The rails must think. Beyond ISO 20022 — the embedded intelligence layer that defines payment infrastructure’s next decade.

Introduction

For the better part of a decade, the modernization conversation in payments has been a deployment conversation. Stand up the new RTGS. Roll out the instant payment scheme. Migrate to ISO 20022. Hit go-live.

That work mattered. But the floor has shifted under it. Real-time payment volumes worldwide reached roughly 266 billion transactions in 2023 and are projected to climb past 575 billion by 2028, per ACI Worldwide’s Prime Time for Real-Time series. The SEPA Instant Payments Regulation has now made instant reach a legal obligation for every eurozone PSP, and ISO 20022 has become the de facto standard for cross-border messaging on Swift as the coexistence period closes in November 2025.

In this landscape, deployment is no longer the differentiator. The competitive frontier has moved one layer up: to the intelligence embedded in the rails themselves. This piece argues that AI and machine learning, applied directly inside the transaction path, will define the next decade of payment infrastructure leadership — and that European operators are the first to face that question in earnest.

From settlement to decisioning: where the value now lives

Modern payment infrastructure carries one of the richest real-time data streams in any industry on earth. Every message contains identity, intent, geography, counterparty, channel, timing, and value — multiplied across hundreds of billions of transactions a year. For most of the rail’s history, almost none of that signal was used in flight. Fraud screening, compliance review, business analytics, and capacity planning all happened elsewhere, often hours or days later, on data that had already gone cold.

That model is breaking. Authorised Push Payment fraud losses in the UK alone reached £341 million in 2023, with instant rails now the dominant attack surface (UK Finance, Annual Fraud Report 2024). India’s UPI now processes well over 15 billion transactions in a single month — a volume no human-mediated review process can plausibly oversee. The rails have to do something fundamentally different: they have to think while they move.

The next generation of payment infrastructure is not a faster transaction processor. It is a real-time decisioning system that happens to settle payments — and the institutions that recognise this distinction first will compound the advantage.

Embedded intelligence as architecture, not feature

The intelligence we are talking about is not a chatbot bolted to a portal. It is AI and machine learning embedded directly in the transaction path: fraud and risk models that score every message in single-digit milliseconds; liquidity intelligence that anticipates intraday positions rather than reporting them after the fact; compliance engines that interpret context instead of matching strings; observability that treats the rail itself as a living system; and analytics surfaced both to the operator and to its participants.

None of this requires speculative technology. Brazil’s Pix, which now clears in excess of 5 billion transactions per month, has built mandatory in-rail fraud signaling because retroactive review at that scale is structurally impossible. The Reserve Bank of India’s directives on real-time fraud risk management in UPI, the European Central Bank’s expansion of TIPS toward a single-digit-millisecond message envelope, and the Federal Reserve’s FedNow risk framework all assume an architecture in which intelligence lives inside the rail, not adjacent to it. ISO 20022 was the data substrate that made this possible. Microservices were the runtime. The deployment era quietly built the foundation. The question now is who wires the intelligence layer in deliberately, and who bolts it on later.

The architectural choice being made today — to design the intelligence layer as a first-class component of the rail rather than as a downstream consumer of it — will largely determine the next decade of competitive standing in payment infrastructure.

Europe as the first post-modernization market

Europe is, by any measure, the most modernized payment landscape in the world. SEPA Instant is mature. ISO 20022 is in production. The interoperability conversations between EBA CLEARING’s RT1, the ECB’s TIPS, and cross-scheme reachability are largely settled architecturally, even where the politics continue. RT1 and TIPS together now clear billions of instant payments annually, and the ECB’s 2024 fee changes were specifically designed to push more participants and volumes onto the pan-European rails.

This is what makes Europe structurally distinct: it is the first major region to face the post-modernization question in earnest. Once everyone is on the same rails, what differentiates the institutions running them? Increasingly, the answer is not the rail itself. It is the intelligence layered into it — and, just as importantly, the intelligence the rail offers back to its participants. European operators that treat their rails as decisioning systems, not just clearing systems, will set the template the rest of the world follows. Those that do not will find their infrastructure indistinguishable from a utility within a five-year horizon.

Europe will not win the next decade of payments by deploying instant infrastructure faster than anyone else. It will win, or fail to win, by deciding what its rails are allowed to know.

Strategic Imperatives

For payment operators, central banks, and infrastructure providers building toward the post-modernization era, three imperatives follow:

1. Own the intelligence layer; do not rent it. Treat AI and ML inside the rail as core infrastructure, not a vendor add-on procured per use case.
2. Wire intelligence in during migration, not after it. ISO 20022 and microservices migrations are once-in-a-decade architectural openings. Use them to install the data plane needed for embedded intelligence — retrofitting later costs materially more.
3. Treat intelligence as a participant-facing product, not just an internal control. The institutions clearing on your rails are customers; what your rails reveal back to them is increasingly part of your value proposition.

The bottom line: The deployment era proved the rails work. The intelligence era will decide whose rails matter.

For institutions that have spent the modernization era on the build side — across European RTGS programmes, instant payment schemes, and cross-border corridors — the lesson is becoming hard to miss. The rails were always the foundation, never the finish line. What gets built on top of them now will define the next ten years of competitive position.

Reach Out To Learn More Here.