Africa’s AI Skills Gap: Can the Continent Train Its Way to 40 Million Jobs?

The African Development Bank and UNDP have pledged $10 billion to catalyse 40 million AI-adjacent jobs by 2035. The continent’s ability to fill those roles depends on a skills pipeline that is, at best, nascent — and unevenly distributed across 54 countries.

In February 2026, the AfDB and UNDP announced the $10B AI Initiative at the Nairobi AI Forum. The headline was infrastructure: compute gaps, data centre capacity, GPU access. But buried inside every projection about AI-driven economic transformation is a human capital assumption — that Africa can produce, in less than a decade, a workforce capable of occupying the roles that infrastructure is meant to create.

That assumption deserves scrutiny.

Africa produces approximately 700,000 STEM graduates per year across its universities and technical colleges. The AfDB’s own modelling suggests that the 40 million AI-adjacent jobs target requires not only graduates, but workers with applied digital skills — a category that current curriculum design in most African university systems does not yet produce at scale.

UNESCO IICBA has flagged the structural dimension of this challenge: Africa’s universities remain largely optimised for industrial-era employment pathways. AI-adjacent roles demand a fundamentally different skill profile — built around problem-framing, tooling selection, and continuous reskilling — rather than the knowledge-transfer model that governs most African university curricula today.

UNESCO’s 2025 data on STEM faculty enrolment at Africa’s top 20 research universities shows that computing and information science fields represent, on average, just 12 per cent of total STEM enrolment — against engineering (31 per cent), natural sciences (29 per cent), and agriculture/life sciences (28 per cent). That distribution reflects curriculum priorities set decades ago.

The Private Sector Has Moved Faster Than the State

The most visible skills investments on the continent are not coming from ministries of education. They are coming from US technology companies with commercial interests in an educated African market.

TeKnowledge — the Microsoft-linked global tech services company — launched the Nigeria AI Skills Phase 2 programme in late 2025, extending its partnership with the National Information Technology Development Agency (NITDA). Phase 2 targets 50,000 Nigerians across foundational AI literacy, cloud computing, and data analytics. In scope and design, it is the largest structured AI skills programme currently running on the continent.

Google’s Africa Developer Training initiative operates through its Google Developer Groups and Women Techmakers networks, with active nodes in Lagos, Nairobi, Cairo, Accra, and Johannesburg. AWS re/Start — Amazon’s cloud workforce development programme — has expanded to eight African cities since 2024, delivering 16-week intensive training cohorts focused on cloud practitioner credentials with employment placement support.

These programmes are significant. They are also structurally constrained by a shared design limitation: they train workers for roles within the technology sector’s formal economy, primarily in countries with existing tech hubs. For the AfDB’s 40 million jobs target to be met, the required skills base is far broader — encompassing AI-augmented roles in agriculture, healthcare, logistics, financial services, and public administration, distributed across all 54 African economies.

Universities Are Adapting, Slowly

At Addis Ababa University (AAU), the Computer Science and Information Technology faculty has moved to integrate applied machine learning modules into its undergraduate curriculum since 2024, supported by a curriculum co-design partnership with Google and Microsoft. A 2025 inaugural cohort of 120 students completed a structured AI fundamentals track — a model that AAU’s faculty leadership intends to scale to 500 students by 2027.

The programme’s stated objective, per AAU faculty documentation, is not to bolt on an AI module but to reconceptualise computing education wholesale: students completing the track are expected to leave not merely as AI tool users but as graduates capable of building evaluation frameworks, conducting algorithmic audits, and applying AI to locally grounded problems.

The University of Nairobi’s School of Computing and Informatics has moved further. Supported by the Kenya CBC policy environment — which made Nairobi the first African capital with a mandated digital literacy curriculum from primary school level — the university is seeing a measurably different cohort entering computing programmes: students who arrive already fluent in computational thinking rather than encountering it for the first time in first year.

Cairo University’s Faculty of Engineering has piloted a data science specialisation at postgraduate level since 2023, with enrolment growing from 45 to 340 students over three cohorts, driven primarily by demand from Egypt’s fintech and logistics sectors.

The TVET Layer Is the Underinvested Story

Much of the policy and investor attention focuses on university-level AI education. The more structurally important pipeline — and the one least resourced — is technical and vocational training.

AI systems in the real economy require support roles that do not require four-year degrees: data annotators, AI output reviewers, prompt engineers, AI systems testers, and algorithmic auditors. These mid-level technical positions are the workforce layer that will determine whether AI applications actually function at production scale across African industries.

Kenya’s Competency-Based Education and Training (CBET) framework — being scaled through the Kenya-China Phase III project, which is equipping 70 TVET colleges with new machinery and training over 1,000 instructors — is among the few continent-wide programmes explicitly building toward this mid-level pipeline. Nigeria’s NITDA bootcamp model, which recently certified 50 unemployed youth in a four-week intensive AI tools programme, demonstrates that short-cycle, employment-linked training can reach populations that university pathways cannot.

The African Union Commission’s analysis of AI labour market readiness has consistently identified the TVET layer as the fastest-leverage point for near-term job creation: prompt engineering, data labelling, and AI quality assurance are roles that African economies can absorb at scale now. The constraint is not demand — it is curriculum depth and equipment access in the TVET sector, which the AU Commission frames as a solvable infrastructure problem contingent on targeted funding.

The Andela Model, Revisited

Perhaps the most consequential experiment in Africa’s AI skills ecosystem is not a government programme or a university curriculum — it is the return of cohort-based training models for software and AI talent at scale.

Andela, which pioneered the model of training African software engineers and connecting them to global remote work, is repositioning around AI tooling. Its newest cohorts, running in Nigeria and Kenya, incorporate applied AI development as a core track. Decagon Institute, operating primarily in Nigeria, has added machine learning engineering to its software engineering curriculum, with cohort sizes of 30–60 students per quarter. Coursera’s government partnerships — with Rwanda, Egypt, and Nigeria — have collectively enrolled over 400,000 learners in AI and data science tracks since 2024, though completion rates and employment conversion remain the harder metrics to track.

The structural challenge for the cohort model is quality control at scale. Training 50 students well is achievable. Training 50,000 workers with skills that meet the hiring bar of serious AI employers requires infrastructure — physical space, instructors, compute access for hands-on training — that the current cohort providers do not have at the requisite scale.

What Ten Years Actually Requires

The AfDB initiative’s 2035 target of 40 million AI-adjacent jobs works out to roughly 4 million workers per year entering AI-relevant roles across the continent. Africa’s current university system produces 700,000 STEM graduates per year. Even accounting for TVET graduates, private sector training cohorts, and professional upskilling, the arithmetic is not comfortable.

The policy lever that could change the calculus most rapidly is not new: it is compelling Africa’s universities to reform curriculum faster than they have historically done, and pairing that reform with the compute infrastructure that makes hands-on AI training possible. A student who completes an AI fundamentals module without access to a GPU to practice on has theoretical knowledge but no applied capability.

The AfDB’s ignition phase, running through 2027, targets the infrastructure and regulatory environment. The skills pipeline depends on it — but the pipeline itself requires investment now, before the compute arrives, in curriculum, instructors, and TVET capacity.

Africa has trained its way into previous industrial transitions. The question for the AI decade is whether the institutional apparatus can move fast enough to matter.

At a Glance: Africa’s AI Skills Landscape