JOHANNESBURG, 2026. Africa’s ability to shape its own AI future depends less on acquiring cutting-edge tools and more on building the unglamorous foundations that make those tools work. That was the clear message from the CNBC Africa AI Summit 2026, where speakers from banking, energy, security and investment spent far less time on AI’s transformative promises than on the preconditions ordinary citizens and institutions need before transformation can happen.
The infrastructure gap is the most immediate obstacle. Much of Africa’s physical backbone still operates on analogue systems, with ageing power grids and communications networks struggling to deliver reliable energy and connectivity across regions. Steven Santini, Vice President of Secure Power for Sub-Saharan Africa at Schneider Electric, offered a concrete illustration during a panel on sustainable energy and smart infrastructure: in some areas, only one traffic light in five functions properly. Equipment that generates no data leaves algorithms with nothing to process, rendering AI investments pointless for the communities those investments are meant to serve. “We need to take a step back and get the fundamentals right before we rush to meet the demand for AI, otherwise we’re just setting ourselves up for failure,” Santini said. Digitization must come first. AI simply accelerates what already exists.
The shift toward more autonomous systems adds a second layer of complexity. Nabeel Rajab, Technical Solutions Architect at Cisco in South Africa, explained that agentic AI presents an infrastructure and security challenge before it becomes a productivity tool. “Moving from chatbots to agents creates a fundamental shift in infrastructure,” Rajab noted. Chatbots remain human-directed, with users providing input and receiving immediate responses. Agents operate differently: a single prompt can trigger a series of automated actions running for minutes before delivering results. That operational model demands different network architecture, greater computing power and security frameworks designed from the ground up rather than retrofitted later. Governance structures must be in place before agents are deployed, and security should be treated as a prerequisite for scaling rather than a problem to solve after systems go live.
Yet infrastructure alone cannot unlock the continent’s AI potential. Johnson Idesoh, Absa’s Group Chief Officer for Information and Technology, identified skills as the real constraint. Organizations can acquire chips, energy and data centers at increasing volumes, and investment capital is no longer the limiting factor. What remains scarce is human capacity to learn, implement and manage these systems effectively.
The numbers tell a stark story. Mark Nasila, Chief Data and Analytics Officer for FNB AI Strategy, revealed during a fireside chat on AI’s employment impact that organizations typically allocate over 40 percent of AI budgets to data scientists and engineers while dedicating less than 2 percent to upskilling the broader workforce. At the same time, AI is reimagining approximately 40 percent of the work those employees perform. Automation is advancing faster than any funded strategy for workforce transition, leaving millions of workers without the preparation they need.
Catherine Botha, professor of philosophy at the University of Johannesburg, articulated a different but equally important risk to citizens. The danger is not cognitive decline but what she termed “dispositional erosion,” the loss of the habit of thinking through problems rather than the loss of knowledge itself. True AI literacy extends far beyond prompt writing. It requires the ability to evaluate what technology produces and to override it when necessary. “We need to create graduates who are able to resist the seduction of AI as a comfortable chair, the comfort of letting it give us the answer before we have even properly thought through the question,” Botha said.
By contrast, Africa’s financial services sector offers a template for what localized AI development can achieve. Hylton Kallner, CEO of Discovery Bank, noted that African financial institutions have already solved many challenges that other regions still confront. Limited capital and infrastructure have forced organizations toward solutions that are affordable, high-value and tailored to local contexts, making them exportable to markets facing similar constraints. Realizing this advantage, though, requires enabling policy, regional cooperation and deliberate choices about which technology layers Africa should control.
Dr Bongani Andy Mabaso, Group Chief Technology Officer at Altron, framed the core decision plainly: “Africa doesn’t need to own every layer of the AI stack, but we must own the strategic ones, especially our data. If we can protect and curate African datasets and use them to train models that serve our own markets, we move from being mere takers of technology to real participants in the AI economy.” The window for establishing that sovereignty is roughly five years, which means the policy choices, budget allocations and skills investments made right now will determine whether Africa’s citizens become architects of the AI economy or simply its consumers.