Frontier Radar: The Four Base Layers Being Replaced

Extension · Frontier (Chapters 30–31, optional reading after the main line) Builds on: Chapter 3 (the hierarchy of money), Chapter 4 (what the merchant gets for the fee), Chapter 6 (institutions determine form), Chapter 7 (the ISO 8583 lesson), Chapter 10 (instant clearing), Chapter 19 (local-in, local-out), Chapter 26 (tokenized deposits) New concepts in this chapter: ISO 20022, instant payment interlinking, wholesale and retail CBDC, Pay by Bank

(Same freshness caveat as the previous chapter: the timelines and project rosters below come from public information as of mid-2026. Check the current state before you rely on any of it.)

1. The Question the Previous Chapter Left Open¶

The previous chapter covered change on the demand side: the payer turning from a person into an agent. Over the same stretch, four base layers on the supply side are being replaced too — the message standard, rail interlinking, the form central bank assets take, and the old idea of bypassing cards, back from the dead on new infrastructure.

None of the four makes headlines the way agents do, but each one lands on a main line of this course. Take them one at a time.

2. The Message Layer Turns Over: ISO 20022 Replaces SWIFT MT¶

Where it plugs into the grid: the card transaction lifecycle covered ISO 8583 — the card message standard from 1987 that rules card transactions to this day, whose field limits became today's product limits. The cross-border version of the same story has now reached its change of act.

SWIFT's MT message format was designed in the 1970s, and its core fields are free text: the payer's name and address are a few lines of whatever someone typed in. That manual review queue from the four cost sources and the compliance skeleton? To a large degree, those few lines of free text are what built it.

A screening engine can only fuzzy-match a block of unstructured text, and the false positives pile up into mountains.

The turnover has already happened: in November 2025, MT messages were retired on SWIFT's cross-border payments channel in favor of ISO 20022. (MT is still in use in other domains — securities, trade finance; what retired is the cross-border payments domain.)

The new standard is structured: payer, payee, address, invoice number each get a field of their own. The next hard deadline is November 2026 — from that point, fully unstructured addresses will be rejected; the hybrid format remains usable, but town and country must go into structured tags.

Verdict: this is the positive print of the lesson from the card transaction lifecycle — the message standard sets the ceiling on what the product can do.

ISO 8583's fixed-length fields saddled card products with forty years of constraints. ISO 20022 runs the same causality the other way, cutting two labor bills that were always expensive: the compliance review that shows up as delay in the four cost sources, and the line-by-line matching work in ledgers and reconciliation.

(Note what it does not cut: the two biggest items of the four. The FX spread and prefunded liquidity are beyond its reach.)

It makes no news, but it changes the unit cost of every cross-border transaction that rides SWIFT.

3. Interlinking the Rails: Instant Payments Go Abroad¶

Where it plugs into the grid: domestic and global described a gap — at home, every country has real-time pipes, fast and cheap; across borders there are only two roads, SWIFT (push) and the card networks (pull), both expensive. This section is about the attempt to fill that gap in.

Start from the starting point. Each country's instant system covers its own country or region and nothing more: Pix in Brazil, UPI in India, SEPA Instant in the euro area (the euro area's instant transfer scheme — ten seconds and the money lands). Fast, cheap, the whole population covered — and unusable one step past the border.

Interlinking means wiring two systems together directly. Singapore and India have already connected their instant systems (you remit to a phone number), and several clusters of Southeast Asian countries have made each other's QR codes mutually accepted.

But pairwise hookups don't solve the root problem — you should recognize it by now: this is the N² problem from clearing and settlement all over again.

Project Nexus builds the "center" that problem calls for: one set of standard interfaces, so that any two countries' instant systems can connect under the same specification — instead of every pair negotiating from scratch. It was incubated at the Innovation Hub of the Bank for International Settlements (BIS, the central banks' cooperative body); since 2025 it has been run by an independent entity jointly owned by five countries — India, Malaysia, the Philippines, Singapore, and Thailand — with Indonesia joining in 2026.

What it means for the cross-border landscape comes through clearest in the language of the Wise model. Break that model back into its three pieces: a domestic payment at each end; someone in the middle doing the FX; and that someone parking a position on each side so the two ends can join up. Wise's moat is that it built all three itself — opening accounts, winning licenses, feeding positions, country after country — heavy investment traded for an exclusive position on corridor after corridor.

Instant-system interlinking = local-in, local-out as public infrastructure. The first of the three pieces has been made public: the two domestic payments at either end now run on each country's public instant system — one interface, near-zero cost.

But the other two — the FX and the positions — did not disappear, and the central bank network has no plan to take them on: it holds no foreign currency, carries no FX risk; what it provides is interfaces and rules, not a pool of funds. Nexus's design turns that seat into a standardized role: private FX providers plug in under uniform rules, quote in the open, and compete with one another. The work Wise does still exists — what changed is the way in: from "build the whole stack or you don't get a seat" to "win the seat by bidding under published rules."

Verdict: once a corridor is covered by interlinking (instant at both ends, near-zero cost, one uniform experience), the middle square flips from an exclusive business opportunity into a standardized role you bid for. The old moat — the accounts at both ends, the licenses, the network itself — has been carted off by public infrastructure, and the only thing left to compete on is FX pricing. The window isn't closed, but it has narrowed: what you earn is no longer a scarcity premium; it is a spread, bounded by rules and contested in the open.

The chapter on what stablecoins solve found one common trait across stablecoins' strong-benefit scenarios: the old rails can't reach them. Interlinking is precisely the business of making the old rails reach more places.

This carries more weight than it appears to. Domestic and global said only two roads go global, and both got there by going around: SWIFT doesn't touch the money, the card networks don't touch the local leg, and the stablecoin road brings along a universal settlement asset of its own. Three detours, and none of them touches national clearing itself.

What Nexus-class projects are attempting is to not go around — to connect the world's clearing systems directly.

That chapter also explained why the earlier roads had to go around: unifying national clearing requires central banks to give up a slice of sovereignty, and for decades nobody pulled it off. So how far this road runs depends on something that has nothing to do with technology — how much of the interface the central banks are willing to hand over. That the project went to a six-country joint venture instead of staying at the BIS is itself one answer to that question.

Signal to watch: every corridor a Nexus-type project brings live rewrites one row of the benefit-scenarios table in the chapter on what stablecoins solve — from "Strongly" to "no longer needed."

4. Central Banks Take the Field: Wholesale CBDC¶

Where it plugs into the grid: the top of the hierarchy-of-money table in the hierarchy of money, plus the price-you-pay table in the chapter on what stablecoins are.

CBDC — central bank digital currency — has to be split into two kinds before anything else; discuss them mixed together and confusion is guaranteed:

Retail is contentious and slow; set it aside for now. Wholesale is the one to watch, because it strikes exactly at the vital point of that table in the chapter on what stablecoins are.

Recall the table: on-chain stablecoins got 24/7 operation, global reach, clearing and settlement completed in the same instant — at the price of the settlement asset being downgraded from a central bank liability to a private company's liability. At the time, the trade looked forced: if you wanted the on-chain form, only private issuers offered it.

Wholesale CBDC cancels the "forced" part: central bank liability (highest tier) plus on-chain form (clearing and settlement in one motion) — both at once.

Verdict: run the tier logic of the hierarchy of money forward — when availability and cost are comparable, whoever can settle in a higher-tier asset will not settle in a lower one.

That qualifier in the first half is not a courtesy. Correspondent banking today runs on commercial bank deposits rather than central bank reserves, and the reason is availability: not everyone gets to open a reserve account. Wholesale CBDC is precisely the narrowest-access asset of the lot — it exists only for institutions that can get onto the central bank's ledger, and that is exactly why stablecoins get to keep some territory.

If wholesale CBDC actually lands, the "cross-border settlement between institutions" row of the benefit-scenarios table in the chapter on what stablecoins solve gets reshuffled once more. Among the institutions that can reach it, the ordering will roughly follow the hierarchy: wholesale CBDC > tokenized deposits > stablecoins.

Stablecoin territory shrinks further in turn, concentrating where the public can reach it, in the cross-border long tail (remittance demand that is high-count, small-ticket, and scattered), and in weakly regulated territory — the places the central bank networks don't go.

Signal to watch: the moment wholesale experiments move from "test environment" to "live settlement."

5. Pay by Bank: The Old Idea of Bypassing Cards, on New Infrastructure¶

Where it plugs into the grid: interchange showed merchants handing over 2–3%; the motive to bypass cards hasn't changed in fifty years. Push and pull and the four US rails showed the rails have existed all along (ACH and instant clearing can both pay straight from a bank account). So why does the bypass never happen?

Because the obstacle was never the rail. It is the experience and the incentives at the two ends.

Pay by Bank means letting the customer pay at checkout straight from their bank account, no card network in between. For it to stand up, these are the missing pieces:

Two markets, side by side — the method is still the rule from interchange: see a difference, look for the institution.

Brazil proved the ceiling. Pix was pushed through by the central bank: institutions above a size threshold must connect, individuals pay nothing, and merchant rates sit at a few tenths of a percent.

Against the two-plus percent that cards take, the merchant-side math takes one glance. The user side gives up little too (Brazil's cash-back ecosystem was weak to begin with) and gets zero cost and instant arrival in exchange. The "what the merchant gets" table from the origin of cards adds up on both sides here.

The US displays the resistance — and at first glance the resistance is a paradox.

Swiping a card earns the user 1–2% cash back; switching to Pay by Bank means giving it up. But interchange already ran this math: the cash-back money doesn't fall from the sky — the merchant's fee gets baked into the sticker price, so cash back is paid for by all customers together, in the shelf price, including the ones who pay with cash or bank transfer. If it's their own money, and Pay by Bank saves that 2% outright so everyone's prices could come down — why won't it budge?

Because under a single sticker price, the money saved and the money given up land in asymmetric places:

And merchants almost never price differently by payment method — surcharging was blocked for decades by card network rules and some state laws, so the single sticker price became the default. Which means whoever switches first loses first, the payoff only arrives if everyone switches, and nobody has a reason to go first. That is not users being irrational. It is a coordination problem: everyone is waiting for everyone else to move.

Three counterexamples point at the same mechanism: when a gas station posts a lower cash price, customers really do choose cash; after Durbin crushed debit interchange, debit cash back all but disappeared; after the EU's IFR caps, the rewards ecosystem withered and account-to-account payment finally got room. Tear out the cash back first, and Pay by Bank can move; leave it standing, and individual rationality locks the status quo in place.

Beyond that, the user also gives up chargeback protection, and the central bank has no power to force banks to connect (the governance structure the four US rails walked through). So Pay by Bank in the US concentrates in bills and subscriptions — scenarios where the user doesn't notice the payment method and the merchant holds the bargaining power — not retail checkout.

Verdict: how far Pay by Bank goes in any given market comes down to three variables — how hard the regulator pushes, how much the user gives up in card perks, and whether the banks themselves have any incentive.

The third is the easiest to miss: what Pay by Bank eats is precisely the issuers' interchange revenue. So in markets with no regulatory mandate, banks have no reason to make this channel work well. The Netherlands' iDEAL is the counterexample — the banks built it jointly themselves, and the first two axes can't explain it.

Run most markets through these three variables and the pace is roughly predictable.

6. The Four Blocks, Back on One Map¶

Put the four together and they point the same way: the old rails are upgrading themselves, annexing the challengers' sharpest selling points — fast, cheap, structured, direct — one at a time.

The chapter on what stablecoins are described the incumbent's playbook: don't ban it — build a version of your own. This is the same playbook rerun at system scale — only this time, the incumbent is the entire central-bank-plus-banking complex.

7. Closing: The Objects Change, the Measuring Sticks Don't¶

What the main line handed you is a set of judgment tools. Everything these two chapters added — agents, the protocol contest, the message turnover, interlinking, CBDC, Pay by Bank — not one of them needs a new tool to be understood:

• Agent credentials are a narrowing of tokenization (the chapter on card-not-present);
• The protocol land grab is the N² problem plus neutrality ("clearing and settlement," "the four-party model," the chapter on where CPN sits);
• ISO 20022 is the positive print of the message standard sets the product ceiling (the card transaction lifecycle);
• Interlinking is local-in, local-out made public (the Wise model);
• Wholesale CBDC is the hierarchy of money re-ranked on-chain (the hierarchy of money);
• Pay by Bank is institutions and incentives determine form (interchange).

This set of measuring sticks is what the course was really built to deliver. From here on, every time a new name appears, don't ask what's new about it first. Ask: Whose liability is it? How far apart are clearing and settlement? Where did the cost move? Who bears the loss? Who sets the rules — and do they play on the field themselves? Ask all five, and its square on the grid develops on its own.

8. Self-check questions¶

1. Why does the ISO 20022 migration directly cut the cost item that shows up as delay in the four cost sources?
2. Using the tier logic of the hierarchy of money, explain why settlement between institutions might stop using stablecoins once wholesale CBDC lands.
3. Pix swallowed a huge share of Brazil's retail payments within a few years, while Pay by Bank in the US crawls. Using the table from the origin of cards and the method from interchange, name the decisive differences.

9. Answers¶

Answer for yourself before reading on.

1. That cost item mostly takes the form of queuing for manual review after sanctions-screening false positives. In an MT message the payer information is free text, so the screening engine can only fuzzy-match the whole block — high false-positive rate, long manual queue. ISO 20022 puts name and address into structured fields matched field against field: false positives fall, the queue shortens — and that is how the delay gets squeezed out. The side benefit lands on reconciliation in ledgers and reconciliation: structured data makes line-by-line matching automatable — fewer breaks, and they go to zero faster.
2. The tier logic: the higher a settlement asset sits, the harder it is, and whoever can use a higher tier won't use a lower one. Institutions accepting stablecoin settlement today are making a trade at bottom — swallow the downgrade to "private company liability" in exchange for on-chain speed and 24/7 operation. Wholesale CBDC is central bank liability plus on-chain form: both advantages at once, so the trade is no longer necessary. The moment it is available, large-value settlement between institutions re-queues by tier: wholesale CBDC first, tokenized deposits next, stablecoins retreating to the scenarios the central bank networks don't cover.
3. The difference is not technical; it sits in institutions and in the user's opportunity cost. Brazil: the central bank forced the whole industry onto Pix and made it free, and users had almost no cash back to give up anyway — the table from the origin of cards adds up on both the merchant side and the user side. The US: nobody can force banks to connect (the governance structure from the four US rails), and a user switching to Pay by Bank hands back 2% cash back plus chargeback protection with no compensation in return — the math from the origin of cards fails on the user side. So it grows only where users don't notice the payment method (bills, subscriptions). When two markets produce two outcomes, look for the institutional and incentive differences first; don't reach for "Americans just like swiping cards."

Previous: Chapter 30 · Agentic Payments: When the Payer Is No Longer Human Next: Appendix A · The Four-Quadrant Solution Comparison