South Africa’s AI Crossroads: Why Access Is Not the Same as Control

A Pattern That Keeps Repeating

There is a story South Africa keeps living through, generation after generation, and it goes something like this: the country gains access to something powerful — a resource, an infrastructure, a technology — benefits from it partially, and then watches the real value drain away to whoever actually owns it. It happened with mining, where the wealth of the earth left the continent while communities above it stayed poor. It happened with electricity and telecommunications, sectors where access expanded but strategic control remained elsewhere. Now, as artificial intelligence reshapes the global economy, the warning signs are flashing again — and the window to do something about it is closing faster than most decision-makers seem to realise.

Understanding how South Africa arrived at this moment requires going back to a choice that seemed sensible at the time. When the public sector adopted a cloud-first digital strategy, the move accelerated transformation across government services and opened up new efficiencies. It looked, for a while, like progress. But beneath the surface, something more troubling was taking shape.

The Cloud Trap and Its Lessons

President Ntuli, Managing Director of HPE South Africa, has been direct about what that cloud-first pivot actually produced. By rushing into adoption without simultaneously building national capability, South Africa handed a small group of global tech giants enormous leverage — over pricing, over terms of service, over the very architecture of the country’s digital future. The lesson Ntuli draws is unambiguous: artificial intelligence cannot be allowed to follow the same script.

That warning landed at a particularly uncomfortable moment for South African policymakers. The Department of Communications and Digital Technologies had published a Draft National AI Policy, positioning the country as a potential ethical leader in the governance of this technology. The document was meant to set the standard. Instead, it became a scandal.

Minister Solly Malatsi withdrew the draft after it emerged that the policy had itself been generated using AI tools — without adequate disclosure and without meaningful human oversight. A framework designed to regulate AI had been produced through the very unaccountable processes it was supposed to govern. The irony was sharp, and the embarrassment was real. It exposed, in the most visible way possible, how far South Africa still has to travel before it can credibly claim to be steering its own AI trajectory.

Sovereignty Is Not Protectionism

The concept of sovereign AI tends to get misread in public debate, often caricatured as a protectionist impulse — a desire to wall off the country from global technology and build inferior domestic alternatives out of nationalist pride. That reading misses the point entirely. The actual argument is far more precise and far more urgent.

What sovereignty in AI genuinely means is owning enough of the stack — the compute infrastructure, the data, the models, the governance mechanisms — to exercise real choice. A country that uses American or Chinese AI systems is not automatically less sovereign than one that builds its own from scratch. The danger emerges when that usage is the only available option, when there is no domestic capability to fall back on, no data governance architecture protecting citizens, and no local talent pool capable of auditing or adapting the systems being deployed.

The distinction between access and control is not abstract. It is the difference between being a participant in the AI economy and being a consumer of someone else’s decisions, someone else’s values, someone else’s strategic interests embedded invisibly into the models shaping South African public life.

The Gap Is Widening

South Africa is not building from zero. Research institutions are active, a developer community is growing, and existing policy frameworks provide at least a foundation to work from. These are real assets, and they matter. But the honest assessment is that the gap between what currently exists and what genuine AI sovereignty would require remains significant — and it is not a static gap.

Global AI development is accelerating at a pace that makes standing still equivalent to falling behind. Every month that passes without deliberate investment in compute capacity, data infrastructure, and local AI talent is a month in which the structural dependency deepens and the cost of reversing it grows. The window for a different outcome is real, but it is narrowing.

South Africa has paid dearly, across its history, for learning the lesson that access is not the same as control. The question facing the country’s policymakers, institutions, and private sector leaders right now is whether they are willing to act on that hard-won knowledge — before, this time, the window closes for good.