Payments are entering another phase of structural change. Real-time payment systems are expanding, tokenised deposits are moving into live institutional use, stablecoins are attracting greater institutional interest and shared-ledger initiatives are being developed to connect regulated digital money across borders.

The proliferation of infrastructure also creates a new problem. Banks and their clients increasingly have more ways to move money, but the value of those alternatives will depend on whether they can operate together rather than develop into separate pools of liquidity and connectivity.

Fabian Khoshbakht, head of Global Payments & Trade, Asia Pacific at BNY, did not expect emerging digital rails to abruptly replace conventional correspondent banking.

“I don't think we're at an inflection point that tomorrow traditional correspondent banking is going to be replaced or shut down because we've got digital asset infrastructure and we're going to use stablecoin to move money around the world,” he said.

Instead, he expected the different infrastructures to coexist for the foreseeable future, with banks providing access to new channels as financial institutions and their underlying corporate clients become ready to use them.

That puts the emphasis less on predicting which form of money or payment rail will ultimately prevail and more on building the infrastructure that allows institutions to use different options as they develop.

New rails add to rather than immediately replace existing infrastructure

The coexistence Khoshbakht described is already becoming visible. BNY took the first step in its tokenised-deposit strategy in January 2026 by enabling an on-chain mirrored representation of participating clients' deposit balances. Beginning with collateral and margin workflows, it creates digital book entries on BNY's private, permissioned blockchain representing clients' existing demand-deposit claims against the bank.

Swift is separately extending its infrastructure into tokenised payments. In July, it said its blockchain-based shared ledger was ready for initial use, with 17 banks, including BNY, exploring to pilot live transactions using tokenised deposits. The ledger provides an orchestration layer for participating banks' tokenised deposits, with final settlement continuing through existing systems.

Khoshbakht saw Swift's move as a logical response to the changing infrastructure. “Swift had to take a look at this,” he said. “We have the infrastructure everyone's plugged into. How do we collectively create the platform or the infrastructure to be able to do this?”

He was encouraged by the participation of banks from Asia Pacific, including Singapore banks, seeing it as evidence that institutions in the region are actively exploring new channels to serve clients faster and more efficiently.

But he remained cautious about predicting how quickly the new infrastructure would develop, preferring to see how the initial activity played out over the following months.

That distinction is important. New payment rails are emerging, but institutional adoption will depend not simply on whether individual technologies work, but on whether they can connect with the wider banking and liquidity infrastructure institutions already use.

Tokenised money could change how liquidity follows the sun

For Khoshbakht, the more consequential opportunity from tokenised deposits and other forms of digital money lies in what they could eventually do for liquidity and working-capital management.

The first stage is the ability to represent cash and other assets digitally and potentially invest excess liquidity. The second is enabling treasurers to move and deploy liquidity more efficiently across an increasingly 24/7 financial system.

“I really see in the future the days of following the sun with your liquidity of the old days is nearing sunset,” he said.

Traditionally, global companies have moved liquidity between regions as markets open and close. Khoshbakht envisaged a model in which tokenised deposits, stablecoins and always-on infrastructure could give treasurers greater flexibility to move funds between regions, invest liquidity when it is not required and retrieve it when needed.

“I'll have tokenised deposits. I'll have stablecoins. I can move money through that infrastructure in different regions quickly, easily invest those funds when I don't need them, and pull them back immediately when I do need them to run my business,” he said.

The significance, he argued, is not that corporate treasury teams themselves need to work around the clock. Rather, their liquidity could remain productive when they are not working.

BNY has separately described real-time, 24/7/365 payments as the foundation of an emerging treasury and working-capital ecosystem, with interoperability, automation and integrated risk and compliance among the requirements for firms operating in an always-on environment.

Khoshbakht also saw demand as increasingly coming from both sides of the bank-client relationship. He described the development as a “marriage” between banks building new capabilities and corporate treasurers looking for greater efficiency and access to liquidity.

While global companies with round-the-clock requirements may have the clearest immediate use cases, he expected the benefits eventually to extend to mid-sized companies and smaller businesses operating digitally across borders.

AI makes a multi-rail payment system more intelligent

More payment options also create more decisions. A corporate or financial institution increasingly has to determine not only where money needs to go but when it should arrive, how urgently it is required and which available channel offers the appropriate combination of cost, speed and security.

Khoshbakht said corporates and financial institutions were looking for smarter ways to make those decisions even before the recent acceleration in artificial intelligence. AI is now expanding what can be done with the information surrounding a payment.

At the corporate level, he envisaged AI working with enterprise resource planning systems to help determine when a payment should be released to make better use of working capital. At the financial-institution level, AI could help determine how a payment should be routed.

“The financial institution is using AI to say, where does this payment need to go? What do I need to do with it? How urgent is it? How non-urgent? And who do I direct that through to do it at the lowest cost, the fastest speed, the safest way possible?” he said.

BNY is also developing the use of AI around the payment information it provides to clients. Khoshbakht pointed to its Client Insights Dashboard, which allows financial institutions to analyse their payment data and flows.

He said BNY was doing additional work to layer AI and agentic AI onto that capability so it could become more proactive. Rather than clients only examining transactions after they have occurred, the objective is to push insights about payment processing, including why transactions became delayed and how flows could potentially be improved.

This becomes more significant as the number of possible payment routes increases. AI is not creating those rails, but it could increasingly help institutions decide how and when to use them.

BNY sees AI as an enabler across the organisation

The expansion of AI also raises questions about the cost of the computing resources required to operate it at scale. Khoshbakht said BNY did not approach AI primarily by attaching a token cost to individual usage. “AI for BNY is primarily focused on being an enabler,” he said.

He pointed to Eliza, BNY's enterprise AI platform, which employees use within defined access and data controls. He said staff were trained before using it and could not use the system to access information from business areas for which they did not have permission.

For Khoshbakht, the relevant measure was whether AI improved processes, enabled faster payment processing and information delivery and ultimately contributed to winning business. “I can't speak for other organisations, but for us, it's an enabler, and it's working,” he said.

His argument suggests a broader distinction in how AI is entering transaction banking. Much of its immediate value lies not in autonomous finance, but in embedding intelligence into existing workflows, data and decision processes and allowing people and institutions to operate them more efficiently and at greater scale.

Trade modernisation follows a slower path

Trade finance illustrates the limits of how quickly technology can transform established financial infrastructure. Khoshbakht said BNY had been using AI to process trade payments and documentation for several years, increasing processing capacity while retaining human involvement where verification and judgement remain necessary. “There is a portion of the trade processing that still needs humans,” he said.

The wider challenge is that trade modernisation depends on more than what an individual bank can digitise. BNY's recent analysis of Asia Pacific trade highlighted uneven digitalisation and interoperability gaps across markets even as banks use AI in document-heavy processes and seek greater straight-through processing.

Khoshbakht described trade modernisation as evolutionary rather than abrupt. “This is one of those industries that needs to evolve. It's an old industry,” he said.

BNY is examining how trade could increasingly operate in a digital environment, including possible use of blockchain, but Khoshbakht stressed that individual institutions cannot determine that transition on their own.

“The whole industry has to want to move too,” he said. “Our goal is always to build the infrastructure and be ready for when our clients are ready, and then they can plug into that.”

Interoperability becomes the question for Sibos

Khoshbakht expected payments to be among the areas of transaction banking that evolve most quickly in the near term. More payments, he said, would eventually move through stablecoins, digital ledgers or other emerging forms of digital money, while deposits and trade infrastructure would also continue to change.

But when asked which industry question most urgently needed a clearer answer as financial institutions prepare to gather in Miami for Sibos, he reduced the challenge to one word: “Interoperability.”

“As everyone looks at all of this, I think we all, as an industry, need to solve interoperability,” he said.

That is increasingly the common problem underlying otherwise separate developments in payments. Correspondent banking, real-time systems, tokenised deposits, stablecoins and shared ledgers can each improve parts of the process, but their broader usefulness will depend on whether money and information can move between them without creating another layer of fragmentation.

Khoshbakht was optimistic that the industry could find a collective solution. “Interoperability is going to be important. I genuinely believe it,” he said. For BNY, he said, the task is to continue developing its own capabilities while ensuring that clients can connect through them to the wider financial system.

“We're playing our part. We're doing what we need to do in order to enable our clients, and then plug into the rest of the world.”

That leaves a more consequential question for Sibos Miami than which new payment technology will prevail: whether an industry that is becoming increasingly capable of moving money instantly and in new digital forms can make those different forms of money and infrastructure work together.