Financial infrastructure is rarely built in one step. Licences, capital and trust arrive in sequence, and a decade is a realistic horizon. The instructive part is not the destination but the order in which things are acquired. Narrow-bank precedents show how phased builds work in practice, and what a long programme should learn from it.

A precedent worth studying

ClearBank is a useful reference point for the narrow-bank model: a clearing bank for fintechs in the UK and the EU. It demonstrates that a bank whose clients are other institutions can be a viable business, with settlement and clearing as the product. The example shows that serving institutions directly — rather than competing for end customers — is a credible path.

The phases

SUPA's own sequence is explicit.

  • Zone (2026–2027). Bring the flow. Units run on licensed partners' rails; no own licence needed to be useful.
  • Principal (2027–2029). Part of the stack moves onto own licences; capacity is bought as principal; direct correspondent relationships begin.
  • Hub (2029–2032). An own banking licence; institutions open correspondent accounts and sign the multilateral reliance agreement.
  • Liquidity cloud (2032–2036). The network becomes a shared balance sheet of participants.

Licensing as sequencing, not a single leap

The licensing route reflects the same logic. A UK HoldCo sits at the top; a Guernsey GFSC sandbox engagement is concierge, with an application in autumn 2026; the Central Bank of Ireland Innovation Hub and a Guernsey Financial Firm Business licence are on the path; an Irish EMI with EU passport follows, with the bank-jurisdiction decision taken later. Cayman was chosen for the first phase as the fastest, cheapest way to test the hypothesis, and the same structure is intended for deployment in Switzerland later.

What a decade should learn

Each phase should reduce the assumptions the next one carries. Own-licence infrastructure should follow demonstrated flow, not precede it. Reliance should start bilateral and become multilateral. Capital should be raised against evidence, not narrative. Sequencing is the discipline that makes a long build survivable.

This content is general information, not legal, tax or financial advice.

In short

  • A decade-long build is normal for financial infrastructure.
  • ClearBank illustrates the narrow-bank model for fintechs.
  • Phases run Zone, Principal, Hub, Liquidity cloud.
  • Licensing proceeds in steps: sandbox, EMI, then banking.
  • Each phase should de-risk the next rather than assume it.

See the full phasing and structure at /about/.