Over the past seven days, the stablecoin market printed a number that should have stopped every desk in Toronto, Singapore, and Zug: Tether and Circle together control roughly 85% of all outstanding stablecoin supply. Concentration is back near historic highs. No exploit. No depeg. No drama. Just two issuers, one dollar, and a market that has quietly decided that's acceptable.
It isn't. But it is profitable, and in this industry those two things are identical until the day they aren't.
I carry a specific bias into this. In 2017 I raised $40,000 from 200 people for a token whose code worked and whose story didn't. The lesson I took away wasn't "don't lie." It was that capital flows toward the most coherent narrative, never the most correct architecture. Eight years later, the most coherent narrative in stablecoins has nothing to do with decentralization. It is a bank account in Manhattan and a treasury operation somewhere else, and the market is paying a premium for it.
Context: How Plurality Collapsed Into Two
Stablecoins were supposed to be plural. The 2018–2021 cycle produced algorithmic experiments, commodity-backed designs, and a wave of "decentralized" clones. Most died quietly. Terra died loudly, and in dying it burned the credibility of an entire design school — the one that promised you could engineer a dollar without ever holding one.
What survived, and then compounded, was the boring answer. Centralized issuance. Fiat reserves. A redemption promise backed by a legal entity that answers to someone. Tether built distribution through offshore exchanges and emerging-market corridors where the local currency was worse. Circle built legitimacy through US regulation, attestations, and a Coinbase partnership that put it in front of retail by default. Between them, the two firms captured the settlement layer of crypto without ever calling it that.
This is where most commentary misreads the situation. Stablecoins are not a product category where competitors fight for share. They are infrastructure — the rail through which DeFi protocols, exchanges, and now tokenized treasuries clear. When 85% of that rail sits with two counterparties, you aren't watching a competitive market. You're watching a duopoly wearing a token wrapper.
Core: What 85% Actually Does
Concentration in the issuance layer is not the same as concentration in the settlement layer. That distinction is the whole game, and almost nobody models it.
A single USDC or USDT unit can be bridged, wrapped, rehypothecated, and lent a dozen times across a dozen chains. The units look diversified. The claims do not. Every one of those wrapped representations terminates at the same two redemption windows, governed by the same two legal entities, subject to the same two sets of banking relationships. I have audited bridge architectures where the underlying collateral was, on a look-through basis, more than 90% one issuer. The chain-level dashboard said diversification. The claim-level view said one door with two hinges.
That is the receipt. Tokens are receipts; memes are the religion. The receipt for most of DeFi's dollar exposure carries two signatures.
Run the concentration math properly and it gets worse. On a Herfindahl-Hirschman basis, stablecoin supply now sits well above the 2,500 threshold economists use to label a market "highly concentrated." In normal markets, concentration is tolerated because it supposedly produces better prices for consumers. Here it produces better prices for the issuers. Reserve income on a float north of $200 billion accrues to two balance sheets. In a rate environment where short-dated Treasuries still yield meaningfully, that is an enormous, quiet subsidy — funded by everyone who holds a dollar in crypto for convenience rather than carry.
Then there is the queue. Every DeFi position I have modeled over the past year — lending markets, perpetual DEXs, liquid staking loops — behaves beautifully in the happy path. The stress path is what matters, and the stress path always terminates at a mint-and-redeem window with business hours attached. In 2023 I built a deliberately crude model of a 15% simultaneous redemption across the top five lending markets. The on-chain liquidations were the fast part. The slow part, the portion that never clears inside a weekend, was converting collateral into bank-settled dollars. At 85% concentration, that queue has two exits and both of them close on Friday afternoon.
Transparency compounds the problem rather than relieving it. Circle publishes attestations with a cadence you can set a calendar to; Tether publishes a different kind of document on a different rhythm, and the market has learned to price the difference as noise rather than as information. Two issuers, two disclosure regimes, one shared systemic exposure. When the underlying collateral is nearly identical in composition — short-dated government paper and cash equivalents — the disclosure gap stops being a governance question and becomes a mispricing. You are not choosing between two risk profiles. You are choosing between two reporting styles for the same risk.
My 2020 work on Compound is relevant here. I spent that DeFi Summer arguing that financializing governance manufactures misaligned incentives. I was early, I was ignored, and then it played out. The same shape is visible now. The community keeps debating decentralized stablecoin design — collateralization ratios, oracle mechanisms, governance tokens, hook-based liquidity modules on the newest DEX primitives — while the market keeps choosing distribution. Design optimizes for resilience. Distribution optimizes for the next block. Distribution wins every cycle, and then the cycle ends and everyone rediscovers the design conversation with fresh sincerity.
The narrative layer makes this legible. Stablecoin stories run in three phases. Phase one: "this is a dollar, it's boring." Phase two: "this is a yield instrument, it's DeFi's oil." Phase three: "this is systemically important, it's too big to fail." We are firmly in phase three. We didn't find a coin; we found a consensus — the consensus that crypto's dollar should be issued by exactly the kind of institution crypto was invented to route around. That consensus is now priced to perfection. Nobody is paying for the tail.
There is an institutional mirror to this that I saw up close. In 2024 I advised a Toronto hedge fund on a $50 million crypto allocation after the ETF approvals. The due diligence conversation was never about whether Bitcoin was sound money. It was about what settles the trade. The answer, every single time, was USDC. That is the moment I understood that stablecoins are the actual institutional entry point into this asset class, and that concentration risk sits buried inside a custody line item that nobody on the investment committee reads closely. Institutions believe they bought Bitcoin exposure. They also bought two counterparties.
The same asymmetry shows up at the exchange layer. Customer balances sit in stablecoins. When stablecoin concentration rises, exchange balance-sheet risk rises with it, and that risk is correlated across every venue because the reserve asset is identical. Layer-2 fragmentation taught us this lesson in a different key — dozens of rollups, the same small user base, liquidity sliced rather than scaled. Stablecoins solved the scaling problem by refusing to fragment. They solved it by concentrating. That worked, and that is precisely why it is now the system's single point of interpretive failure.
Contrarian: The Risk Isn't a Depeg. It's a Kill Switch.
The consensus tail scenario is a Tether or Circle failure — reserves come up short, a run starts, the dollar breaks. That's the horror movie, and it is one of the least likely versions, because both issuers now sit inside the regulatory perimeter with banking partners, attestations, and political exposure. A disorderly collapse would damage the very people who supervise them.
The realistic tail runs the other direction. Nothing breaks, and the state uses the concentration. Two issuers holding 85% of supply is not a fragility problem from a supervisor's perspective. It's an interface. It means a single phone call can freeze, sanction, or slow the crypto dollar in ways that would be impossible across a thousand permissionless protocols. From Washington or Brussels, concentration isn't the bug. It's the control panel.
That reframes everything. The danger is not that Tether dies. It is that Tether and Circle stay healthy, compliant, and increasingly conditional. Address screening already does this at the edges. The next increment is programmability at the issuance layer, and programmability requires a duopoly to implement. Fragmentation would make it unenforceable.
Chaos is the alpha, but coherence is the asset. The coherence of the stablecoin market is precisely what makes it capturable.
So the contrarian trade is not "short the dollar." It is to stop valuing decentralized stablecoins as products and start valuing them as options. You aren't buying a better dollar. You are buying the right to settle without permission in a world where permission quietly becomes the default. That option throws off almost no current cash flow, which is why it looks absurd on every spreadsheet I have built. Options aren't priced on cash flow. They're priced on regime volatility.
The market is sideways. Chop is for positioning. This is the environment where you accumulate optionality nobody wants and forget about it for a cycle.
Takeaway
Watch three things and ignore the headlines. Stablecoin concentration falling below 70%. Decentralized stablecoin TVL breaking 20% of total stablecoin TVL. Any regulatory framework that formalizes issuer-level programmability.
The first is bullish for the duopoly. The second is bullish for the option. The third makes the first two irrelevant, because once the dollar on-chain becomes a permissioned rail, the question stops being which stablecoin and becomes who still has an exit.
I'll be honest about where I stand. I don't carry a large decentralized stablecoin position. Almost nobody does, and that is exactly the point. The trade isn't obvious, it isn't liquid, and it isn't popular.
Which is the only place alpha has ever lived.
