Every major technological revolution eventually becomes a conversation about power.
The industrial revolution was built around machinery and energy; the telecommunications revolution around networks and spectrum; and the internet age around connectivity and data. Artificial intelligence (AI) brings these together.
AI requires enormous computing power, data, electricity, water, land and capital. The countries that understand this early will not simply consume artificial intelligence; they will own parts of the infrastructure that produces it.
South Africa enters this new era with an important advantage. We already possess the continent’s most developed data-centre ecosystem, supported by sophisticated financial markets, extensive telecommunications infrastructure and multiple international submarine cables connecting us to the global digital economy. Johannesburg has become an important African cloud hub, while Cape Town provides proximity to major international cable routes.
This places South Africa in a strong position to become one of Africa’s principal centres of AI computing.
But the cloud does not actually live in the cloud. Behind every AI model are physical servers housed in enormous buildings, consuming electricity and generating heat.
As AI adoption accelerates, the demand for computing infrastructure will rise dramatically. This creates an obvious question for a country that has experienced both electricity shortages and water constraints: who will provide the resources required to power Africa’s AI ambitions?
The answer should not be to resist data centres, but to rethink how we build them. South Africa is a vast country of approximately 1.22 million square kilometres, with exceptional renewable-energy resources.
The Northern Cape has enormous solar potential, while the Eastern and Western Cape possess strong wind resources. We also have more than 2,800 kilometres of coastline. Our geography should therefore become part of our AI industrial strategy.
Instead of simply asking Eskom to accommodate enormous new electricity consumers, hyperscale developments should increasingly be linked to additional generation. Solar, wind and battery storage can be developed alongside computing infrastructure.
The AI economy should help expand South Africa’s electricity system rather than merely place additional pressure on it.
The same thinking should apply to water. Desalination is energy intensive and is not a substitute for responsible management of scarce freshwater resources.
But when renewable energy, desalination, industrial water recycling and modern cooling technologies are considered together, different possibilities emerge. South Africa could develop coastal digital-industrial zones combining renewable energy, water infrastructure, submarine cable connectivity and data centres.
Communities should never have to compete with computers for drinking water. Major technology investments should instead contribute towards expanding the resource base on which both communities and industry depend.
There is another reason South Africa should compete aggressively for this infrastructure: distance still matters. Data travels quickly, but it still travels through physical networks. Locating computing infrastructure closer to users reduces latency and becomes increasingly important as AI moves beyond chatbots into financial services, healthcare, manufacturing, logistics, smart cities and autonomous systems.
Africa, with more than 1.4 billion people, cannot permanently depend on computing infrastructure located thousands of kilometres away in Europe, America or Asia.
South Africa’s existing fibre networks, submarine cables, financial system and established data-centre market give us an opportunity to become a computing gateway for the continent.
Yet hosting infrastructure and owning infrastructure are two different things.
South Africa could become Africa’s largest AI data-centre market while South Africans own very little of the computing capacity, platforms and intellectual property underpinning it. That would make us an attractive location for the AI economy without necessarily making us meaningful participants in it.
Ownership must therefore become central to our national AI discussion. Foreign investment should be welcomed, but every major investment should also raise developmental questions. Who owns the data centres? Who owns the fibre connecting them? Who produces their electricity? Who provides their cybersecurity? Who develops the applications running on them? And will South African entrepreneurs and researchers have affordable access to the computing capacity required to build their own AI companies?
For transformation, this is particularly important. Black participation in the AI economy cannot end with construction, security, catering and other peripheral procurement opportunities around infrastructure owned elsewhere.
Transformation in the AI age must progressively include ownership of networks, data centres, computing capacity, software businesses and intellectual property. We cannot enter a new technological revolution while reproducing the ownership patterns of the previous one.
This is also why the State Information Technology Agency (SITA) should play a central role in the sovereignty discussion.
Government holds some of the most strategically important information in the country, covering citizens, public finances, health, education, infrastructure and the administration of the State. As AI develops, the value of large datasets increases because increasingly powerful systems can analyse and correlate information at a scale previously impossible.
South Africa therefore needs to think seriously about a state-led sovereign hyperscale data-centre capability, with SITA playing a leading role, where strategically sensitive government data and critical AI workloads can reside under South African jurisdiction, security oversight and operational control.
This does not mean cutting ourselves off from global cloud providers. South Africa needs partnerships with the world’s leading technology companies. But technological partnership should not require surrendering control over strategically important national data. In the AI era, protecting government data is increasingly a question of national sovereignty.
A sovereign computing capability could ultimately do more than store government information. It could support universities, researchers and public institutions and, over time, provide computing capacity to South African innovators. Access to compute may become to the AI economy what access to spectrum became to telecommunications: an essential productive resource that determines who can meaningfully participate.
South Africa should therefore stop thinking about data centres merely as large property developments and start viewing them as part of a national industrial strategy. Our land, renewable-energy potential, coastline, submarine cables, telecommunications networks, universities and existing data-centre ecosystem give us advantages that few African countries currently possess at comparable scale.
Those advantages will not last indefinitely. Other African economies are investing aggressively in digital infrastructure. The geography of Africa’s AI economy is being determined now.
South Africa has the land. We have the sun and wind. We have the ocean and international cables. We have the financial institutions, engineers and the continent’s leading data-centre ecosystem.
The question is whether we will use those advantages simply to host the infrastructure of the artificial-intelligence revolution, or whether we will ensure that South Africans own a meaningful part of it.
That distinction will determine whether South Africa becomes Africa’s AI capital merely in name, or in substance.
Luvo Grey is the Secretary-General of Progressive Blacks in ICT (PBICT). He writes in his personal capacity.
**The views expressed do not necessarily reflect the views of the National Media Group.