Everyone is watching the ETF tape. Watch the float instead.
Aggregate supply of dollar-pegged tokens crossed roughly $310 billion this quarter, per issuer disclosures and on-chain supply counts. Three years ago the number was under $130 billion. That is not growth. That is a new balance sheet inserted into the plumbing of global dollar funding, and almost nobody on the buy side is treating it as one.
Here is the part that should stop you. The single largest issuer of that float โ roughly sixty percent of it โ still publishes attestations, not audits. A signed snapshot from an accounting firm confirming that on one date, at one hour, certain accounts held certain balances. No opinion on internal controls. No opinion on whether the liabilities are complete. No examination of the affiliated entities that hold the reserve assets.
The market has decided this is fine. The market has decided this for four years. The float is now large enough that "fine" is no longer a market opinion โ it is a systemic assumption.
To understand why the float matters, you have to stop reading stablecoins as crypto products and start reading them as money market funds with a token wrapper.
A money market fund holds short-dated paper and issues shares at a stable net asset value. A stablecoin issuer holds short-dated paper and issues tokens at a stable peg. The functional difference is regulatory: the fund is examined, the issuer is attested to. The fund reports monthly holdings line by line. The issuer reports aggregate categories, sometimes quarterly, sometimes with a lag that widens precisely when you would want it to narrow.
The second difference is distribution. A money market fund reaches institutional treasuries through custodians and broker-dealers. A stablecoin reaches anyone with a wallet, in any jurisdiction, in about eleven seconds on the right chain. That is a genuinely superior product. It is also a genuinely superior channel for transmitting a reserve shock into offshore credit markets without a single regulator seeing the wire.
Third difference: redemption. Fund shares redeem into a bank account during business hours. Tokens redeem into the issuer's own compliance queue, at the issuer's discretion, subject to the issuer's judgment about whether your address is clean. That is not a red flag in itself โ it is a design choice โ but it means the peg is a promise about process, not a promise about math. When the queue lengthens, the peg does not break at the midpoint. It breaks at the door.
Now place that structure inside the current macro frame. Dollar funding is not tight, but it is not loose either. The front end has repriced lower, the curve has normalized, and the marginal dollar of leverage is once again hunting for carry. That is precisely the environment where a $310 billion float stops being a settlement tool and starts being a funding tool. Watch the flow, ignore the noise.
Let me give you the mechanism, because the mechanism is where the money is.
Every dollar of stablecoin supply is a dollar of reserve assets sitting somewhere โ Treasury bills, repo, bank deposits, and in some cases commercial paper or secured lending. When supply grows by $50 billion in a quarter, that is $50 billion of new demand for short-dated paper arriving from an entity that is not a bank, not a fund, and not a sovereign. On the Treasury curve, that shows up as marginal demand at the bill end. On the repo market, it shows up as a new cash lender who does not care about the policy rate nearly as much as it cares about liquidity and settlement speed.
That is the benign read, and it is mostly right. Stablecoin issuers have become a structural bid for bills. They smooth the front end. They absorb issuance. In a heavy bill-supply quarter, they are the difference between a soft auction and a sloppy one.
Here is the non-benign read, and it is the one your risk committee has not priced.
The float has become a monetary transmission channel that no central bank controls. When an issuer shifts reserve composition โ from deposits to bills, from bills to repo, from one custodian bank to three โ that shift is a portfolio decision made by a private company with no mandate, no published framework, and no obligation to pre-announce. A $10 billion rotation is a rounding error to the issuer and a meaningful liquidity event to the counterparty losing the deposit. Multiply that by four issuers running the same optimization at the same point in the curve, and you have an uncoordinated portfolio shift large enough to move a funding spread. Nobody voted on it. Nobody was asked.
I have seen this movie. In 2022 I spent six months auditing the failure mechanics of algorithmic stablecoins after the Terra collapse, and the lesson that stuck was not about algorithms. It was about redemption ordering. The peg did not die because the math was wrong. The peg died because the redemption queue was longer than the reserve's ability to convert, and because the reserve's composition was disclosed at a granularity that made the gap invisible until it was terminal. Every stablecoin failure is a disclosure failure first and a solvency failure second.
Which brings me to the regulatory layer, because 2026 is the year the wrapper finally gets a rulebook.
The reserve frameworks landing across the US, the EU, and several Asian jurisdictions converge on the same three requirements: eligible reserve assets limited to cash, bills, and repo; periodic disclosure of composition; and a redemption right at par. Good. Necessary. Insufficient.
Here is what the frameworks do not solve. First, they govern issuers domiciled inside the perimeter and say almost nothing about the offshore float that clears the same corridors. Second, they require disclosure of composition but not disclosure of counterparty concentration โ you can be fully compliant holding bills through three custodians that all bank with the same institution, and the disclosure will look pristine. Third, and this is the one that will matter in the next stress event, they do not standardize the attestation itself. Two issuers can both be compliant while one publishes a full examination and the other publishes a photograph of a signature.
Now layer the DeFi side on top, because the float does not sit still.
Roughly a third of stablecoin supply is parked in on-chain venues earning yield. Lending markets, liquidity pools, delta-neutral basis structures, tokenized bill wrappers. Headline APRs on these venues run anywhere from four to double digits, and the gap between the on-chain rate and the risk-free rate is the entire pitch. DeFi yields are traps, not gifts. Not because the protocols are broken โ most of the audited ones are not โ but because the yield is compensation for a risk the depositor cannot see from the interface.
That risk has a name: duration mismatch inside a pegged wrapper. A venue paying nine percent on a dollar stablecoin is doing one of four things. It is lending against volatile collateral at a liquidation threshold that has never been tested in a genuine gap. It is running a basis trade that is short funding. It is recycling emissions from a token whose float is thin. Or it is holding a claim on off-chain credit that settles in weeks while the deposit redeems in minutes. In three of those four cases, the depositor is long a bank run and short the queue.
I have run this trade. In 2020 I structured a leveraged delta-neutral position across two lending venues to capture a spread that looked like free money, and it worked โ twenty-two percent annualized, which in that regime was not extraordinary. What made it work was not the spread. It was rebalancing discipline and the willingness to eat gas costs that consumed a third of the gross. Arbitrage closes; liquidity remains. The spread I captured in 2020 does not exist in 2026, and anyone advertising it now is advertising a different risk wearing the same clothes.
The Layer 2 dimension makes this worse, not better. Rollup operators are running proving infrastructure at a cost structure that only works in a fee environment we have not seen since the last cycle. ZK proving costs are absurdly high relative to the revenue a rollup actually captures, and the gap is being closed with token subsidies and sequencer margins that are themselves subsidized. When the subsidy ends, venue economics tighten, and the first thing that tightens with them is the yield paid on parked stablecoins.
So you have a float that is a monetary channel, sitting on reserves disclosed at coarse granularity, distributed into venues whose yields depend on subsidy structures, all of it denominated in a unit that promises par. That is not a crypto problem. That is a dollar funding problem with a crypto distribution layer.
The consensus thesis right now is decoupling: crypto is becoming its own macro asset class, driven by institutional allocation, ETF flows, and a regulatory tailwind. I think that thesis is directionally right and tactically dangerous.
The reason is simple. The more institutional the float becomes, the more correlated it becomes to the funding market it sits inside. Every treasury desk that adopts a stablecoin for settlement is a desk that will pull that balance the moment its own liquidity coverage ratio gets uncomfortable. That behavior is rational at the desk level and destabilizing at the system level, and it is exactly what happened in March 2023, when a single bank failure forced a reserve composition change that took the peg to ninety cents on one venue for a weekend.
The decoupling thesis assumes stablecoin demand is a crypto-native phenomenon. It is not. It is a dollar convenience phenomenon. Convenience demand is the most flight-prone demand there is.
The second blind spot: everyone is auditing the smart contracts and nobody is auditing the custody chain. I have read a dozen protocol audits this year. They are competent, thorough, and completely silent on the question of who holds the reserve and under what legal agreement. The code is not the risk surface. The signature authority on the reserve account is.
So what do you actually do with this.
Position for the float, not the price. If you are running institutional capital, treat issuer reserve disclosure quality as a hard eligibility criterion, the same way you treat a counterparty's capital ratio. If you are running a treasury, stress the redemption queue rather than the peg โ ask what happens if your redemption settles in eleven days instead of eleven seconds, and whether your liabilities can wait. If you are running a venue, stop paying yield you cannot explain in one sentence to a risk officer.
The float will keep growing. It is too useful to shrink. The question is not whether $310 billion becomes $600 billion. The question is whether the disclosure regime grows as fast as the balance sheet does.
My read: it will not. Not this cycle. The rulebooks are being written by people who understand money market funds, and the float is not a money market fund. It is something older and less examined โ a private bank, wearing a token.
Watch the attestation dates. When the reporting lag widens, the queue is already forming.