What M-Pesa Knew in 2007

In 2007, a Kenyan mobile operator launched a way to send money by text message. The project had started as something narrower — a donor-funded pilot for repaying microloans — until the team noticed users doing something unplanned: sending airtime and balances to family upcountry. Safaricom renamed the product around what people were actually doing. The tagline was three words: Send money home.

Within eight months, M-Pesa had a million users. Within a decade it had most of Kenya’s adults, and the annual value flowing through it was equivalent to roughly half the country’s GDP. A nation where the overwhelming majority had never held a bank account became one of the most financially connected places on Earth — leapfrogging the branch, the card, and the check entirely. It remains the most successful money product of the mobile era.

And the technology was almost comically primitive: SMS and SIM cards. No smartphones, no apps, no internet. Two-thousand-and-seven phones. The breakthrough wasn’t in the phone at all.

The product was never the phone

M-Pesa’s actual product was its agent network: the corner shops, kiosks, and market stalls — eventually numbering in the hundreds of thousands — where a person could hand cash to another person and watch it become balance, or hand over a phone and walk away with cash. For scale: Kenya has on the order of 1,500 bank branches and a few thousand ATMs. M-Pesa put a functioning teller within walking distance of nearly every Kenyan, by deputizing infrastructure that already existed.

Look closely at what the agent actually is, because the design is deceptively deep. The shop already existed — no real estate to build. The trust already existed — people had bought airtime and groceries from the same vendor for years; the first deposit, the terrifying one, happened because a familiar face was accountable for it. The cash-handling skill already existed — counting, securing, and banking paper was already the shopkeeper’s daily job. And the incentives aligned: agents earned commissions on every deposit and withdrawal, and the foot traffic lifted the rest of the shop’s sales. M-Pesa didn’t build a banking system. It recruited one that was already standing there, wearing an apron.

The numbers nobody believed

The economists went looking for effects and found them everywhere. The most cited study, published in Science, estimated that access to M-Pesa lifted roughly two percent of Kenyan households out of extreme poverty — with the strongest effects among female-headed households, who gained a private, safe place to hold money for the first time. Households with access absorbed shocks better: when illness or drought hit, money arrived from a wider network, faster, and consumption didn’t collapse. Savings moved out of mattresses and livestock into balances. None of this required a new economy — it required removing the friction from money the economy already had.

For our purposes, though, the most important number is the simplest: the agent count. Adoption tracked the network, corridor by corridor, year by year. Where you could walk to an agent, M-Pesa worked. Where you couldn’t, it didn’t — no matter how good the phone menu was. Distribution wasn’t a channel for the product. Distribution was the product.

Why it stayed home

Here is the puzzle: if M-Pesa was that good, why isn’t it everywhere? Vodafone tried. Exports to India, South Africa, and Eastern Europe mostly failed and were wound down. The reasons are instructive. The agent economics that worked in Kenya — commissions, float, density — didn’t transplant automatically. Incumbent banks and card rails fought back in markets where they were entrenched. And the product was structurally provincial: a closed loop, run by one telecom, denominated in one national currency, stopping at the border.

That last limitation matters most. M-Pesa digitized the Kenyan shilling — it never offered an escape from it. A balance that inflates alongside the paper it replaced protects no one’s savings. Transfers across borders still meant the remittance industry and its tolls. And every country needed the miracle rebuilt from scratch, inside its own telecom, under its own regulator, in its own silo. M-Pesa proved the model and hit its ceiling in the same decade: the right distribution, wired to money that couldn’t travel.

The internet dollar has no corner shop

Now invert the picture. Stablecoins are money with the opposite profile: global instead of national, dollars instead of a depreciating local unit, open rails any company can build on instead of one operator’s ledger. In every dimension where M-Pesa was structurally limited, the internet dollar is structurally unlimited.

Except one. Stablecoins have world-class everything above the ground — issuers, chains, wallets, law — and almost nothing at street level. The street layer that exists is a bad joke by Kenyan standards: crypto ATMs number a few tens of thousands worldwide, concentrated in rich-country cities, and routinely charge double-digit spreads — toll booths dressed as doors. Exchanges serve the banked by design. Nineteen years after a Kenyan telecom proved that the last meter is the entire product, the most advanced money network ever built still hasn’t shipped its corner shop.

M-Pesa put a teller in every corner shop. The internet of money needs the same.

Deputize, don’t build

The blueprint hasn’t changed, because the physics of cash haven’t changed. You don’t build branches; you deputize what exists. The shop that takes cash all day is a cash-in point waiting for software. The vendor the neighborhood already trusts is the onboarding flow. The commission that makes an agent say yes is the entire go-to-market. What’s different this time is what flows through the network: not a national currency in a closed loop, but dollars on open rails — money that holds value, crosses borders for cents, and works with any wallet, any app, anywhere.

That is what Guap’s teller network is: the M-Pesa move, applied to the open dollar. People and machines, embedded where cash already changes hands, turning paper into internet dollars and back — no account, no paperwork, no distance. The agent earns like a Kenyan agent earned. The user gets what the Kenyan user never could: money that leaves the island.

The rails are done. The law is written. The wallets are good. What the internet of money needs now is what Kenya had in 2007: somewhere to walk to.