The absorption gap, seen across four sectors.
The same pattern, from government to financial services.
Spend time across South African industries and the same gap appears in every one. The technology is available. The capacity to absorb it is not. We call it the absorption gap, the distance between what AI could do for an organisation and what the organisation can actually take up. It is remarkably consistent from one sector to the next, which tells you the problem is structural, not specific. Here is the same pattern, seen four times.
Financial services
South Africa's banks and insurers are the country's most advanced adopters. They use AI for fraud detection, customer service, credit scoring, and compliance, and the largest banks are investing heavily, with over half planning to spend more than twenty million rand where most other institutions plan under one million. And yet, when the Financial Sector Conduct Authority and the Prudential Authority surveyed more than 2,100 institutions in late 2025, the single biggest constraint was not capital or technology. It was a shortage of skills. Even the most resourced sector in the country is held back by absorption, not access.
Government
The public sector sits at the other end. Here the gap is widest, because the data is fragmented and the processes are still largely manual. The work AI is best suited to, sorting, summarising, routing, and triaging high volumes of cases, is exactly the work that piles up in government. The technology to clear it exists and is not expensive. What is missing is the capability and the data foundation to put it to work. Where that gap has been closed, even partly, the results are dramatic, precisely because the starting point was so manual.
Healthcare
Healthcare shows the gap sharpened by sensitivity. The potential is enormous, from triage to administration to patient communication, but health data is special personal information under POPIA, and the Information Regulator issued draft rules for it in 2025. The absorption gap here is not only about skills. It is about trust and governance. A clinic cannot adopt what it cannot govern, so the constraint becomes the ability to deploy AI safely, not the AI itself.
Manufacturing and mining
On the floor, the gap is about people and digital maturity. Large operators run predictive maintenance and analytics, but the country's many small and medium manufacturers adopt slowly, held back by weak digital foundations, thin technical skills, and a workforce that is understandably wary. The machines can be instrumented. The question is whether the people running them can use what the instruments produce. That, again, is absorption.
The pattern
Four sectors, four starting points, one shared constraint. In each, the binding factor is not the model, the compute, or even the budget. It is the organisation's capacity to take the technology up, and that capacity is built, not bought.
This is why South Africa's position is so revealing. The country holds more than two-thirds of Africa's data-centre capacity and leads the continent on AI infrastructure. By the measure of access, it is ahead. By the measure of absorption, the same gap shows up everywhere you look. The infrastructure is necessary, and it is not the thing holding anyone back.
The practical implication is the same in every sector. If the constraint is absorption, then the highest-return investment is not another tool. It is the capability to use the tools you can already reach. Close the absorption gap, and the technology stops being the story.
Sources: FSCA & Prudential Authority, Artificial Intelligence in the South African Financial Sector (Nov 2025); Mastercard, Harnessing the Transformative Power of AI in Africa (2025); OECD, Africa Capital Markets Report (2025).
