That scene, mundane in its details, is quietly extraordinary in its implications. It represents the convergence of two of the most powerful forces reshaping Africa’s economic life: artificial intelligence, and the tentative but accelerating movement toward open banking, a framework in which financial data flows, with the customer’s consent, between banks, fintechs, and third-party providers through standardised digital interfaces known as application programming interfaces, or APIs.
Together, these forces are doing something that decades of donor programmes, microfinance initiatives, and bank-expansion drives failed to accomplish at scale: they are bringing the formerly unbanked and underbanked into the formal financial system, on terms that actually suit them.
| KEY STATISTICS: Africa’s Fintech Landscape | |
| Registered mobile money agents in Africa (2024) | ~28 million |
| Untapped credit demand across the continent | USD 330 billion+ |
| Projected African fintech revenue by 2030 | USD 65 billion |
| AI’s potential addition to Africa’s GDP by 2030 | USD 2.9 trillion |
| Additional GDP from inclusive AI deployment by 2035 | USD 1 trillion |
| African banking sector CAGR (2020-2024, constant currency) | ~17% |
| African banking sector ROE (2024) | 19% |
| African countries with functioning interbank transfer systems | ~26 of 54 |
THE INFRASTRUCTURE BET
For most of Africa’s history, the financial system was built around a simple assumption: that a credible customer was one with a physical address, a payslip, and a relationship with a branch manager. That assumption excluded the majority of the continent’s population.
Mobile money cracked that model open. M-Pesa, launched in Kenya in 2007 and now operating across seven countries, demonstrated that with a mobile phone and an agent network, it was possible to move money reliably at the margins of the formal economy. By 2024, Africa had approximately 28 million registered mobile money agents, according to industry data, creating a financial infrastructure that existed largely outside the traditional banking sector.
Open banking represents the next structural shift. Where mobile money built new rails for payments, open banking proposes to unlock the data those transactions generate, and make it available, with user consent, to any provider capable of using it. The theory is elegant: if a lender can see, in real time, that someone has been consistently receiving and spending money through a mobile wallet for three years, that record should function as a credit history. The lender no longer needs a salary slip. The algorithm has what it needs.
Nigeria became the continent’s most advanced jurisdiction in 2025, when the Central Bank of Nigeria released its Open Banking Implementation Framework, complete with technical standards, governance models,and defined rollout phases. According to fintech commentator Adedeji Olowe, the first real open banking APIs were expected to go live by August or September 2025,
representing, in his words, “a rare case of follow-through in a region that has grown used tostalling halfway.” South Africa has issued draft policy positions. Kenya’s Central Bank has articulated open banking as a strategic objective within its National Payment System Vision and Strategy 2021 to 2025, and has begun working to define API standards and mandate robust data portability. Ghana has used a regulatory sandbox, launched in 2023, to test open banking products in a controlled environment.
| “AI is no longer a tool; it has become a true engine of transformation.” — Attijariwafa Bank, AFIS 2025 Awards application |
The patchwork nature of regulation across the continent’s 54 countries is simultaneously the biggest obstacle and, some argue, the most important opportunity. Where regulation has been clear, markets have moved. Where it has been vague or absent, the results have been uneven.
THE CREDIT PROBLEM, SOLVED DIFFERENTLY
The problem that AI is solving in African finance is not new. It is the problem of information asymmetry, of lenders not knowing enough about potential borrowers to price risk accurately and therefore refusing to lend at all, or lending only at rates that make repayment nearly impossible.
Traditional credit scoring relies on a borrower having a documentable financial history. In much of Africa, that history either does not exist or exists only in analogue form, in a shopkeeper’s notebook, in the memory of a rotating savings group, in the receipt of a years-old market transaction. None of it was legible to a bank’s underwriting system.
AI changes the inputs. Machine learning models can be trained on non-traditional data: mobile phone usage patterns, the frequency and timing of airtime purchases, the consistency of mobile wallet top-ups, the geographic movement implied by location data, even utility payment records. These signals, invisible to a conventional loan officer, are highly predictive of creditworthiness when processed at scale.
The results in practice have been striking. In Ethiopia, more than 380,000 micro, small and medium-sized enterprises were able to access capital amounting to $150 million through uncollateralised credit facilities driven by AI credit scoring, according to 2025 data cited in an OECD report on Africa’s capital markets. In Zambia, a fintech firm has deployed AI-powered algorithms to provide uncollateralised credit by analysing data sources ranging from mobile transactions to digital footprints. In Kenya, firms are using AI to extend credit to digital financial services consumers and to support debt management.
| EXPLAINER: WHAT IS OPEN BANKING? Open banking is a regulatory and technical framework that allows customers to share their financial data, such as account balances, transaction histories, and payment records, with third-party providers via secure digital interfaces known as APIs. In theory, this breaks the monopoly banks have historically held over customer data and allows fintechs, insurers, and other providers to build services on top of it. The customer controls what is shared and with whom. Open banking differs from screen scraping, an older technique in which third-party apps accessed bank data by logging in as the customer, which carries significant security risks. Most advanced open banking frameworks, including Nigeria’s, require standardised APIs and formal consent management. |
Carbon, a Nigerian fintech, uses AI-powered chatbots and a mobile application to facilitate account creation and financial services without requiring customers to visit a physical branch. The company’s model, which analyses spending patterns and other non-traditional data signals, illustrates how the technology stack has become accessible even to relatively young companies operating in markets with limited formal data infrastructure.
The financial logic is compelling. Africa’s untapped credit demand stands at a minimum of $330 billion, according to industry estimates. If AI can reduce the cost and risk of lending to previously excluded populations, that demand does not simply represent a market opportunity for lenders. It represents, in the aggregate, an enormous expansion of productive capacity across the continent’s economies.
THE REGULATORY RACE
The tension between the speed of technological change and the pace of regulatory development is not unique to Africa. What makes the African context distinctive is that the countries building open banking frameworks are doing so largely without the luxury of stable legacy infrastructure to fall back on. There are no entrenched card-payment networks whose interests must be balanced. There are no decades-old banking relationships between affluent customers and trusted high-street institutions. The slate is not entirely blank, but it is far less encumbered than in Europe or North America.
That creates an opportunity for bolder choices. Nigeria’s framework is widely regarded as the most structurally complete on the continent. The Central Bank’s policy documents cite improving competition and access to financial services as primary objectives, and the operational guidelines are explicit about promoting innovation. According to the Nigeria Central Bank’s Policy Insight Series 2025, the country’s approach involves consolidating innovation policy, financial inclusion, system integrity, and cross-border ambition into a single framework, a degree of integration that observers say no other African regulator has attempted.
South Africa’s Financial Sector Conduct Authority has emphasised innovation, customer experience, and competition, particularly for fintechs, as policy objectives. The framework also aims to eliminate security risks associated with screen scraping, which remains prevalent in the market. Kenya’s Central Bank has committed to defining API specifications covering identification, verification, authentication, customer account information, and transaction initiation, anchored in the country’s Data Protection Act of 2019.
| “This is not incremental reform. This is the first time an African regulator has attempted to consolidate innovation policy, financial inclusion, system integrity, and cross-border ambition into a single framework.” — Capital Business, citing Nigeria’s CBN Policy Insight Series 2025 |
Kenya made further moves in 2025, launching its National Artificial Intelligence Strategy 2025 to 2030, which laid out a vision for building AI skills, infrastructure, ethical frameworks, and safety mechanisms. A draft AI Code of Practice and a forthcoming Robotics and AI Bill are expected to require registration for certain AI systems and impose transparency and documentation standards.
Rwanda and Ghana have developed regulatory sandboxes that allow fintechs to test products within defined parameters before full market release. According to data from the Datasphere Initiative, at least 25 sandbox programmes now operate across 15 African countries, spanning fintech, data governance, and artificial intelligence. Startups participating in sandboxes experience a 30 to 50 percent reduction in time to market, research suggests, and attract higher investment confidence because regulatory risk is demonstrably lower.
INFRASTRUCTURE MEETS INTELLIGENCE
For AI to work in open banking, the underlying infrastructure must be capable of generating and transmitting the data that models need. That requirement exposes a deep structural challenge. According to the OECD’s Africa Capital Markets Report 2025, the majority of financial institutions across the continent are working from fragmented or incomplete data records. Processing capability is often available only through third-party service providers. The diversity of regulatory frameworks across jurisdictions further compounds the problem by making it difficult to establish standardised approaches to data management and privacy.
The legacy systems problem is particularly acute. Many African banks were built on core banking platforms that were never designed to share data in real time. Integrating those systems with modern open APIs requires investment that smaller institutions, in particular, have struggled to justify.
Amsterdam-headquartered banking platform provider Backbase launched what it described as the world’s first fully AI-powered banking platform in 2025, building on an Intelligence Fabric technology it introduced in 2024. The platform is designed specifically to address the problem of fragmented data in legacy systems, translating customer behaviour, transaction records, and operational patterns into actionable intelligence in real time. The company identified Nigeria, Kenya, Egypt, and South Africa as priority markets, and made the platform available for African deployment later that year.
South Africa’s banking sector has moved further along the AI adoption curve than most. Banking institutions are the leading adopters of AI in South Africa’s financial sector, with adoption rates exceeding 50 percent, according to a 2024 survey conducted jointly by the Prudential Authority and the Financial Sector Conduct Authority. The survey covered some 2,100 respondents across banking, insurance, investments, payments, pensions, fintechs, and lending. The banking sector demonstrated a strong inclination toward substantial AI investment, with more than half of banks planning to commit more than R20 million to AI in the same period.
| AI ADOPTION IN SOUTH AFRICA’S FINANCIAL SECTOR (2024) | |
| Banks using AI in financial operations | >50% |
| Payments institutions with AI adoption | ~50% |
| Banks planning AI investment of R20m+ (2024) | >50% |
| Respondents in joint PA/FSCA AI survey | ~2,100 |
| Primary AI application in banking | Operations & IT |
| Primary GenAI application in banking | Sales & Marketing |
Morocco’s Attijariwafa Bank, operating in 27 countries with recent expansion into West Africa and Egypt, won the Bank of the Year title at the AFIS 2025 Awards in part on the strength of its AI strategy. The bank’s submission stated that AI had become a true engine of transformation rather than merely a tool. The bank recorded net banking income growth of 15 percent in 2024, reaching the equivalent of $3.3 billion, and consolidated net profit rose 29 percent to $1.1 billion. While those gains cannot be attributed solely to AI, the bank’s leadership has been explicit about the technology’s role in driving operational efficiency and customer acquisition.
South African neobank TymeBank passed 10 million customers in 2024, according to McKinsey’s snapshot of African banking, achieved largely through a network of more than 1,000 in-store onboarding kiosks and around 15,000 retail points. Approximately 85 percent of new accounts were opened at kiosks rather than through a traditional branch, a model that depends entirely on digital identity verification and AI-driven onboarding automation.
THE PAN-AFRICAN AMBITION
The most ambitious version of the open banking and AI story in Africa is not about individual countries or individual lenders. It is about the possibility of a connected, interoperable financial system spanning the continent, in which data, credit histories, and payment rails flow across borders as freely as they do within the most advanced domestic markets.
That ambition is still largely aspirational, but the building blocks are being laid. The Pan-African Payment and Settlement System, backed by the African Export-Import Bank and the African Union, is designed to enable instant cross-border payments within the African Continental Free Trade Agreement framework. Nigeria’s Central Bank has moved to remove transaction limits on the system and permit authorised dealer banks to source foreign exchange through the Nigerian Foreign Exchange Market.
The African Development Bank’s December 2025 report on AI productivity gains projected that inclusive AI deployment could generate up to $1 trillion in additional GDP by 2035, equivalent to roughly one third of the continent’s current economic output. The three-phase roadmap outlined in the report begins with an ignition phase running from 2025 to 2027.
Ousmane Fall, Director of Industrial and Trade Development at the African Development Bank, put it bluntly in the report: “Africa’s challenge is no longer what to do. It is doing it on time.”
| “Africa’s challenge is no longer what to do. It is doing it on time.” — Ousmane Fall, Director of Industrial and Trade Development, African Development Bank |
UNESCO has estimated that while AI could add $2.9 trillion to Africa’s economy by 2030, the continent currently captures just 2.5 percent of the global AI market and a tiny 0.3 percent of projected worldwide AI investment. That gap is the central challenge facing policymakers, investors, and technologists who see the opportunity clearly but have yet to translate it into the infrastructure and capital that would make it realisable.
RISKS AND FAULT LINES
None of the optimism around AI and open banking in Africa is unchallenged. The risks are real, and some of the most credible warnings come from practitioners who are enthusiastic about the technology’s potential but clear-eyed about what can go wrong.
Data quality is the foundational problem. AI models are only as good as the data on which they are trained. In markets where financial records are patchy, where identity verification is unreliable, and where the data that does exist reflects historical inequities, models trained on that data will reproduce and potentially amplify those inequities. An AI credit-scoring system trained primarily on urban, male, formally employed borrowers will almost inevitably underserve rural women running informal businesses, regardless of how creditworthy they actually are.
The collapse of Nigerian fintech startup Okra in 2025 was a sobering reminder that brilliant technology without a viable ecosystem model can still fail. Okra had positioned itself as a central infrastructure provider for open banking in Nigeria, aggregating financial data on behalf of fintechs and developers. Its shutdown forced a reassessment of the business models underpinning the open banking ecosystem, and in particular of whether the value chain was structured in a way that could sustain companies at every layer.
Cybersecurity is another persistent concern. As financial data becomes more interconnected and more valuable, it also becomes a more attractive target. The South African Reserve Bank flagged cyber risks and related costs as a significant impediment to successful AI deployment in financial services in its 2025 assessment. Kenya’s Central Bank has called explicitly for clear risk management frameworks from every open API participant.
There is also the question of who benefits. The history of technology-driven development initiatives in Africa is full of innovations that reached the top of the income distribution and did not descend further, or that extracted value from low-income users without proportionately improving their welfare. The risk that AI-powered open banking becomes primarily a tool for harvesting data from previously excluded populations, rather than genuinely including them, is not hypothetical.
| EXPLAINER: THE CONSENT CHALLENGE Open banking frameworks live or die on consent infrastructure: the systems and interfaces through which customers understand what data they are sharing, with whom, for how long, and for what purpose. In markets with high financial literacy gaps, complex consent interfaces can become instruments of manipulation rather than protection. Several African frameworks, including Kenya’s developing approach, emphasise user-centric consent design, including explicit, granular permissions and consent dashboards. The effectiveness of those designs in practice depends heavily on the quality of their implementation and on the digital literacy of the populations they are meant to serve. |
WHAT THE NUMBERS CANNOT CAPTURE
The McKinsey snapshot of African banking published in early 2026 noted that the continent’s banking sector expanded at a compound annual growth rate of approximately 17 percent on a constant-currency basis from 2020 to 2024, compared with a global average of 7 percent. African banks delivered a return on equity of 19 percent in 2024 and 17 percent in 2025, well above the global average of 10 percent. The sector’s revenues grew from $81 billion to $99 billion in dollar terms over the same period.
These are impressive numbers, but they do not capture the texture of what is actually changing. They do not capture the women in Eastleigh’s Garissa Lodge market who now access short-term credit without a male guarantor. They do not capture the smallholder farmer in Zambia who received an uncollateralised loan because an algorithm assessed his airtime purchase history and found it creditworthy. They do not capture the small-business owner in Lagos who, for the first time, can share her transaction history with a new bank without having to physically collect and present months of statements.
The African fintech market is projected by Global FinTech to generate $65 billion in revenue by 2030, a thirteenfold increase from current levels. The African Development Bank’s analysis suggests that AI alone could contribute up to $1 trillion in additional GDP to the continent by 2035. Those figures describe magnitude, but they do not describe meaning.
The meaning is in the access. In the question of who gets to participate in the formal economy, on terms that do not require them to already be wealthy or already be documented or already be known to a bank manager. The algorithm, at its best, asks a different question: not who do you know, but what do you do? It is a question that, for the first time in the history of African finance, large numbers of previously excluded people can answer.
Back in Eastleigh, Amina Hassan repaid her first loan in full within three weeks, slightly ahead of schedule. She has since taken out two more, each slightly larger than the last. Her credit profile, built entirely from digital transaction data, is being updated in real time. Somewhere in a server stack, the model that lent her money is learning from her.
She does not know the name of the algorithm that assessed her. She does not know which bank sits behind the fintech application on her phone, or which open banking framework made the data sharing possible. She knows that she needed capital, that she asked for it through a phone, and that it arrived. The machinery of artificial intelligence and open banking, in her daily life, is simply invisible infrastructure, working quietly, in the background, on her behalf.
Whether it continues to work on her behalf, rather than merely through her, is the question that will define the next decade of financial technology in Africa.
No comments:
Post a Comment