"Eighty-five percent." The number went up on the screen in a Zurich conference hall, and the panel kept talking over it.
I was there for the institutional track — the one with the tailored suits and the custody slides. The moderator asked how many stablecoin issuers the industry could realistically support. A managing director laughed. “Two,” he said, “and we should be grateful for that.”
Nobody in the room flinched. That is what unsettled me. Tether and Circle now control roughly 85% of the stablecoin market, a concentration level sitting close to its all-time high. Here is the part the applause line skipped: that 85% is not a statistic about code. It is a statistic about two companies that can mint, freeze, and redeem the dollars the entire market settles in. Chasing the alpha while the market sleeps means noticing the thing the market refuses to price. This is it.
Stablecoins stopped being a crypto curiosity a long time ago. They are the settlement layer — the unit of account inside exchange order books, the collateral in lending markets, the rail for cross-border payments and, increasingly, the cash leg of tokenized treasuries. When you swap two assets on a DEX, the quote you see is denominated in a stablecoin. When a fund moves capital on-chain, it moves in a stablecoin. When a market maker hedges, it parks inventory in a stablecoin. The maturity is real: these are now regulated-adjacent instruments with routine attestations, deep secondary liquidity, and payment integrations that would have sounded absurd in 2018.
That makes issuance the most economically significant position in the industry, and it turns the concentration question into a systemic one rather than a branding one.
I have been auditing token mechanics since the 2017 whitepaper boom — fifty ERC-20s in six months, back when the Golem and Bancor economic models had holes you could read from across the room. From ICO hype to on-chain truth, the habit has not changed: read the issuance logic before you read the roadmap. So let me do the arithmetic most coverage skips.
Convert the market share into a Herfindahl-Hirschman Index and the picture sharpens. Take Tether at roughly 65% and Circle at roughly 20%, with the remaining 15% scattered across a long tail of smaller issuers. Square and sum: 4,225 + 400 plus a tail contribution of roughly 100 points. That is an HHI north of 4,600. Antitrust economists in the United States treat anything above 1,800 as "highly concentrated." This market runs at two and a half times that line, and it has done so for years.
Now locate the risk correctly. USDT and USDC are not protocol-level assets with governance votes and upgrade timelocks. They are claims on off-chain reserves, administered by private companies, with three powers that matter: mint, freeze, redeem. All three are policy decisions executed outside the chain. The blockchain is the display layer. The ledger records what two compliance teams permit. Two private balance sheets now perform a function central banks guard jealously: deciding how many dollars exist on-chain.
That distinction has teeth. When the US Treasury sanctioned a mixer in 2022, Circle froze roughly $75 million of USDC in a single transaction — no fork, no vote, no exit. Tether has frozen billions cumulatively across law-enforcement requests. Those actions are defensible on their own terms. They are also proof that the base unit of DeFi has an admin key.
And this is where the industry's proudest recent engineering runs into a wall. The last three years of DeFi innovation have been about making the spigot programmable. Uniswap V4 hooks let developers bolt custom logic onto pools — dynamic fees, on-chain limit orders, oracle-gated curves. It is genuinely elegant. Turning a DEX into programmable Lego is a real achievement. I would make a different bet, though: the complexity spike will scare off the majority of developers who try to write a hook that does not leak value. Even so, suppose it works perfectly.
A hook still executes after you already hold the token. It cannot manufacture the token. You can build the most sophisticated AMM in existence on top of a base asset whose supply curve is set by two compliance departments. That is not decentralization with extra steps. That is a permissioned core wrapped in programmable packaging.
Two more technical realities get almost no airtime. Start with the redemption mismatch. Crypto settles 24/7. The reserves behind the largest stablecoins — short-dated Treasuries, money market funds, bank deposits — settle in TradFi hours, through TradFi intermediaries, inside TradFi settlement windows. The binding constraint on a stablecoin is not block space. It is a wire transfer on a Monday morning.
Then there is the composability assumption. DeFi's entire architecture presumes a stable unit that can be moved permissionlessly. If 85% of that unit can be paused by a court order from a single jurisdiction, then the money leg of every DeFi stack carries a single point of policy failure — and no amount of auditing addresses it, because there is nothing in the code to audit. The unaudited thing is the boardroom.
Here is where I part company with the consensus read. Most analysts frame this concentration as a regulatory risk — the thing that detonates if Washington turns hostile. I think it is the opposite. It is a regulatory output. Regulation-by-enforcement, with no clear statutory framework for issuance, does not produce a diverse market. It produces a duopoly. When the rules are written as settlements rather than statutes, the rational strategy for any issuer is to stay large, stay boring, and stay inside the blast radius of the biggest regulator. Clarity would open room for a dozen licensed issuers, each competing on reserves, transparency, and speed. Ambiguity rewards incumbency. The 85% is not a market failure the SEC inherited. It is a policy the SEC produced.
The second blind spot is correlation, and almost nobody prices it. Traders treat USDT and USDC as diversification. They are not. Look at the reserve side: both hold short-duration dollar assets — T-bills, repo, government money market funds. Their reserve portfolios are, functionally, the same trade. Long dollar duration, short crypto beta. If a rates shock or a Treasury-market dislocation hits one, it hits both, because the underlying asset is identical. You cannot diversify away a risk that lives in your reserves by holding two different brands.
The decentralized-stablecoin crowd has spent years arguing that the technology is ready. It mostly is. The problem is demand. The market wants a dollar that every exchange accepts, every auditor can verify, and every off-ramp will touch. A dollar that cannot be frozen is philosophically superior and commercially quieter. Capturing the fleeting spirit of the herd means accepting that liquidity follows acceptance, not architecture. And the grant committees funding the alternatives — I have watched a lot of them up close — rarely back the design that survives contact with a bank account.
Watch three signals over the next two quarters. The share of supply held by issuers outside the top two. The on-chain TVL of fully collateralized, governance-minimal designs. And the first real statutory framework that licenses new issuers rather than settling with existing ones.
If the rules arrive and concentration stays above 80%, the duopoly was never about regulation at all — it was about distribution. If it drops below 70%, someone should ask why it took a statute to create a market that was supposed to be permissionless. That gap is the number the next cycle will be decided by. Meanwhile, the two companies that mint the dollars everyone trades remain the least priced systemic exposure in crypto. Human faces behind the blockchain code, as always. Speed meets substance in the void — and the void, for once, has a settlement window.
