Section 04

The clock

Everything in the preceding sections is a snapshot. This section asks the question that decides whether any of it is worth acting on: how long does each advantage last, and what specifically ends it. The answer is shorter than it looks, and one number governs the rest.

A test worth applying before anything else

Strategy documents in this market are full of the word moat. It is used loosely, usually to mean "something we would have and a competitor would not". That definition is useless, because it is satisfied by anything at all — including simply having started earlier.

The distinction that does work is between two quite different things.

A moat is an asset that grows as you operate and that a funded, competent competitor cannot buy on any timescale that matters. A head start is a cost or a delay that a funded competitor simply pays.

The test is not whether something is difficult. It is whether difficulty can be converted into money by somebody who has money. A licence that takes fourteen months and two million euros is a head start: a competitor with fifteen months and two million euros arrives anyway. A record of every payment you have ever made to a given recipient is a moat, because no amount of money buys somebody else's history.

Applying that test honestly to this position is uncomfortable. Most of what looked like protection turns out to be a head start with a date on it, and this section puts those dates on one axis. But it starts somewhere else, because there is a clock that runs faster than any of them and it is not a competitor.

The clock that governs the others

Prices in this market fall on their own. Not because of any single competitor, any single announcement, or any regulatory intervention — they simply fall, year after year, in the published accounts of the firms doing the work.

This is the most important single fact on this site, and it is also the best-evidenced thing on it. It does not rest on our modelling, on a vendor's claim or on an analyst's projection. It is three sets of audited accounts.

THE ONLY MEASURED NUMBER IN THIS SECTIONWhat three listed firms kept out of every hundred euros they moved, at the start of a financial year and at the end of it.Each is disclosed by the firm itself. None of them announced a price cut; this is what the accounts show.020406080100120BASIS POINTS KEPT PER PAYMENTWise58 → 52 basis pointsdown 10% in one financial yearCorpay71 → 59 basis pointsdown 17% in one financial yeardLocal105 → 84 basis pointsdown 20% in one financial year, gross-profit basisThree firms, three business models, three sets of accounts, one direction. That consistency is why the rest of this sectiontreats ten to twenty per cent a year as a fact about the market rather than a forecast about it.SOURCED · C3 B1 B2 · 2026-08-11
Each firm's own disclosure of what it kept per unit of payment volume, at the start of a financial year and at the end of it. Three different business models, one direction.

Ten to twenty per cent a year, compounding. Stated as a claim: The master clock is not a competitor but arithmetic: take rates compress 10 to 20% a year, so the price halves in three to seven years.

It is worth pausing on what that does to a plan, because the effect is larger and faster than it sounds. A business intending to charge somewhere between sixty and a hundred and twenty basis points in 2027 — bounded below by what the cheapest credible competitor already charges today, and above by what an inattentive buyer still pays — is planning against a moving floor.

THE SAME PAYMENT, PRICED FORWARDThe band is what a business like this could charge, compounded at the rate measured above — ten per cent a year along the top edge, twenty along the bottom.Nothing in it is a forecast of a decision. It is what happens if nobody does anything.02040608010012020262027202820292030203120322033BASIS POINTS CHARGEDBy 2030 the band is 25 to 79By 2033 it is 13 to 64TO EARN THE SAME MONEY2026the baseline20301.5 to 2.4 times the volume20332.1 to 4.8 times the volumeThe band starts as an estimate — bounded below by what the cheapest credible competitor already charges, and above by what aninattentive buyer still pays today. The compression applied to it is not an estimate.MODELLED · A1 C3 B1 F1 · 2026-08-11
The same opening band compounded forward at the measured rate. The right-hand column is the consequence stated the other way round: how much more volume is needed to earn the same money as before.

The consequence is a treadmill, and it is the hardest constraint in this work. Corridor volume has to roughly double as fast as the price halves, simply to hold revenue flat. Growth in this business is not upside — a substantial part of it is the cost of standing still, and any plan that treats volume growth and revenue growth as the same line is counting it twice.

Notice what the compression does not depend on. Not on anyone entering the market, not on a rail being commoditised, not on a regulation landing. It happens whether or not any of the events in the next figure occur — which is why those events matter less individually than they appear to.

Every defence, with the event that ends it

Now the rest of them. Each candidate advantage below has been dated to the earliest published event that removes it: a scheme go-live, a regulation entering force, a competitor capability becoming standard. Where no such event exists the bar runs to the edge, and that should be read carefully — no dated end is not the same as lasting forever. It means nobody has yet announced the thing that ends it.

MOAT OR HEAD START — THE SAME QUESTION, PUT ON A TIME AXISEach bar runs from today to the earliest published event that removes the defence. A bar reaching the right edge has no dated end, which is not the same as lasting forever.Two of the nine are things we would have to build. One was given up by choosing this position at all.20262027202820292030203120322033The record of who we have paid beforeno external clock — and we do not have it yetCompliance as a fixed cost to clearscales with us; bars the next startup, not the incumbentsMeasurements nobody else publishesa scheme could mandate publication — none has announced itOwning the conversion offshoreno event removes it; the price it earns compressesThe licence stack as a barriera payment institution will run token routes on notificationCorridor coveragethe schemes’ own interlinking programme targets go-liveCharging less than the incumbentthe price halves on the arithmetic aboveChoosing the route wellsuppliers ship good-enough routing within about two yearsHaving no rail of our own to favourforfeited the day we become principal on the conversionMODELLED · A1 D2 B4 B5 C3 C7 · 2026-08-11
Nine candidate defences against the events that remove them. Bars reaching the right edge have no dated end; two of those are things that do not exist yet.

The same content follows with the reasoning attached, because why each item sits where it does matters more than the ordering.

The defenceWhich is itWhat ends it, and roughly when
The record of who we have paid before — verified recipient records, identity checks, sanctions decisions with the reasoning kept, and what actually happened on every past payment to that recipient Moat. The only one with no external clock Nothing external. Only our own failure to capture it from the first transaction. It cannot be retrofitted, because it is a decision about what to record rather than a feature to add later, and the internal deadline is roughly twelve months from first volume.
Compliance as a fixed cost to clear — six hundred thousand to one and a quarter million euros a year before a single transaction, three-quarters of it fixed at low volume Head start, and only against those behind us Nothing removes it, but it bars the next startup rather than the established firms. Every competitor named anywhere on this site already pays it.
Measurements nobody else publishes — how long payments really take at the tail, how often the payee-verification check disagrees, what the per-institution value limits actually are Moat, small — defensible because it decays and must be re-measured A scheme or regulator mandating publication; none has announced one. Separately, any incumbent with the volume could sell this data today and none does, because it is not their business — a commercial fact rather than a structural one.
Owning the conversion offshore Revenue, not protection No event removes it. The price it earns compresses at the rate above, which is a slower death but not a different one.
The licence stack Head start Ten to sixteen months and one to two million euros of first-year cash — which is exactly what a funded competitor pays. There is a dated step-down too: from around 2028 a payment institution will be able to run token-settled routes on forty working days' notification, roughly halving the stack.
Corridor coverage Head start, short The payment schemes' own cross-border interlinking programme, targeting go-live in 2027.
Charging less than the incumbent Rent, not advantage The arithmetic in the previous figure. Around 2029 on the central case.
Choosing the route well Not a defence at all Suppliers ship good-enough routing themselves within about two years. This has already happened once, in card payments, to an entire category of companies.
Having no rail of our own to favour Forfeited by choosing this position The day we become principal on the conversion, which is the first day. It is the correct trade — but it should be made deliberately, and we should stop citing it.

The defence that points the wrong way

One argument recurs in this market and sounds strong: in corridors where the currency cannot be traded offshore, permission to convert it is legally scarce, and legal scarcity is the best kind of barrier there is. It is hard to argue with, and it is true.

It is also, on inspection, an argument for somebody else.

WHERE THE LAW PROTECTS THE CONVERSION, IT IS NOT PROTECTING USAcross, how hard it is for anyone to obtain permission to convert the currency. Up, how far we could actually hold that permission ourselves.A defence worth having would put a corridor in the top right corner. Nothing is in it.THE CORNER A MOAT WOULD LIVE IN — EMPTYscarce permission that we could nevertheless hold ourselvesthe nearest thing to it is capped per transactionOPEN TO ANYONERESERVED TO BANKSHOW SCARCE THE PERMISSION TO CONVERT ISNOT OURSFULLY OURSMexiconothing bars anyone, so nothing bars usBrazilours to hold, up to about €86,500 a timeKenyaa partner’s to earnPhilippinesreserved to banks at any commercial sizeNigeriathe bank converts and the regulator caps the markupRead the diagonal. The corridors where the conversion is legally scarce are exactly the corridors where somebody else holds it, andthe one corridor we can own outright is open to everyone. The protection is real; it points the other way.MODELLED · D4 A1 E2 · 2026-08-11
The four priority corridors and one comparison, placed by how scarce the conversion permission is and by how far we could hold it ourselves. The two are almost perfectly opposed.

Read the extremes. In the Philippines, selling currency at commercial size is restricted to banks and their subsidiaries — a genuine, durable, legally enforced barrier, enforced against us. In Nigeria the bank performs the conversion and the regulator caps the markup, so even the residual margin is administered. Brazil is the one corridor where a foreign-controlled non-bank can lawfully own the whole conversion, and it is capped per operation at roughly the point where business payments begin.

And Mexico, the corridor this work recommends entering first, is available to us for precisely the reason that it is available to everyone: the peso settles through CLS and is deliverable offshore, so no permission is required and none is scarce.

Where the conversion is legally protected, we are on the wrong side of the protection. Where we can own it, it is not protecting anything.

This is not an argument against Mexico. The money there is real and it is an order of magnitude larger than anything routing earns. It is an argument that entering Mexico buys revenue rather than protection — and that whatever protection this business ends up with has to be built out of something else while that revenue is arriving.

The one threshold that changes what kind of business this is

There is a way through the compression, and it is not an idea. It is a number.

The price at which a firm can buy wholesale currency improves as its volume grows — not smoothly, but in tiers. The distance between the bottom tier and the top of that ladder is worth considerably more than any routing gain discussed anywhere on this site: tens of basis points, against the thirteen to seventeen that choosing the best available route earns.

That matters because it changes the kind of advantage being held. Today the margin comes from the customer not having shopped around, which is why The core economics are procurement-inattention rent rather than a manufacturing advantage. Rent is defended by being the firm the customer buys from, and by the customer not knowing what the wholesale price was. Neither a licence nor a router protects it, and the compression above is exactly the sound of it being competed away.

Cross the volume threshold and the margin instead comes from manufacturing the conversion more cheaply than a competitor can. That compresses too, but from a floor the competitor cannot reach. It is the transition from a three-year position to a ten-year one, and it is the only such transition identified anywhere in this research.

No plan in this work has named a date for crossing it. That absence is the most consequential thing missing from the planning, and it is recorded as an open question rather than quietly resolved.

What this section does and does not settle

Taken together: The position clears the feature bar decisively and does not clear the durable bar — a good three-to-seven-year business whose one path to durable runs through volume rather than cleverness.

Both halves of that are meant. A three-to-seven-year business that clears the feature bar is a real business, and a materially better base case than the comparable category in card payments ever managed — that category produced no public listings in fifteen years, on more than half a billion dollars of funding. This category does produce companies. It produces compressing, capital-hungry, low-multiple companies that are eventually bought by larger ones.

What follows from that is a question about strategy, which section 07 answers and this section only supplies the evidence for. A reader who thinks the compression rate is wrong, or that a defence here has been dated too early, should disagree at this point — before meeting our reading of it, and against three sets of accounts rather than against our argument.