The number everyone is quoting is 77.5%. The number that actually matters is 1.2%.
Chainalysis published a cross-border stablecoin flow dataset this week. Total volume: $220.3 billion, up 77.5% year over year. Two lines deeper in the same release sits the figure the coverage skipped — 4,708 new inter-country transfer paths, carrying a combined $2.64 billion.
Run the division. $2.64 billion across 4,708 corridors averages roughly $561,000 per path. That is not a corridor. That is a test transaction with a country tag stapled to it.
The growth is real. It is also structurally narrow. The gap between those two facts is where the trade lives.
Context: a measurement, not a protocol
Start with the calibration. This is a single-source industry note. One data provider, one disclosure, no protocol architecture, no token model, no team, no audit. Chainalysis is one of the largest on-chain analytics firms, it sells compliance and monitoring products to exchanges, banks, and governments, and its geographic attribution is a proprietary process that has never been fully open-sourced. Every number below inherits those properties.
That does not make the data wrong. It makes it unverified by a second party, which in my world is the same thing as unaudited. Trust the audit, verify the stack, ignore the hype.
The comparative frame is useful for scale. The World Bank puts global remittance flows somewhere north of $800 billion annually. SWIFT settles in the neighborhood of $150 trillion per year across its messaging network. Against SWIFT, $220.3 billion is less than 0.2%. Against the remittance market, it is a clear minority but a fast-growing one. Both statements are true. Both are routinely omitted from the headline.
Now define the unit of measurement, because the whole report turns on it. A cross-border transfer path is a directed pairing — country A to country B — inferred from the geographic tag attached to the sending and receiving service. Path count measures corridor breadth. Path value measures corridor depth. The two are not the same metric and they are not moving at the same rate.
What the dataset does not contain is equally important. There is no breakdown by chain, no split between B2B settlement and person-to-person transfer, no disclosure of how exchange-to-exchange internal transfers are treated, and no stated period definition for the growth figure. That last omission is a real defect — the disclosure is dated September 25 with no year attached, so the 77.5% may already be stale arithmetic by the time it is quoted.
Core: the arithmetic of concentration
4,708 new paths. $2.64 billion carried. $220.3 billion total. The three numbers define a ratio: 1.2%.
That means 98.8% of all cross-border stablecoin value moved along corridors that already existed before this measurement period. Growth of 77.5% happened almost entirely on rails that were laid years ago. This is not network expansion. It is throughput deepening on an existing network.
Break it into a script, because the arithmetic is the entire argument:
total_flow = 220_300_000_000 # $220.3B
new_flow = 2_640_000_000 # $2.64B
new_paths = 4_708breadth_ratio = new_flow / total_flow # 0.01198 -> 1.2% avg_path_size = new_flow / new_paths # ~$561K per corridor ```
Two outputs. A breadth ratio that says the long tail is dormant, and an average corridor size that says the new tail is not yet substantial enough to register on the tape.
Now put 4,708 against the theoretical maximum. Roughly 200 sovereign jurisdictions produce a directed matrix of around 40,000 country pairs. 4,708 new paths is on the order of 12% of that space — and those paths carry 1.2% of value. The number of corridors is growing. The economic weight of those corridors is not yet moving with it.
This is the signature of trunk-route economics. Liquidity attracts liquidity. Pre-existing corridors with FX depth, established compliance relationships, and merchant integration absorb the marginal dollar faster than any new corridor can be created. A new route has to be built twice — once by the sender, once by the receiver — before it carries a single dollar. An old route only needs to be deepened.
Where the 98.8% actually moves
The dataset does not name chains. But the payment-flow footprint of stablecoins is not a mystery to anyone who has run a settlement desk. USDT on Tron has dominated high-frequency, low-ticket transfer volume for years because the fee structure is predictable and cheap relative to Ethereum mainnet. USDC on Ethereum and its L2s carries the institutional and DeFi-adjacent legs. Solana has been capturing retail settlement throughput. Whichever way the split lands, it lands on a small number of settlement layers.
That produces a second-order consequence nobody quotes. If 98.8% of cross-border value is concentrated on a handful of corridors, and those corridors settle on a handful of chains, then the global payment network claim is being carried by a stack of maybe three chains and a few dozen corridor pairs. That is not a distributed network. That is a hub-and-spoke system with crypto branding.
Gas demand is the quiet beneficiary. Every dollar that moves on Tron or Ethereum or Solana pays a fee in the native asset, and that fee accrues to block producers and validators. Cross-border stablecoin growth is, structurally, a demand event for the settlement layer. It is not a demand event for any application token that happens to sit on top of it. Yield is the interest paid for patience and risk — and the network fee is the toll paid before any yield can exist.
The methodology problem
Here is the technical claim that deserves more scrutiny than it is getting. Chainalysis assigns a country to a transfer primarily through address clustering and service-provider tagging. The tag library attaches a jurisdiction to an exchange, a custodian, or a payment processor based on where that entity is registered or licensed.
That is not the same thing as the physical location of the user. A trader in Warsaw moving USDT to an exchange registered in Seychelles generates a Poland-to-Seychelles flow on the map. A market maker rebalancing inventory between two of its own desks generates a cross-border flow with no payment intent behind it at all. A settlement desk routing order flow through a liquidity venue in Singapore books that leg as a cross-border transfer.
The result is that cross-border payment volume and cross-border trading and market-making volume may be blended inside the same headline figure. If that blending is material, the $220.3 billion overstates genuine B2B and B2C payment demand, and the 77.5% growth rate is partly a reflection of exchange inventory management rather than merchant settlement.
Chainalysis has not published the exclusion rules. Until it does, the number is a measurement with an unknown error bar.
I have seen this pattern before. In 2018 I spent roughly 120 hours tracing variable dependencies in the Solidity v0.4.24 implementation of an early collateralized debt position contract, hunting for an integer overflow in the oracle feed calculation that could have drained collateral during a flash crash. The documentation described a clean price feed. The code described something with an unhandled edge case. The documentation was not lying — it was simply not describing the mechanism. Code doesn't lie. Prose does.
Who captures the value
Follow the flow of funds, not the flow chart. When cross-border stablecoin volume grows, the most certain beneficiary is the issuer, and the mechanism is the float.
A stablecoin issuer holds reserves — predominantly short-dated Treasury instruments — against tokens in circulation. It pays nothing, or a token amount, to holders. The yield on that reserve portfolio accrues to the issuer. Every dollar of cross-border flow that expands circulating supply expands the reserve portfolio, and therefore expands what is functionally a modern seigniorage stream. This is the single most structurally certain value capture point in the entire chain, and it is not a token. It has no price chart and no unlock schedule.
The second beneficiary is the settlement layer, through gas and priority fees. The third is the compliance infrastructure vendor — which, in this case, is the firm publishing the data. That is a structural conflict worth naming plainly. A company that sells monitoring and compliance products publishes a dataset highlighting high concentration across a small number of corridors and jurisdictions. Concentration is precisely the condition that makes monitoring valuable and regulation plausible. The data can be accurate and the framing can still be directional.
The fourth category is the one the market keeps trying to price and cannot: nothing. There is no tradable instrument inside this dataset. No protocol, no governance vote, no emission curve. Anyone building a position off this headline is building it on a narrative, not on a cash-flow claim.
That is not a criticism of the dataset. It is a statement about what the dataset can support. It supports a thesis on infrastructure concentration. It does not support a token thesis, because there is no token in the text.
The dispersion test, five thousand euros, 2020
I have run this class of experiment before at a smaller scale, and it taught me the difference between a model and a market.
In 2020, during DeFi Summer, I allocated €5,000 into a Curve ETH/USDC pool specifically to test whether automated daily rebalancing beat static holding once impermanent loss and gas were priced in. I wrote a Python simulation, then deployed the strategy live, because the simulation is not the test. The gas bill is the test. Over three months the rebalanced strategy beat static holding by roughly 14% during the volatile stretch, netting around $800. The theoretical edge was real. The realized edge was smaller, because every rebalance is a gas event paid in a volatile asset.
The lesson transfers directly. The theoretical claim — stablecoins are expanding into new markets — can be true in corridor count and false in economic weight. You have to price both sides. The 4,708 new paths are the theoretical expansion. The $2.64 billion is the realized expansion, after compliance friction, FX depth, and counterparty onboarding are subtracted. The realized number is 1.2% of the story.
The 2022 signal, applied
In May 2022 I exited stablecoin positions roughly 48 hours before the Terra depeg was fully priced, not because I had a view on a team, but because the on-chain stablecoin inflow pattern into a small number of venues was inconsistent with the mechanism the protocol claimed to run. That mechanism required continuous incentive issuance to hold the peg. The chain showed the incentives being consumed faster than the mechanism could absorb. The structural weakness was visible in flow, not in sentiment.
The same discipline applies to concentration. A system where 98.8% of flow runs through a small set of corridors is efficient, and efficiency is the same thing as fragility viewed from a different angle. One regulatory action against a dominant corridor pair, or one congestion event on a dominant settlement chain, transmits to the whole total. The dataset shows resilience in the aggregate and concentration in the anatomy. Those are not contradictory findings. They are the same finding read at two different resolutions.
Contrarian: the narrative runs backwards
The prevailing story is that stablecoins are democratizing cross-border payments — that the rails are spreading outward, that the unbanked are being reached, that the network is broadening. The data says the opposite. The network is deepening, not broadening.
Here is the smart-money reading. If 98.8% of value rides existing trunk routes, the marginal dollar goes to whoever already owns those routes: the large issuers, the dominant settlement chains, and the compliance vendors who price access to the corridors. The retail reading is that 77.5% growth means the whole space is expanding and any project adjacent to it will benefit. That reading buys the long tail, which carried 1.2% of the flow.
The blind spot is the average corridor size. A new path averaging $561,000 is not evidence of adoption. It is evidence of a pilot. The market is treating 4,708 as a network effect. It is a contact list.
There is a sharper angle underneath. This concentration data is being published by a firm whose revenue depends on the perceived need for compliance monitoring. The most commercially useful version of this dataset is one showing high concentration on few corridors with mixed jurisdictional exposure. The most surprising version would be one showing breadth expanding and geography dispersing. You may reasonably ask which version gets the press release. The market rewards those who read the source code — and equally those who read the source's incentives.
Takeaway
Watch one number: the breadth ratio. New-path value as a share of total cross-border flow. Today it is 1.2%. If it clears 5% over the next two to four quarters, the dispersion narrative becomes real and long-tail corridor beneficiaries start to matter. If it stays in the low single digits, the trunk-route winners keep winning and every stablecoin-for-the-unbanked pitch remains a story about a test transaction.
Two secondary signals. First, the treatment of exchange-to-exchange transfers in the next Chainalysis release — a methodology change there moves the entire series. Second, the licensing frontier, because MiCA in the EU, the US stablecoin framework, and the Hong Kong, Singapore, and UAE regimes all land directly on these corridors. Regulation is the mechanism that either forces dispersion or locks in the incumbents.
The $220.3 billion is a real number, measured by one vendor, with an undisclosed methodology and a commercial interest in the framing. The 1.2% is the same number, read properly. Which one you trade depends entirely on whether you price the mechanism or the headline.
