The number is $482 billion. It grew 45 percent. That is the entire dataset.
No reporting period. No sample frame. No adjusting methodology. No named statistician, no regulator, no filing, no counterparty. The claim arrives as one sentence and a citation to nobody. I have spent nine years building models that live or die on exactly those four metadata fields, and I can tell you what happens when they go missing: the number stops being a measurement and becomes a mood.
Start with the arithmetic, because the arithmetic is where the sentence breaks. If $482 billion is an annual figure, stablecoin payments are roughly two percent of what the two dominant card networks clear in a year. If it is a monthly figure, the same number annualizes to $5.8 trillion, call it a fifth of that base. Same four words. Two readings, a factor of ten apart in strategic consequence. One earns the phrase reshaping the financial system. The other earns a footnote.
The source does not tell you which. It also does not tell you whether the $482 billion counts a market maker shuffling USDC between two wallets it owns, whether it counts an MEV bot cycling through a router four hundred times a minute, or whether it counts the same dollar eleven times before lunch. Those are not edge cases. In every volume audit I have run since 2020, they were the majority of the raw number.
So this is not a piece about an article. It is a piece about a data point stripped of provenance and released into a market that is sideways, starved for direction, and therefore unusually willing to believe anything with a percent sign attached.
What the Source Establishes, and Where It Stops
The payload is thin. Visa, Mastercard and Stripe are building bridges between crypto and cash. Stablecoin payment volume grew 45 percent to $482 billion. The framing: this may reshape the financial system.
Three company names. One growth rate. One dollar figure. One civilizational adjective. That is the whole object.
What is missing is everything an auditor needs. No chains named. No issuers named. No corridors, meaning no statement of which countries, which currencies, which settlement legs. No counterparty institutions. No transaction counts. No unique address counts. No distinction between gross and net. No baseline period, which means the 45 percent has no denominator and cannot be verified even in principle. No statement of whether the number is on-chain gross transfer value, on-chain adjusted transfer value, fiat notional at the on and off ramps, or merchant acquiring volume.
Those four are not synonyms. They are four different quantities measured by four different parties with four different error terms. In my own reconstructions, the spread between the loosest and the tightest definition of the same underlying activity ranged from three to twelve times. A headline that does not specify which one it used is not a measurement. It is a range with the error bars deleted.
I want to be exact about what I am not claiming. I am not claiming the growth is fake. I am claiming the growth is unmeasured in any form that survives an audit. Those are different assertions, and conflating them is how analysts lose money in both directions.
What I can add from my own coverage, flagged explicitly as background rather than as anything the source established: on-chain settlement is not a new capability for the card networks. Visa has been settling with issuers in USDC over public chains for years, with publicly described pilot volumes in the low single-digit billions annualized. Mastercard has run a multi-token settlement network connecting banks and issuers. Stripe acquired a stablecoin infrastructure company in 2024 and has been shipping payout rails since. None of this appears in the source; I carry it from watching these three institutions move. [Background, medium-high confidence.]
Which reframes the 45 percent. This is not a story about a capability appearing. It is a story about a capability scaling. Those two stories have completely different risk profiles. The first is a technology bet. The second is an adoption and unit-economics bet, and adoption is where the real work sits.
The Transmission Chain: How a Number Loses Its Denominator
A figure like this rarely originates as a lie. It originates as a line item inside a longer document, surrounded by definitions. Then it gets extracted. The extraction is where the damage happens.
Watch the sequence. An institution publishes a report with a scope statement, a methodology appendix and a reconciliation of adjusted versus raw. A trade publication quotes the headline number and drops the appendix, because the appendix does not fit a headline. An aggregator rewrites the trade piece and drops the scope statement, because scope is boring. A social account posts the number with an adjective attached. By the fourth hop, the number has a vibe, a growth rate, and no owner.
This is not unique to crypto. It happens to GDP estimates, to retail sales, to every monthly print that gets quoted in a headline before anyone reads the release. What makes crypto different is the base rate. In official statistics, the underlying agency has a legal obligation to publish methodology and revise errors. In crypto media, nobody has that obligation, and the volume of summary-shaped content is climbing fast.
The source item here is a summary that repeats itself, carries no byline, no data source and no detail. Content of that shape is increasingly common. Its rise is itself a signal: the information noise ratio in this sector is going up, and for anyone who prices media flow into positioning, that is a structural negative that nobody marks to market.
There is also a selection effect worth naming. The publication chose to run this headline now, in a flat tape, with no new disclosure behind it. That choice tells you where narrative supply sits. When a real trend starts being retailed through de-specified numbers, the fundamental increment has usually been priced and the narrative increment is still being spent.
Settlement Is Not Payment
This is the technical distinction the entire claim hinges on, and it is the one most readers will skip.
A card network settling with an issuing bank in USDC is B2B treasury clearing. One institution moving money to another institution, against a contractual obligation, with near-zero dispute exposure, in sizes that make consumer payments look like rounding errors. A consumer paying a merchant in stablecoin is retail acquiring. Different legal regime, different risk surface, different unit economics, different adoption curve.
The source uses one word, stablecoin payments, for both. That single word choice is where the $482 billion loses its meaning.
Here is the taxonomy I use internally.
| Business | Who pays whom | Typical ticket | Settlement urgency | Dispute exposure | Regulatory path | |---|---|---|---|---|---| | B2B treasury clearing | Issuer to network, bank to bank | $10M to $500M per leg | Same-day desirable | Near zero, contractual | Banking, payments licensing | | Merchant acquiring | Consumer to merchant | $5 to $5,000 | Instant | High, chargeback regime | Consumer protection, scheme rules | | Cross-border remittance | Individual to individual | $100 to $10,000 | Minutes | Medium | Money transmission, FX controls | | Internal venue transfer | Exchange to exchange | Any size | Instant | None | Custody, VASP |
Only the second row is what a normal reader hears when they read the word payments. Only the second row carries consumer protection obligations, chargeback rights and the merchant acceptance problem. And the second row is exactly where the card networks are weakest, because their entire moat is the thing that row needs most: dispute handling.
Rows one and four are where the volume is. They are also the rows that can grow by an order of magnitude without changing anything a consumer experiences. A settlement desk can triple its on-chain throughput next quarter and no merchant will notice, no card will be swiped differently, and no fee will fall for anyone.
So when a headline reports that stablecoin payments grew 45 percent, the first question is not how. It is which row. A 45 percent increase in row one is a plumbing statistic. A 45 percent increase in row two is a genuine competitive event. The source does not distinguish, and the difference in market implication is total.
Throughput Is Not the Bottleneck
The technology commentary around this theme is also misaligned, and it is worth thirty seconds of correction because it keeps producing bad forecasts.
Card networks clear in the tens of thousands of transactions per second at peak. Public chains settle at fifteen per second on the base layer for Ethereum, low thousands sustained on the faster L1s, with rollups pushing aggregate capacity higher at the cost of a trust assumption on the sequencer. Stablecoin transfer marginal cost is effectively zero. Cross-border wires run T+1 to T+3 at tens of dollars per leg.
The real advantage of a stablecoin rail is therefore settlement latency and cost, not throughput. And the real bottleneck is not throughput either. It is reconciliation, compliance screening and dispute handling. Those are three problems that no amount of blockspace solves, and all three of them are boring, expensive and staffed by humans.
Every roadmap I have reviewed that promised to fix payments with throughput was solving the wrong constraint. It is the equivalent of optimising a database query while the accounting department is still reconciling by hand at month end. The bottleneck is where the humans are.
How $482 Billion Gets Manufactured
Four mechanisms inflate raw on-chain stablecoin volume. I have measured all four, each in a different year, each with a different method, and each with a number I can defend.
Vector one: self-transfers and change addresses
In 2020, for my undergraduate thesis, I manually traced $45 million of Uniswap V2 liquidity flow across roughly 12,000 Ethereum transactions. The first thing I had to build was not a liquidity model. It was a clustering heuristic, because a material share of the flow was one entity moving value between addresses it controlled. Change addresses, sweeps, wallet rotation. Naive counting overcounted by a double-digit percentage.
Once I applied clustering and excluded same-entity movement, something else became visible: a persistent arbitrage inefficiency caused by slippage tolerance settings. Small trades were being systematically mispriced against the curve, and the mispricing was invisible in aggregate volume. It only appeared after the volume was cleaned. That is the general lesson and it has never stopped being true. Raw volume is a gross number. Analysis lives on the net.
Vector two: exchange internal netting
When a large venue rebalances between hot and cold storage, or moves funds between two omnibus wallets it controls, that is simultaneously a real on-chain transfer and a zero-economic transfer. It also happens to be the single largest source of stablecoin transfer count in most months.
On-chain counters see it. A payment system does not. If your denominator includes exchange treasury operations, you are not measuring payments. You are measuring the internal logistics of custodians and calling it commerce.
Vector three: MEV and routing loops
In 2021 I analysed 8,500 secondary sales on OpenSea for a prominent profile-picture project. Forty percent of reported volume traced to five connected wallets. That was wash trading with a social motive, designed to produce a chart.
The 2026 version has no motive at all. It is arbitrage. A router that splits one swap into six hops produces six transfers and one economic event. Search-and-backrun activity multiplies that again. The volume is real in the sense that gas was paid and blockspace was consumed. It is not real in the sense that anyone intended to move value from a payer to a payee.
Vector four: autonomous agents
In 2026 I designed and ran an experiment in which autonomous agents executed 10,000 micro-transactions on a new L2 to measure gas fee volatility. I collected terabytes. What I did not expect, and what I now consider the most under-priced measurement problem in the sector, is this: agent-driven flow is statistically indistinguishable from human flow inside any aggregate counter. It scales with compute budget rather than with commerce. And it produces exactly the kind of predictable liquidity gaps that other agents then harvest. That is a closed loop. It generates volume. It generates fees. It generates zero end-user payment.
Which means the marginal unit of on-chain stablecoin volume in 2026 is increasingly produced by a machine with no commercial intent, and no counter on earth separates it from the volume produced by a family sending money home.
The cumulative effect
Here is my working set of deflators. These overlap, so they do not simply add. That is precisely the point: any single one of them destroys the credibility of an unadjusted headline.
| Filter | What it removes | Typical reduction, my estimates | |---|---|---| | Self-transfer and change | Same-entity movement | 8 to 20 percent | | Exchange internal netting | Omnibus rebalancing | 25 to 45 percent | | MEV and router loops | Multi-hop, single economic event | 10 to 30 percent | | Agent and bot flow | Non-commercial automated volume | 5 to 25 percent | | Non-payment economic transfer | Collateral, treasury operations | 20 to 50 percent | | Cumulative, raw to adjusted | All of the above | often 3x to 12x |
[Estimates, medium confidence. These are my ranges from audits across four chains and several years, not a published benchmark. Nobody has published a clean benchmark, which is the actual scandal here.]
Take only the middle of that cumulative range and a raw $482 billion becomes something between $40 billion and $160 billion of economically intended payment. And the 45 percent growth rate, measured on the raw series, could easily be negative on the adjusted series if the growth came from agent flow and exchange churn rather than commerce.
I cannot assert that. I can assert that nobody has shown otherwise, and that the burden of proof runs the other way when the metadata is missing.
Three Plausible Provenances
Since the source provides no origin, the honest move is to enumerate what could have produced the figure and price each branch.
Provenance A: a vendor or industry body published a marketing-adjacent report, using gross on-chain transfer value, annualised from a strong quarter. Implication: the 45 percent is a cherry-picked base effect and the $482 billion is gross. Probability weight, my judgement: highest of the three.
Provenance B: a data provider published adjusted transfer value with a defined methodology, and the media dropped the methodology on extraction. Implication: the underlying number is defensible, the headline is not, and the true payment figure is materially smaller than $482 billion.
Provenance C: the figure is fiat notional at the on and off ramps, meaning it measures conversion activity rather than transfer activity. Implication: it is a liquidity statistic, not a payments statistic, and it would double count every round trip.
All three branches produce the same sentence and three different worlds. That is the definition of an uninformative number, and it is why I will not take a position on the level. I will take a position on the direction of the error, which is unambiguously downward.
A Five-Filter Reconstruction, and One New Metric
Here is how I would rebuild the number from scratch. Five filters, applied in order, with a named failure mode for each.
Filter one: entity resolution. Cluster addresses by common-input heuristics, gas funding patterns, deposit-address reuse and behavioural fingerprints. Failure mode: custodial exchanges look like one entity, which is correct for netting and wrong for counting users. I treat exchanges as a single node and net them out.
Filter two: change and sweep detection. Remove value returning to an address controlled by the sender within the same block window. Failure mode: genuine refunds get caught. I accept a small false-positive rate and flag those for manual review.
Filter three: loop detection. Identify paths where value returns to its origin within N hops, or where a single intent generates multiple transfers through routing contracts. Failure mode: legitimate split payments get collapsed. Threshold selection is where most of the judgement lives.
Filter four: automation classification. Separate flow whose inter-arrival times, gas price selection and nonce patterns are consistent with programmatic agents from flow that is not. This is a probabilistic classifier, not a rule. Failure mode: high-frequency human trading gets labelled bot. In practice the classifier matters less than the direction of the correction, which is always downward.
Filter five: economic purpose. Classify the surviving flow as treasury, collateral, remittance or commerce. This is the step nobody publishes, and the only step that answers the question the headline pretends to answer.
Run those five and you get a number. It will be smaller than $482 billion. It will also be the only version comparable across quarters.
The metric I actually track
Here is the thing I built out of frustration, and I think it is the useful contribution of this piece.
Define the Settlement-Adjusted Volume Ratio:
SAVR = entity-resolved settlement-layer stablecoin transfer value / end-user payment value after automation and purpose filters.
Read the regimes. A SAVR near 1 means the on-chain ledger is mostly commerce and the payment story is the real story. A SAVR of 5 to 10 means the ledger is mostly treasury and arbitrage, and the payment story is a rounding error inside a plumbing story. A rising SAVR means volume is growing faster than payments, which is the regime that produces headlines like this one: an impressive dollar figure attached to a business that is not the business being described.
My working estimate, flagged as an estimate rather than a published figure: for the major stablecoin pairs on major chains, SAVR has been well above 1 and rising over recent quarters. Volume growth and payment growth have been diverging. The headline metric is not merely unverifiable. It is structurally the wrong metric, because it moves with the thing that is not payments.
If you take one operational rule from this article, take that one. Track SAVR. When it falls, payments are winning. When it rises, plumbing is winning and the narrative is borrowing credibility from the wrong series.
Finality Versus the Chargeback Window
Code does not care about your feelings, and it especially does not care about your refund policy.
A stablecoin transfer is final. A card transaction is reversible for up to 120 days. That asymmetry is not a user-experience detail. It is the load-bearing wall of the consumer payments business, and it is the reason merchant adoption of stablecoin rails lags treasury adoption by years.
Walk the merchant economics. Card interchange for a mid-size merchant runs somewhere in the range of 150 to 250 basis points, and a meaningful slice of that buys dispute handling, fraud screening and consumer protection. A stablecoin rail might save 100 basis points on processing. It then adds an uninsured dispute cost, because the merchant must self-insure the reversal window, plus a working capital drag, because final settlement into an operationally awkward asset still has to be converted and reconciled.
For low-ticket, low-dispute categories the trade can work. For high-ticket, high-dispute categories it is a loss, and no amount of enthusiasm changes the arithmetic. This is why the volume concentrates in B2B legs, where disputes are contractual and settlement is already near-final.
I have priced this for a fund. The self-insurance line item is not small, and it does not appear in any growth narrative I have read.
Reconciliation Is a Systems Project
An on-chain ledger settles in seconds and is immutable. A corporate ledger closes monthly and is accrual-based. Mapping one onto the other requires deterministic treatment of chain reorganisations, failed transactions, gas as operating expense versus cost of goods, and depeg marks on the settlement asset itself.
I built this once. It is a systems project measured in quarters. It is also the reason institutional settlement volume grows in steps rather than curves: each step is a reconciliation pipeline going live. Every one of those steps is invisible in a headline and decisive in a model.
What the Card Networks Are Actually Buying
Follow the smart money, not the hype.
The smart money here is not betting that stablecoin payments replace cards. It is betting that if replacement happens, the toll booth gets rebuilt in the same place. That is a different thesis from the one being retailed, and the difference matters for anyone trying to express a view.
The interchange pool is the prize and the exposure simultaneously. If stablecoin rails scale at merchant checkout, the first revenue line to compress is card interchange. So the incumbent move is to own the rail that would otherwise route around them. Absorb the disruptive technology, charge for it, keep the customer relationship. Textbook incumbent behaviour, and rational.
Which yields a clean conclusion. The card networks are beta exposure to this theme, not alpha. They will participate in the volume and they will not capture the convexity, because their participation is defensive by construction.
Stripe's incentive is different again, and the source's decision to list all three as a bloc conceals it. Stripe is a payment service provider. Its win is shortening the settlement chain, improving float economics on merchant balances, and owning cross-border B2B where the margin is fattest. A card network wants to preserve network effects. A payment service provider wants to shorten the path. Those interests diverge the moment stablecoin rails get good enough to bypass a scheme.
Three parties, one sentence, three incompatible strategies. When a source flattens that into a single trend, it is not summarising. It is smoothing over the only analytically interesting part.
The Structural Fact Nobody Prices: There Is No Token
Run the value capture chain and see where it terminates.
Issuance revenue goes to the issuer, as interest on reserve balances. Settlement fees go to the networks and the payment service providers. Gas and MEV go to validators, searchers and the chain, in amounts that are large in absolute terms and negligible against payment notional. What remains for a token holder? In this vertical, usually nothing, because the three named participants have no token at all.
That is the most important structural fact in the sector and it is almost never stated plainly: the value capture endpoint of stablecoin payments is equity in listed card networks, equity in the issuer, and equity in a private payments company. Not a tradeable token.
So when a retail reader sees a 45 percent growth headline and goes looking for a token to express it, they are searching for an instrument that does not exist. They will find a substitute that is correlated in narrative and uncorrelated in cash flow. Exit liquidity is someone else's entry. The narrative needs buyers on the other side of it, and the absence of a direct instrument guarantees those buyers end up somewhere adjacent.
Velocity Is the Enemy of Float
This is the part I would put in front of any portfolio manager, and it inverts the headline.
Issuer revenue is approximately float multiplied by the short rate. Payment volume is approximately float multiplied by velocity. Therefore a 45 percent increase in payment volume against a flat float is a 45 percent increase in velocity. Each dollar of float is doing more work. The same throughput requires less float. Revenue per unit of volume compresses.
Read that again, because it is the opposite of what the narrative implies. Volume growth is not automatically good for the issuer. Float growth is. A stablecoin that circulates fast is a stablecoin being used as money, and money is a brutal business when your revenue is the interest on idle balances.
Work a simple case. An issuer holds $50 billion of float at a 4 percent short rate. Gross reserve revenue is $2 billion annualised. Now payment volume doubles while float stays flat, because merchants accept and immediately convert. Reserve revenue is unchanged. Costs are up. The growth headline is a cost centre.
Now the short rate falls 150 basis points. Revenue drops to $1.25 billion. No payment volume figure on earth offsets that, because the business is a rate bet wearing a payments costume. This is the interest rate sensitivity every stablecoin narrative quietly omits, and it is the largest single swing factor in the model.
Follow it through to beneficiaries. Whoever charges on volume captures volume growth: networks, payment service providers, ramp operators. Whoever earns on float captures float growth, and only float growth. The naive reading, that a rising tide lifts everyone in the sector, fails at the first line of the model.
Who Holds the Kill Switch
One more structural point that the decentralization framing cannot absorb.
The decisive decisions in this rail are made inside the compliance and risk departments of a handful of companies. Who gets an account. Which merchants are permitted. Which addresses get frozen. Which corridors are open this quarter. None of that is on-chain, none of it is governed by a token vote, and none of it is disclosed on a schedule.
A stablecoin issuer can blacklist an address. A card network can terminate a merchant. A payment service provider can close an account with a notice period measured in days. The decentralization content of this vertical is close to zero, and that is not a bug in the implementation. It is the product. Institutions participate precisely because someone can freeze something.
Which produces the sentence that most annoys the maximalists: the more successful stablecoin payments become, the more clearly they are an extension of traditional finance rather than a replacement for it. That is not a criticism. It is a description of what was actually built, and it is the reason the growth is real.
The Real Competitor Wears a Bank Charter
Tokenized deposits. A regulated bank issues a digital claim on a deposit it holds, backed one-for-one by central bank money, with no reserve attestation theater, no depeg risk, and direct access to the settlement backbone that already exists.
Put yourself in the chair of an institutional treasurer choosing between that and a public-chain stablecoin. The tokenized deposit wins on credit risk. It wins on regulatory certainty. It wins on integration, because it plugs into infrastructure the treasurer already has. The public-chain stablecoin wins only where the banking system does not reach: retail in weak-currency jurisdictions, and permissionless composability for developers.
Which suggests the end state is not stablecoins eating banking. It is stablecoins confined to the perimeter. Public chain for the unbanked and the composable. Tokenized deposits for the institutional core. The two coexist, and the one carrying the institutional volume is the one with a bank charter.
Crypto media almost never discusses this. It is boring. It has no ticker. It does not pump. Transparency is the only security, and on the dimensions a treasurer cares about, the tokenized deposit is the more transparent instrument. I hold this view at medium confidence, contingent on how fast the regulated perimeter builds. Regulated perimeters build slowly until they do not.
The Enforcement Gradient
The fastest-growing part of this volume is also the least legally stable part, and the two facts are related.
The genuine demand for dollar-denominated settlement comes from cross-border B2B and remittance, which is exactly where capital controls, FX management and sanctions compliance intersect. Corridors with the strongest demand for dollar access are frequently the corridors with the tightest capital accounts. Growth concentrates there. Enforcement risk concentrates there too.
A licensed rail is a rail that can be switched off. That is the trade institutions accept in exchange for legal certainty, and it means the growth curve in the highest-demand corridors is not a smooth function. It is a step function with a discretionary downside. Any model that extrapolates it linearly is mis-specified, and every model I have seen extrapolates it linearly.
What Would Change My Mind
An analyst who cannot state falsification conditions is doing astrology.
One. A published methodology from a named institution, with a reporting period, a chain list, an issuer list, and an adjusted-versus-raw reconciliation. If the number survives that, the growth is real and larger than my estimate implies.
Two. A chargeback or escrow product for stablecoin merchant acceptance, underwritten by a named insurer, priced in basis points. That would close the finality gap and unlock the merchant row of the taxonomy. It would be the most important structural development in the sector and it would not trend on social media.
Three. A merchant cohort disclosure. Not aggregate volume, but the number of merchants above a materiality threshold, with retention measured at twelve months. Retention is the only honest adoption metric in payments, and nobody publishes it.
Absent those three, the 45 percent stays a sentence.
Risk Matrix
| Category | Risk | Level | Probability | Impact | |---|---|---|---|---| | Data provenance | Figure has no period, source or methodology; possibly marketing-derived | High | High | High | | Mapping | Total volume misread as benefit to a specific instrument | High | High | Medium | | Rate sensitivity | Float revenue compresses in an easing cycle regardless of volume | Medium-high | Medium | High | | Technical | Finality versus chargeback conflict caps merchant penetration | Medium | High | Medium | | Operational | On-chain to ledger reconciliation errors, timing mismatch, refund gaps | Medium | Medium | Medium | | Regulatory | Cross-border corridors hit FX controls or sanctions enforcement | Medium | Medium | High | | Credit | Issuer reserve event or depeg propagates into settlement rails | Medium | Low | High | | Competition | Tokenized deposits and CBDC displace public-chain stablecoins institutionally | Medium | Medium | High | | Narrative | Reshaping-finance framing against single-digit penetration | High | High | Medium | | Disintermediation | Card interchange erodes if rails scale at checkout, the incumbent paradox | Medium | Medium | Medium |
Aggregate: medium-high. The technical and competitive risks are normal for the sector and manageable. The data risk is abnormal. For anyone acting on this, the failure mode is not being wrong about direction. It is making a size error because the denominator was never defined.
Five Signals for the Next Fourteen Days
Sideways markets are for positioning, not conviction. Here is what I am watching, with thresholds.
One. The next issuer reserve report. Compare average outstanding supply against any disclosed payment volume. If volume grows and float is flat, velocity is rising and the float thesis is weakening. Cleanest available test of the whole narrative, and it needs no new data source.
Two. The next card network settlement disclosure. Read the vocabulary. Settlement volume or payment volume. Chain named or not. Issuer counterparties named or not. Vocabulary is the tell, and it changes before the numbers do.
Three. Any chargeback or escrow pilot. A named underwriting partner priced in basis points would be a genuine structural unlock. Continued absence confirms the finality gap is open.
Four. Central bank and BIS tokenized deposit announcements. This is the competition that does not appear in crypto media and will decide the institutional layer.
Five. The next appearance of this number. Check whether a reporting period finally shows up. If a full media cycle can cite $482 billion repeatedly without anyone asking per what, that is your measurement of the information noise ratio in this sector, and it is the most actionable figure in this piece.
The question is not whether stablecoin payments grew. Something grew. The question is who needed you to believe a number nobody can define, and what they were holding while you did.