President Ntuli, MD of HPE South Africa, put the problem plainly: when South Africa adopted a cloud-first approach across the public sector, it accelerated digital transformation but entrenched deep dependency on a small number of global providers. By failing to build meaningful national capability alongside adoption, the country surrendered influence over pricing, terms, and ultimately its own digital trajectory. AI, Ntuli argues, cannot be allowed to follow the same path.
The Policy Vacuum at the Worst Possible Moment
The timing of South Africa’s sovereign AI debate is made more urgent by what can only be described as a governance embarrassment. The country’s Draft National AI Policy, published by the Department of Communications and Digital Technologies, was withdrawn by Minister Solly Malatsi on 26 April 2026 after it emerged that the document contained fictitious, AI-hallucinated academic citations. A policy designed to establish South Africa as an ethical AI leader had itself been produced through exactly the careless AI use it sought to regulate. An Independent Expert Review Panel, chaired by Professor Benjamin Rosman, was subsequently appointed to rescue the process.
The irony is sharp, but the consequences are real. While the state was busy with a document that had to be recalled, the private sector moved on without it. By mid-2026, analysts were already describing South Africa as having shifted from AI pilots to operational deployment across customer service, coding, cybersecurity, fraud detection, and financial services. The country ranked 46th out of 147 economies in Microsoft’s Global AI Diffusion Report for Q1 2026, placing it ahead of every other African nation in the study. Generative AI usage had climbed to 23.1% of the working-age population. The market was not waiting for Pretoria to get its framework right.
What Sovereignty Actually Means in Practice
The argument for sovereign AI is sometimes misread as protectionism, a desire to shut out global technology companies and build inferior domestic alternatives behind a regulatory wall. That is not what the argument actually says. The more precise case is that South Africa needs to own enough of the AI stack , the compute, the data, the models, the governance to exercise genuine choice rather than being structurally dependent on decisions made elsewhere.
The distinction matters. A country that uses American or Chinese AI models is not necessarily less sovereign than one that builds its own. The problem arises when that usage is the only option, when there is no domestic capability to fall back on, no data governance architecture to protect citizens, and no local talent ecosystem capable of adapting or auditing the systems being deployed.
South Africa is not starting from scratch. UCT’s AI compute initiative is expanding access to the processing power needed for genuine research. The University of Pretoria ranks first in South Africa for AI research output. The AI Hub for Development’s Compute Accelerator programme is helping local builders access tools and expertise. Pick n Pay has launched a generative AI shopping assistant built on Google’s Gemini platform. Standard Bank, FirstRand, and Absa are among Africa’s most advanced AI adopters in financial services. These are real building blocks, not aspirational slides from a policy summit.
The Infrastructure Problem Nobody Wants to Price
Where the picture darkens is on the fundamentals. Only 26% of South African households own a computer, according to the ITU. Generative AI usage among the unemployed has reached over 90% globally, because access to AI is increasingly how people navigate a disrupted labour market. In South Africa, for the majority of that cohort, that access runs through a smartphone with intermittent data and unreliable electricity. Those are not edge cases. They describe most of the country.
The IMF estimates that AI could lift sub-Saharan Africa’s economy by around 4% over the next decade but only if the region materially improves electricity supply, internet access, and digital skills. In South Africa’s case, compute cost is directly linked to energy cost, and energy cost is linked to Eskom’s ongoing structural instability. You cannot build a sovereign AI economy on a grid that cannot guarantee uptime.
The Model Worth Borrowing
The Siemens Energy example that Ntuli cites is instructive precisely because it is not about rejecting global partnerships. Working with HPE, Siemens built an AI-enabled engineering platform that gave it control over its data, intellectual property, and critical systems, while using high-performance computing to accelerate innovation across digital twins, simulations, and predictive maintenance. AI was embedded in how the business operates, not bolted on as a vendor subscription. South Africa’s strategic infrastructure operators, Eskom, Transnet, the major metros could and should be pursuing the same approach.
The countries moving fastest on AI development are not necessarily those spending the most. They are the ones building the ecosystems the talent, the regulation, the compute access, the data governance , that make AI an endogenous capability rather than an imported service. South Africa has the university depth, the data assets, and the private sector dynamism to compete for that position. What it lacks, still, is the policy coherence to make those pieces work together.
The intelligence layer of the economy is being built right now. The decision about whether South Africa is building it, or merely buying it, is also being made right now. Those two decisions are not the same, and the country cannot afford to confuse them again.
Written by:
*Dr Iqbal Survé
Past chairman of the BRICS Business Council and co-chairman of the BRICS Media Forum and the BRNN
*Sesona Mdlokovana
Associate at BRICS+ Consulting Group
Africa Specialist
**The Views expressed do not necessarily reflect the views of Independent Media or IOL.
** MORE ARTICLES ON OUR WEBSITE https://bricscg.com/ (https://bricscg.com/)
** Follow @ (https://x.com/brics_daily)brics_daily (https://x.com/brics_daily)on Twitter for daily BRICS+ updates and instagram @brics_daily (https://www.instagram.com/brics_daily?igsh=bmhvbTd0YzA4a2wx)
