The dataset reports $482 billion in stablecoin payments over nine months. Over the same nine months, aggregate stablecoin supply grew 0.4 percent, to roughly $305 billion. Payment volume rose 45 percent. Transaction count rose 71 percent. Average ticket fell 15 percent, from $547 to $463.
Four numbers, one source, no methodology.
That is the entire evidentiary basis for the current claim that stablecoins have become payment infrastructure. The direction is correct. I have reached the same conclusion from raw chain data, and I do not doubt it. But the precision being asserted here — to the billion, to the tenth of a percent — is not supported by anything the data provider has published.
Data does not negotiate; it only reveals.
The report comes from Allium, an enterprise on-chain data platform, and it is the only named entity in the document. No stablecoin issuer is named. No payment company is named. No settlement chain is named. What the report asserts is that, after stripping out arbitrage, self-transfers, and automated trading, $482 billion of stablecoin movement across nine months was genuinely payment activity.
That filter is the product. Everything downstream depends on it, and nothing about it is disclosed.
The framing matters more than the figure. For a decade, stablecoins were measured by market capitalization and total value locked. This report measures them by throughput. That is not a cosmetic adjustment. It is an admission that issuance-based metrics have stopped describing how the asset is actually used — and, more quietly, that issuance itself has stopped growing. When an industry switches its scoreboard, it is usually because it is losing on the old one.
I have audited enough of these datasets to know where the bodies are buried. The number is not necessarily wrong. The number is unauditable.
Let me separate what survives scrutiny from what does not.
The arithmetic is internally consistent. Annualizing nine months of $482 billion gives roughly $642.7 billion, a turnover ratio near 2.1x against a $305 billion float. The growth decomposes cleanly: 1.71 in transaction count multiplied by 0.846 in average ticket yields 1.45 in net value. The average ticket derived independently from the raw figures — $482 billion divided by 1.04 billion transactions — lands at $463, matching the reported decline. These are not independent confirmations; they are the same four numbers rearranged. But the rearrangements do not contradict each other, and that is more than most vendor datasets accomplish.
The denominator does not survive. "Identifiable payments" is a filtered quantity. The filter rules are undisclosed. The address-clustering method is undisclosed. The false-positive rate is undisclosed. Without a denominator, the growth rate is not comparable across periods, and a growth rate that is not comparable across periods is not a growth rate. It is a press release with a decimal point.
In 2022 I spent three months reconstructing circular trading in TerraUSD, mapping ten thousand wallets by hand and by script to isolate roughly $40 billion of artificial volume. The lesson was not that the fraud existed. The lesson was that clustering heuristics are a modeling choice, and modeling choices have owners. Every rule that excludes a wallet from the payment category is a judgment. Allium's judgments may be reasonable. They are also unexamined, and they are commercially motivated.
The value capture is where the market is simply wrong. Issuer revenue is a function of reserve size and reserve yield: supply multiplied by short rate, multiplied by one minus the distribution share. Turnover does not appear in that function. A stablecoin that changes hands twenty times a year generates the same interest income as one that sits in a vault. The 45 percent payment growth therefore accrues to gateways, acquirers, processors, and settlement layers — not to issuers. When a headline says payments grew 45 percent and an investor marks up an issuer's equity on that basis, the investor has committed a category error. The report's own author conceded the point: transaction volume is not profit. That concession deserves more weight than the headline it accompanies.
The 0.4 percent is ambiguous, and the ambiguity is load-bearing. Flat supply alongside surging throughput admits two readings with opposite implications. Either idle balances are being activated — a genuine efficiency gain — or new payment demand is being captured by substitutes: tokenized money market funds, deposit tokens, tokenized settlement cash. The first reading is constructive for issuers. The second is terminal. The dataset cannot distinguish them, and the report does not attempt to.
A magnitude check. At $642.7 billion annualized, identifiable stablecoin payments remain a single-digit percentage of global cross-border flow, which is measured in tens of trillions. The correct frame is zero-to-one validation, not displacement. Any reading of this dataset as evidence that stablecoins are disrupting card networks is an order-of-magnitude error.
Retention data is absent, and its absence is a signal. Ten point four billion transactions across nine months is about 3.8 million per day. Card networks clear hundreds of millions daily. That is a two-order-of-magnitude gap, and it means the live question is whether this can scale, not whether it has. But the more damaging omission is repeat usage. Without retention or repurchase data, one-time incentive-driven volume is indistinguishable from durable commercial flow. A vendor that publishes average ticket to two significant figures but not retention has made a disclosure decision.
Data quality flags precede methodology flags. The document timestamps its data to September 2026. If that is a typographical slip for 2025, the year-over-year comparisons hold. If it is not, the entire dataset is forward-looking, and forward-looking vendor data is a different product category with different reliability characteristics. An analyst who will not resolve a four-character date discrepancy will not resolve an address-clustering ambiguity either. I read the metadata before the claims.
Regulatory exposure is unaddressed. A $482 billion cross-border flow touches foreign exchange controls, anti-money-laundering rules, and sanctions screening at every hop. The report discusses none of it. KYC posture, legal structure, reserve attestation: all blank. I have written before about custody providers marketing decentralization while running unpatched legacy banking infrastructure. The same gap applies here, one layer up.
Data does not negotiate; it only reveals.
Here is what the bulls got right, and it is more than the skeptics admit.
The framing itself is the contribution. By measuring stablecoins through payment throughput rather than market capitalization, the report performs a methodological correction the industry needed and resisted. The narrative that stablecoin growth requires continuous new issuance is false, and this is the first serious attempt to demonstrate that with numbers. The argument stands even if every decimal supporting it is wrong.
The second thing the bulls got right is subtler. The report was published at the precise moment supply growth stalled. Framing a flat-supply period as evidence that growth does not need issuance is, structurally, a narrative repair operation. But a repair operation that tells the truth is still telling the truth. Motive does not invalidate a finding.
Where the bulls are wrong is the strategic read. Payment companies wiring card rails and bank accounts into on-chain settlement is not crypto achieving adoption. It is incumbent financial infrastructure co-opting the rails. That pattern historically commoditizes the technology and concentrates value in the existing distribution network. For crypto-native payment projects, it is a competitive threat wearing the costume of a validation. This is the same logic that explains why PayPal built PYUSD: better to become a regulatory partner than to wait to be regulated. The incumbents are not joining the movement. They are absorbing it.
The next version of this dataset will either publish its filtering methodology or it will not. That single decision will tell you more about the reliability of the $482 billion than any subsequent quarter of growth. Until then, treat the figure as a directional indicator with an undisclosed error bar, and treat every valuation built on top of it as resting on a number nobody outside Allium can reproduce. Ask for the denominator. If it is not provided, the absence is the answer.