African Banks Ramp Up AI Spending Even As One-Third Don’t Know If It’s Valuable
African banks are accelerating investment in artificial intelligence even as nearly one in three institutions cannot say whether the technology is generating value, a new report has found, highlighting a growing disconnect between spending and accountability as lenders race to modernise.
A survey of 277 senior banking executives across 37 African countries found that 83.2% of banks plan to increase AI investment over the next 12 months, according to the report by African Banker magazine in partnership with Backbase, a global leader in AI-powered banking platforms. Yet only 67.1% of institutions formally measure the return on those investments.
More strikingly, 82% of banks without any formal AI return-on-investment (ROI) framework still plan to expand spending, suggesting many lenders are committing more capital before proving existing projects are paying off. The findings point to a new phase in Africa’s AI adoption, where banks are shifting from experimentation to deployment under increasing pressure from boards and investors to justify technology spending.
“The question is no longer whether to invest in AI, but what the return on that investment is,” the report said, arguing that rising cloud costs, foreign exchange pressures and tighter data localisation rules are forcing banks to become more disciplined about digital investments.
Banks that do measure returns are largely being rewarded. Among institutions with formal ROI frameworks, 85.1% reported that AI projects either met or exceeded their original financial projections, while more than half said returns surpassed expectations altogether. Only about 15% said AI investments had failed to deliver anticipated value.
The report suggests that the problem is therefore not AI itself, but how banks govern and evaluate it. Institutions working with third-party AI vendors measure returns at more than twice the rate of those building entirely in-house, at 71.7% versus 31%, a gap the report calls the “partner premium”.
One of the more surprising findings is that senior executives responsible for approving technology spending are among the least likely to measure its success. Finance departments recorded the strongest accountability, with 82% tracking AI returns, followed by technology and innovation teams at 63.1%. Executive leadership measured ROI only 50% of the time, while risk and compliance teams performed even worse at 48.1%, despite being heavily involved in implementing AI systems.
Legacy architecture remains the sector’s single biggest constraint. Half of all respondents cited integration with existing systems as the primary internal obstacle. On average, 55.7 cents of every IT dollar spent by African banks goes toward maintaining legacy systems, even as nearly half of respondents rate those same systems as highly or fully capable of supporting AI, a gap the report identifies as a potential blind spot.
“African banks don’t have an AI problem; they have an architecture problem,” said Aymen Daoud, Regional Vice President for Africa at Backbase. “The institutions that treat integration as the plumbing to fix before scaling agents will spend less, comply more easily, and be the ones still standing when the current generation of models is inevitably replaced by the next”.
Fraud detection and transaction monitoring emerged as the most impactful AI use case, followed by credit scoring and alternative assessment for thin-file customers, an application the report identifies as a credible route to bringing more of Sub-Saharan Africa’s unbanked population into the formal financial system.
Despite the challenges, sentiment about AI’s role in African banking remains strongly positive, with 86.9% of respondents positive or very positive about its role over the next two years.
