Red Hat and Temenos executives discuss a new reality in which different payment rails are emerging for different needs, and why interoperability is the key to functionality
The payments industry is not short of rails, but it is short of coherence between them.
Instant payment schemes, correspondent banking networks, emerging stablecoin infrastructure and legacy batch processing systems all coexist, but they all serve different use cases and operate under different regulatory regimes. And, for the most part, they are unable to communicate with each other in any standardised way.
For banks operating across multiple markets, that fragmentation has real consequences, as modernising one part of the stack while leaving legacy architecture in place elsewhere creates bottlenecks that constrain the value of the modernisation itself.
As a result, one of the big questions facing institutions today is not simply which new rails to adopt, but how to build infrastructure that can work across all of them.
The case for moving first
Sai Rangachari, Chief Product Officer at Temenos, points to BIL in Luxembourg as an illustration of what early modernisation can unlock. The bank moved quickly on SEPA Instant and has since supported other banks’ access to the scheme.
“If you have the right infrastructure you can go and offer this as a service to those who are behind, and almost open up a new revenue stream,” Sai said.
The flip side is that modernising piecemeal carries its own risks. Legacy ACH and wire systems, built for batch processing and largely unchanged for two decades, can only absorb so much bolt-on innovation. “You can only bolt on so much before everything comes apart,” Rangachari said, arguing that banks need to commit to architectural change rather than continue patching.
A similar case is made by Ramon Villarreal, Payments Lead at Red Hat, who says: “If you build the right foundation with the right level of resilience, you will be able to deliver new capabilities to the market faster and more efficiently, maximising innovation.”
For Red Hat, that foundation is OpenShift, a ‘containerised’, multi-cloud environment which helps organisations address the demands of DORA in the EU and the Operational Resilience Act in the UK, while remaining flexible enough to run any AI model and integrate it into existing software delivery cycles.
“You don’t want to be dependent on a single system,” Villarreal said. “You want to be able to integrate any AI model on the same platform.”
Both Rangachari and Villareal allude to a similar underlying point, that banks which have pushed hard on innovation without maintaining the foundations are now feeling the consequences. “The only organisations that will be in trouble are the organisations that forgot the operational basics: resiliency, performance, security, and scalability,” Villarreal said. “Payments systems cannot fail.”
Why instant payments adoption remains uneven
Despite growing adoption, around 85% of US banks have yet to use instant payment rails. Part of the explanation is structural, a result of legacy infrastructure, regulatory uncertainty, and implementation cost.
But Rangachari points to something more structural, in that a bank’s nature is to prioritise liquidity, and when funds move through a chain of intermediaries, every party in that chain has an incentive to hold them for as long as possible.
“Everybody wants to receive money fast, but nobody wants to send money fast,” he says. “As an institution, you want to hold on to the cash as long as possible.”
Where the use case for instant payments is strong enough, though, adoption has followed. Gig economy payouts, Uber drivers receiving same-day earnings, and peer-to-peer transfers via Venmo have driven genuine uptake in the US. But penetration there remains patchy.
Contrast this to Brazil, where regulators have mandated Pix, instant rails have become the default payment method. In India, UPI dominates for the same reason. “Wherever the regulators have made it mandatory to roll out RTP schemes, they’ve gone live,” Rangachari said. “Where they’ve kept it optional, it takes its own time.”
There is operational complexity beyond implementing the rails themselves. Real-time settlement removes the clawback window that ACH provides, which means fraud controls need to operate at the same speed as the payment. “Instant payments really means instant controls,” Rangachari said, “and banks are not ready for those.”
A transaction flagged for review that takes 48 hours to investigate is not an instant payment in any meaningful sense.
Cross-border: A coordination problem as much as a technology one
Those control gaps are compounded further when payments are made cross border, where the regulatory complexity multiplies at every step.
Rangachari describes a transaction moving from the US to India, passing through at least three separate regulatory regimes, each with its own KYC and AML requirements. Correspondent banking adds cost and opacity at each step. “I might pass KYC in the US, but the correspondent banking leg might be in the UK, and the recipient might be in India, and it may fail on UK KYC and AML obligations when you and I have nothing to do with it,” he said.
This is where Villarreal’s argument about interoperable infrastructure becomes most pointed. A portable, standards-aligned platform is not just a resilience play – it is what makes it practical to operate across the fragmented corridors that cross-border payments still depend on. “Partnerships are going to become more and more important,” he said. “It’s not only about what you can deliver, it’s about who you can partner with to expand your capabilities and get to market faster.”
Speaking at a MoneyLIVE Summit 2026 panel, Dovile Naktinyte, Product Design Lead at the Bank of England, acknowledged that progress under the G20 cross-border payments roadmap has been real but that “more needs to be done to bring the real benefits to the end users.”
The BoE is extending CHAPS settlement hours to 01:30 from 2027, with ambitions toward 24/7 availability, and is developing a synchronisation capability within RTGS that would allow atomic settlement across different ledgers, meaning tokenised assets and traditional funds settle together or not at all. Naktinyte described the broader ambition as a multi-money ecosystem in which different forms of digital and traditional money can coexist and interoperate, anchored by central bank money as the risk-free settlement asset.
Data quality is crucial to all of it. Speaking on the same panel, Vijay Lulla, EMEA Head of Cross-Currency Payments at J.PMorgan, said: “For AI to work, you need to have the right level of data. Otherwise it’s garbage in, garbage out.”
Today, ISO 20022 is becoming the standard through which richer payment data is being normalised across the industry.
“As long as you don’t get to that space where everyone is ISO native, you will not be able to get the value out,” Lulla said. That collective dependency is itself an interoperability problem, and it remains unresolved across much of the industry.
AI and digital assets: Grounded expectations
On AI, Rangachari believes narrow, deterministic use cases are where the returns are real, and scope creep is where the risk lies. “AI still hallucinates,” he said. “If you create a grandiose use case – let’s have AI manage end-to-end payments – I don’t think we’re there yet.”
The value of that scoped approach shows up in practice too. A Temenos client application identified that ACH transactions initiated on Fridays failed at a higher rate, because customers could spend funds over the weekend before collection. Routing Friday transactions via same-day ACH fixed it.
Villarreal’s view is that the model matters less than the integration. Smaller, inference-optimised models built for specific functions are more practical in payment environments where latency and determinism are non-negotiable.
“It’s not always about what the latest and newest technology is. It’s about what is the safest and most reliable technology,” he said. Swift‘s sanctions screening work is the kind of targeted, workflow-embedded application he holds up as the right model.
On digital assets, Villarreal notes that many organisations are beginning to implement crypto custodian infrastructure, though he is cautious about what consolidation of digital assets will bring.
“I want to see how it shakes out when the giants fall,” he said. And while stablecoin settlement can be fast on-chain, converting value back to fiat in another jurisdiction under a different regulatory framework is where speed runs out. “While stablecoins and tokenised deposits can move at the speed of light, liquidity cannot move at that speed of light,” Lulla said. Most current commercial use cases remain intra-bank, which is, again, an interoperability problem deferred rather than solved.
The value of faster, more capable payment infrastructure depends on how much of the ecosystem can access it on the same terms. So while infrastructure is improving and rails are multiplying, the harder things to evolve are regulatory alignment, data standards and a willingness to migrate fully rather than maintain translation services that mask a deeper incompatibility.
