In schools like Kaalfontein, Meadowlands and Ahanang primaries, a coding and robotics session can mean 40 learners, a handful of tablets and a teacher who has never written a line of code. That is the room Veli Mokhabodi walks into most weeks. It is also, if you follow the argument he makes, the room where South Africa’s AI future is actually being decided.
In November, Johannesburg hosts the first South African AI Awards. Nominations opened in March, shortlisting runs through September, and winners will be named at a black-tie gala. There are 24 categories, from AI Startup of the Year to Best AI in Healthcare to Best AI for Multilingual and African Language Innovation. One of them is Township AI Innovator of the Year, and Mokhabodi’s organisation is precisely the kind of outfit it was written for.
He has not entered. The reason is partly logistical and partly a matter of principle, and both are worth hearing.
The country is not the laggard here
Start with what the data says, because it complicates the usual story.
The 2026 Global Outsourcing AI Readiness Index, published by Ataraxis, ranks the world’s top 25 outsourcing destinations across four measures: population AI adoption, workforce AI literacy, enterprise AI adoption and the AI education pipeline. South Africa lands at number 8 globally, scoring 66.5 out of 100, and number 1 in Africa by a distance. Egypt, the continent’s next-best performer, sits at 49.35. That is a gap of 17.35 points, and roughly 24 points above the African average.
Look at enterprise adoption and the gap turns into a chasm. South Africa scores 65. Egypt manages 42. Then Morocco on 39, Kenya on 35, Nigeria on 34, Ghana on 31, Uganda and Ethiopia on 14 apiece. South Africa is the only country on the continent above 50. Corporate AI deployment in Africa is, functionally, a South African story. Ataraxis spokesperson Camilo Izquierdo has said the country has “moved well beyond the pilot stage”.
South Africa is also one of a small handful of destinations where every single sub-index clears the halfway line: population adoption 78, workforce literacy 63, enterprise adoption 65, education pipeline 53. Globally it slots between Czechia on 66.9 and Bulgaria on 62.8, and it ranks fifth in the world on the companion Global Outsourcing Talent Index.
Now the crack in the wall. That education pipeline score of 53 ranks South Africa 13th out of 25 on that measure alone. It is the country’s weakest dimension by a wide margin, and it is the one that describes the next decade rather than this one. Ataraxis analysts are blunt that closing it will determine whether South Africa keeps its continental lead. The full index and methodology are at ataraxismgmt.com.
Which brings us straight back to the classroom with 40 learners and four tablets.
What 6000 learners looks like
United Siyafunda NPC has been running since 2020. Since inception it has reached over 6000 learners aged 5 to 17 through in-school coding and robotics programmes, plus more than 200 out-of-school youth and 300 teachers. On a conservative estimate, the teacher training alone touches another 12,000 to 24,000 learners indirectly. Facilitators are certified through the UNISA I-Set programme, sessions run in seven of the country’s eleven official languages, and in 2023 the organisation won the African Impact Challenge community impact stream.
“AI does not work without people who can use it, question it, build on it, and teach the next generation to do the same,” Mokhabodi says. Trained teachers, in his framing, are multipliers. They become the network that builds AI literacy from the ground up rather than waiting for it to trickle down.
The policy backdrop supports him. The Department of Basic Education gazetted a Coding and Robotics CAPS curriculum for Grades R to 9 in June 2024, but implementation is not compulsory. Schools take it up where capacity exists, grade by grade, from 2026 onward. The subject is real on paper and optional in practice, and the space between the two is being filled by organisations like his.
He is notably generous about teachers, which is not the default tone in this conversation. “The willingness is already there,” he says. Teachers know the curriculum requirement is coming and most want the training. So United Siyafunda meets them where they are, working within the constraints of what is available, whether that is missing equipment or zero prior experience. “We work with what we get,” he says. Teachers are encouraged to bring colleagues along, which turns a workshop into a culture.
The category question
So why no entry? Timing, first. Between July and August the team set out to activate thousands of learners across South Africa and Lesotho, and they are already approaching 3000. Awards admin loses to a robotics session, every time.
The second reason is sharper. Mokhabodi is not sure he wants the township category at all. He calls it important, and says it would signal that the work already exists and create visibility the sector needs. But he also says it “suggests that our innovation might not make it in the open category”. His follow-up is the question the awards should probably be sitting with: if we are doing the work, shouldn’t the metrics used to judge any other AI innovation apply to ours too?
It is a fair challenge. The published criteria for the township category are innovation, community impact, scalability, adoption and ethics. Those are close cousins of the criteria in the open categories. If a programme reaching 6000 learners directly and up to 24,000 indirectly can clear them, the ring-fence starts to look less like a platform and more like a smaller stage.
Shortlisting closes in September, so there is still time. And a first-year awards programme is a rare thing, because nobody has decided yet what winning is supposed to look like.
Ask Mokhabodi who belongs on that stage in November and he does not name a company. “Anyone who is building the human infrastructure that makes AI meaningful at scale, not just the people building the models,” he says. “The end-users and impact of these AI innovations are as central to the AI story as the server rooms and language models.”
South Africa has the enterprise adoption. It has the population uptake. What it has, at 53 out of 100, is a pipeline problem. The people fixing it are in classrooms in Kaalfontein and Meadowlands and Ahanang, working with what they get.
Worth a nomination, at minimum.
Figures cited from the 2026 Ataraxis Global Outsourcing AI Readiness Index: https://ataraxismgmt.com/global-outsourcing-ai-readiness-index
