The Float Business: A Digital Asset Market View
1. Executive Summary
Between July 2025 and July 2026, XRP fell roughly 70 percent, from a peak of $3.66 to roughly $1.08. Over the same twelve months, institutional investors marked Ripple, the company most closely associated with that token, from a $40 billion valuation to $50 billion. The two moved in opposite directions, and the pattern was not confined to Ripple. Across the industry, the past eighteen months delivered nearly everything advocates had spent a decade asking for: the first federal stablecoin law, the end of the SEC’s long-running case against Ripple, spot ETFs beyond bitcoin and ether, and the beginning of bank charters for crypto firms from the Office of the Comptroller of the Currency (OCC). Through all of it, most token prices fell.
This report treats that divergence as deeply informative and attempts to explain it. The reading offered here is that regulation resolved the existential question of whether digital assets would be permitted inside the U.S. financial system, and in resolving it, revealed where the durable economics actually sit. They sit with companies, in the interest earned on reserves, in custody and prime brokerage fees, and in the compliance infrastructure the law now requires, while token holders have so far captured little of it.
Mainsheet holds this view with moderate confidence. The sections below lay out the evidence, flag where it is thin, and describe the ways the view could be wrong.
2. A Paradox Worth Explaining
The Ripple case rewards a closer look, because it is the cleanest natural experiment the industry has produced.
For the token, the news could hardly have been better. The SEC concluded its lawsuit against Ripple in August 2025, ending five years of litigation. The SEC approved the first wave of spot XRP ETFs in November 2025, and those funds have accumulated roughly $1.4–1.5 billion in inflows since launch. Ripple secured conditional approval for a national trust bank charter. And yet the price fell by roughly 70 percent from its peak, as Figure 1 shows.
Over the same period, the company moved in the opposite direction. Ripple raised $500 million in November 2025 at a $40 billion valuation in a round backed by affiliates of Citadel Securities and Fortress Investment Group, then completed a $750 million share buyback in early 2026 at a $50 billion valuation. It deployed roughly $3 billion in disclosed acquisitions. Sophisticated institutional investors were, in effect, paying more for the company while the market paid less for its token.
One explanation is that the market is simply wrong about one of the two. A better explanation is that the two prices measure different things. A token’s price reflects speculative demand for the asset itself. A company’s valuation reflects claims on cash flows. In 2025 the cash flows became large enough, and visible enough, to price separately. The rest of this report is about where those cash flows come from.
3. What a Stablecoin Actually Is
It is worth being precise here, because the mechanics explain the economics.
A dollar stablecoin works like this: a customer wires dollars to an issuer, the issuer mints an equal number of tokens and hands them over, and the customer can now move those tokens anywhere in the world in seconds, at any hour, for fractions of a cent. The issuer, meanwhile, takes the customer’s dollars and invests them in short-term U.S. Treasury bills. The Treasuries pay interest. The token holder receives none of it.
In substance, this is a money market fund in which the manager keeps the entire yield. That arrangement might sound like a temporary quirk that competition would erode. It is now federal law. Why would Congress entrench it? Part of the answer is competitive: stablecoins that paid interest would compete directly with bank deposits, a point the banking industry pressed throughout the drafting. Whatever the full intent, the GENIUS Act, enacted in July 2025, prohibits permitted issuers from paying interest or yield to holders of payment stablecoins. Put differently, Congress wrote the issuers’ business model into statute. For anyone underwriting the sector, this is the most consequential sentence in the legislation, and it has received remarkably little attention relative to its importance.
The scale of the resulting income stream is straightforward to approximate. Total stablecoin supply is somewhere around $280 billion as of mid-2026, though estimates published within weeks of one another by different market-data aggregators have ranged from roughly $230 billion to just over $300 billion depending on what gets counted. A bottom-up check narrows the range: Tether’s roughly $186 billion of USDT and Circle’s roughly $77 billion of USDC alone imply a floor near $263 billion before any smaller issuer is counted. At short-term rates in the 3.5–4 percent range, that float generates on the order of $10–11 billion in annual interest, paid not to the people holding the tokens but to the companies that issued them.
4. The Evidence
Two firms allow this model to be tested against reported numbers, one private and one public.
Tether, the largest issuer, reported net profit exceeding $10 billion for 2025 against roughly $186 billion of USDT in circulation, with total U.S. Treasury exposure of about $141 billion, a position that places it among the largest holders of U.S. government debt in the world. That profit figure was actually down from approximately $13 billion in 2024, which is itself instructive: the decline tracked falling short-term yields rather than any drop in demand. Tether’s earnings behave like what they are, which is interest income on an enormous, cheaply gathered deposit base.
Circle, which went public in June 2025, provides audited detail that Tether does not. In the first quarter of 2026, Circle reported $694 million in total revenue, of which $653 million, or 94 percent, was interest on the reserves backing USDC. Full-year 2025 revenue was $2.7 billion, up 64 percent, on USDC circulation that ended the year at $75.3 billion. Circle’s filings also expose a part of the value chain that issuer profit figures obscure: of that $694 million in quarterly revenue, $407 million went out the door as distribution, transaction, and other costs, the largest share of it to distribution partners. Whoever controls the customer relationship extracts much of the float. The issuance business is real, but it is not evenly shared.
Table 1 compares the two issuers; the assurance row matters as much as the financial rows.
Table 1. The two largest dollar stablecoin issuers, year-end 2025 and Q1 2026.
Two conclusions follow from these numbers. The core revenue engine of the stablecoin industry is verifiable, large, and growing with supply. It is also a leveraged bet on short-term interest rates, a point taken up in the risk section.
5. What Regulation Resolved, and What It Revealed
The regulatory architecture assembled since mid-2025 has three main pieces. The GENIUS Act established the federal framework for payment stablecoins. Its one-year statutory deadline for six agencies to finalize implementing rules passed on July 18, 2026, without a coordinated set of final rules; several comment periods remain open into August, the Federal Reserve has yet to propose its own framework, and partial finals with interim guidance now look like the base case. The OCC conditionally granted national trust charters to Circle, Paxos, and three other nonbank firms in December 2025, with Ripple holding its own conditional charter approval. The CLARITY Act, which would settle market structure questions beyond stablecoins, passed the Senate Banking Committee in May 2026 but has not yet received a full Senate vote, and its timing remains uncertain.
The practical effect of all this is that compliance became a moat. Reserve requirements, monthly attestations, capital floors, and Bank Secrecy Act obligations are expensive at small scale and manageable at large scale. The framework favors incumbents and well-capitalized entrants, and Mainsheet expects it to drive consolidation among smaller stablecoin-adjacent firms.
It also invited in a new class of competitor. JPMorgan, Citigroup, Bank of America, and Wells Fargo have discussed a shared tokenized deposit network targeted for 2027. Tokenized deposits differ from stablecoins in kind: bank money on new rails, with deposit insurance and a lender of last resort behind it. For institutional settlement, that is a serious substitute, and Mainsheet regards bank entry as the most credible long-term threat to public stablecoin issuers, precisely because the banks are competing for the same float.
6. Tokenization Is Real, and Smaller Than the Headlines
The second verifiable use case is the tokenization of traditional assets, led by U.S. Treasuries. BlackRock’s BUIDL fund, launched in March 2024, has grown to roughly $2.5 billion in assets and operates across multiple public blockchains. Ondo Finance, Franklin Templeton, and Circle’s own tokenized money market product each manage comparable amounts. In May 2026, JPMorgan, Mastercard, and Ondo settled a tokenized Treasury transaction in real time on the XRP Ledger, a meaningful signal that large institutions are willing to run production transactions on public infrastructure, beyond the pilot stage.
To be sure, the direction is credible. The caveat is one of magnitude, and the measurement problem is severe enough to deserve a table of its own. Table 2 shows how estimates of the same market, drawn from the same most-cited tracker in the same weeks, vary with counting choices: freely tradable tokenized assets excluding stablecoins stood at roughly $31–33.5 billion in early July 2026, the tracker’s separate pipeline measure of committed but not yet tradable assets ran near ten times that, and the Treasury subsegment alone spans a two-to-one range depending on whether money-market-style products are included. Against several hundred trillion dollars of traditional securities outstanding, even the largest of these figures is small. A full treatment of tokenized private credit, funds, and equities is beyond the scope of this report. Widely circulated projections that this market reaches $16 trillion or more by 2030 are extrapolations, not forecasts Mainsheet would underwrite, and how quickly the Treasury-led segment compounds is not something the current evidence can establish.
Table 2. Estimates of the tokenized asset market drawn from rwa.xyz data, June–July 2026.
7. What Can Be Verified, and What Cannot
Because this industry produces an unusual ratio of narrative to audited fact, a research report is obliged to separate the two explicitly.
The following can be verified with reasonable confidence: reserve attestations from major issuers, which are prepared monthly or quarterly by large accounting firms; the financial statements of public companies such as Circle and Coinbase; enacted legislation and charter approvals, which are matters of public record; and the existence of on-chain transactions, though not always their economic meaning, since raw volume figures are inflated by automated and circular activity.
The following cannot be independently verified: the revenue and transaction claims of private companies, including Ripple’s statements that it has processed more than $100 billion in transactions and holds 75 regulatory licenses, which may well be accurate but rest on the company’s own disclosure; user counts, such as Tether’s claim of 530 million users; the substance behind most partnership announcements; and total value locked as a measure of real economic activity.
One distinction deserves emphasis because it is routinely blurred. An attestation is an accountant’s confirmation that stated assets existed at a moment in time. It is not an audit, which examines controls, liabilities, and the full financial picture over a period. Tether has never published an audit. Even the most-cited data source produces figures an order of magnitude apart in the same week, depending on whether pipeline commitments are counted (Table 2), and one widely circulated aggregator analysis put BlackRock’s BUIDL at ten times the roughly $2.5 billion corroborated by primary-adjacent sources. Mainsheet treats self-reported and aggregator metrics as directional inputs only, and readers are encouraged to apply the same discount.
8. The Ripple Question
Ripple is the natural test of this report’s thesis. Mainsheet holds exposure to the company, disclosed below.
The bull case for the company is that it has converted regulatory clarity into acquired infrastructure faster than any peer. The roughly $3 billion in disclosed acquisitions since early 2025 bought a prime brokerage clearing on the order of $3 trillion annually (Hidden Road, now Ripple Prime, acquired for $1.25 billion), a treasury management platform serving over a thousand enterprise clients (GTreasury, $1 billion), and stablecoin payment infrastructure (Rail, $200 million), alongside smaller custody and licensing deals. Its RLUSD stablecoin grew from roughly $132 million to between $1.5 and $1.75 billion in circulation in about a year, with reserves held at BNY Mellon and attested monthly. That is fast growth from a small base; RLUSD remains under one percent of total stablecoin supply, and caution is warranted against reading its trajectory as settled.
The complication is that Ripple’s balance sheet is itself heavily exposed to the token this report’s thesis discounts. The company holds a very large XRP position, has historically funded operations partly through periodic XRP sales, and weathered the recent drawdown in part because of that treasury. An investor in Ripple equity therefore owns two things at once: an operating business in payments, custody, and prime brokerage, and a large, volatile token treasury. Mainsheet’s view is that the operating business is becoming the larger share of the value, and the pattern of institutional buyers paying rising prices for the equity while the token fell is consistent with that view. But Ripple does not publish audited financials, so the mix cannot be demonstrated from public data. That is a genuine limitation of the thesis, and it belongs in the body of the argument rather than a footnote.
9. Ways This View Could Be Wrong
Four things, principally.
Interest rates are the first. If 94 percent of Circle’s revenue is reserve interest, then issuer economics are a function of Federal Reserve policy. Circle’s reserve return rate already fell 66 basis points year over year to 3.5 percent. A sustained cutting cycle of 200 basis points would compress the industry’s core revenue by roughly half at constant supply. Supply growth can offset this, but the offset is not guaranteed.
Bank tokenized deposits are the second. If institutional settlement migrates to insured bank instruments, public stablecoin float, and the income on it, could plateau well below current projections.
Private valuations are the third. Ripple’s $50 billion mark rests on secondary transactions and a buyback rather than audited earnings, and marks established in thin private markets can be slow to reflect deterioration.
And the fourth is the possibility that Mainsheet has simply mistimed the token market. Tokens have repriced violently upward before from conditions that looked like this one. If that happens, the equity-over-token framing will look, in hindsight, like an argument for having bought the tokens. Mainsheet’s judgment is that the asymmetry still favors the equity layer, because the companies earn income in both scenarios and the tokens in only one, and that judgment is probabilistic rather than certain.
10. Where Mainsheet Lands
Mainsheet’s position mirrors the one it took in The Embodied AI Inflection (May 2026): when a technology’s trajectory is uncertain, the most defensible exposure is to the infrastructure that gets paid regardless of which specific asset wins. In digital assets, that means the equity layer: issuers earning float, prime brokers and custodians earning fees, and the compliance and tokenization plumbing that regulation now requires. Direct token exposure requires a further belief, that speculative demand for a particular asset will return, which the past year’s evidence does not support and which Mainsheet is not currently willing to underwrite.
Investors weighing exposure to the sector may wish to apply the same split this report does: underwrite the fee and float streams that survive both token outcomes, and treat token upside as a separate decision requiring a separate belief. The near-term signposts worth watching: how the agencies resolve the GENIUS rulemaking they did not complete by the July 18 statutory deadline; whether the CLARITY Act reaches a Senate floor vote this year; whether institutional settlement consolidates on public stablecoins or bank tokenized deposits; and whether any major private issuer, Ripple included, moves to audited public disclosure. Each of those developments would sharpen or weaken this view more than any movement in token prices, and Mainsheet expects to revisit the thesis as they resolve.
Disclosures.
This report is prepared by Mainsheet Ventures for informational purposes only and does not constitute investment advice, an offer to sell, or a solicitation of an offer to buy any security. Mainsheet Ventures structures SPV-based co-investment vehicles on a deal-by-deal basis; this report is not an offer of any such vehicle. The information contained herein is based on sources believed to be reliable as of July 2026 but has not been independently verified. Past performance is not indicative of future results. All investments involve risk, including the possible loss of principal. Mainsheet Ventures has structured an SPV with exposure to Ripple, disclosed at mainsheet.ventures; that vehicle is closed to new investors, and this report is not an offer of any vehicle. This report may not be reproduced, distributed, or transmitted without the prior written consent of Mainsheet Ventures.
Mainsheet Ventures | July 2026