Perspective on how cross-border payments actually work — insights and hot takes on the landscape, and how industry shifts play out.
Cross-border payments run on infrastructure that is invisible when it works. The discipline is to find its fragility before a failure does: concentrated correspondent banks, weak last miles, and the private bridges holding the middle together.
A corridor rated high complexity can mean three different things: elaborate but predictable, hostile and exclusionary, or prone to sudden shocks. Confusing them leads to the wrong decision on whether and how to enter.
The average price in a corridor tells you almost nothing. What matters is the spread between the cheapest and most expensive provider, the split of fixed fee versus FX markup, and who the competitors actually are.
The first two layers describe a corridor as it stands today. Demand momentum asks where it is heading, and it is directional and multi-sided: it means little on its own but changes how you read every other layer.
A single currency corridor decomposes into four operator archetypes, each a distinct business with its own customers, ticket sizes, and price sensitivity. A provider built for one is structurally incapable of serving another.
The headline flow tells you a corridor is large, not whether you can win it. Reading market intelligence properly means unpacking what the flow is made of, how much you can actually reach, and who is moving it.
Evaluating a corridor is more than a market-size number or a customer feature request. Reading it properly means looking through six distinct layers, where a single adverse factor can break the opportunity.
A '2.9% + 30 cents' rate looks like it all belongs to the processor. It doesn't. A card transaction has four parties, and the processor controls only a thin margin of the price it charges.
Partnerships are the only way to acquire payments customers at scale, and the most poorly executed GTM channel in the industry. The mechanics of structure, commercial terms, and post-launch management decide whether they work.
The cost of weak compliance is rarely on the compliance team's report. It is spread across churn, onboarding drop-off, support tickets, and banking partner notices, which is why it is consistently underestimated.
Compliance in payments is not a legal function or a cost centre. It is operational infrastructure, built as a five-layer stack from KYC through regulatory reporting, and how you build it early decides what you can scale later.
'Supporting' 150 countries and 'operating' a corridor are different things. Cross-border payments is dozens of businesses stitched together, each with its own costs and margins, and a company profitable on USD-GBP can lose money on USD-NGN at the same fee.
There are only three ways to acquire customers in cross-border payments: platform partnerships, direct self-serve, and enterprise sales. Each needs a different team, cost structure, and product, and trying to run all three at once is one of the most expensive mistakes a payments company can make.
Prices in cross-border payments fall for two reasons. One is healthy: lower costs. The other is a self-reinforcing spiral where aggressive acquisition pricing, thin margins, and retention discounts train customers to negotiate and leak your floor to competitors.
Retention in payments is not one problem. It is five structural forces holding customers in place and a set of sudden trigger events that push them out, which is why companies are so often surprised by churn.
The price on the pricing page has almost nothing to do with what you actually pay, and what you pay has almost nothing to do with what it costs to move your money. Payment pricing is five interlocking components, each controlled by a different actor.
A wave of payments M&A is framed as strategy, but the simpler reason is that valuations are down and buyers are moving on cheap assets. What every announcement skips is the multi-year cost of integration.
Fintech borrowed subscription pricing from SaaS, but payments has variable costs SaaS does not. A flat subscription forces the company to absorb those costs and inverts exactly when it wins larger customers.
A competitor can copy the UI, onboarding, dashboard, and fee in a year. A licence, and the compliance infrastructure that makes it effective, takes years. The product on top is the wrapper; trust is the real product.
Fintech decks lead with better UX as the advantage. In payments, UX is table stakes, never a moat. The real moats are licensing, banking relationships, pre-funded local accounts, and integration depth.
Around year three or four, payments companies start offering loans. The data advantage and margin maths are real, but the ease of building the first loan hides the discipline lending actually demands through the credit cycle.
When a B2B payments customer discovers their real all-in cost through a reconciliation, they don't leave over price. They leave because they feel misled. That is a trust problem, and the retention playbook cannot fix it.
A sales demo is not a solution to a confusing product. It is a symptom. Consumer payments solved self-serve a decade ago; B2B has normalised a workflow where sales teams exist to translate products that cannot explain themselves.
India broke the zero on UPI after six years of avoiding it. The fee framework is narrow by design, but crossing the free-to-priced line sets a precedent that matters more than the revenue it raises.
The November 2026 ISO 20022 deadline looks like a data schema change: free-text addresses become structured fields. Underneath it sits a large data migration, a customer-outreach problem, and a shift in where value is captured.
Stripe's acquisition of OpenRouter is not a crypto or AI side bet. It is the same infrastructure playbook it ran with Bridge, extended to own the routing layer of the AI economy.
The Nuvei-Payoneer deal is told as acceptance meeting payouts. That story is accurate and boring. The more interesting layers are the disputed premium, the interest income inside EBITDA, the overlapping headcount, and a stablecoin bet.
FX & Float