A 2026 survey of 277 banking executives across 37 African countries found that 83.2% of institutions plan to increase AI spending over the next 12 months. Yet nearly one-third do not formally track the technology’s return on investment (ROI). Among banks that do not measure AI returns, 82% still intend to increase their budgets.

The findings highlight the tension at the centre of Africa’s multi-million-dollar banking automation push. Banks are investing in AI despite dollar-denominated cloud bills, legacy core-banking integration constraints and a shortage of machine-learning engineers. In this edition of Delve into AI, EMMANUEL NWOSU examines the gap between faster operations and measurable balance-sheet returns.

Loan documentation has historically been an administrative bottleneck, including at well-capitalised financial institutions. At Standard Bank, a vetting process that previously took roughly an hour can now be completed in about 60 seconds using artificial intelligence.

The South African lender disclosed the figure in its March 2026 Capital Markets Day presentation. The metric applies to the bank’s commercial asset-finance division and illustrates how Africa’s largest bank by assets is applying machine intelligence to established processes.

Standard Bank also reported that its conversational bot handles 65% of incoming digital customer queries. By August 2026, more than 39,000 employees—about 72% of its workforce—were regular enterprise users of generative AI.

These examples show the immediate operational potential of automation. Administrative backlogs can be reduced to minute-long tasks, customer friction can decline, and employees can be redirected towards advisory work that requires human judgement.

The financial benefit, however, remains less certain. Deploying AI in regulated financial environments requires data-governance controls, software licences, systems integration and workforce reskilling. To justify those costs commercially, banks must demonstrate that the financial gains exceed the expense of implementation and operation.

Faster work, uncertain financial returns

The State of AI in African Banking report, a 2026 joint study by banking-software provider Backbase and African Banker, surveyed executives in 37 African countries. It found that 83.2% of institutions planned to increase their AI allocations during the following 12 months, while only 67.1% formally tracked ROI from those deployments.

Among the roughly one-third of respondents whose banks did not track AI returns, 82% nevertheless planned to increase their budgets. The figures indicate that many banking leaders are expanding a technology whose commercial contribution they have not yet quantified.

The absence of formal measurement does not necessarily mean that AI projects have failed. Algorithmic systems can shorten processing times, improve fraud monitoring and enhance customer service without immediately reducing costs on the income statement. But the difference between operational efficiency and financial return becomes more important as AI moves from isolated pilots into systems used daily by banks.

Standard Bank has built its AI capabilities on technology investments made over several years. In its March investor presentation, the bank said conversational AI was handling 65% of digital queries on its personal-banking app. It also reported that engineering productivity had improved by more than 20% over the previous year, while AI was beginning to accelerate lending-related work.

The commercial asset-finance example, in which loan-documentation processing fell from roughly an hour to about one minute, demonstrates the potential improvement in some administrative workflows. Standard Bank has also expanded access across its workforce. In August, it reported more than 39,000 active users of generative AI operating through approved enterprise tools.

Reducing a documentation process from an hour to a minute frees the employees who previously performed that work. The bank might process more applications with the same workforce, reduce overtime or assign employees to other responsibilities. Those operational benefits become measurable financial savings only when they produce avoided expenses or additional income that can reasonably be attributed to the investment.

Standard Bank said it planned to redirect time saved through AI towards customer relationships and work requiring more complex decisions. That approach looks beyond direct cost-cutting and seeks to generate more value from the existing workforce. The statement does not, by itself, specify revenue generated or costs saved.

In lending, AI could allow employees to process more applications without a corresponding increase in staffing costs. If used to assess borrowers’ creditworthiness, it could also support lending decisions and potentially reduce defaults. These are separate benefits, and Standard Bank has not established that its AI investments have delivered either one financially.

The bank’s non-performing-loan ratio was 5.6% at the end of June, compared with 6.2% a year earlier. Credit-impairment charges, which reflect expected losses on loans, fell 12% to R7.13 billion ($435.5 million) in the first half of 2026. Standard Bank has not attributed those improvements to AI use.

Nedbank, another tier-one South African lender, has reported measurable operational improvements. In its 2025 integrated report, the bank said it implemented 30 advanced AI use cases during the year. Its broader programme, covering AI, generative AI, analytics, machine learning and robotic process automation, saved approximately 312,000 employee-hours across media and content creation, document translation and personal productivity. Global payment-processing times also fell by about 30%.

Nedbank has used predictive models to identify customers more likely to purchase investment products. It has also applied analytics to target potential credit-card and loan customers. These applications reflect two broad ways banks expect AI to generate returns: by reducing the cost of existing services and by helping them serve customers more effectively.

The cost of making AI work

AI does not operate independently of a bank’s existing technology. Models require reliable data, secure computing systems and access to the applications used by employees and customers. Banks also need procedures to test outputs, restrict access to sensitive information and monitor decisions made with AI.

As a result, the cost of deploying AI can be considerably higher than the price of a software subscription. At its 2026 Capital Markets Day, Standard Bank reported total technology spending of R23.5 billion ($1.44 billion) in 2025, equal to 27.7% of operating expenses.

The bank said 88% of its legacy servers had been decommissioned and that 71% of computing workloads eligible for migration had moved to the cloud when the presentation was delivered. It later reported that cloud migration had reached 78% of eligible workloads by the end of June 2026.

Standard Bank said technology investments over several years had helped improve efficiency and contributed to a decline in its group cost-to-income ratio from 59.1% to 50.2%. In practical terms, the bank spent about R50 ($3.05) on operating expenses for every R100 ($6.09) of operating income, compared with roughly R59 ($3.60) previously.

Technology spending continued to increase as the bank expanded its AI operations. Standard Bank spent R23.51 billion ($1.43 billion) on technology in 2025, up 4.9% from R22.4 billion ($1.36 billion) in 2024. During the first half of 2026, software, cloud and technology-related expenses rose 6% to R7.16 billion ($436.5 million). The bank attributed part of the increase to expanding AI capabilities, along with investments in specialised technology expertise, core-banking systems and cybersecurity.

Overall technology spending rose by only 2% to R11.83 billion ($721 million) during the six months because lower amortisation and other technology costs offset part of the increase. Standard Bank has described its approach as a “save-to-invest” strategy, using efficiencies from earlier technology investments to fund newer capabilities, including AI.

The modernisation programme illustrates the work required to introduce new capabilities into an established banking business. In the Backbase and African Banker survey, 50.2% of respondents identified integration with existing technology as their biggest internal obstacle to AI adoption.

Banks may need to connect customer records, transaction histories and other information stored in systems developed at different times. A lender introducing AI into credit assessment, for example, requires reliable access to the financial information used to evaluate borrowers. Consolidating that information can require upgrades to legacy systems, additional computing infrastructure and specialised employees to maintain it.

These requirements increase both the initial and ongoing cost of AI. Banks that purchase cloud services or software in foreign currencies may also face higher local-currency expenses when exchange rates move against them. The financial case for AI therefore depends on more than the time a system saves. Banks must weigh those gains against the full cost of introducing and operating the technology.

Why the returns remain difficult to measure

The Backbase and African Banker research suggests that some African banks are achieving the financial results they expected. Among respondents whose institutions formally measured returns, 85.1% said their investments had met or exceeded projections.

Those banks were measuring performance against their own targets. The survey does not establish how much each institution invested, the financial value generated or whether returns were calculated consistently across organisations.

AI-powered lending illustrates the difficulty. If a bank approves more applications after introducing AI, the additional loans could increase interest income. The financial benefit would still depend on funding costs, the performance of those loans and whether the technology improved lending decisions. Similarly, the value of an AI fraud-detection system lies in losses prevented, not simply in the number of suspicious transactions it identifies.

The survey found that 71.7% of respondents working with third-party AI providers measured returns, compared with 31% of those developing AI entirely in-house. The difference reflects the varying ability of banks to build proprietary AI systems, which can itself become a significant cost centre.

In June 2026, Evident’s inaugural AI Index for Banks in the Middle East and Africa assessed 25 major banks on AI talent, innovation, leadership and transparency. Standard Bank ranked second and Nedbank fourth, making them the only two South African banks among the region’s four leaders. FirstRand, Capitec and Absa were also among the top 10.

Evident attributed the comparatively strong performance of these banks partly to larger pools of AI talent, proximity to research universities and engagement with emerging central-bank regulation. The results suggest that South African banks have developed stronger AI capabilities than many regional peers.

Across the 25 institutions, however, Evident found that specialist AI-development talent relative to workforce size was only half the level recorded in its benchmark of major global banks. The shortage has financial implications. Banks without sufficient internal expertise must either invest in developing their own capabilities or rely on external providers, while continuing to pay for integration and maintenance.

Some banks outside Africa have provided a clearer account of AI’s financial contribution. Singapore’s DBS Bank reported approximately S$1 billion ($783 million) in economic value from data analytics and AI in 2025, up from S$750 million ($587 million) in 2024. By the end of that year, DBS had deployed more than 2,000 models across over 430 use cases.

DBS has gone beyond reporting the number of tasks AI completes or the time employees save. It has assigned a monetary value to the technology’s contribution to the business, although it is unclear whether the bank is profitable on its AI usage.

The comparison clarifies the financial question facing African banks. Standard Bank and Nedbank have demonstrated productivity improvements, and many institutions surveyed by Backbase report meeting their AI investment targets. But banks provide limited information about the cost of individual AI programmes or the financial returns they generate.

Productivity is a major AI use case in banking and in other non-financial organisations. The challenge is to establish what that productivity is worth and how much of it can be attributed to AI. As banks expand beyond pilots, the clearest evidence of progress will be more consistent disclosure of programme costs, measured outcomes and the link between those outcomes and financial performance.

Source: TechCabal