African Fintech: The Choice
Africa is building financial infrastructure from scratch, and the early returns look like payday lending with better UX.
FairMoney is Nigeria's most profitable neobank. It earned ₦121.9 billion in 2024, up 62% from the year before. Ninety-five percent of that revenue comes from lending interest. Monthly rates run 2.5-30% with origination fees of 3-15%, translating to effective APRs of 46.8-108%.
Its non-performing loan ratio is 86.8%.
I wrote earlier this year that financial inclusivity is a myth because the Western financial system is structurally designed to benefit people who already have money. The argument was about the US: overdraft fees as a poverty tax, credit scoring built on access to credit, payday lending that traps 12 million Americans in annual cycles at roughly 400% APR. Tokenization doesn't fix it either — the same KYC gates and walled gardens re-form on new rails.
Africa was supposed to build something different. No legacy baggage. Over 40% of Sub-Saharan adults are unbanked, and the infrastructure is still being designed from scratch. But the early returns on what's actually getting built look uncomfortably familiar.
The math didn't change
Eight million Kenyans are negatively listed with Credit Reference Bureaus, most from unpaid mobile loans. Digital borrower default rates hit 50.9%. M-Shwari's facilitation fees push annualized rates to 395%. Collection practices include public shaming, contacting employers, and mining phone contacts without authorization to pressure defaulters through social humiliation.
This is the payday lending cycle. Short terms, high rates, repeat borrowing, aggressive collection. The same math, running on M-Pesa instead of a storefront on Martin Luther King Boulevard.
The best estimate of microcredit's impact on poverty, synthesized across six randomized controlled trials by Innovations for Poverty Action, is zero. Microcredit expanded some business activity in some contexts. It did not reduce poverty. It did not empower women. The poorer the client, the more likely they fell deeper into poverty.
Milford Bateman calls microfinance "a powerful institutional barrier to sustainable local economic and social development." Grameen Bank, the original model, charges 2-5x commercial bank rates to serve the poor. A Bangladeshi court questioned in 2026 why those rates shouldn't align with commercial banks. Poverty levels in Bangladesh haven't moved since the 1970s despite decades of microfinance.
"Financial inclusion" is the phrase that makes this sound like progress. But inclusion into what? A system where 86.8% of borrowers can't repay is a system that profits from default, not from lending.
What prepaid got right by accident
Mobile money wallets don't have overdrafts. You can't spend more than you have, and there's no minimum balance, no monthly maintenance fee, and no penalty for low activity.
American banks extracted over $12 billion in overdraft and NSF fees in 2025. Twenty-three percent of households earning under $25,000 paid overdraft fees versus 6% of those earning over $150,000. The CFPB tried to cap overdraft fees at $5 in 2024. Congress reversed it in 2025.
Africa's mobile money base layer avoided this entirely. Not by policy choice, but by accident of architecture. A prepaid wallet can't overdraft because the product wasn't designed to extend credit. The constraint that seemed like a limitation is actually structural protection.
Wave charges 1% flat on transfers. Free cash-in, free bill payments. Compare that to US check-cashing fees of 2-5% or the $3,000 per year the average unbanked American pays in alternative financial product fees.
The question is whether the layers built on top of mobile money preserve that protection or undermine it. M-Shwari, Fuliza, and their equivalents are credit products bolted onto prepaid infrastructure. They reintroduce the extraction that the base layer excluded.
Three architectures that don't extract
Alternatives exist, some old and some already scaling.
Community savings groups predate formal banking across Africa. Three hundred thousand chamas in Kenya hold roughly $4 billion in assets collectively. Stokvels in South Africa manage R50 billion (roughly $2.5 billion at current exchange rates) annually. Tontines are ubiquitous across WAEMU.
These systems run on social trust instead of collateral. The group builds assets together rather than borrowing against future income.
MoneyFellows in Egypt digitized the gam'eya, Egypt's version of the ROSCA, to 8.5 million users and reached profitability. Average payout per user nearly doubled from $453 to $906. It raised over $60 million.
The revenue model is subscription and facilitation fees on group savings, not interest on loans.
Parametric insurance removes the mechanism that makes traditional insurance extractive. Satellite data triggers automatic payouts. No claims process. No adjuster. No denial. Pula has worked with 4.3 million farmers across 13 African markets and paid out over $40 million to 900,000 farmers. Research shows insured farmers increase investment by 16%, improve yields by 56%, and increase savings by up to 170%.
The product works because the trigger is external and verifiable. The insurer can't game the payout because the satellite data is public.
Cooperative finance returns profits to members instead of shareholders. Kenya's SACCOs hold KSh 1.21 trillion ($9.3 billion) in assets with 7.39 million members. 267 digital SACCOs offered 407 digital financial products in 2025.
Interest rates are set by member governance rather than venture capital return requirements, and the borrowers and the capital providers are the same people.
The design choices still open
Africa's financial infrastructure is still being designed. Architectural decisions being made now will determine which model wins.
Payment rail ownership is one. UPI, Pix, and PI-SPI are public infrastructure, free or near-free for small transfers. When India built UPI as a public good, hundreds of fintechs built on top of the free rail.
WAEMU's PI-SPI mandates free P2P transfers up to roughly $14 per day. The alternative is M-Pesa's model, where a private platform owns the rail and extracts at the infrastructure layer.
Interest rate regulation is another. Kenya capped rates in 2016 and MSME loan growth dropped from 15% to 6%. GDP fell 1.4%. Banks shifted to government securities. The cap was repealed in 2019.
BCEAO caps WAEMU rates at 14% for banks and 24% for others. Neither extreme works on its own, because caps protect borrowers while excluding them from credit, and removing caps permits the kind of predation that produces 86.8% default rates. The answer is the architecture around the number: data portability, borrower rights, mandatory cooling-off periods, CRB reform.
Data ownership is the unsettled question. Fintechs harvest mobile money transaction patterns, call logs, app usage, typing speed. Forty-five African countries now have data protection laws, and Nigeria fined Meta $220 million.
But consent models assume literate, digitally savvy users. When a first-time mobile money user taps "agree" on a lending app that mines their contacts for collection leverage, the consent is technically given and practically meaningless.
Inclusion versus extraction
Can infrastructure built for inclusion survive contact with capital markets that demand extraction?
VCs want returns. The African VC recovery of 2025 came with hardened expectations around profitability. FairMoney delivers: profitable, growing 62% year-over-year.
The unit economics work by lending at 46-108% APR to borrowers who default at 86.8%. That attracts growth capital.
MoneyFellows is also profitable. On group savings facilitation. With $60 million raised. But it took longer and the revenue per user is lower. The VC model has a preference, and the preference is margin.
Parametric insurance, cooperative finance, and digitized savings groups can generate sustainable returns. They probably can't generate the 10x multiples that make a VC fund's portfolio math work. That constraint shapes what gets built. The infrastructure that VCs fund is the infrastructure that scales.
What the architects should build
Default to savings, not lending. MoneyFellows proved the model at 8.5 million users. Digitize chamas, esusu, tontines, stokvels. Build credit histories from savings behavior, not from debt. When someone demonstrates two years of consistent contributions to a savings circle, that's a stronger credit signal than a nano-loan repaid under social coercion. Fund this path. It's slower to scale and harder to pitch to a growth-stage VC. It also doesn't produce 86.8% default rates.
Make payment rails public goods. PI-SPI, not M-Pesa. When India built UPI as public infrastructure, the rail became free and innovation happened above it. When the rail is privately owned, the owner extracts at the infrastructure layer. WAEMU's PI-SPI mandates free P2P transfers under $14 per day. Every African central bank building a real-time payment system should follow this model. The rail is a utility. Charge for what runs on it, not for the pipe.
Parametric by default for insurance. No claims process. Satellite triggers. If the rainfall index drops below the threshold, the payout hits the farmer's M-Pesa within 72 hours. Pula has proven this works across 13 markets and 4.3 million farmers. The structural advantage is that the insurer can't deny what a satellite already measured. Scale this to health, livestock, and property.
Data ownership as a right, not a checkbox. Users own their financial transaction history. Portable between providers. Not a consent form that a first-time mobile user taps through without reading, but actual ownership: the ability to export, delete, and control who accesses your financial behavior. Nigeria's open banking framework gestures at this. Make it mandatory, enforceable, and real.
Mandatory APR transparency. Every lending product publishes its effective annualized rate in a standardized format. No hiding 395% behind "facilitation fees." No splitting the cost across origination charges, monthly rates, and penalty structures that take a calculator to untangle. Let borrowers compare. Let the 86.8% default rates explain themselves in public.
None of this requires new technology. The mobile money rails exist. The savings groups exist. The satellite data exists. The open banking frameworks exist in draft. What's missing is the decision to build infrastructure that serves users instead of extracting from them.
Africa is still designing its financial infrastructure from scratch. The choices being made right now, in licensing decisions, in regulatory architecture, in which business models get funded, will determine whether the result serves the people using it or extracts from them the way the systems they were supposed to replace always have.
Sources
- FairMoney 2024 Revenue Data (TechCabal) - Revenue breakdown, lending interest share, NPL ratio
- Innovations for Poverty Action: Evidence on Microcredit - Synthesis of six RCTs on microcredit's impact on poverty
- Burden of Digital Lending in Kenya - CRB blacklisting, default rates, collection practices
- NCLC: Overdraft and NSF Fees Rise Above $12 Billion - US overdraft fee data, demographic breakdown
- MoneyFellows Raises $13M (TechCrunch) - ROSCA digitization, profitability, user growth
- Pula Series B and Farmer Impact Data (TechCrunch) - Parametric insurance scale, payout data, yield improvements
- Kenya's SACCOs and Fintech Partnerships - SACCO asset data, digital transformation
- The Africa Report: Microfinance Is Menace Not Saviour - Bateman's critique, Grameen rate analysis
- Interest Rate Caps Have Caused More Harm to SMEs (CBK) - Kenya rate cap experiment outcomes

Writes Ledger Drift. Engineer building AI systems and financial infrastructure. Open source contributor to x402, ACK, and MPP – agentic payment protocols.
Frequently Asked Questions
Built by Trio, a fintech-native engineering partner helping teams build the next generation of financial technology and infrastructure.
Subscribe to Ledger Drift for high-signal insights into how modern fintech is built, from systems to code to teams.