I keep a small notebook of numbers that sound too clean. Last month a treasury friend texted me a screenshot of a stablecoin dashboard and asked, half joking, whether his company had just joined a trillion-dollar payments club. The chart was impressive. The invoice sitting next to his laptop was not. A token had moved. The supplier had not been paid yet. That gap is the whole story, and it is wider than the slogans suggest.
Stablecoins can cross a public ledger in minutes. A migrant sending money home, or a factory settling a parts invoice, needs something duller: a person or a firm who can actually use the money. Those are not the same event. If you only count wallet-to-wallet motion, the market looks enormous. If you ask who got paid, for what, and in which currency they could spend it, the picture shrinks. It does not vanish. It just stops pretending to be the whole river.
Why A Token Transfer Is Not A Payment
A dollar token sent from one address to another leaves a clear record. It does not leave an invoice, a residence, a beneficiary name, or the rate paid to turn that token into spendable local cash. I have found that people skip this point because the chain feels complete. The hash arrives. The balance updates. Done, right? Not if the economic unit is an obligation between real parties.
Recent onchain classification work puts observable stablecoin payments for the first eight months of 2026 somewhere between $401 billion and $527 billion. Businesses appear to have received roughly 58% to 64% of that classified value. The range is analytical uncertainty, not two official ledgers fighting each other. Substantial, yes. Proof that every transfer settled a good or a service? No.
Perhaps the most interesting aspect is how ordinary the confusion is. We do the same thing with bank wires if we are sloppy. A sweep between two accounts at the same firm is not a sale. A deposit into a trading venue is not payroll. Stablecoins just make the mistake easier, because the ledger is public and the labels are not.
What Sits Between A Transfer And A Finished Payment
Picture a customer who hands dollars to an intermediary. The intermediary buys a dollar token, sends it to a receiving partner, and that partner sells it for pesos. One obligation. Several blockchain hops. The customer may never hold a token. The merchant may only see a bank credit. Flip it, and a business can settle an invoice wallet to wallet in a single visible transfer while compliance and accounting happen somewhere else entirely.
The same ledger also records exchange deposits, liquidity shifts, lending collateral, swaps, bridges, issuer mints and redemptions, and treasury sweeps between wallets owned by one firm. Some of that is economically real. Most of it is not a completed payment to an independent recipient. A wallet address is not a country. Inferring an international transaction takes reliable information about counterparties and where they sit, and that information usually lives with payment companies, not on a block explorer.
Counting every token movement as commerce is like counting every time a banknote passes a till, a vault, and a courier, then calling the sum retail sales.
Central bank researchers looking at 2025 activity on one major smart-contract chain examined hundreds of millions of token transfer events inside a smaller set of transactions for three dollar stablecoins. Nearly 60% of those transfer events sat inside complex transactions. Treat each event as a standalone payment and you misclassify a large slice of the tape. That finding belongs to the assets and period they sampled. It is not a universal percentage for every coin and every chain. Still, it is a useful warning against lazy addition.
Cross Border Is A Second Question, Not A Free Upgrade
Even a clean payment label leaves another problem. Was it cross border? An American company can pay an American contractor on a globally accessible chain. Two domestic exchange addresses can sit in the middle of an international remittance that actually settles off the chain. Public records alone cannot close either case.
I keep coming back to this because market decks love the phrase borderless. Borderless infrastructure is not the same thing as an international payment. The border is a fact about people, licenses, and accounts. The chain does not stamp a passport on the hash.
The Hundreds Of Billions Have Edges
That $401 billion to $527 billion figure covers eight months, not a full year, and it mixes use cases. An earlier joint estimate put stablecoins at about 0.31% of a $44.3 trillion retail cross border payments market in 2025, up from roughly 0.2% the year before. Do not divide those numbers as if they share a denominator. One is a payment estimate across purposes and a different window. The other is a share of a defined retail international market.
A separate industry study offers a sharper warning about definitions. It compares more than $62 trillion of annual stablecoin transfers with about $4.2 trillion of activity described as real-economy payments, near 7%. Its conservative count of observable bilateral payments for goods and services in 2025 is only about $350 billion to $550 billion. The $4.2 trillion bucket is broader than that bilateral subset. The conservative method also leaves out some offchain use, including internal exchange settlement and stablecoin card payments. Quote the 7% and the bilateral range as if they were the same total and you hide the difference on purpose.
In my experience, this is where commentary goes soft. A writer picks the largest number in the room, calls it adoption, and moves on. The honest version is messier, and more useful.
| Measure | What it roughly captures | What it does not prove |
| $401B to $527B, eight months of 2026 | Classified payment-like flows | That every public transfer was commerce |
| 0.31% of $44.3T in 2025 | Share of a defined retail cross border market | Wholesale treasury or domestic payroll share |
| About $4.2T real-economy filter | Broader filtered activity | A tight bilateral goods-and-services count |
| $350B to $550B bilateral, 2025 | Conservative observable payments for goods and services | Offchain card use and some internal settlement |
| Above $62T annual transfers | Gross address-to-address motion | Completed beneficiary payments |
Labels Change The Story Without Changing The Past
One widely watched onchain dashboard sorts transfers into payments, decentralized finance, exchange flows, investment and trading, store of value, issuance, routing, and infrastructure. Adjusted volume strips labeled exchanges, contracts, bots, bridges, and similar activity, then uses heuristics for unlabeled wallets. A consumer payment, a payroll transfer, and a person-to-person send can all land in the payments bucket. A dollar sent to an exchange normally does not. The model is an inference from labels and behavior, not a global receipt file.
That inference moved on 18 September. A refreshed identity set expanded from about 15 million labeled addresses to roughly 600 million. Adjusted volume fell because more exchange, contract, and automated addresses could be excluded. One automated program on a high-speed chain had been pushing tokens through thousands of throwaway wallets and inflating the tape. The definition stayed nominally the same. The data used to apply it got better. Historical comparisons should stick to one dashboard vintage, or they will invent a crash that never happened.
A decline after a methodology change is not customers walking away. Rapid reported growth can also be better identification of payment processors rather than brand-new commerce. Both mistakes are common. Neither is innocent if you are pricing a business on the chart.
Where Live Payment Evidence Actually Shows Up
Classification models are not the only evidence. In October, a global card network said more than 160 stablecoin-linked card programs were running, with program payment volume up nearly 200% year over year. About 17% of that card volume in the fiscal year to date came from business and commercial programs. These are network figures for those programs. They are not the share of all global commerce, and they are not proof that each merchant accepted a token. Cardholders can spend on familiar rails while issuers and settlement partners use stablecoins behind the curtain.
The same network has reported an annualized stablecoin settlement run rate above $20 billion. Annualized is a pace extrapolated from recent activity, not a calendar-year total. Settlement between a network and its partners is a different economic step from each retail purchase. Add card purchases and the matching stablecoin settlement and you can count one commercial cycle twice. Easy to do. Hard to unwind once the slide is in the board pack.
- Card volume measures spend through programs, not direct merchant token acceptance.
- Settlement run rate measures obligations among network partners, often later in the chain.
- A single customer purchase can generate both a card transaction and a later settlement step.
- Business and commercial programs are a minority of that card volume, not the whole story.
One discrete test is easier to bound. A British banking group and the card network ran a seven-day pilot that settled $750,000 of cross border obligations in a major dollar stablecoin, with the reported transfer arriving in under an hour, including over a weekend. That shows a working route between participants. It does not show that banks can repeat the economics at mass scale, across every corridor, with identical compliance load. Fiat entry, token movement, and local payout each carry cost and operational dependency.
An earlier card-issuer trial moved a $20,000 intercompany payment between U.S. and Mexican entities in another dollar token, with a reported seven-minute settlement. Seven minutes is a measure of that trial. A corporate treasury team still cares about prefunding, conversion spread, beneficiary account availability, weekend liquidity, and transaction monitoring. Speed at the blockchain layer is one component of delivered payment quality. It is not the whole grade.
Why Business Flows Dominate The Measured Dollars
Onchain estimates suggest businesses receive most classified payment value, with business-to-business settlement around $137 billion to $153 billion in the period covered. A single supplier invoice can outweigh thousands of consumer purchases. Business dominance by value does not mean most people are paying with stablecoins at the shop counter. Enterprises also have repetitive international obligations where a faster funding cycle, or plain access to dollar liquidity, can be worth the operational hassle.
Stablecoins can move at any hour. Traditional bank transfers may wait on business hours, intermediary banks, and domestic payment windows. A stablecoin still needs an issuer, a reserve, and a redemption path. A business often needs a regulated exchange or payment provider to convert it. The recipient needs local liquidity and, sometimes, a bank account. A fast transfer into a wallet with no economical off-ramp is a poor payment. I would rather have a slower credit I can spend on Monday than a token I cannot exit without a ugly spread.
Central bank commentary has been blunt on this point. Entry and exit costs can make the total cross border cost as high as, or higher than, a bank transfer in some cases. A quoted foreign-exchange rate is not the delivered route. Anyone who has watched a remittance desk knows the trick: a tiny visible fee, then the spread does the earning.
The Corridor Decides Whether Speed Is Worth Paying For
Where dollar banking is scarce and local transfers are slow, access to a liquid dollar token can shorten a treasury cycle. Where domestic instant payments and competitive foreign exchange already work, the gains may be smaller. A service charging a blockchain fee of cents can still earn a wide exchange spread or require collateral that ties up capital. A credible comparison prices the whole route, from the sender’s funding instrument to the beneficiary’s usable balance.
Onchain inference and card-network observation are complementary, not additive. One reads blockchain data. The other sees activity inside card programs. Sum them without mapping overlap and you will double-count the fashionable part of the market. Perhaps that is the tell. If a figure only works when you refuse to define it, it is a marketing number.
The Final Mile Decides Whether Speed Has Value
A remittance sender cares about the amount the recipient can actually use. The chain can confirm a token transfer while an off-ramp delays identity checks, waits for banking hours, or finds thin local-currency liquidity. An intermediary can quote a friendly token fee and recover the cost in the exchange rate. A business recipient may prefer a dollar balance and keep the stablecoin. Another may need local currency for payroll or tax. Those are different services even when the onchain transactions look alike.
This is why market share based on visible wallets is fragile. Some payment providers net thousands of customer obligations internally and put one settlement movement onchain. Others route the same obligation through more than one address. Card payments can use stablecoin funding without a merchant wallet. Conservative bilateral measures omit some offchain processes. Broad gross transfer measures count infrastructure several times. Actual commerce probably sits between measures with different coverage. No honest estimate erases that uncertainty by adding decimal places.
A payment system is judged by whether the right party receives usable money under reliable rules, not by whether a hash appears in a block explorer.
A practical test worth keeping
Concentration matters too. If a handful of high-value business corridors account for most dollars, headline growth can coexist with limited adoption among ordinary consumers. A large treasury transfer is not a million separate remittances. Conversely, a corridor with many small payments may have modest dollar volume and a significant effect on the people using it. Count of completed beneficiary payments, median amount, and total value together would give a fuller picture than value alone. We rarely get all three.
Control Points Sit Outside The Public Chain
The practical control points are often off the public blockchain. An issuer decides who can mint and redeem. A bank provides reserves and fiat settlement. Exchanges and payment firms manage liquidity, screening, and account access. A recipient may have recourse against a service provider for a mistaken payout. An irreversible token transfer cannot resolve the dispute by itself. Institutions considering the rail have to test fraud handling and operational recovery alongside processing speed. Skip that test and the pilot will look brilliant until the first exception.
For future comparisons, researchers should disclose which chains, tokens, and offchain routes are included, how wallets are labeled, whether transfers involving exchanges and smart contracts are excluded, and whether a classification change restates prior periods. The September label refresh shows why. More labels lowered adjusted volume even though the underlying historical transactions did not disappear. Treat that dip as customers leaving and you will misread the market. Treat every jump as new commerce and you will misread it the other way.
Four Numbers That Answer Four Different Questions
The global supply of stablecoins measures the outstanding token balance at a point in time. One analytics estimate put that stock at about $303 billion in August 2026. Supply can support many payments during a year because a token can change hands repeatedly. Comparing annual payment value with supply measures turnover, not the fraction of coins assigned permanently to commerce. A token held on an exchange as trading collateral can circulate among traders without paying a supplier once.
Gross transfer volume is a flow across addresses. It is closer to counting every movement of a banknote between tills, vaults, intermediaries, and customers than to counting only purchases. The $62 trillion annual figure is a broad record of such transfers. The roughly $4.2 trillion real-economy classification uses filtering. The $350 billion to $550 billion conservative bilateral goods-and-services estimate narrows the definition further. All are flows. Their scopes differ materially.
The cross border share introduces a new denominator. Classified stablecoin use set against $44.3 trillion in retail international payments for 2025 produced that 0.31% estimate. It says little about wholesale treasury settlement, exchange transfers, or domestic payroll. A higher value in a later year may reflect more genuine usage, improved identification, or a changed market size. A comparable time series needs the same definition on both sides of the fraction.
Card volume is a network measure. It is based on transactions through card programs, some of which use a stablecoin for funding while the merchant is paid through conventional acquiring and settlement. The nearly 200% growth refers to those programs, not a 200% rise in all stablecoin payments, and not the fraction of every card purchase. The $20 billion annualized settlement run rate is another measure of obligations among network partners. One customer purchase can generate both a card transaction and a later settlement step.
Four questions, four answers: Supply asks how much token stock exists. Gross transfers ask how much the tokens moved. Classified payments ask which moves look like commerce. Card and settlement figures ask what a network actually processed.
Keeping these four categories separate prevents a common error in market stories: adding supply, transfers, completed purchases, and wholesale settlement into one supposed adoption total. The values are related. They are not mutually exclusive. A token can be issued, sent to an exchange, passed through a payment provider, redeemed, and used to settle a card obligation in a series of entries around one commercial event. The useful question is where the independently completed obligation sits in that chain.
How A Processor Could Actually Prove The Claim
A processor can report the number and value of customer instructions it received, the number actually completed, the origin and destination corridors, and the local currency credited to beneficiaries. It can disclose median end-to-end time and the distribution of delays, rather than the fastest blockchain confirmation. If it quotes costs, those should include funding, foreign-exchange spread, blockchain fees, compliance charges, redemption, and payout. The quote should say who bears each cost. That kind of disclosure would allow a comparison with bank and card alternatives without exposing personal transaction details.
The difference between a test and recurring business is visible in the seven-day banking pilot. A $750,000 settlement shows the route can work under the conditions of that trial. It does not report a quarterly count of independent customers or a completed end-user remittance program. A bank’s internal obligation to another institution is a legitimate cross border settlement. It answers a different question from whether a family can send a small amount abroad more cheaply and reliably.
- Count instructions received, not just tokens observed.
- Count instructions completed, with corridor and currency credited.
- Publish median time and the slow tail, not the highlight reel.
- Itemize funding, spread, chain fees, compliance, redemption, and payout.
- Say who pays each line, so a cheap fee cannot hide an expensive rate.
Merchant acceptance is similarly easy to overstate. A stablecoin-linked card can let a consumer spend a token balance at a conventional terminal while the retailer receives local currency from an acquirer. That is meaningful utility for the cardholder and the issuer. It is not direct stablecoin acceptance by the merchant. Direct acceptance requires the seller to receive and manage the token, or to contract a processor to do it on the seller’s behalf. The settlement chain decides where the stablecoin actually changes hands.
Public researchers face limits that processors do not. A processor can match an invoice, verified counterparties, and a payout record. An independent analyst sees token movements and may infer intent from labels and behavior. Better labeled addresses improve the model. They do not make an unlabeled customer wallet self-describing. The September revision is evidence of how much improved identity information can change adjusted volume after the transactions have already happened.
Redemption And Rules Shape The Route
A payment token’s usefulness depends on its issuer’s promise to redeem and on the firms willing to accept it. A dollar stablecoin may stay close to one dollar across major exchanges, yet a recipient in a thin local market can face a wide spread when cashing out. Banking relationships can halt deposits or withdrawals even while the blockchain runs continuously. Providers must satisfy rules for identity, sanctions, travel information, and consumer protection in the jurisdictions where they operate. The blockchain transfer is one part of that regulated service, not a substitute for it.
For a company, an international payment has an accounting life after settlement. It needs the invoice, the exchange rate, any fees, the identity of the payee, and a view on tax or reporting obligations. A recipient who keeps a stablecoin on the balance sheet takes issuer and custody exposure until redemption. A firm that converts immediately needs reliable market depth and a payout partner. These constraints explain why a fast pilot may grow slowly despite sound technical performance. The technology can be ready while the operating manual is not.
The institutional route can still have advantages. A company that routinely pays suppliers in several time zones may reduce prefunded balances and weekend delays if stablecoin liquidity and local payout are reliable. A recipient without practical access to dollars through banks may value a dollar-denominated token. The advantage is corridor specific. It has to be demonstrated through delivered amounts and repeat transactions, not extrapolated from total transfers on the issuing chain.
What The October Evidence Does And Does Not Settle
The data available in October establishes that classified payments exist at scale and that some card and bank programs are live. It does not establish the fraction of every public transfer that represents a cross border purchase. That question gets sharper as providers share completed transaction data and researchers keep classifications stable across time. Until then, anyone quoting a single adoption percentage is choosing a definition and hoping you will not ask which one.
I have sat through enough product briefings to know the next sentence in the script. Growth is real, programs are multiplying, weekend settlement works in a pilot. All of that can be true. None of it licenses the claim that headline transfer volume is payment activity. The two can rise together. They are still different animals.
What Is Worth Watching Next
The next useful disclosure is a consistent series of completed payments by purpose and corridor, with the sender’s total cost and the beneficiary’s received amount. Payment processors can publish those operational measures without exposing private customer identities. Updated classifications and methodology logs will show how much a changed label set alters historical totals. More bank and card programs can demonstrate repeat use. A pilot should remain a pilot until it reports volume at commercial scale.
The direction of travel is clearer than an exact global market share. Real payment cases exist, and classified volume has grown. The trillion-dollar transfer headlines mostly describe the motion of tokens through trading and financial plumbing. The estimate of commerce depends on classification. The best cross border test includes both ends of the payment and all the costs between them.
A Working Way To Read The Next Headline
When the next chart lands, I would ask five plain questions before believing the adjective attached to it. Is this supply, gross transfers, classified payments, or network settlement? Which months, which tokens, which chains? Were exchanges, bots, and bridges removed, and with which label set? Does the figure include offchain card spend or only what a public ledger can see? And what did the beneficiary actually receive, in which currency, after spread and delay?
If the answers are fuzzy, the number is a conversation starter, not a market share. That is not a knock on the technology. It is a refusal to let plumbing masquerade as checkout. Stablecoins can be a serious payment rail in specific corridors, for specific users, once redemption, screening, and local payout hold up on a Tuesday afternoon and not only in a demo. Until the disclosures match that standard, the honest summary is the one the notebooks keep missing. Billions are moving. A smaller, still meaningful slice looks like payment. The rest is the machinery that makes the slice possible, and machinery is not the same thing as the sale.
Does a cheap chain fee make a cheap remittance? Not by itself. Funding, foreign exchange, redemption, compliance, and local payout decide what the sender pays and what the recipient gets. Can blockchain data alone tell you a transfer crossed a border? Only with reliable information about counterparties and location. Addresses have no nationality. Is every stablecoin transfer a payment? No. Trading, collateral, bridging, issuance, internal treasury moves, and genuine payments all leave similar footprints. The footprint is not the obligation.
Measurement improves when labels stay consistent, methodology revisions are published, and processors release completed-payment data by corridor, cost, and beneficiary amount. Until that arrives, treat the hundreds of billions as a serious lower bound on classified activity, not as a census of global commerce. The river is real. The receipt is the part worth counting.
