South African CIOs are entering the agentic era under conditions that leave little room for experimentation without purpose. AI spending in South Africa is rising quickly, but organizational maturity remains early: IDC projects local AI-related spending to exceed $7 billion by 2030, yet more than half of South African organizations still sit in an ad hoc, experimental stage of AI maturity today. That tension matters. It means the opportunity is real, but so is the risk of fragmented investment, inflated expectations, and stalled value realization.

IDC’s global research shows that 2025 launched the agentic era, but 2026 is the year enterprises started to industrialize it. A shift is underway towards systems that can reason, act, and coordinate across workflows. Yet only 45% of AI projects to date have delivered any measurable business value at all, never mind achieving their promised ROI. There is plenty of interest in AI, but the challenge is in converting AI activity into repeatable business outcomes. Executive boards should not fund AI – or any technology investment – for its own sake; they should invest in measurable outcomes.

For CIOs, the first imperative is controlled value. IDC’s research shows that fewer than 5% of CXOs currently view fully autonomous agentic AI as appealing, while 54% of C-suite leaders rank operational efficiency and productivity as the most important short-term outcomes for agentic AI. The market is not ready for grand promises about autonomous enterprises, but it is more than ready for credible progress in bounded, governed, high-value use cases. That means using agentic capabilities where they can improve speed, quality, and decision-making without creating unmanaged risk.

At a time when global markets are in turmoil and supply chains in flux, business risk tends to dictate a conservative approach to innovation, and geopolitical instability is raising the importance of sovereignty, control, and trust.

Virtually all of the South African organizations IDC has engaged with have highlighted the role of central governance in building AI roadmaps. The organizations that will move ahead are the ones that successfully impose discipline: start with strategy, prioritize use cases, map the value, model the cost, adjust for risk, and then optimize continuously. Curated pilot projects with clearer business ownership and more demanding baselines. In a constrained environment, the right roadmap is a ranked portfolio of viable, measurable business outcomes, and governance must be the steering system that allows innovation to scale safely, rather than a barrier to progress.

Even so, we expect friction along that journey. Many AI project failures, in practice, are enterprise-readiness failures as much as they are model failures. IDC’s research shows that data accuracy and availability are top inhibitors to successful agent deployment, but systemic issues include cultural barriers, data privacy and security concerns, unpredictable AI costs, and immature business cases.

There is also a critical workforce dimension to this transition. In IDC research, 45.5% of organizations identified an AI-ready workforce as a top-three AI adoption priority for 2026, while 81% say they are struggling to find and retain talent because of the shift to AI-enabled work models. In a skills-constrained environment, the winning narrative is human plus agent. Agentic systems should be framed as force multipliers that lift productivity, improve insight, and create room for innovation. If leaders reduce the conversation to headcount substitution, they will weaken adoption, trust, and the future talent pipeline.

This is where the CIO role is changing most dramatically. IDC increasingly describes the emerging CIO mandate as one of balancing efficiency and innovation while shaping the organization’s operating model for AI. I would go a step further: the CIO, and their colleagues among the broader AI governance board, are the orchestrators of autonomy. That means deciding where autonomy belongs, where human judgment must remain central, how decision boundaries are defined, and how value, risk, and accountability are continuously monitored. We are reshaping business architecture, enabled by increased machine agency, not dictated by it.

This will be one of the defining themes at IDC CIO Summit 2026 in Johannesburg. As South African organizations move from AI experimentation to enterprise-scale adoption, the central leadership question is how to build a roadmap that is mature, measurable, governable, and resilient. The CIOs who get this right will be the ones who most effectively translate AI ambition into business value, workforce capability, and trusted execution.