The payments layer buys its plumbing

A pattern has repeated three times this month in different markets: the company that sells the customer experience buys the infrastructure it had been renting.

Kredete, which provides payments, savings and stablecoin-based products to emerging-market and diaspora customers across more than 60 countries, said on Monday that it has acquired Gravv, a settlement network connecting banks, blockchains and local payment rails. Terms were not disclosed.

The scale figures explain the logic. Kredete says it serves more than 6.5 million users and processes $18bn of payments on an annualised basis, against $24.75m raised. Gravv says it handles more than $3bn annualised for over 1,000 businesses in more than 100 countries, with seven global banking partners.

A cross-border payment is not one transaction. It is a chain: collect in one currency, convert, move value across a border, deliver into a local rail, reconcile, and hold liquidity at both ends so nobody waits. The parts that look like technology are mostly parts that look like treasury.

That is why owning the settlement layer matters more than it sounds. A company that routes through somebody else's network pays a spread on every leg and inherits that provider's banking relationships, currency coverage and outages. Buying it converts a variable cost into a fixed asset.

The stablecoin element is the reason this can be bought at all. The settlement leg that used to require a correspondent bank in each corridor can now be a token transfer, which is what makes a network like Gravv assemblable by a startup rather than by a consortium.

The word the companies use for the result is "agentic" — software that watches liquidity, picks routes, manages currency exposure and reconciles without a person deciding each time. Adeola Adedewe, Kredete's chief executive, put the thesis in one line: "Infrastructure is the moat, and intelligence is the operating layer."

Treat the second half of that sentence as a claim rather than a fact. Automated routing across fragmented rails is genuinely hard, and the failure mode is not a bad exchange rate but stranded liquidity in a jurisdiction on a Friday afternoon.

The regulatory backdrop is less friendly than it was a week ago. Brazil has just barred stablecoins from settling bulk currency flows, and any operator whose economics depend on the token leg has to assume more of that, corridor by corridor.

Which is the argument for buying the rail rather than renting it. A company that owns its settlement network can reroute a corridor through licensed currency transactions when a regulator closes the token path. A company that resells someone else's waits to be told what its product does next.