Send money from a bank account in Mumbai to a bank account in Singapore and, in most corridors, you're still looking at a correspondent-banking chain that takes a day or two, charges a spread you can't fully see, and gives the recipient no idea when the funds will land. Meanwhile, sending money from one UPI account to another UPI account inside India takes about three seconds. The gap between those two experiences is not really about technology — domestic instant payment systems have existed for a decade — it's about the fact that each country's instant rail was built to stop at the border. Interlinking is the project of removing that stop.
What "interlinking" actually means
Over the last fifteen years, more than 70 countries have built some version of a domestic instant payment system (IPS): UPI in India, PayNow in Singapore, PromptPay in Thailand, DuitNow in Malaysia, InstaPay in the Philippines, PIX in Brazil, FedNow in the US, and dozens more. Each one solves the same problem within its own borders — moving money between bank or wallet accounts in seconds, 24/7, using a phone number, national ID, or QR code instead of a long account number.
The problem is that these systems were designed independently, on different rulebooks, different messaging formats, different participant models, and different regulators. A UPI transaction and a PromptPay transaction don't speak the same technical language, and even if they did, no bank sits inside both networks with settlement accounts in both currencies ready to clear a payment instantly.
Interlinking is the layer that sits between two (or more) domestic IPS and lets a payment initiated in one system be received and completed in another, in something close to real time, without either country's core payment rail being rebuilt. There are two broad architectures in use:
- Bilateral linking: two countries' systems are connected directly to each other, with a dedicated technical and legal agreement covering just that pair. India-Singapore's UPI-PayNow link, launched in 2023, is the best-known example.
- Multilateral hub-and-spoke linking: a single shared platform sits in the middle, and each country connects to the hub once, gaining access to every other connected country through that one integration. This is the model the Bank for International Settlements (BIS) has been building under Project Nexus.
The hub model is the more consequential of the two because it changes the economics of interlinking. A bilateral link solves one corridor. A hub, once a country joins it, opens up every other participating country at the marginal cost of one connection instead of N.
How the plumbing actually works
It helps to walk through what happens, layer by layer, when a payment crosses one of these links.
1. Address resolution
Instant payment systems let you pay someone using a "proxy" — a mobile number, virtual payment address, or national ID — instead of an account number and bank code. For a cross-border payment to work, the sending system needs to resolve the recipient's proxy into routing information inside the receiving country's system. This requires either a shared directory or a real-time lookup call across the link at the moment of payment, so the sender can see the recipient's name confirmed before hitting send — the same anti-fraud check domestic IPS already do.
2. Message translation
Domestic systems use different ISO 20022 message variants, different field requirements, and different limits on what data can travel with a payment. The interlinking layer has to translate a payment instruction from the sending system's format into the receiving system's format without losing the data both regulators require — sender and beneficiary identity, purpose codes, and screening fields.
3. FX conversion
Unlike a domestic instant payment, a cross-border one almost always changes currency mid-flight. The link needs an FX conversion step with a rate that's transparent to the sender before they confirm the payment — this is one of the explicit design goals of hub models like Nexus, which specify that the sending payment service provider (PSP) must quote an all-in rate upfront rather than let the recipient's bank apply a hidden spread later.
4. Settlement
This is the part that's easy to overlook and hardest to solve. Moving the message in real time is a solved problem; moving the money in real time, safely, across a currency pair is not. Most links use a model where PSPs on each side pre-fund settlement accounts (often via a local settlement bank or the central bank's real-time gross settlement system), and the interlinking platform nets or matches transactions against that pre-funded liquidity so the end customer sees instant completion even though the underlying interbank settlement may batch or net on a slightly different cycle.
5. Compliance screening
Every leg has to clear sanctions and AML screening. Cross-border links generally require both the sending and receiving PSP to run their own screening, plus the link operator may run a shared check, which is one reason "instant" cross-border payments still typically complete in single-digit seconds rather than true milliseconds — the screening call has to return before the payment does.
A quick history of how we got here
Domestic instant payment systems didn't arrive all at once, and neither will their interlinking. The rough sequence is worth knowing because it explains why interlinking is only now becoming practical rather than something that could have happened a decade ago.
The first wave of domestic IPS launched between roughly 2011 and 2016 — the UK's Faster Payments, India's IMPS (the precursor to UPI), and similar systems in a handful of other markets. These were built purely for domestic use, often on messaging standards and participant rules that made sense only within one country's banking structure. Nobody was designing for an external connection point because there was no obvious partner to connect to yet.
The second wave, roughly 2016 to 2021, is when the current generation of proxy-based systems arrived: UPI's full launch in India, PromptPay in Thailand, PayNow in Singapore, DuitNow in Malaysia, PIX in Brazil. This is also when regulators started explicitly discussing cross-border interoperability as a policy goal rather than a hypothetical, partly because remittance corridors between exactly these countries — large migrant-worker populations moving money home — were an obvious and politically visible use case for cheaper, faster transfers.
The third wave, from about 2021 onward, is interlinking itself: first as bilateral pilots (Singapore-Thailand's PayNow-PromptPay link went live in 2021, India-Singapore's UPI-PayNow link in 2023), then as the BIS began formalizing a multilateral approach through Project Nexus. What made this wave possible wasn't a technology breakthrough — proxy resolution, real-time messaging, and API-based FX quoting were all things individual systems already did internally. What changed was that enough domestic systems had matured, stabilized, and proven their own fraud and liquidity models that regulators were willing to open an external door into them.
Why this matters now
Project Nexus has moved from a BIS Innovation Hub research paper into live implementation. The initial group of connecting countries — India, Malaysia, the Philippines, Singapore, and Thailand — is working toward operationalizing the platform, with Indonesia participating as a special observer. This is the shift that matters: Nexus was, for several years, a technical proof of concept published by the BIS Innovation Hub Singapore Centre. It's now a platform with named central banks committed to actually connecting their live IPS to it, which is a very different (and much harder) undertaking than a pilot.
The reason this is worth watching is what it implies for the region. India, Singapore, Malaysia, Thailand, and the Philippines already have some of the highest per-capita instant payment volumes in the world domestically. Migrant labor flows and trade between these economies are large and persistent. A working hub connecting all five (plus future joiners) would mean a bank or fintech in any one of them could reach real-time payment users in every other connected country through a single integration to Nexus, rather than negotiating a separate bilateral deal — and separate technical build — for each corridor.
It also matters as a test of the hub model itself. If Nexus reaches production volume across this group, it becomes the template other regions point to when deciding whether to build their own bilateral links (slower to scale, but simpler to negotiate) or wait for a hub to reach them.
Bilateral links vs. the hub model
The two architectures aren't mutually exclusive — several countries run both simultaneously — but they trade off differently.
| Dimension | Bilateral links | Multilateral hub (Nexus-style) |
|---|---|---|
| Integration effort per new country | Full new technical + legal build for each pair | One connection to the hub, reuses shared rulebook |
| Time to add the Nth corridor | Roughly linear — each pair repeats the work | Falls sharply after the hub itself is built |
| Governance | Two central banks / regulators negotiate directly | Shared rulebook governs all participants; harder to agree upfront, easier to extend after |
| Examples live today | India–Singapore (UPI–PayNow), India–UAE, Singapore–Thailand (PayNow–PromptPay) | Project Nexus (BIS), in build phase |
| FX and liquidity model | Negotiated per pair, often bank-specific | Standardized across the hub, common settlement approach |
| Failure isolation | An issue in one link doesn't touch others | A hub-level issue can affect every connected corridor |
Neither model has "won." Bilateral links are live and processing real transactions today because two willing central banks can agree on a pairwise deal faster than a five-or-more-party group can agree on a shared rulebook. The hub model is a bigger lift to stand up but scales better once it's running, which is exactly the trade Nexus is making a bet on.
Practical implications for businesses and builders
For teams building in payments, remittances, or cross-border commerce, interlinking changes a few things that are worth planning around rather than reacting to after the fact.
- Remittance providers face margin compression on covered corridors. The economics of traditional remittance (correspondent banking, FX spread, agent networks) get harder to defend once a corridor has a direct instant rail with transparent FX. Providers with existing corridor licenses and compliance infrastructure are better positioned to add interlinked rails as a feature than newcomers are to compete purely on speed.
- PSPs need to think about proxy resolution and screening as product surfaces, not backend plumbing. Confirming a recipient's name before the sender commits, and showing an all-in FX rate upfront, are now baseline user expectations set by domestic UPI/PayNow-style UX — that expectation is transferring to cross-border flows.
- Settlement liquidity becomes a genuine operational function. Any PSP participating in an interlinked corridor has to pre-fund and manage settlement accounts in a currency it may not otherwise hold much of, which is a treasury and forecasting problem, not just an engineering one.
- Compliance requirements don't relax — they combine. A cross-border instant payment has to satisfy both jurisdictions' AML/sanctions regimes plus whatever the link operator layers on top, so screening latency budgets need to be designed in from the start rather than bolted on.
- Corridor coverage will be uneven for a long time. Builders shipping remittance or payout products should expect to support both "fast interlinked" corridors and "slow correspondent banking" corridors in the same product for years, and design the user experience to be honest about which is which rather than promising uniform speed.
Limitations and open questions
Interlinking is real progress, but it isn't a solved problem, and some of the hardest parts are still being worked out in public.
- Governance at scale is unproven. Bilateral deals involve two regulators; a hub involves negotiating a shared rulebook among five or more, with more expected to join later. Getting consensus on liability allocation, dispute resolution, and data privacy across that many jurisdictions is a slower and more political process than the technical integration itself.
- FX risk doesn't disappear, it moves. Someone in the chain — a PSP, a settlement bank, a liquidity provider — is holding intraday FX exposure to make instant settlement work. As volumes grow, that exposure grows with it, and the current model leans on PSPs pre-funding accounts, which ties up capital.
- Fraud typologies change with speed. Instant, irrevocable cross-border payments are an attractive target for social-engineering and account-takeover fraud precisely because they're instant and hard to claw back. Domestic IPS have spent years building fraud controls tuned to their own user base; interlinked systems need those controls to work across a much more varied population of senders and receivers.
- Not every country's IPS is technically or legally ready to interlink. Some domestic systems weren't built with an external connection point in mind, and retrofitting one is nontrivial engineering plus a fresh round of regulatory approval.
- Interlinking doesn't equal interoperability of everything. A live link typically covers person-to-person and modest-value payments first; larger B2B flows, trade finance, and bulk payments tend to lag because the compliance and liquidity requirements are heavier.
What to watch next
The signal to track isn't announcements of new countries joining a hub — those are relatively cheap to make. It's whether transaction volume on live corridors actually grows, whether settlement works smoothly at scale without central banks having to step in, and whether FX pricing on interlinked corridors stays meaningfully better than existing remittance channels once real volume, not pilot volume, is flowing through them. Also worth watching: which non-participating countries start building their own bilateral links in the meantime rather than waiting for a hub to reach them, since that's the market voting on which model it trusts to arrive faster.
FAQ
What is Project Nexus?
Project Nexus is a multilateral platform, developed through the BIS Innovation Hub, designed to connect multiple countries' domestic instant payment systems through a single shared technical and governance layer instead of requiring separate bilateral links between every pair of countries.
How is instant payment interlinking different from a wire transfer or SWIFT payment?
A traditional cross-border wire moves through a chain of correspondent banks, each adding time, fees, and opacity about the final exchange rate. Interlinked instant payments connect domestic real-time rails directly, so a payment can resolve the recipient, quote an FX rate, and settle in seconds rather than days.
Which countries have live interlinked instant payment systems today?
Several bilateral links are already operating, including India's UPI linked with Singapore's PayNow, and Singapore's PayNow linked with Thailand's PromptPay. Project Nexus's initial connecting group includes India, Malaysia, the Philippines, Singapore, and Thailand, with Indonesia as a special observer.
Does interlinking eliminate FX fees?
No. It makes FX pricing more transparent by requiring an upfront, all-in rate shown to the sender before they confirm payment, but currency conversion and the cost of holding cross-border liquidity don't disappear — they're just priced more visibly than in traditional remittance corridors.
Is interlinked instant payment the same as a central bank digital currency (CBDC)?
No. Interlinking connects existing domestic payment rails and existing currencies; it doesn't require either country to issue a digital currency. CBDC interoperability is a related but separate line of central bank research that sometimes overlaps with the same institutions working on interlinking.
What happens to fraud protection when payments move faster across borders?
Interlinked systems still run sender- and receiver-side AML and sanctions screening, plus proxy resolution that confirms the recipient's identity before the sender commits. But because interlinked payments are fast and hard to reverse, fraud controls have to be tuned for a much more diverse population of users than a single domestic system was originally designed for.
Will interlinking replace correspondent banking?
Not in the near term. Interlinking currently covers person-to-person and lower-value payments on a limited number of corridors; correspondent banking still handles the bulk of cross-border volume, including most B2B and trade-related payments, and will likely continue to for corridors and payment types that interlinked rails haven't reached yet.
Teams building cross-border payment or remittance products who want help evaluating which corridors and integration paths make sense for their use case can reach out to Woyce Technologies.