Nick Fernando: the bank infrastructure tokenisation still needs

Research by Aqua Global Solutions found that 51 per cent of banking IT leaders link rising operational costs to inefficiencies in payment processing, that only 13 per cent of banks report full automation across core payment functions, and that 65 per cent spend more time repairing payment data than producing it. Tokenised deposits and stablecoins are being offered as the next generation of cross-border rails. The question is whether the systems underneath banks can carry them.

Nick Fernando is co-founder and director of Aqua Global Solutions, with more than four decades in banking technology, financial messaging and global payments. In written answers to Disrupts, he described what is going wrong inside bank payment operations, where tokenisation is already doing real work, and what a bank should fix first.

Many banks, Fernando said, still run payments on systems built decades ago, "when settlement took days and nothing much changed year to year". Each new requirement tends to be bolted on rather than built into the core. "This means sanctions screening in one place, a translation tool for ISO in another, and a separate tool for investigations."

The result is data split across platforms that cannot reference each other. "A payment can pass one system and break in the next with nobody spotting it until it is rejected. As a result, someone ends up repairing failed messages by hand, chasing missing beneficiary details across correspondents by email and matching nostro statements line by line." Staff time goes on manual processes, and longer investigations leave cash sitting idle at a correspondent bank.

It is hard to fix because the technology has revenue running through it every day. "Most banks cannot afford the possible downtime that comes with a move. So they put it off, and the budget goes to the latest regulation with the most pressing deadline instead."

Where tokenisation is real

Tokenised deposits, Fernando said, "are already doing real work. They reduce reliance on long correspondent banking chains and move richer transaction data alongside the payment. That means faster settlement, easier reconciliation and more efficient compliance."

Where the promise runs ahead of reality is "everything the payment touches on the way out". Legacy systems cannot properly handle or share the structured data arriving from digital rails, "so the payment can end up fragmented, manually reconciled or delayed". He acknowledged the momentum behind digital money, from UK Finance's tokenised deposit pilots to Swift's work on digital interoperability, "but stablecoins alone will not transform cross-border payments. The bigger shift comes when banks modernise the infrastructure underneath them so these digital rails can actually work end to end."

Asked how ready that infrastructure is for tokenised assets and DeFi-style settlement, his answer was short: "Not very." Manual processes are a poor fit for rails designed to settle in near real time, because any speed gained is lost if payment data still has to be checked, repaired or reconciled by hand. "Tokenised deposits also still have to talk to existing rails like Swift and BACS, and most legacy systems were never built for that level of data exchange. You end up with a fast rail bolted to a slow bank."

Sanctions and fragmentation

Real-time settlement in a geopolitically fragmented world runs into a system built over fifty years. Cross-border payments still pass through correspondent banks and separate domestic systems, each with its own processes, technology and regulatory requirements, and a transaction may face different AML, KYC, sanctions and identity checks depending on where it is going. "Every one of those adds a step. That is what makes a truly joined-up real-time flow so hard to build."

Sanctions are "the sharpest version of the problem". The lists have diverged over the past few years, so a payment that is legitimate in one jurisdiction can be stopped in the next, and screening throws up false matches that need a person to review them. Deep integration across currencies and clearing systems is not something even the best technology can deliver overnight. What banks can realistically do, he said, is "pick the payment corridors that matter most to their customers, build properly for those, and be clear with everyone else about when money will actually arrive".

Start with visibility

If a bank can fix only one thing, Fernando's answer is visibility. "Most banks cannot actually see their own payments from end to end. A transaction goes through six or seven systems that do not report into one place, so there is no live picture of what is failing or where."

The advantage is that it does not require ripping out the core. "They can add a layer that brings payment status, screening and exceptions together across the systems they already have. That is much more achievable than a full transformation programme." Once a bank can see where payments are breaking, it can start automating and fixing the problems that happen most often "rather than guessing where to focus".

Looking three to five years out, he expects banks to keep joining what he called the tokenised cross-border settlement train, but said the UK's underlying infrastructure "is not ready to support it at scale just yet". With payment delays and slow investigations already unresolved, tokenised deposits "could very much be a case of banks running before they can walk".

The next 18 months matter most, because that is when the pilots run in live conditions. "What will separate the leaders is not who moves first on tokenisation. It is who has built the data foundations underneath it. Banks that can automate reconciliation and matching across both old and new rails will be able to pick up whatever comes next without much drama. The ones still repairing payments by hand will keep announcing pilots and not much else."

UK Finance's tokenised deposit pilots and Swift's digital interoperability work are the live tests Fernando expects to settle the question over that period.