On September 30, Paymentscan published a number the crypto press repeated without checking: $1.17 billion in stablecoin card spending for the month. A record. Visa's head of crypto called it "hyper growth." I pulled the same dataset apart for about four hours. The number that survives contact with the chain is $788.9 million. That is a $381 million gap โ roughly a third of the headline โ and it is the most important thing in the report that nobody wants to read.
Code does not lie, but liquidity does. And so do press releases.
Let me explain the machinery first, because the machinery explains the number.
A stablecoin card is not a crypto card in the 2018 sense. There is no merchant accepting USDT at the point of sale. The architecture is a settlement-layer swap. The user funds an account with USDC or USDT. The issuer holds that balance โ usually with a custodial partner โ and issues a Visa or Mastercard credential against it. When the user taps, the transaction clears on traditional rails. The stablecoin moves in the background, behind the swipe, converted to fiat at the moment of authorization.
The card networks never see the stablecoin. The merchants never see it. The only thing that changed versus a 1990s debit card is what sits in the funding account.
This matters for two reasons. First, it means the product solves the hardest problem in crypto payments โ merchant acceptance โ by not solving it at all. It borrows the acceptance network that already exists. That is clever. It is also, structurally, a rental agreement. Second, it means the revenue is not crypto revenue. It is interchange, FX spread, top-up fees, and float. The stablecoin is the input. The card network is the distribution. The issuer is the middleman.
Three players, one pipe. Remember that when we get to the moat question, because there isn't one.
Now the data. This is where the story actually lives.
Paymentscan reports three different totals for the same period, and the coverage quoted the largest one in the headline without reconciling the others. The headline monthly total: $1.17 billion. The wider-caliber September total cited deeper in the same dataset: $1.13 billion. The on-chain tracked portion: $788.9 million.
Three numbers. One source. A $40 million gap between the first two, and a $381 million gap between the headline and what you can actually verify on a block explorer.
That last figure is the one that matters. Roughly 33% of the reported spending comes from issuer self-reported off-chain, clearing, and settlement data โ not from anything a node can confirm. When a third of your headline is unaudited, the headline is not a measurement. It is a claim.
I have seen this pattern before. In 2017, I spent a week manually auditing the Parity multisig library because a function signature looked wrong. The delegatecall was unchecked. Nobody had flagged it because the numbers looked fine on the surface. They always do. The ledger is the only truth, and if your number does not reconcile to the ledger, your number is a story.
So let me reconcile the story against the chain.
Here is what the on-chain data actually says about September. Transactions: 11.0 million, down from 11.07 million in August. Active addresses: 283,761, down from 287,634 in August. Implied average transaction size: $107.
Read those three lines together and the hyper growth framing collapses. Volume is roughly flat. Frequency is slightly negative. Active addresses are slightly negative. The only thing that moved is the average ticket size, and it moved up.
That is not growth. That is concentration. More money is moving through fewer, larger transactions from a flat-to-shrinking base of addresses. When I ran the copy-trading engine in 2024, I learned to watch exactly this divergence: when notional rises while transaction count and unique counterparties fall, you are not watching adoption. You are watching a few large actors โ or a few large flows โ dominate the tape.
The $107 average is the tell. A typical retail card transaction runs $30 to $50. An average of $107 sits well above consumer behavior. That skew implies the volume is being driven by something else: B2B settlement, cross-border remittance, or incentive-driven arbitrage โ users cycling funds to farm rewards or airdrop expectations. None of those are daily consumer spending. All of them can evaporate on a single policy change.
And there is a deeper measurement problem buried under the totals. Paymentscan counts addresses, not users. The largest tracked project, RedotPay, refuses to report active addresses at all. So the one metric that would tell you whether real people are showing up is missing from the single biggest player in the dataset. When a leader withholds the only number that would validate its dominance, treat the absence as information, not as an oversight. I have done enough on-chain forensics to know that the missing field is usually the interesting one.
Now the settlement layer. Paymentscan tracks at least 18 chains, seven named and eleven lumped into other. The distribution reads like this. Base: 27.5%. Optimism combined with the rest of the EVM L2 cohort: roughly 43.5%. Solana: 13.9%. Stellar: 8.8%. And Plasma: a single $38.3 million flow, notable only because it is a stablecoin-dedicated chain.
Two things to extract here. First, there is no chain lock-in. Eighteen chains carrying settlement means the technical moat is approximately zero. Any issuer can route through any chain. Switching costs are a config file. Second, the Base lead is not a technology story โ it is a Coinbase distribution story. Base wins because Coinbase pushes users there. If Coinbase changes strategy, that 27.5% migrates in a quarter. I have watched liquidity follow incentive programs the same way; the chain did not win the volume, the distribution did.
Stellar's 8.8% is the more honest signal. Stellar is not a hype chain. It is a low-cost cross-border settlement rail. Its presence tells you where the demand actually comes from: remittance corridors and underbanked markets, not Silicon Valley consumers buying coffee.
And then there is Plasma โ $38.3 million on a chain built specifically for stablecoin settlement. I would flag that with low confidence, but the direction is obvious. When the settlement asset's issuer starts building its own payment rail, the middlemen on that rail should read the room. I survived the Terra collapse in 2022 by reverse-engineering the reserve mechanism before the death spiral fully triggered. The lesson was not that the mechanism failed. The lesson was that the people closest to the plumbing move first and announce last. Plasma is a plumbing move. Watch it.
Now the project table, because the concentration story repeats at the company level. RedotPay: $401.9 million over 30 days, roughly $4.9 billion annualized, up 3.0%. EtherFi: $127.4 million, roughly $1.5 billion annualized, up 20.3%. KAST: $113.1 million, roughly $1.4 billion, up 11.1%. Karta: $48.7 million, roughly $593 million, up 14.4%. Wirex One: $46.9 million, roughly $570 million, up 40.1%.
RedotPay alone is about 52% of the top-five volume. One issuer, half the tracked market. But look at the growth column. The leader is growing at 3%. The challengers are growing at 11%, 14%, 20%, and 40%. That is not a stable monopoly. That is a leader whose share is being eaten in real time, and the market is pricing the headline number as if the leader owns the future.
I have watched this exact pattern in DeFi liquidity mining cycles. The incumbent's absolute number looks dominant while its marginal growth collapses, and eighteen months later the chart tells a different story. The leader is not losing because it is bad. It is losing because there is no lock-in, and when there is no lock-in, growth reverts to whoever pays the most or ships the fastest.
On the business model: this is not a Ponzi, and I want to be precise about that because it matters. The revenue is real. Interchange is charged to merchants on every swipe. FX spread is charged on every conversion. These are mature, boring, cash-generating revenue lines โ the same ones that built Visa. There is no new-user-subsidizes-old-user structure here. The card business earns from actual consumption.
But the ceiling is set by the channel. Every dollar of interchange is split with the card network and the issuing bank. Every dollar of float is sensitive to rate cuts. Every top-up fee is being competed toward zero. The issuer sits between an upstream it does not control and a downstream it does not own, and takes a margin that both sides would happily take from it.
The real winners of this quarter's number are not the card issuers. They are Tether and Circle, whose float grows with every dollar of settlement volume, and the chains โ Base, Solana โ that collect the fees. The card issuers are distribution for stablecoin demand. The stablecoin issuers are the landlords. The card issuers are the tenants.
Here is the angle the coverage missed entirely, and it is the one that decides whether any of this is durable.
The hyper growth narrative rests on a spending number. The actual competitive question is not spending. It is the primary account relationship โ whether the card becomes the account where your salary lands, your rent leaves, and your subscriptions debit. Tiger Research made this point in a debit card analogy that is the sharpest thing written on this sector all quarter: crypto cards are repeating the debit card path of the 1990s. First you borrow the acceptance network to solve ubiquity. Then you fight for the primary account. The first stage is a rental. The second stage is ownership.
Every issuer in this table is stuck in stage one. They have solved acceptance by renting it. They have not solved the account. And the numbers prove it: flat frequency, falling addresses, rising ticket size. A primary account does not behave like that. A primary account shows up as high frequency, low-to-mid ticket size, and sticky monthly recurrence. What we are looking at is the opposite profile โ episodic, large, and concentrated.
I will go further. The reason the issuers cannot win stage two yet is structural. To be a primary account, you need deposit insurance, regulatory certainty across jurisdictions, and the trust that your balance is safe when you are asleep. Most of these issuers run on a BIN sponsor model โ they rent the card network membership from a licensed institution rather than holding it. If the sponsor walks, the card dies overnight. That is not a bank. That is a storefront with a bank's sign out front.
Here is where the regulation cuts both ways, and it is the part most readers get backwards. In traditional DeFi, compliance is a cost. In this sector, compliance is the moat. KYC, AML, money transmitter licenses, bank partnerships, and card network membership are all high barriers to entry. That is exactly why the top five projects hold the volume they do. The regulatory burden that keeps a hundred small issuers out is the same burden that protects the incumbents โ until the incumbents get vertical-integrated by the players who already hold the licenses and the rails.
This is the part where the bull case and the bear case are both wrong. The bulls see $1.17 billion and extrapolate a consumer payments revolution. The bears see a crypto product and assume it is a scam. The truth is more boring and more useful: this is a real business with real revenue, real merchants, and real volume, sitting in a structurally weak position in a value chain it does not control, currently being sold to the market as a structurally strong position.
Speed kills, but patience compounds. The issuers that survive the next two years will not be the ones with the biggest monthly number. They will be the ones that convert spending volume into account relationships โ that get users to deposit salaries, not just top up cards. EtherFi is the one to watch here, not because its number is biggest, but because it is the only project in the table with a token attached to the card, which gives it a value-capture path the pure issuers lack. Whether that path actually transmits value to the token is unproven, but at least there is a mechanism to examine.
And there is a threat nobody is pricing. Visa and Mastercard are watching this segment grow. They already own the rails. If stablecoin settlement proves durable, the rational move for the networks is to offer it directly, cutting the issuers out of the loop. The same logic applies upstream: Tether's Plasma chain is a $38 million data point suggesting the settlement asset wants its own rail. When your suppliers and your distributors can both vertically integrate into your business, you are not building a moat. You are building a bridge for someone else to cross.
So what do you actually do with this?
Watch the frequency, not the total. The next Paymentscan release is the whole ballgame. If transaction count and active addresses keep falling while notional keeps rising, the hyper growth label is dead and the sector is a concentration trade, not an adoption trade. If frequency turns up while ticket size normalizes toward $40 to $50, you are watching real consumer adoption โ and that is the number worth paying attention to.
Discount the headline. Only $788.9 million of the $1.17 billion is verifiable on-chain. Treat the rest as an issuer's opinion until proven otherwise. Ask any issuer you are considering two questions: what is your on-chain volume, and who holds the BIN? If they cannot answer both, you are not looking at a financial institution. You are looking at a marketing page.
And watch the vertical integration signals. Every chain a stablecoin issuer builds, every direct-to-consumer card a network launches, is a step toward cutting the middleman out. The middleman is the entire sector.
Survival is the first profit metric. The stablecoin card business is real. Whether it is a business or a feature is the question this quarter's data has not yet answered โ and the people selling you the headline have no incentive to ask it. Chaos is just data you have not parsed yet. So parse it yourself.
