale with the value of their collateral, which makes them procyclical: strongest when the market is strongest, weakest when stability is needed most. When ETH drew down in 2022, DAI's supply contracted alongside it. A stablecoin that shrinks during stress is not a safe haven; it is a leveraged long.
Algorithmic models told a cleaner story and delivered a shorter lesson. UST reached tens of billions in supply before it reached zero in a week. The failure was not bad luck. It was a structural inevitability: a peg maintained by reflexive incentives has no external anchor, and a peg without an external anchor is a promise to keep promising.
The 15% tail is not the seed of future competition. It is the pressure-release valve that lets the market pretend competition exists.
Kill Switch
Every review I publish includes explicit failure conditions. Here are the ones for the stablecoin concentration regime.
Condition 1 — Attestation failure. A single missed or qualified attestation from either issuer, without same-week remediation, triggers a credibility cascade. Precedent: the March 2023 USDC depeg to $0.87 following the Silicon Valley Bank disclosure. That depeg resolved in roughly 48 hours. A longer window in a more leveraged market would not self-correct.
Condition 2 — Regulatory reserve segregation. If a major jurisdiction mandates full segregation with daily cryptographic proof, issuers face a compliance cost that compresses interest income. Revenue declines; the incentive to defend the par fiction weakens.
Condition 3 — A sustained depeg exceeding 2% for more than 48 hours. Below this threshold, arbitrage restores parity. Above it, the market begins pricing the claim as an equity claim rather than a cash claim, and the redemption queue becomes a run.
Condition 4 — Further concentration. If the combined share rises above 90%, the duopoly becomes a de facto monopoly with a subordinate satellite, and the systemic risk premium reprices across all of DeFi at once.
Each condition is observable. None requires a forecast. Code does not lie, but it often omits the truth. So do attestation reports.
—
Here is what the bulls got right, and it deserves stating without sarcasm.
Stablecoins work. Settlement is fast, final, and available outside banking hours. A USDC transfer clears in seconds for fractions of a cent; a correspondent-bank wire clears in days for tens of dollars. Tether and Circle did not capture 85% through conspiracy. They captured it through liquidity network effects that decentralized models have repeatedly failed to replicate.
Liquidity is a moat. The deepest pool attracts the most volume; the most volume deepens the pool. A new stablecoin with perfect collateralization and clean audits still fails if no exchange lists it, no protocol accepts it, and no market maker quotes it. The 15% tail has spent years proving that soundness without distribution is commercially irrelevant.
So the bull case is legitimate: concentration is the price of efficiency, and the market chose efficiency. Circle's compliance posture has made it the institutional default; Tether's operational resilience has made it the emerging-market default. Both outcomes are rational.
The blind spot is not that concentration exists. It is that the market has stopped pricing it. A risk that is known and unhedged is not a managed risk; it is a deferred loss. The bulls are correct about the moat and wrong about the margin of safety.
—
Watch the ratio, not the price. The signal that matters is not whether bitcoin prints a new high; it is whether the decentralized stablecoin share crosses 20% of total supply. Until it does, every DeFi yield is underwritten by a duopoly, and every duopoly is one attestation away from a repricing. The question is not whether the 85% constant holds. It is who is holding the claim when it breaks.