Login
Free Float, Rented Float: Measuring Revenue Defensibility in Stablecoins
A framework for figuring out whose stablecoin revenue actually survives 2026.
In September 2025, some of the best-capitalized companies in crypto stood in a public auction and competed to give their revenue away. Paxos bid 95% of the reserve yield. Ethena bid 95% plus $75 million in ecosystem incentives. Frax and Agora bid 100%. The prize was the right to issue USDH, Hyperliquid’s native stablecoin, and the bidding logic ran in reverse: whoever promised to keep the least deserved to win the most.
That auction was not an anomaly. It was the stablecoin business model meeting honest price discovery for the first time, and it poses the question this piece tries to answer: for each dollar of a stablecoin’s supply, what does the issuer have to pay to stop it from leaving? The answer determines which issuers keep their revenue over the next three years, and it has almost nothing to do with the market-cap league table.
A commodity business that prints $13 billion
A fiat-backed stablecoin is a float business. The issuer holds customer dollars, invests them in short-dated Treasuries at 4 to 5%, and pays holders nothing. Revenue equals float times the spread kept. Tether ran this machine to over $13 billion of profit in 2024.
Which should be impossible. The product is a perfect commodity; every stablecoin is a token worth one dollar, and any rival can offer holders a share of the yield and pull float away. Dozens of yield-bearing alternatives exist. Tether pays nothing and keeps growing anyway.
Profits persisting in a commodity market mean something blocks the competition, and the industry’s shorthand for that something, “the moat,” conceals a category error. The moat is not a property of the coin. It is a property of each individual dollar of float, set by who holds that dollar and why.
Every dollar of float has a deposit beta
Banking solved this classification problem a century ago. Balances spread across millions of small checking accounts are core deposits: sticky, cheap, indifferent to rates. Balances from a few large yield-seeking institutions are hot money, and banks funded by hot money periodically die overnight; Silicon Valley Bank’s depositor base was small, sophisticated, and coordinated through the same group chats, and it moved as one organism in March 2023.
Bank analysts quantify the difference as deposit beta: the fraction of market interest rates an institution must pass through to retain a deposit. Empirical studies put the average US bank’s pass-through at roughly 30 to 46%, but the average hides the point; checking accounts sit near zero, brokered institutional money near one, and a bank’s franchise value is largely the value of its low-beta book.
A stablecoin’s circulating supply is functionally a deposit base, so the transplant is direct. Every dollar of float is either free float, which stays without being paid, or rented float, which stays only for as long as the yield is handed over. An issuer’s defensible revenue is free float times the Treasury rate. Everything else is assets under management wearing a stablecoin costume.
The float segments, from most expensive to keep to cheapest:
B2B reserve float. Another issuer’s treasury, like USDtb sitting in BUIDL. Beta of roughly 1.0. Nobody owns the relationship; the buyer shops.
Platform-captive float. Traders on one venue, like USDC on Hyperliquid. Beta of roughly 1.0 once the venue negotiates. The platform owns the relationship.
DeFi incentive float. Yield farmers. Beta of roughly 1.0, owned by whoever pays most this month.
Exchange-distributed retail. CEX users, like USDC on Coinbase. The beta is contractual: it is the rev-share. The exchange owns the relationship.
Fragmented transactional float. Small wallets and emerging-market commerce, which in practice means USDT. Beta of roughly zero, and the issuer owns the relationship structurally.
The bottom segment deserves a sentence of respect, because it is the only one that cannot be bought. A $200 balance forgoes about fifty cents a month by not chasing yield; nobody rewires their financial life for that. And in the corridors where USDT circulates as the unit of account, from Lagos to Buenos Aires to Istanbul, leaving requires your whole economic neighborhood to leave with you. Coordinating millions of strangers is the hardest problem in economics, which is exactly why this float is defensible.
Exhibit 1: the rent, measured in GAAP
Rented float is not a metaphor. Circle is a public company, and the ransom it pays to keep its float sits on the income statement as “distribution, transaction and other costs,” most of it the Coinbase revenue share (Coinbase collects 100% of reserve income on USDC held on its platform and 50% elsewhere; $908 million of Circle’s $1,011 million in 2024 distribution costs went to Coinbase).
The trajectory is the finding. In 2022, Circle paid out 37 cents of every revenue dollar for distribution. By 2024 it was 60 cents. Across 2025’s reported quarters it held near 61 cents, on much larger revenue. USDC’s supply roughly doubled over that period, which is the point: the supply grew and the share of it Circle gets paid to hold shrank. Growth composed of rented float raises revenue and dilutes revenue quality at the same time, and no supply chart will ever show you that.
Exhibit 2: the deposit base, on-chain
If float quality is real, it should be visible in holder structure. I pulled the holder data for three dollar tokens on Ethereum mainnet from Blockscout on July 5, 2026.
Three tokens, one product, three different animals:
USDtb ($727M supply) is the pure B2B case: 92% of supply sits in ten addresses, and the single largest, an Ethena custody wallet, holds 43% by itself. Behind the labels, this is essentially one treasury desk’s allocation decision. Its stated beta is one by construction; USDtb’s model passes reserve yield through, and BlackRock’s BUIDL, which backs it, keeps that relationship only by remaining the highest bidder among interchangeable tokenized T-bill funds.
USDC ($50.1B on Ethereum) shows an institutional-DeFi profile: 8.1 million holding addresses, with 27% of supply in the top ten, led by Sky’s peg-stability module and a cluster of institutional smart accounts. Fragmented enough to look retail, but the retail is largely intermediated, and the intermediaries, as Exhibit 1 shows, send Circle an invoice.
USDT ($97.1B on Ethereum, more than 16.7 million holding addresses) is the interesting one, because its top-10 concentration (50%) is higher than USDC’s, and the composition explains why that’s a strength rather than a weakness: the big addresses are almost entirely exchange custody wallets (Binance and OKX dominate the list) holding on behalf of millions of end users, plus the issuer’s own treasury and a bridge lockbox. And Ethereum is USDT’s institutional venue; the retail long tail lives on Tron, where USDT’s supply exceeds $86 billion on a network that passed 389 million total accounts in June, built almost entirely on cheap USDT transfers in emerging markets. Tether’s deposit base is the shape a bank would pay a premium for.
One honest complication: exchange custody blurs wallet counts in both directions. A Binance hot wallet is one address and millions of users. The framework handles this cleanly, though, because custody is precisely the case where the platform owns the relationship, and platform-owned float is where the next section’s repricing happens.
2026 keeps running the experiment
Hyperliquid separated the two moats. Native Markets won the USDH ticker in September 2025. Eight months later USDH had stalled near $100 million while USDC on Hyperliquid doubled to roughly $5 billion; the challenger has since faded toward $20 million on its way to sunset.
Liquidity gravity won; every order book and habit was denominated in USDC, and no governance vote could repeal that. But look at the terms of victory. In May 2026, Coinbase became USDC’s official treasury deployer on Hyperliquid under AQAv2, committing to share the vast majority of the reserve yield with the protocol, and bought the USDH brand to retire it. The incumbent kept 100% of the float and surrendered nearly 100% of the income on it.
The lesson generalizes: the moat that defends supply and the moat that defends margin are different moats. Network effects answer whether the dollars stay. Deposit beta answers who keeps the yield on them. A platform doesn’t need to replace your stablecoin; it needs a credible threat of replacing it, and the yield reprices on its own.
Open USD is a beta-of-one stablecoin by design. The 140-plus-partner consortium announced June 30 (Visa, Mastercard, Stripe, BlackRock, Coinbase, Google among them) routes reserve earnings to distribution partners from day one. Circle’s stock fell 16% on the announcement. Wherever distribution owns the customer, issuer margin is being declared zero in advance.
Regulation is running the same experiment from the other side. The GENIUS Act bans issuers from paying yield to holders, legislating retail float’s beta to zero inside the compliant perimeter; the yield now leaks through distribution deals instead of holder payments. Same leak, different plumbing. Meanwhile MiCA enforcement pushed USDT off regulated EU venues and Tether still lacks a GENIUS reciprocity determination, so the market is splitting into a regulated zone where a dozen interchangeable compliant issuers fight over rented float, and an offshore zone containing nearly all the free float in existence.
Re-rank the table
Total stablecoin supply sits near $312 billion; USDT holds roughly $184 billion of it. Adjust each issuer’s supply for float quality and the standings deform. USDT’s number is disproportionately free float, earning the full spread. USDC’s skews toward exchanges, platforms, and institutions: float Circle demonstrably rents, some of it repriced in public this year. Reserve-backing stablecoins and consortium coins carry betas near one by construction; their supply is real and their defensible revenue rounds to a management fee.
The uncomfortable conclusion is that the industry is celebrating the wrong number. Every consortium launch, yield-share deal, and platform extraction grows the supply charts and shrinks the share of that supply anyone is paid to hold. Distribution always captures the margin in commodity markets; crypto spent five years believing it had built an exception.
The only exception that exists is the float that can’t be bought: fragmented, transactional, unit-of-account money, accumulated over a decade of ground-level adoption, priced in USDT because everything around it is. That is most of the defensible revenue in a $300 billion industry, and one issuer owns it, from outside the regulated perimeter.
Watch one variable from here: whether wallets begin abstracting yield away from retail holders, auto-converting idle balances into yield-bearing equivalents in the background. That would raise the deposit beta of the last low-beta segment without any holder ever “deciding” to switch, and it is the single biggest threat to the franchise this piece describes.
Free Float, Rented Float: Measuring Revenue Defensibility in Stablecoins was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.
