Two issuers now control roughly 85% of the entire stablecoin market. Not 85% of trading volume on a quiet Tuesday, not 85% of one chain's bridged liquidity — 85% of the dollar-denominated supply that DeFi borrows against, that exchanges quote in, and that most people I know quietly treat as the safest place to sit out a bear market. Concentration is near an all-time high.
No governance vote produced that number. No protocol upgrade shipped it. Two companies grew, and the rest of the market contracted around them. I did the arithmetic on the back of an index card in a Washington coffee shop, because I wanted to see it without a dashboard's reassuring gradient underneath: an Herfindahl-Hirschman Index in the low four thousands depending on how you split the two, two to three times past the threshold that competition economists describe as highly concentrated.
That is not a price signal. It is a structural fact about where your assets actually sit when you believe they are sitting in your wallet. And in a bear market, structure is the only thing that survives long enough to matter. Price action is a rumor; the issuance table is a fact.
A stablecoin is not a cryptocurrency in the sense a Bitcoin maximalist means the word. It is a redemption obligation — a contractual claim against a private balance sheet — mirrored onto a public ledger. Tether began in 2014 as Realcoin; USDC arrived in 2018 under the Centre consortium as Circle's answer to the same problem. Both solved one question: how do you move dollars across a permissionless network without asking a bank for permission each time?
The answer they found was to become the bank. Today they are the numeraire of the industry: the collateral in lending markets, the settlement leg in most over-the-counter flows, the base asset on nearly every pair a person actually trades. When the bear market arrived, the question changed. Nobody asks what yields fifteen percent anymore. They ask whether the thing they are holding can be redeemed, by whom, and on what terms.
That shift is why concentration matters more now than it did in 2021. In a bull market, depth hides fragility: new deposits paper over withdrawals, and everything looks solvent because everything is growing. In a contraction, the exit queue is where the risk lives. A redemption is a bank run with a web form instead of a sidewalk, and the shape of that queue depends on how many issuers are actually standing behind the door.
So the 85% figure is not an abstraction for regulators to argue over. It is the answer to a practical question a reader sent me last month: if my stablecoin breaks, who is on the other side of the trade? Four times out of five, the answer is the same two companies — whether you chose them or not.
Based on my audit experience reading Solidity line by line — including six months I spent on consensus-level code in 2017, when I declined advisory money to do the unglamorous work — the two dominant stablecoin contracts are not complicated. They are simple in a way that should unsettle anyone who repeats the word immutable.
USDC is deployed behind an upgradeable proxy. Balances stay put while the implementation logic beneath them can be replaced, which is a polite way of saying the rules are editable. Alongside ordinary transfers, the contract exposes blacklist and unblacklist functions, a pause that halts every transfer in the token, and mint authority granted to a small set of issuer-controlled addresses. Tether's Ethereum contract carries a comparable toolkit: functions to add addresses to a blacklist, to destroy the funds held at those addresses, to adjust fee parameters, and to transfer ownership of the contract itself. Neither is a bug. Neither is hidden. Both are documented, deliberate, and shipping in production.
Those powers get used. Addresses tied to sanctioned mixing services were frozen administratively, without a court order in the loop and without any on-chain governance process. That is the part of the story infrastructure builders prefer to skip: what you hold is a revocable license with a market price attached. Truth is immutable, unlike the price action — but on this stack, truth has an administrator. Immutable settlement, revisable issuance.
There is a defensible reason the market chose this architecture. Liquidity is the product. Reversibility is the feature exchanges cannot operate without; a subpoena needs an address to answer to. Custody teams will not hold an asset that cannot be frozen, and treasury departments will not settle in an instrument whose supply is voted on by strangers. Concentration is not a conspiracy. It is the visible cost of solving a real problem, and pretending otherwise makes the analysis useless.
The real risk is not that these functions exist. It is where the risk sits during a contraction: redemption, not price. In March 2023, news that part of USDC's reserves sat with a failed California bank pushed the token to roughly eighty-seven cents in a Curve pool within hours. DAI, which held a large share of its collateral in USDC through its peg stability module, followed it down — a decentralized protocol depegging because a centralized issuer's balance sheet had a hole in it. That was a small, contained shock. Model the same reflexivity across 85% of supply and the containment assumption stops holding.
Composability does the rest. Because nearly every major pair routes through these two assets, avoiding them carries a measurable tax: thinner books, wider spreads, fewer counterparties, worse execution. Sovereignty has a slippage cost, and in a bear market most participants pay the spread instead. That is how concentration becomes self-reinforcing — not through marketing, but through the ordinary arithmetic of doing a trade.
Then there is the verification gap. Reserve reports arrive as periodic attestations — statements that on a given date, a given set of accounts held a given set of assets — not as continuous, cryptographically verifiable proofs. I spent much of 2025 working on exactly this problem, drafting guidelines for how zero-knowledge proofs could let an issuer demonstrate solvency without exposing its counterparties. The hard lesson from that work is that a proof of reserves proves a snapshot. A bank run is a movie. Proving the first frame tells you almost nothing about the frames that follow.
Consider what the reserves are. The largest issuer's backing has moved heavily toward short-dated government debt and repo, which is more transparent than the commercial paper era that frightened people in 2022. But short-dated Treasuries remain a duration position with a market price, and a run that forces liquidation into a disorderly rate environment realizes losses that a par-value report will never show. Solvency and liquidity are different questions, and attestations only answer the first.
The second issuer's model is cleaner on paper: segregated cash and short Treasuries, monthly reports. Its concentration risk is arguably lower. But it is also the asset institutions actually custody, which means its freeze powers get exercised more often, its issuer answers to banking supervisors, and its peg depends on the same handful of custodial banks that failed in 2023. Two issuers. Two different fragilities. One shared point of failure underneath them both.
This is where the ETF conversation and the stablecoin conversation quietly converge. Institutional custody, tokenized treasuries, and dollar stablecoins are the same plumbing seen from three angles. When one custodian holds the fund's coins and the same small set of banks clears the reserves, the distinction between crypto exposure and stablecoin exposure becomes largely cosmetic.
I made this argument in 2024, after the ETF approvals, and it cost me some professional goodwill. Two thousand emails arrived from readers holding the same doubt with no venue to voice it. The bear market has not changed the structure. It has only removed the noise that made the structure easy to ignore.
None of this is a stablecoin problem in isolation. It is a routing problem that surfaces in every leverage loop built on top. When the collateral everyone posts is issued by two entities, every liquidation cascade inherits their operational and legal risk. A pause on one contract is not a pause on a market; it is a pause on that market's capacity to settle at all. That is what systemic risk means in practice — not that prices fall together, but that settlement stops together.
What makes the measurement itself tricky is that stablecoin market share is not a single number. It depends on whether you count supply, transfer volume, or the share of pairs where the asset is the quote currency. Set the metric to transfer volume and the picture can swing by double digits in a week, because a handful of large desks can move more notional than millions of retail wallets. Set it to circulating supply and the figure is stubbornly stable, which is the tell: supply reflects accumulated trust, and trust does not reprice quickly.
The other thing that rarely gets said plainly: control of issuance is monetary policy, executed by private firms. When a mint address is opened, dollars enter the on-chain system; when tokens are burned at redemption, they leave. These are open-market operations without a committee, a mandate, or a published reaction function. If you would not accept that arrangement from a central bank you distrust, it is worth asking why the same arrangement is acceptable when the operators are incorporated in a jurisdiction you find convenient.
Here is the uncomfortable part for anyone who shares my instincts. The reflexive purist answer — rotate into decentralized stablecoins — does not obviously survive a genuine run. The canonical algorithmic experiment collapsed in 2022 precisely when demand for the exit peaked. The remaining tail of supply is not a resilient alternative so much as a set of correlated, thinly traded, over-collateralized positions, several of which hold the very assets they exist to hedge against. The tail inherits the head's risk while offering none of its depth.
Concentration is also pro-cyclical. Under stress, liquidity migrates toward the deepest venue, which means market share rises exactly when diversification would be most valuable. A portfolio constructed around the alternatives is therefore most likely to fail in the one scenario it was designed for. Markets do not reward the principled position; they reward the position that settles.
And regulation, sold as the corrective, is more likely to lock the structure in place. Reserve, custody, capital, and audit requirements are fixed costs that scale with compliance headcount rather than with ambition. The frameworks taking shape in Europe and the United States are achievable for firms already holding billions and effectively impossible for a five-person protocol. Consumer protection written into law becomes a moat written in the same ink. The plausible trajectory of today's 85% is 90%, not 60%.
So the question I keep returning to, with the ledger open on a cold morning, is not whether that number climbs. It is what we believe we are holding. If the dollar's on-chain rails can be paused, frozen, and upgraded by two boards of directors and a handful of commercial banks, then the decentralization we spent a decade building sits one layer above the money — and the money was never the part we fixed. Which layer did you actually come here for?