Cross-border payments are a $150 trillion market annually. The cost of moving that money — fees, FX spread, correspondent banking charges, settlement delays — runs somewhere between $400 billion and $500 billion per year. That number has barely moved in a decade despite significant technological change. That gap is not shrinking on its own.
We think about this a lot. The infrastructure that moves international money is genuinely old. SWIFT, the messaging network that coordinates most correspondent banking, was built in the 1970s. The correspondent banking relationships layered on top of it were designed for a world of paper instructions and end-of-day settlement. None of it was built for speed, transparency, or the kind of programmatic access that modern software demands.
The cost structure of a cross-border payment isn’t obvious until you follow the money through each hop. A typical international wire from the US to Southeast Asia might pass through three correspondent banks, each charging lift fees, before reaching the destination bank. FX conversion happens at one or more of those hops at a spread the sending company never negotiated. Settlement takes between one and five business days depending on the corridor.
The problem is not any single link in that chain. It’s the chain itself. Each institution in the correspondent network is protecting a margin on a transaction they’re touching for maybe thirty seconds of actual processing time. The collective rent extraction across that chain is what creates the $400 billion figure.
For large corporates doing high-volume treasury operations, some of this gets negotiated away. For mid-market companies, small businesses, and individuals — the categories that represent the majority of transaction volume by count — the rack rate applies. There is almost no transparency and very little leverage.
Three approaches are making a dent:
| Approach | Mechanism | Corridors That Work |
|---|---|---|
| Pre-funded liquidity networks | Local accounts on each side, netting positions | High-volume corridors (US-MX, US-PH) |
| Stablecoin rails | Mint/burn on either end, near-instant settlement | Corridors underserved by traditional banks |
| Real-time payment network interoperability | Linking domestic RTP systems across borders | Connected countries (UK-India via UPI linkage) |
The pre-funded model has been around the longest and works well at scale. The challenge is capital efficiency — you need to pre-position liquidity in dozens of local currencies, and that capital has a cost. The companies doing this well have built sophisticated treasury operations that optimize how much they hold where.
Stablecoins are genuinely interesting as infrastructure. We’re not talking about speculative cryptocurrency here. We’re talking about dollar-denominated tokens that settle in seconds and can be minted and burned at local endpoints to convert back into local currency. In corridors where traditional banking is expensive or slow, this is a real solution for real businesses. We wrote more about this in our piece on stablecoin rails.
The opportunity we’re most excited about in 2026 is real-time payment network interoperability. Most developed economies now have a domestic RTP network — FedNow in the US, Faster Payments in the UK, PIX in Brazil, UPI in India, PayNow in Singapore. Each of these networks settles domestic payments instantly and cheaply. But they don’t talk to each other.
Building the bridges between these networks is technically complex but not technically impossible. The interesting companies are not building new rails — they’re building the routing, FX, and compliance layer that connects existing rails. Done right, this means a US company can pay a Brazilian supplier through PIX at FedNow cost, without a correspondent bank in the middle.
Several central banks are working on exactly this. BIS Project Nexus is one initiative we follow. But the commercial infrastructure layer — the routing, the FX optimization, the compliance tooling, the developer API — is where we believe private companies will build enduring businesses.
Speed and cost are not the only challenges. Compliance is often the actual constraint. KYC/AML requirements differ across jurisdictions. Sanctions screening is mandatory at each hop. OFAC, UN, EU, and local watchlists all need to be checked, often in real time. A payment that fails compliance at the destination bank after a two-day float is a genuine business problem for the company that sent it.
We see compliance infrastructure as a payments infrastructure investment, not a separate category. The companies that get the cost-of-compliance right — and can demonstrate clean compliance records to banking partners — earn better correspondent relationships and pricing. That advantage compounds.
The gap in cross-border payments infrastructure is not a technology problem. The technology mostly exists. It’s a coordination, regulation, and incentive problem. That’s harder to solve, but it also means the winners will be harder to displace.
Our data shows that the companies in our portfolio touching cross-border payment flows are seeing corridor-specific win rates that correlate almost entirely with how well they’ve solved the local liquidity and local compliance problem. The global solution matters less than being genuinely good in the corridors you’re actually in.
Working on payments infrastructure? We invest in this category.
