Your AI strategy is not a strategy if governance is an afterthought

Without decision-making clarity and proper oversight, AI programmes drift, duplicate, and fail quietly.

Article link: https://open.substack.com/pub/johanosteyn/p/your-ai-strategy-is-not-a-strategy?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Audio summary: https://youtu.be/178Rzgh4YFI

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

South Africa is not short of AI enthusiasm. We have leaders speaking confidently about innovation, competitiveness, and “not being left behind”. But confidence is not the same as capability, and capability is not the same as execution. In a recent Daily Maverick piece, Kara Le Roux makes the blunt point that our constraint is not belief, but follow-through. That framing matters, because it shifts the national conversation away from hype and towards the unglamorous work of delivery: cleaning data, modernising processes, securing systems, aligning incentives, and building the skills to operate AI safely at scale. That is where South African organisations will either win or waste money.

CONTEXT AND BACKGROUND

AI adoption has entered a new phase globally. In the early wave, many organisations could “try AI” with minimal disruption: a chatbot here, an analytics model there, a small automation pilot in one department. Today, the conversation is moving towards embedded AI, where organisations depend on always-on systems for customer service, fraud prevention, workforce productivity, and decision support. This shift raises the stakes for execution, because failure now means operational risk, not merely a disappointing demo.

In South Africa, that tension is amplified by the realities of infrastructure, skills, and uneven digital maturity. It is easy to be impressed by the promise of AI while underestimating what it takes to implement it responsibly across messy data estates and legacy systems. TechCentral practically captured this adoption reality: the hard part is not whether people are excited about AI, but whether organisations have the data foundations and organisational capability to make AI deliver real value through people, process, and governance.

INSIGHT AND ANALYSIS

Execution fails in predictable places. First, data: many organisations still do not know what data they have, who owns it, how trustworthy it is, or whether it can legally be used. Second, operating model: AI cannot remain a side project owned by one innovative manager; it must be integrated into business workflows with accountability for outcomes. Third, risk: model errors, privacy breaches, procurement shortcuts, and “shadow AI” usage become inevitable when governance lags behind adoption.

This is why “strategy to execution” is not a slogan; it is a capability gap. TechCentral recently put it plainly: many South African organisations are trying to move beyond AI pilots into production, but struggle at the handover where experiments must become operational, governed systems that deliver measurable outcomes.

Then there is the physical layer that business leaders too often ignore. AI is not weightless. Compute requires power, cooling, and resilience. Even when your organisation is not building data centres, you are increasingly dependent on suppliers and platforms that are. South African commentary has started to connect these dots: in a November 2025 piece, Energize warned that AI data centres are power-hungry, heat-intensive and water-dependent, and that South Africa’s energy constraints make this an execution and competitiveness issue, not just an IT conversation.

Finally, the grid conversation matters because it shapes priorities. In Europe, regulators are already acting on the reality that power-hungry data centres can outpace network build-out: Ireland’s energy regulator has published a dedicated connection policy for new data centres, linking grid access to conditions like location constraints, renewables, and even requiring onsite or proximate generation/storage for larger loads. You do not need to get lost in engineering detail to understand the implication: when AI services become “always-on”, economies start redesigning infrastructure and policy around those loads.

IMPLICATIONS

For South African business leaders, the immediate takeaway is straightforward: stop treating AI as a procurement event and start treating it as an organisational change programme. If you cannot explain which business process is changing, who will be trained, how performance will be measured, and how risk will be governed, you are not executing; you are experimenting.

For policymakers and public institutions, the execution lens is even more important. The state is a massive custodian of sensitive data: education records, health information, social services, and identity systems. When AI is introduced into these environments, weak execution becomes a child-safety issue as much as a technology issue. Poor data controls, unclear accountability, and vendor-driven deployments create long-term consequences that are hard to unwind.

For the wider economy, execution becomes a jobs and competitiveness story.

IT-Online recently highlighted research showing executives are betting heavily on AI-driven growth while simultaneously worrying about integration into core operations, and it flagged just how quickly skills are expected to shift and become obsolete. If we do not invest in practical delivery skills, change management, and governance, we risk importing solutions while exporting value.

CLOSING TAKEAWAY

South Africa does not need to “believe in AI” more. We need to implement better. That means building the muscle for delivery: data readiness, governance, security, procurement discipline, and human capability. It means being honest about constraints, including infrastructure, and designing responsibly around them. If we take the execution challenge seriously, AI can become a genuine lever for productivity and better services. If we don’t, we will be left with a country full of pilots, a trail of vendor invoices, and very little to show for it.

Author Bio: Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net