Ten Jurisdictions in Detail: What Each Licence Lets You Do

Part I · The Map (Chapters 0–1) Builds on: Chapter 0 (what each of the six columns covers, the three rules, the coordinate vocabulary and marking conventions; the 10×6 reference grid in its Appendix A is what this chapter opens up row by row) New concepts in this chapter: how passporting works, the line between a PI and an EMI, mobilisation, the severing of direct connections

1. Jurisdiction by Jurisdiction¶

This chapter goes jurisdiction by jurisdiction. If what you are after is the businesses and the products, you can skip it.

The reference grid in Appendix A of the previous chapter gave each of its 60 cells one line, and the three rules arrived as conclusions with the evidence stripped out. This chapter takes every cell apart: for each licence, the regulator, what it lets you do, where it stops, and who holds it — the evidence for the rules is all here. The ten sections stand alone and can be read as look-ups. The EU section is the longest, because the line between a PI and an EMI (question two) and why a bank must stand behind an EMI (question three) are two mechanics the whole course reuses.

2. The EU: One Licence, the Whole Bloc¶

The EU is the only one of the ten jurisdictions where the country that licenses you is not the same thing as the territory you may trade in: get authorised in any member state and passporting carries you across the whole EU/EEA, about 30 countries. That is why Banking Circle made its home in Luxembourg. Marqeta, the US card issuing platform, took the same route into Europe — rather than apply for its own licence it bought TransactPay, an authorised e-money institution, and covers the continent on that one licence.

Question one: what business a single EMI can hold up

The entire front end of a neobank. Open wallets and issue users a balance; issue debit and prepaid cards (an EMI can join Visa and Mastercard as an issuing member in its own name, or ride a BIN sponsor); collect and pay out over SEPA; acquire; remit and convert currency. Revolut before it got its banking licence, and Marqeta's TransactPay to this day, are pure EMI operations. Since 2025 that front end has grown by one more piece: the EU changed the law to let PIs and EMIs participate directly in central bank clearing (T2 for large value, TIPS for instant — eligible from 2025-04-09, with access live from that October), so an EMI no longer has to hang off a bank. This is the EU's statutory version of Rule 3 in the previous chapter, the clearing door sinking.

There is exactly one thing it cannot do: touch the customers' money. Balances have to be segregated 1:1 (parked at a bank or held in government-grade assets), cannot be lent out and cannot pay interest, which locks the business model onto the fee side — the FX spread, interchange (the issuer's cut of card spend, paid over by the acquirer), subscriptions — and shuts it out of a bank's net interest margin. That also explains the standard evolution path for a neobank: start on an EMI to prove out customer acquisition, then trade up to a banking licence once the scale is there to earn a spread. Revolut took exactly that route, on a Lithuanian banking licence. The boundary holds across all nine jurisdictions that have a payment licence at all (Switzerland has none and runs a different line — see the Swiss section below): the Philippines' GCash likewise earns from segregated balances plus fees, and has to stand up a separately licensed company to lend.

Question two: with only a PI and no EMI, what can the product look like

The short answer first: no, you cannot build the stored-value wallet where money is topped up and then spent down at leisure — issuing a balance is issuing e-money, and that capability belongs to the EMI alone. The line between a PI and an EMI comes down to one question: may the money stop and sit as stored value. Client funds in a PI's hands are only ever allowed to be in transit as part of executing a payment; they cannot become a standing balance.

But no stored value does not mean no product. A PI's products come in three shapes, ordered by how much they touch the money:

Account-shaped already looks like a wallet — and the real line is not in the interface, it is in what the money is.

An EMI's balance is stored value that has been issued: the user pays money in and gets back a claim they can hold indefinitely and spend with any third party at any time, so the sitting still is itself the product. Money in a PI account is always a payment in transit: every pause has to attach to a payment purpose, and when the regulator comes asking, the PI has to be able to say why this money is sitting here.

Card issuing splits the same way. A PI's card is an instrument for initiating a transfer out of a payment account, the construction of a bank debit card. An EMI's card spends a stored-value balance, the construction of a prepaid card. The tap is identical; what gets written on the ledger is two different things.

The mnemonic: a wallet sells the stop, a payment account sells the pass-through. Whether the stop is itself the product is the line.

How thin that line is: one company, Wise, runs the same multi-currency account as an EMI through its UK entity and as a PI through its EU entity in Belgium — thin enough to stand on both sides, with the call turning in practice on how the product is characterised and how each national regulator reads it. The EU's legislative direction is to stop maintaining the line at all: PSD3 repeals EMD2 and folds EMIs into a single payment institution framework (see the reform note at the end of this section). Two licences become one, and after 2028 this question becomes a historical one.

Question three: EMIs can issue balances now — so why must a bank still stand behind one

The short answer first: an EMI cannot be a settlement bank. A settlement bank is the bank that holds funds on behalf of another institution and completes the final transfer for it inside the clearing system — other people's money comes to rest in its accounts. An EMI has to park even its own customers' funds at a bank, so it is in no position to be the resting place for anybody else's. Issuing a balance is only bookkeeping; the money itself still needs somewhere to sit and rails to move on, and an EMI has neither. It rents both from a bank. Broken out, that is three things:

1. A resting place for the money. The balance an EMI writes for a user is an IOU, the EMI's liability to that user, and the regulator requires the real money behind the IOU to be segregated 1:1 — held in a segregated account at a bank or in government-grade assets, and nowhere else. So behind every EMI licence there stands at least one safeguarding bank: the balance the user sees is on the EMI's ledger, and the money itself sits at the bank.
2. A seat in clearing. A balance has to actually take money in and pay money out to the outside world: the user tops up (SEPA in) and withdraws (SEPA out), and both legs run through a clearing system. Traditionally an EMI had to route through a bank. Since 2025 it can apply to participate directly in T2 and TIPS (see question one), but that covers the euro only and the operational bar is high; for every other currency — Danish krone, Swedish krona, Polish złoty, and sterling and Swiss francs outside the bloc — it still has to go country by country finding a bank to connect it to clearing. Renting from a rail wholesaler like Banking Circle, a bank that packages multi-country clearing access and sells it on, gets the lot in one stop.
3. The funding execution behind card settlement. TransactPay is an issuing member of the card schemes, and membership settles the question of who is entitled to issue and who answers to the scheme. But the daily net settlement with the scheme is a real movement of money, and it has to be executed through a bank account.

The division of labour between Banking Circle and Marqeta is cut along exactly those three lines: TransactPay supplies the name (issuer identity, the balance ledger), Banking Circle supplies the sitting and the moving (accounts across 30 countries, the resting place for the money, local clearing rails). In the language of Rule 1: the EMI sits at tier ②, the bank at tier ④ — an EMI can keep the books, and that does not change the fact that there is still a layer above it. Whether TransactPay's settlement bank with the card schemes and its safeguarding bank are in fact Banking Circle is not named in any public source [?].

Reform in flight: PSD3/PSR reached political agreement on 2025-11-27 and the final compromise text was published on 2026-04-23, with publication in the Official Journal expected in the second half of 2026 and application from around 2028. At that point EMD2 is repealed and EMIs fold into a single payment institution framework. Anyone applying today still applies under PSD2/EMD2.

3. The UK: Where the Door First Sank¶

Its own bloc since Brexit: the licence structure is inherited from the old EU regime (PI and EMI still separate, and the PSD3 merger has nothing to do with Britain), but on two counts it leads the world — in 2018 it was the first anywhere to let a non-bank participate directly in central bank clearing, and in 2026 it pulled the whole crypto line into an FCA authorisation regime.

4. Switzerland: The Country with No Payment Licence¶

What marks the Swiss row is its empty cells: no EU-style payment or e-money licence, and no unified crypto licence either. Moving money for a living runs on an AML self-regulatory registration, and the moment holding client funds crosses a red line you land straight in the Banking Act. That gap is exactly what produced Switzerland's most distinctive licence — the FinTech licence.

Reform in flight: on 2025-10-22 the Federal Council opened a consultation on amending FinIA (closed 2026-02-06), proposing two new licences to fill the two empty cells above — a payment institution licence (replacing today's FinTech licence, dropping the CHF 100 million cap and allowing Swiss stablecoin issuance) and a crypto institution licence (custody and trading of crypto-assets). In force by 2027 at the earliest [?] — a consultation-stage estimate, not a committed date. How existing FinTech licence holders transition is left open in the draft [?]; the Fiat24 analysis later in the course takes today's 1b licence as given.

5. The US: Two Tracks, and One Licence per State¶

The US has no single national financial regulator. A bank charter comes on two tracks, federal or state, take your pick; the payment licence has no federal version at all, so moving money means filing state by state across nearly 50 states. The key to reading the US row: whether you are a bank can be answered at either level, but whether you may move money is for the states alone to say.

6. Singapore: One Act, Seven Activities¶

Singapore (regulated by MAS, the Monetary Authority of Singapore) put all seven payment-related activities — opening accounts, domestic transfers, cross-border remittance, acquiring, issuing e-money, digital token services and currency exchange — inside a single Payment Services Act: one licence, tick the activities you need, two bands by size. It is the most integrated design in the ten jurisdictions.

7. Hong Kong: Front-runner on Crypto Issuance¶

Hong Kong has moved fastest of anyone in the crypto issuance column: the stablecoin ordinance came into force 2025-08-01, and the first two issuer licences were granted 2026-04-10. Its banking cell keeps a distinctive three-tier system of deposit-taking institutions, and its payment cell is the textbook case of slicing by product — one licence for stored value, another for currency exchange, under two different regulators.

8. Mainland China: Forced Intermediation, the Reverse Case¶

The Mainland China row reads backwards from everywhere else. Its payment cell was once the largest open experiment in the world (Alipay, WeChat Pay), but the direction of the rules is forced intermediation: every payment institution has to go through a single clearing organisation, and a direct connection to the central bank is not on the table. Its two crypto cells are the only pair among the ten jurisdictions that are both banned outright.

9. Brazil (South America, Part One): Payment Institutions Wired Straight into PIX¶

Brazil is the largest economy in South America, its licensing regime is complete and written down, and it has one design feature that is rare elsewhere: a payment institution does not have to hang off a bank, it can connect directly to PIX, the central bank's instant payment system. This is the soil the Nubanks of the world grew out of.

10. Argentina (South America, Part Two): An Ecosystem of Registrations¶

Argentina is the specimen case of an ecosystem built on registrations: no payment licence, only a PSP registration; no crypto licence, only a filing with the securities regulator. And yet, on the strength of one addressing invention, the CVU, it hung non-bank accounts off the national transfer network and grew a giant in Mercado Pago.

This row also carries a constraint no other one does: what really decides whether a business is viable here is not licensing but capital controls. Argentina has spent years restricting how residents buy foreign currency and how money leaves the country, and locally the whole apparatus is called the cepo (Spanish for "pen" or "corral") — it sets how many dollars an individual may buy each month and how long a company waits to pay an overseas supplier. So the first question in Argentine cross-border payments is not which licence to obtain but whether the money can be converted and got out at all, and when. That constraint is also what drove dollar stablecoin usage here to among the highest in the world. The cepo was largely dismantled in 2025-04, which loosened the limits, but FX remains managed.

11. The Philippines: The EMI Licence That Built the National Wallets¶

What to watch in the Philippine row is the payment cell: the EMI licence grew GCash and Maya, two wallets at national scale. The banking cell is run as a scarce quota (digital bank licences capped at 10 in total), and the crypto cell is frozen.

12. Wrapping Up¶

With ten jurisdictions read through, the three rules from the previous chapter have something solid under them: the four-tier yardstick is in the can-do and cannot-do of every section; geographic reach is in the contrast between the EU and the other nine; and the clearing door sinking is in the chronology of doors opening across the UK, Hong Kong, Switzerland, Brazil, Singapore and the EU. The next chapter starts the product half of the course, and the first company up is Banking Circle — a bank that turned taking the heaviest licence in every jurisdiction into a strategy.

13. Open Questions¶

1. Whether the OCC's final GENIUS rule appeared before the statutory deadline of 2026-07-18, and how far along the first federal stablecoin licence applications are [?] — this chapter verified only the proposed rule from early 2026 and the comment deadline in early May.
2. Whether Singapore's stablecoin legislation is in force [?] — officially it was still being drafted as of 2025-12, and "in force by mid-2026" is a press expectation.
3. The exact reference number and effective date of the Philippine SEC's CASP rules [?] — the text gives only the year.
4. The outcome of the three new Philippine digital bank applications [reported] — as of 2026-07 the central bank said they were still being assessed.
5. The parliamentary timetable for the Swiss FinIA amendment — "in force by 2027 at the earliest" is a consultation-stage estimate, not a committed date.
6. Whether TransactPay's card scheme settlement bank and its safeguarding bank are Banking Circle [?] — the BC × Marqeta announcement describes the division of labour as BC supplying accounts and rails, and never names a settlement bank.
7. The legislative milestones for Hong Kong's supplementary custody and OTC regimes beyond the VATP [reported].
8. The exact reference numbers for Argentina's PSAV capital and compliance rules [?] — the text gives only the year.

14. Sources¶

• EU PSD3/PSR timeline: Norton Rose Fulbright, Morrison Foerster; PI/EMI direct participation in clearing: the ECB's access policy, ECB Decision (EU) 2025/222
• The UK crypto authorisation regime and stablecoins: the FCA's new cryptoasset regime, FCA policy statements overview, FCA press release; Revolut's full UK licence: Revolut's own announcement
• GENIUS implementation in the US: the OCC's proposed rule (Federal Register), OCC Bulletin 2026-3, Gibson Dunn's analysis; Wise's trust charter refusal: OCC decision CD#1381
• Hong Kong's first stablecoin licences: HKMA press release (2026-04-10), Charltons, Central Banking
• The Philippines: the BSP's VASP freeze, digital bank licence progress (Philstar), ICLG Fintech 2026, Philippines, Wise's InstaPay direct access announcement
• Brazil: the BCB's official note on the new virtual asset rules, Chambers analysis, Res. 561 and eFX (Ledger Insights), Wise's Brazilian payment institution licence announcement
• Singapore: the MAS payment licensing page, MAS on digital token service providers, the MAS stablecoin framework, Wise's FAST access announcement, BC Payments' MPI licence
• Switzerland: Bär & Karrer: Fintech 2026, Switzerland, Chambers Fintech 2026, Switzerland, the FINMA FinTech licence page
• The [reported] items in this chapter (Hong Kong's extended custody and OTC regimes, the offshore renminbi stablecoin discussion, Argentine stablecoin usage, the Philippine digital bank applications, the expected commencement date in Singapore, Revolut's Argentine acquisition) come from mainstream financial press and industry reports rather than primary sources; they are not linked line by line, and the Open Questions list carries the corresponding numbers.
• The Mainland China and Argentina sections, and every other unmarked fact (what each licence may statutorily do, capital thresholds, which institutions hold what, and clearing access rules), come from the regulators' own websites and the text of the law.

Previous: Chapter 0 · Overview: What Payment Licences Exist Next: Chapter 2 · Banking Circle: The Heaviest Licence in Every Jurisdiction