Four numbers get quoted as card spend. They count different things, and one of them is not spending at all.
Three figures went out this year under the heading of stablecoin card spend: $18 billion, $759 million and $20 billion. All three are sourced, current, and defensible. Any deck that puts them on the same slide is still wrong. The first is an annualized estimate of crypto card volume as of late 2025. The second is a single month of onchain flows through one tracked set of programs. The third measures settlement between Visa and its clients, which is a different leg of the same transaction and moves on its own schedule.
The category has reached the size where the difference matters. Product teams are pricing programs off these numbers, investors are sizing the market off them, and even after you put them on the same annual basis, the figures quoted this year still spread from about $9 billion to $20 billion. So before the next round of headlines, here is what each number counts, how far apart the sources sit, and which one belongs in which sentence.
Definitions, before the numbers
Four things travel under the same name.
Card spend is what cardholders buy. It is measured at authorization, in dollars, and it is the figure that maps to merchant categories, interchange and everything a payments team recognizes as volume.
Onchain program flows are tokens moving into and out of the contracts and treasuries that fund cards. They are a proxy for spend, visible to anyone with a block explorer and a list of addresses, and they miss any program whose funding never touches a public chain.
Network settlement is what an issuer owes the network at the end of a cycle, paid in stablecoin instead of fiat. It sits downstream of spend, it can move on a different clock, and it includes the same dollars the cardholder spent plus whatever else the issuer settles that way.
Company-reported payment volume is a firm’s own total. It usually bundles card authorizations with top-ups, transfers, payouts and anything else the platform moves, and no two companies draw the boundary the same way. The same discipline applies here that applies to supply and volume on the token side: the labels look interchangeable and the numbers are not.
One purchase, four numbers
Put an $86 grocery run on a stablecoin card and watch where it lands. Card spend counts $86 at the moment the terminal gets its approval. Onchain flows count the tokens that leave the cardholder’s balance to fund it, which is $86 if that balance sits on a public chain and nothing at all if the program holds the funds internally. Settlement counts the issuer’s payment to Visa at the end of the cycle, which contains that $86 alongside every other transaction the program cleared that day. And a company total can count the $86 purchase, the $200 top-up that funded it, or both, depending on how the firm defines volume.
One transaction, four legitimate readings, and a spread that widens with every program added to the count.
The tape
Visa’s own accounting is the anchor for the network side. In 2025 its stablecoin-linked cards processed about $5.2 billion, a 319% increase on the year before, across more than 130 programs in more than 50 countries. Visa expected the program count to roughly double during 2026, and by early September it reported more than 160 programs live with payment volume on them up nearly 200% year over year.
The monthly series comes from Paymentscan, whose data a16z published on August 7. Crypto card volume reached $759 million in July, about 2.5 times the $306 million of July 2025, across nearly 9 million purchases at an average of roughly $86. Those figures cover onchain activity attributable to the programs Paymentscan tracks, with the largest program, RedotPay, self-reporting its numbers instead of being observed onchain.
The annualized estimate comes from Artemis research published in January, which put crypto card volume above $18 billion a year by late 2025, growing from about $100 million a month in early 2023 to more than $1.5 billion a month, a compound rate of 106%. The same report credited Visa with more than 90% of onchain card volume.
The settlement number is Visa’s too. On September 8 it said its stablecoin settlement run rate had passed $20 billion annualized, more than fifteen times where it stood a year earlier, when the figure was $3.5 billion.
One provider, two July numbers
The clearest warning about all of this came from a single month of one dataset. On August 7, a16z published Paymentscan’s July figure as $759 million across nearly 9 million purchases, with USDC at about 58% of spend and USDT at about 26%. On August 23, CoinDesk put the same July month at $1.04 billion across more than 10 million transactions, with USDC at 50.8% of volume and USDT at 20.3%.
Sixteen days, one provider, one month, and the headline number moved 37%. The average purchase held at about $86 in both, which suggests the revision came from programs and transactions entering the count, not from a repricing of what was already in it.
None of that makes the data bad. Onchain series backfill as addresses are identified and programs are added, which is ordinary for young datasets and is exactly why the number needs a date attached every time it is used. A figure quoted without one will be wrong within a quarter, and usually low.
What settlement counts, and what it does not
Visa’s $20 billion is the number most often misquoted, because it is the largest and because the word volume appears near it. It measures the leg between an issuer and the network: the issuer pays Visa in USDC instead of wiring dollars. Consumer purchases sit upstream of that, and the two figures are related without being the same.
Visa has not published a split between settlement tied to card programs and settlement for other flows, so the honest statement is that $20 billion annualized is the size of the stablecoin settlement business and not the size of the card market. Quoting it as spend inflates the category by somewhere between two and four times, depending on which spend series you trust.
The same September disclosure showed why settlement is worth watching on its own terms. Credit Coop has built a stablecoin-denominated revolving facility secured by settlement receivables, and Rain has drawn about $2 billion through it to fund daily Visa obligations, across more than 2,000 borrow events and 7,000 repayments, with no defaults. That is a working capital market forming around the settlement leg, and it will have its own numbers soon enough.
The self-reported layer
Then there are the figures programs publish about themselves, which are the easiest to find and the hardest to add up.
RedotPay announced its Series B in December 2025 with more than 6 million registered users and over $10 billion in annualized payment volume. KAST raised in March 2026 on more than a million users and nearly $5 billion in annualized transaction volume. Rain, which sits in the card enabler layer underneath the consumer brands, told The Block in August that it processes more than $3 billion a year.
Add those three and you get $18 billion, which is larger than the entire tracked market for the same period. The arithmetic fails for two reasons. Enablers process for programs, so an enabler’s volume and its clients’ volume are frequently the same dollars counted twice. And payment volume at a wallet company includes top-ups, transfers and payouts alongside card authorizations, so the card share of a headline figure is usually unstated and always smaller.
This is not a reason to ignore company numbers. They are the only visibility into individual programs, and they move first. It is a reason to keep them in their own column, never summed, and never mixed into a tracker total.
Where the two trackers part company
Artemis put crypto card volume above $1.5 billion a month in late 2025. Paymentscan put it at $759 million, later revised to $1.04 billion, in July 2026. Read literally, the market halved while every other indicator in the category went up.
It did not. The two series count different populations. Artemis measured crypto card volume broadly, without publishing which programs sit inside the number. Paymentscan measures onchain activity attributable to the programs on its list, which leaves out any card funded from a custodial balance that never moves on a public chain, and that description covers a good share of exchange cards.
Neither provider publishes a full method, and until one does, the useful move is to pick a series, say which one you are using, and track its direction more than its level. Direction is where the two agree: both show roughly two to three times growth a year, and Visa’s own program volume, up nearly 200%, sits in the same range.
Which coins get spent, and which get settled
The coin mix is the part of this data that has moved most.
On the spend side, dollar tokens funded about 70% of tracked transactions in July. USDC leads at somewhere between 50.8% and 58% depending on which July print you take, and USDT sits between 20.3% and 26%, up from roughly 7% a year earlier. That is the fastest share shift in the dataset, and it tracks where cards are being issued rather than where crypto trades.
On the settlement side, USDT is absent. Mastercard’s June settlement expansion listed USDC, PYUSD, USDG, USDP, RLUSD and SoFiUSD across eight blockchains. Visa’s US settlement launched in USDC. So the token a cardholder spends and the token an issuer settles in are increasingly different tokens, with a conversion in between that somebody is pricing and nobody is publishing.
Chains tell a similar story about where this activity actually lives. As of July, Optimism carried about 29% of tracked card spend, with Solana and Base at about 19% each, which is a different distribution from stablecoin supply and a hint that card programs pick chains for cost first and liquidity second.
Read more: Rain’s $250 million round prices stablecoin card plumbing • What $309 billion in stablecoins does and doesn’t tell us
The scale check
Set the biggest credible spend figure against the network carrying it. Visa’s payments volume in fiscal 2025 was $14.2 trillion. The $5.2 billion of stablecoin-linked card spend inside it is about four hundredths of one percent.
What makes the category worth watching is the rate of change. The 2025 figure grew 319%, the monthly series roughly 2.5 times year over year, and Visa’s program volume nearly 200%. Growth also concentrates outside the markets that already have good cards: StraitsX data cited in the same CoinDesk report showed gross transaction value up about 600% in lower-GDP markets between March 2025 and February 2026, against 150% in higher-GDP ones.
A category compounding at that rate from four hundredths of a percent is interesting and small at the same time, and both halves of that sentence belong in any serious forecast.
Which number to use for what
For consumer adoption, use card spend, and say which series and which month. For an enabler’s revenue, start from spend and apply interchange, because settlement and company payment volume both overstate the base. For treasury, liquidity and working capital questions, settlement is the right number, and it is the one that explains why a credit market is forming around card programs. For momentum, use the monthly series and accept that it revises.
And for anything that will be read a year from now, write the date and the source into the sentence. Every number in this piece will move.
What this column will track
From here, this site will publish one card spend series each month rather than whichever figure is loudest that week. The series will name its provider, state what the provider says it counts, mark revisions to prior months instead of quietly overwriting them, and keep company-reported figures in a separate column that is never added to the tracker total. Settlement will be reported as settlement.
Four things would change the picture quickly. Visa’s next quarterly disclosure, which has been the category’s most reliable scoreboard. Any figure at all from Mastercard, which has published none. An enabler publishing spend volume the way processors publish total payment volume, which would put a floor under everyone else’s estimates. And the January 18 effective date under the GENIUS Act, which decides which programs operate in the United States and therefore which dollars end up inside any of these numbers.
Until then, the most accurate thing anyone can say about stablecoin card spend in 2026 is that it runs somewhere between $9 billion and $18 billion a year, that nobody can close that gap from public data, and that it is growing faster than any of the people arguing about it can measure.