Two addresses. That is the entire market.
Data does not negotiate; it only confirms. And what the mint ledger confirms this quarter is a number that should stop anyone who has ever run a concentration check on a counterparty book: Tether and Circle together account for roughly 85% of all outstanding stablecoin value — a share that reads near the highest level ever recorded in the category. Not a cycle high. An all-time high.
The number has been reported all week as a risk flag. Systemic risk. Regulatory challenge. The usual vocabulary. Read the headlines and you get a story about two companies that got too big.
That framing is wrong. It is wrong in a way that will cost money.
Eighty-five percent concentration inside a single competitive market would at least imply pricing pressure, switching costs, and a functioning arbitrage. What we actually have is two near-monopolies that occupy different jurisdictions, serve different customers, and share almost no rails. The market cap number adds them together. The plumbing does not.
The risk is not the duopoly standing. It is the thin, unhedged, largely manual seam running between them — a seam that a handful of off-chain desks control. Almost nobody is pricing that seam. That is the trade.
What Actually Changed
Start with the structure, because the structure is the story.
Tether issues USDT. Circle issues USDC. On a spreadsheet, they are two rows in the same column. In practice, they are two different financial products wearing the same ticker suffix.
USDT is the offshore dollar. It lives on Tron, on Ethereum, on a dozen smaller chains, and it is the settlement asset of choice in markets that either cannot access or do not want the US banking perimeter — Turkey, Argentina, Nigeria, Vietnam, and a meaningful slice of OTC crypto trading that never touches a US-regulated venue. Tether's reserve base is dominated by short-dated US Treasuries, repo, and — a detail the market keeps forgetting — a Bitcoin position and a gold position that do not behave like a money market fund. Tether's attestations are quarterly. They are attestations. Hold that word.
USDC is the onshore dollar. It is the stablecoin that Coinbase distributes, that Circle has built into a regulated money market wrapper, and that became the first major issuer to fully restructure around MiCA after the EU regime reached full applicability in December 2024. Circle went public in June 2025. Its economics are now public, auditable, and — this matters — its reserve income is shared with distribution partners, which means its margin is a function of a distribution agreement, not a technology moat.
The regulatory landscape hardened into law rather than guidance. The US stablecoin legislation passed in mid-2025 set the template: one-to-one reserve backing, monthly disclosure of reserve composition, no yield paid to holders, and a dual federal-state chartering regime. Every major issuer now has a compliance department that looks like a bank's.
That is the context. Two issuers. Two jurisdictions. Two compliance postures. One number — 85% — that has been flattened by analysts into a single "concentration risk" label.
The label is the analytical error. The label assumes the 85% is one market. It is not.
The Reserve Is the Product, and the Reserve Is a T-Bill Ladder
I want to be specific about where the risk physically sits, because "stablecoin risk" is the kind of phrase that lets people avoid thinking.
Under the current regime, a compliant stablecoin is a pass-through vehicle for short-duration US government debt with a zero-duration liability attached to it. The issuer takes customer dollars, buys 0–3 month Treasuries and repo, and owes the customer a par redemption on demand. Margin is the spread. That is the whole business.
Which means the concentration question is not really a crypto question. Scale it up and ask the real one: how large a position in the front end of the US Treasury curve does this represent?
The answer is large enough that Tether, on its own disclosed reserve composition, sits in the same conversation as sovereign holders of US paper. A single private company, incorporated in the British Virgin Islands, is one of the largest holders of short-dated US government debt in the world. Its redemption obligations are denominated in dollars. Its assets are denominated in dollars. Its liquidity transformation is nominal, not structural.
On paper that is the safest balance sheet in crypto.
In practice, it is the most procyclical position in the system, because the front end of the curve is exactly where funding stress shows up first. September 2019 repo. March 2020 dash-for-cash. The 2023 debt-ceiling episode, when bill yields gapped on default-adjacent headlines. August 2024, when a yen carry unwind forced global deleveraging in a single session.
Every one of those events hit the 0–3 month sector hardest. Every one of them is a stress test on stablecoin reserves — run through the exact instrument that backs the stablecoin.
I spent the 2020 DeFi summer building break-even models for yield farms that were paying 400% APR on emissions that could not survive their own unlock schedule. The lesson was not that high yield is fake. The lesson was structural: yield is not income; it is risk repackaged. The stablecoin float is now the largest single repackaging of that risk ever assembled — $200B-plus of par-redeemable claims sitting on top of a duration ladder that repriced violently four times in six years.
Nobody calls it a money market fund. It is a money market fund without the SEC's Rule 2a-7 liquidity buffers, without a board empowered to gate redemptions, and without a sponsor obligated to break the buck.
Silence in the Ledger Speaks Louder Than Hype
Here is what I keep coming back to. The audit trail never lies, only the auditor can.
Tether has operated for over a decade without producing a full financial audit. It publishes attestations — point-in-time confirmations of reserve composition, signed by an accounting firm. Attestations are not audits. An attestation tells you what the balance sheet looked like at a moment. An audit tests existence, completeness, internal controls, and the reconciliation between the two across a period. The gap between those two documents is not semantic. It is the difference between "the assets were there on the last day of the quarter" and "the assets are controlled by the entity and the numbers are complete."
Circle has published audited financials since its public listing. That is a real, structural, non-narrative difference between the two halves of the 85%.
And yet the market assigns them roughly the same risk premium, because both are priced on liquidity and adoption, not on disclosure quality. That is a mispricing with a long history. In 2017 I spent 72 hours reverse-engineering a token contract before a public launch and found three reentrancy paths that nobody had disclosed. The contract was not lying. The documentation was. Same pattern here, different layer.
The disclosed reserve composition is the contract. The absence of an audit is the missing documentation.
Sit with what is not in the ledger. There is no quarterly reconciliation of customer liabilities to on-chain mint events published by either issuer — you have to reconstruct it. There is no public reserve buffer policy — no stated minimum excess over 1:1. There is no disclosure of the redemption queue depth or the internal thresholds that trigger a manual review before a large redemption is processed. Tether's redemption terms are contractual, not algorithmic: minimums, fees, verification tiers, business-day settlement.
Silence in the ledger speaks louder than hype. And in a bull market, the silence is what nobody is reading.
The March 2023 USDC de-peg is the cleanest available case study. When it became known that a portion of USDC's reserves sat as uninsured deposits at a failed bank, USDC traded to roughly $0.87. The recovery did not come from a smart contract. It did not come from an on-chain mechanism. It came from a phone call, a weekend, and a decision by the federal government to backstop depositors.
Read that again. The largest onshore stablecoin in the world was rescued by a policy decision, not by its code. Every line of Solidity in USDC executed perfectly. The reserve was the risk, and the reserve was a bank deposit.
The Real Exposure Is the Seam, Not the Duopoly
Now the part that the 85% headline actively obscures.
Concentration is usually measured on stock — how much of the float does each issuer control. Stock is the wrong lens. What moves markets and breaks pegs is flow: the rate at which value crosses from one issuer's liability to the other's, and the depth of the channel that carries it.
There is no protocol for USDT-to-USDC conversion. There is no atomic swap between the two at scale. There is no shared redemption facility, no joint liquidity backstop, no netting arrangement. What exists is a set of OTC desks, market makers, and a handful of cross-chain bridges, each quoting a spread and each carrying its own counterparty risk.
So the correct description of the stablecoin market is not "an 85% duopoly." It is two near-monopolies, each with a captive customer base, connected by a thin manual seam controlled by a dozen firms.
That is strictly worse than a duopoly. In a duopoly, competition prices the product. In a partitioned market, each issuer sets its own redemption terms, its own fee schedule, its own minimums, and its own jurisdictional posture — and the customer cannot switch without paying the seam.
And the seam gets thinner every cycle, because the routing layer is being rebuilt off-chain.
Intent-based architectures — the UniswapX, Across, and 1inch Fusion model — have replaced AMM routing with a competitive solver auction. The user signs an intent; solvers bid to fill it; the winning solver executes and captures the spread. This is genuinely better UX. It is also a direct transfer of price discovery from a transparent, on-chain order flow into a small number of off-chain solver books.
I have said this before and I will keep saying it: intent-based architectures do not remove MEV, they relocate it. The extraction does not vanish. It moves from a public mempool to a private auction among a solver set that is measured in single digits. Apply that to stablecoin FX — the USDT/USDC cross — and you get the following: the seam between the two halves of the 85% is being priced by an off-chain oligopoly.
When the seam holds, nobody notices. When it doesn't, there is no on-chain mechanism to fall back on. The 2023 episode proved the fallback is a phone call.
Settlement Costs Are About to Reprice, and Nobody Has Modeled It
One more layer, because it changes the cost base of the entire thesis.
A meaningful share of stablecoin payment volume is settling on Layer 2s. Issuers have spent two years building there — mint/burn contracts, native issuance, cross-chain messaging standards — on the assumption that L2 settlement is permanently cheap.
That assumption has an expiry date.
The cost collapse after EIP-4844 came from blob space. Blobs are a finite, per-block resource, and the demand curve for that resource is not flat — it is driven by every rollup competing for the same block space at the same time. Post-Dencun blob data will be saturated inside two years, and when it is, rollup gas fees double. Not because anyone changed the rules, but because a fixed supply met an expanding demand and the fee market did what fee markets do.
Stablecoin issuers are building their settlement layer on a subsidized cost curve. The subsidy is temporary. When it expires, the per-transaction economics of L2-based payments change materially — and the issuers who modeled permanent sub-cent settlement will be repricing their unit economics in public.
This is the same error as the 2020 yield farms, one layer down. The emissions schedule was always going to run out. The blobs were always going to fill.
The Contrarian Read
So here is the position, stated plainly.
The 85% concentration is not the systemic risk. The systemic risk is that the 85% is not one market, it is two, and the conversion channel between them is a manual oligopoly with no fallback mechanism.
The consensus trade is to short the duopoly narrative — to assume that concentration at a historic high implies mean reversion, that decentralized stablecoins will absorb share, that competition will restore balance. Watch what actually happens.
Examine the incentive structure. Every new regulatory framework that lands makes the compliance burden heavier, and heavier compliance burdens favor scale. MiCA did not fragment the European stablecoin market; it pushed non-compliant issuers out and consolidated the remainder into the two names that could afford the licensing. The US regime does the same thing with its reserve and disclosure requirements: it converts stablecoin issuance into a regulated financial activity with a fixed cost floor. Fixed cost floors are moats.
I did the same exercise in 2024, standardizing several hundred pages of SEC filing language into a scored approval framework. The pattern was consistent: regulation does not level the playing field, it fences it. The players already inside the fence get larger.
Which means the 85% is not a peak. It is a plateau with upside.
Watch what PayPal did. They launched their own dollar token rather than routing payments through the incumbent float. Read the decision carefully, because it is the clearest strategic signal in the sector: PayPal chose to become a regulatory partner rather than wait to be regulated as a customer. That is not a product decision. It is a positioning decision, made by a company that understood the fence was coming and bought a seat inside it.
If concentration were going to mean-revert on its own, that move would not have been necessary.
The second half of the contrarian read is about what the concentration actually threatens — and it is not crypto. It is the front end of the Treasury curve. The stablecoin float is now a structurally captive bid for 0–3 month bills. That bid is price-insensitive in the way that matters: issuers buy bills because the regulation requires short duration, not because the yield is attractive. Remove or stress that bid and the marginal buyer of the world's risk-free asset changes composition.
That is a macro story, and the market is trading it as a crypto story.
The third piece, and the one I would put money behind: the seam is where the first failure will appear. Not a de-peg headline. A widening spread. Watch the USDT/USDC cross on the OTC desks. Watch the number of solvers quoting it. Watch how the depth changes when a mid-size issuer or a mid-size exchange runs into trouble.
The Next Signal
Track three things, and track them weekly, not quarterly.
Effective concentration as measured by HHI on the float — if it breaks below roughly 70%, the fencing thesis is wrong and the duopoly is genuinely eroding. Flow concentration across the conversion seam — if the number of independent solvers quoting USDT/USDC materially expands, the seam is getting more resilient. And first-party disclosure quality — whether either issuer moves from attestation to full audit, because that single document would reprice the risk premium on roughly half the float overnight.
I have run this checklist before. I ran it on Terra four hours after the de-peg, and the answer was in the redemption terms, not the price chart.
The number is 85%. The question is not whether two companies are too large. The question is who prices the gap between them, and what happens to that price when the two halves stop agreeing on what a dollar is worth.
Speed without structure is just noise. Structure is what you build the model on.
Build the model. Then wait for the seam to widen.
