Dion Harvey, regional GM at Red Hat sub-Saharan Africa. African organisations face unique challenges – including legacy infrastructure, rising complexity, skills shortages and cost pressures – that must be addressed before they can meaningfully benefit from AI, according to Red Hat regional executives.Speaking at a media roundtable in Johannesburg this week, Dion Harvey, regional GM at Red Hat sub-Saharan Africa, said the company's recent summit in Atlanta was not just about "pushing the boundary of the next big thing or AI at all costs". He said Red Hat had developed a deeper appreciation for what local customers are experiencing."AI is not the only thing on people's minds," Harvey said. "Our customers are still trying to modernise, still trying to get to grips with how to leverage the cloud and how to keep the lights on. The level of complexity is only increasing."He cited software upgrades and vendor lock-in as major headaches. CIOs are forced to "constantly play the upgrade game" because vendors may stop supporting older versions, even when an organisation's critical applications cannot move to newer versions.At the Red Hat Summit 2026, Matt Hicks, CEO of Red Hat, described a widening gap between infrastructure complexity and available resources. "Your virtualisation costs are up significantly, your regulatory environment is expanding beyond your compliance team's ability to keep up, and your board is probably pushing for ROI on AI projects that might not match your enterprise reality," he said."That gap, that chasm between the complexity of your environment and the resources you have – that is the real story of enterprise IT right now, not just AI."Despite these challenges, Harvey noted that agentic AI is already being implemented in SA's financial services sector. He said the goal for local enterprises is to "not only do more with less but do it in such a way that it doesn't compromise future growth".Red Hat announced that its AI 3.4 platform now offers MaaS (model-as-a-service), providing a single interface for developers to access curated models and manage prompt quality and security.However, Harvey stressed that the right foundation matters more than speed.Hicks told attendees at the summit: "The organisations that will have the most success will be the ones that build on the right foundation. Virtual machines, containers and AI agents are not competing with each other — they are converging. The answer is not going to be one cloud or one vendor or one model; it will be the right platform with a broad ecosystem that supports it."