Nelson Mandela Bay Audits The Curriculum

Africa is deploying manufacturing AI faster than it is training the people who will keep it running. That is the finding to carry around for the rest of this piece, and it holds up whichever end of the continent you walk in from.

Start in the south. On 7 September 2026, Engineering News reporter Irma Venter reported the Nelson Mandela Bay Business Chamber's audit of 55 manufacturers employing 12,662 people in the automotive-heavy Port Elizabeth cluster. The Hard-to-Fill Index came in at 30.4. Eight of ten recruitment searches failed. Vacancies sat open at a 7.5 out of 10 on the chamber's duration measure. Youth retention in technical roles was 17.8 per cent, and 80.4 per cent of surveyed companies said their workforce lacks the foundational digital skills to step into technology-enabled roles. The missing crafts, in order of pain: toolmakers, millwrights, CNC programmers, instrumentation technicians, quality specialists, CMM programmers, frontline supervisors, and IT/OT integration specialists. That final line is the industrial-AI line. Someone has to glue the OPC-UA feed from a stamping press into a model that flags a tooling drift before the part goes out of tolerance, and that someone, in Port Elizabeth right now, is not on the roster.

Chamber chief executive Denise van Huyssteen's quote was blunt. "We cannot continue treating these as two separate problems." She meant 33.1 per cent regional unemployment on one side of the audit and unfilled industrial vacancies on the other. There is a second reading. The AI curriculum and the manufacturing floor are the two separate problems, and the chamber is drafting a Regional Manufacturing Skills Observatory precisely because the national pipeline is not sending it the hands it needs.

The deployment layer

Move east. On 26 September 2025, Creamer Media's newsletter listed the ten semifinalists for the Milken-Motsepe Prize in AI and Manufacturing, each with fifty thousand dollars of unrestricted funding and a shot at a million-dollar grand prize. The named deployments were specific. DataProphet in Johannesburg delivers production intelligence to machine builders. INDOS in Egypt converts manufacturing environments into real-time quality systems. BleagLee in Cameroon uses computer vision to sort plastic, agricultural, and electronic waste. Freshpack Technologies in Tanzania pushes AI-driven cooling into informal food markets. Green Building Design Group Africa runs climate-smart infrastructure optimisation out of South Africa. Toto Safi runs circular manufacturing out of Rwanda.

The Milken Institute's Next-Gen Industry Prize report, published 21 April 2026 by Terry Mulligan, Emily Musil and Adoma Addo, puts the headline opportunity at $2.9 trillion in continental GDP by 2030 if the deployments scale. That is Milken's number on Milken's model, and the usual discount applies: a continental projection seven years out, resting on scaling assumptions no one can audit, deserves to be read as a direction of travel rather than a settled gauge. The semifinal list is the sturdier evidence. The operators are already building. The hands they need are the ones Nelson Mandela Bay cannot find.

The training layer

Move west. Nigeria's 3MTT programme, run through NITDA, targets three million technical talents in multiple phases, with 300,000 participants planned across phases one and two. Businessday Nigeria's 2 September 2026 explainer on the country's AI talent race adds a Google.org commitment of ₦2.8 billion, a Meta-backed AI Academy for 20,000 young Nigerians, a Microsoft-sourced AI adoption rate of 10.1 per cent, and a 2027 data-centre capacity target north of 150 megawatts of IT load. Read the 3MTT curriculum against that build-out and the problem names itself. The twelve tracks are AI/machine learning, animation, cloud computing, UI/UX design, data analysis and visualisation, data science, DevOps, game development, product management, quality assurance, software development, and cybersecurity. There is no manufacturing track. There is no OT-security track. There is no plant-floor instrumentation track. There is no OPC-UA, SCADA or MES glue track.

That is a design choice rather than an oversight to be fixed by a syllabus update next quarter. It reflects where the country's growth hopes have been pointed for a decade. It leaves the industrial-AI build with no state-funded pipeline for its roles.

The last row is the one to sit with. The attack surface for a model running a cement kiln in Lagos or an engine line in Kariega is not the one a cybersecurity bootcamp graduate was trained to defend. It is a separate specialism, and continent-wide, nobody is paying tuition to grow one.

The IMF counterargument

The strongest disconfirming reading comes from the International Monetary Fund. In a 21 July 2026 Reuters piece carried on Engineering News, the Fund estimated that AI could add roughly four per cent to Sub-Saharan Africa's economy over the next decade, with Martin Schindler as lead author. His co-author Andrew Tiffin pushed the binding-constraint argument the other way. "It's hard to have anything without electricity," he said. Roughly half the region lacks reliable power. Internet penetration sits at 38 per cent against a 68 per cent global baseline. On that reading, skills are a second-order problem. Fix the plant's power supply and its backbone connectivity, the argument goes, and the models can be run on remote compute by a thinner local layer.

Answer it on the merits. Electricity is prior in the sense that nothing runs without it, and no one who has watched a load-shed cycle in Gqeberha disputes it. The weaker form of the IMF line is correct. The stronger form, that fixing power and connectivity thins the local skills requirement, understates what a model in production asks of the people around it. When a quality-prediction model misfires on a VW Polo bodyshell on the Kariega line, the person who diagnoses the misfire is in the shop with a laptop plugged into the stand, not in Johannesburg or Nairobi reading a dashboard. The local roles are the ones the Fund's frame absorbs into "deployment at scale" and the audits in Port Elizabeth confirm are the hardest to hire for.

A regional answer

The policy architecture exists on paper. The African Union's Continental AI Strategy, adopted on 17 June 2024, names skills as a pillar. The African Development Bank's June 2024 partnership with Intel committed to training three million Africans and thirty thousand government officials in AI and Fourth Industrial Revolution competencies. The applied domains the AfDB named were agriculture, health and education. Manufacturing was not among them.

The regional moves are closer to the plant. The Nelson Mandela Bay chamber's own proposal, embedded in the same September 7 audit, is to run industry-higher-education strategic skills panels, stand up a centralised skills database, open artisan knowledge-transfer programmes and seed Industry 4.0 technical hubs tied directly to the region's manufacturers. Hannover Fairs MENA's Industrial Transformation Africa 2026 event, held in Casablanca from 29 September to 1 October, ran the same play one coast north, with a Smart Factory Academy on-site providing hands-on upskilling for engineers and industrial talent across Morocco, Tunisia and West Africa.

The HCAIM and PANORAIMA curricula, drafted first for European universities and now being mirrored by African partners, meet the same question the Nelson Mandela Bay chamber has turned into an audit: who, specifically, do you train, against which plant, in which city, running which protocol? The answer is not a lecture series. It is a dual-track apprenticeship that puts a bachelor's-level AI engineer on the same shift as a journeyman toolmaker for eighteen months, with the model's logs and the lathe's ticket going into the same incident file.

The Cisco–Carnegie Mellon whitepaper on AI and the Workforce in Africa laid out the baseline last year: a continental AI professional base of about five thousand, growing at forty per cent a year, against 230 million jobs that will need digital skills by 2030, and less than a quarter of African tertiary students entering STEM in the first place. That is the floor the regional moves above are building on.

Eighty-point-four per cent. That is the share of Nelson Mandela Bay manufacturers reporting that their workforce lacks the foundational digital skills to step into technology-enabled roles. The AI is already on the floor.

Tarry Singh is the founder and CEO of Real AI, an enterprise AI advisory and deployment firm working with global enterprises on production agent systems, model risk, and AI sovereignty strategy. He also leads Earthscan for Energy AI startup, and is a founding contributor to the EU-funded HCAIM and PANORAIMA programmes for responsible AI education across European universities. He writes at tarrysingh.com.