Forecasting payments is usually a matter of extrapolation. A more useful exercise is to name a small number of plausible paths and identify the signals that would confirm each. Over the coming five years, three variables matter most: how rails evolve, how licensing unfolds, and how quickly agent commerce arrives.

Scenario one: incremental rails, shared layer

In this path, existing rails improve and a connective layer emerges between them. The G20 targets, set in 2021 for end-2027, are partially met; progress remains uneven, as the FSB's 2025 report already suggests. Netting and routing become standard practice because they are the cheapest way to improve cost and speed without replacing infrastructure.

Signals: growing netting volumes, more routing across rails, and reliance agreements standardising across jurisdictions.

Scenario two: controlled, licence-led build

Here the pace is set by licences. A phased programme moves from operating on partners' rails to buying capacity as principal, to direct correspondent relationships, to an own banking licence. Under SUPA's published sequence, this runs from a Zone phase in 2026–2027 through Principal in 2027–2029 and a Hub phase in 2029–2032.

Signals: sandbox engagements progressing (a Guernsey GFSC sandbox application is expected in autumn 2026), an Irish EMI with EU passport in view, and the bank-jurisdiction decision taken later.

Scenario three: agent commerce arrives early

Agent-initiated payment volumes grow faster than expected, driven by platforms embedding units through APIs and protocol work in x402, ACP, AP2/UCP and MPP. Demand shifts from moving money to governing it: identified principals, revocable agent identities and policy enforced below the model.

Signals: rising agent-unit registrations, more platform embedding, and policy-layer requirements appearing in procurement.

How to read the signals

The scenarios are not exclusive. A licence-led build can proceed while agent commerce accelerates, and shared-layer improvements are compatible with both. What they share is a test: whether the gap between policy targets and observed performance narrows.

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

In short

  • Three variables matter: rails, licensing and agent commerce.
  • Shared-layer improvements can advance without new infrastructure.
  • A licence-led build follows a defined phase sequence.
  • Agent commerce may arrive faster than expected.
  • The shared test is closing the gap to the G20 targets.

Track the scenarios and the phasing at /investors/.