Real-time cross-border payments are now operating at meaningful scale. More than 25 banks committed to implementing Swift’s new retail framework by June 2026, and Swift reports that over 90% of payments across its network reach the beneficiary bank within an hour. The remaining drag increasingly sits inside institutions, where reconciliation, screening, and liquidity processes still struggle to match the speed of the rails.
That shift changes what a payments leader should be working on. For a decade, the sensible posture was to wait for the networks to mature. In 2026, a bank that cannot post, screen, and reconcile a payment in near real time can become the slowest component in its own corridor. Customers can now see that delay through end-to-end tracking services such as Swift GPI.
This article covers what went live in 2025 and 2026, where implementations break inside the enterprise, and a practical sequence for fixing the gaps.
What changed for instant payments in 2025 and 2026
Three developments moved real-time cross-border payments from roadmap to live implementation.
Swift’s retail payments framework: In March 2026, Swift announced that more than 25 banks had committed to processing consumer and SME payments under its new framework by June. The initial rollout covers corridors across Australia, Bangladesh, Canada, China, Germany, India, Pakistan, Spain, Thailand, the UK, and the US. The framework is designed to provide certainty on cost, full-value delivery, and end-to-end traceability, with instant settlement where the receiving market supports it. Five of the world’s ten largest remittance markets are included in the first wave, and Swift expects more markets to be activated before the end of 2026.
Regional instant-payment linkages: In October 2025, six payment networks from five ASEAN countries signed the George Town Accord to develop common technical and operational standards for non-card retail payments. Together, the participating networks serve more than 538 million people. The Bank for International Settlements’ Project Nexus addresses the same interoperability challenge through a multilateral model for connecting domestic instant payment systems, avoiding a separate bilateral integration for every corridor. For a comparison of how these schemes relate to the incumbent network, see how Swift and real-time payment systems compare.
One-leg-out settlement over domestic rails: Another model for cross-border payment corridors routes one leg of a transaction through a domestic instant system, such as UPI in India or Pix in Brazil. The US picture is covered in FedNow and the US instant payments build-out. Once funds reach the destination market, domestic instant rails can credit them in seconds. End-to-end speed still depends on the slowest participant, which may now be a bank’s internal batch process rather than the network.
The G20’s 2027 targets for speed, cost, transparency, and access continue to push this work forward. However, although network capabilities have advanced quickly, performance remains uneven across institutions, regions, and corridors.
Why bank infrastructure has not kept pace
What is happening: Payment networks have modernized through coordinated changes such as ISO 20022 migration, GPI tracking with unique end-to-end references, and shared service levels. In contrast, modernization inside most banks has happened system by system, leaving real-time payment channels connected to batch-based posting, screening, or reconciliation processes.
Why it matters: When settlement moves in seconds but internal ledgers and controls update later, the delay shifts inside the institution. Payments may reach the bank before they can be posted, screening queues can hold transactions promised quickly, and treasury teams may lack a current view of intraday positions. That can force larger liquidity buffers and increase settlement risk.
How it shows up in practice: The clearest sign is a widening gap between network arrival and customer credit. It appears in larger exception queues, more manual reconciliation, and tracking data that pinpoints the delay after the payment has reached the bank.
Where enterprise implementations break down
Four failure points account for most of the gap between corridor speed and delivered speed.
Reconciliation built for end-of-day: When settlement happens continuously but reconciliation still runs overnight, the bank can carry intraday positions before its records catch up. Continuous reconciliation depends on event-level posting and matching, which makes this a data-engineering challenge as much as a payments one.
Screening at transaction speed: Sanctions and fraud controls built around batch windows can create review queues that absorb the speed gained on the network. Cleaner, better-structured payment data can reduce false positives and manual review before a bank considers replacing the screening engine itself.
24/7 liquidity management: Instant corridors settle outside traditional operating hours. Treasury teams working from prior-day positions may need to hold larger nostro buffers or accept a higher risk of failed settlement. Intraday liquidity visibility therefore becomes a core requirement for real-time payments.
The batch-based core: Some banks route instant payments around the core ledger and reconcile them later. That can support lower volumes, but it becomes harder to sustain as traffic grows. The institution must eventually decide which flows can remain on a modernized edge and which require deeper ledger changes. Autonomous agents initiating payments add another layer to that decision, raising questions fintech leaders should be asking about agent-initiated payments before transaction volumes increase further.
A practical sequence for payments modernization
The order of payments modernization affects how quickly a bank can reduce delays without committing too early to a full core replacement.
- Instrument the payment flow first. Build event-level visibility across the existing systems before changing them. That gives teams a clearer view of where transactions slow down, fail, or move into manual review.
- Address posting and reconciliation next. Continuous posting and matching can reduce the intraday blind spots created by overnight batch processes, while also lowering exception volume.
- Tune screening with better data. Cleaner ISO 20022 data can help reduce false positives and manual review without changing the institution’s risk appetite or replacing the screening engine.
- Make the core decision with evidence. Once the first three steps are in place, the bank can see which payment flows genuinely need deeper ledger changes and which can continue through a modernized edge. The same pressure appears across regulated industries, where legacy modernization has become the AI readiness test for banking and insurance.
An edge-first approach can make sense when volumes are concentrated in a few corridors, the core remains stable, and time-to-market is the immediate pressure. Core-first modernization becomes more difficult to avoid when the ledger cannot support event-level posting or reliable real-time updates.
What real-time readiness looks like
A bank that has modernized these areas should see lower exception rates on instant corridors, stable reconciliation staffing as volume grows, on-demand visibility into intraday liquidity, and corridor-level tracking data it can show customers with confidence.
Progress can begin without replacing the core. Reliable data flows, event-level posting, and production integration across the existing stack often determine how much delay the bank can remove first.
Fulcrum Digital’s AI and machine learning engineering practice supports this work through payment data engineering and event-driven integration carried into production within the client’s own environment. The objective is a governed engineering foundation the bank controls and can extend as payment networks and regulatory requirements evolve, without being tied to a single vendor’s roadmap. For institutions facing a similar backlog, the practical starting point is to identify where internal delays are concentrated and which layer should move first.
Start that conversation today.
Key Takeaways
- More than 25 banks committed to processing consumer and SME payments under Swift’s new retail framework by June 2026. The initial corridors span 11 markets, including five of the world’s ten largest recipients of remittances.
- Swift says 75% of payments across its network reach the beneficiary bank within 10 minutes, and over 90% arrive within an hour. More than 80% of the average end-to-end journey occurs after the payment reaches the bank.
- The George Town Accord brings together six payment networks from five ASEAN countries to develop common standards for non-card instant retail payments, covering more than 538 million people.
- Regulatory reporting, foreign exchange controls, limited 24/7 infrastructure, and manual processing remain major sources of delay during the final domestic leg.
A practical modernization sequence begins with event-level visibility, followed by posting and reconciliation improvements. Better screening data comes next, giving the bank evidence to decide how much core change is required.
Frequently Asked Questions
What are real-time cross-border payments?
Real-time cross-border payments are international transfers designed to move and make funds available within seconds or minutes rather than through traditional multi-day processing. They may use linked domestic instant-payment systems, faster correspondent-banking services, or a domestic real-time rail for the final leg.
How fast are cross-border payments in 2026?
Swift reports that 75% of payments across its network reach the beneficiary bank within 10 minutes, and over 90% arrive within an hour. That measures the bank-to-bank journey. Performance still varies across cross-border payment corridors because final credit can be delayed by local infrastructure, operating practices, regulatory checks, or currency controls.
What is the difference between Swift GPI and Swift’s new retail framework?
Swift GPI improves correspondent-banking payments through end-to-end tracking, payment-status visibility, fee transparency, and service-level monitoring. Swift’s new payments scheme adds a common rulebook for consumer and SME transactions, with commitments covering upfront cost, full-value delivery, traceability, and instant settlement where domestic systems allow.
What are the G20 targets for cross-border payments?
The G20 targets for cross-border payments cover cost, speed, transparency, and access across wholesale, retail, and remittance payments. By the end of 2027, 75% of retail and remittance payments should make funds available within one hour, with the remainder completed within one business day. The retail cost target is a global average of no more than 1%, with no corridor above 3%.
Why do banks struggle to support instant cross-border payments?
Many banks still connect faster payment channels to batch-based posting, reconciliation, screening, and liquidity processes. ISO 20022 can improve the structure and quality of payment data, but banks still need internal systems that can use that data in real time. Delays can also arise from regulatory reporting, foreign exchange controls, limited 24/7 infrastructure, and manual checks after the payment reaches the beneficiary bank. End-to-end tracking helps institutions identify which part of the payment journey is creating the delay.
