Making AI Work in African Universities: Readiness, Infrastructure, and Funding

When IREX and Development Gateway published From Ambition to Adoption: Insights into University AI Readiness from Around the World earlier in the year, a report exploring the state of AI readiness across universities around the world, one finding drew most of the attention: only about one in three institutions surveyed had a clear AI strategy linked to academic and operational priorities. The gap between ambition and institutional practice is real, but easy to misread. Across the continent, faculty and students are already using AI in teaching, research, and daily academic work, mostly having taught themselves through peers, social media, and their own exploration. This, and recent insights from ecosystem experts, has led us to think that the focus should not really be on whether universities are adopting AI. The onus should be on whether that use is supported, resourced, and governed well enough to matter at an institutional level.

This was the focus of a regional webinar we convened alongside IREX and the African Research Universities Alliance (ARUA) on September 9, 2026, building on the global webinar we hosted in June on AI readiness in higher education and extending the conversation into the African context. Three issues carried the most weight in the conversation: institutional readiness, infrastructure & talent, and the investment needed to move from experimentation to sustained practice.

AI readiness is an institutional capability, not a strategy document

An institutional AI strategy is relatively straightforward to produce. Implementing one is much harder, because implementation depends on licenses, bandwidth, and faculty time, and that is where most universities are struggling on the continent.

Structured institutional support is largely missing, so learning stays informal with little opportunity to test what works and assess whether approaches being adopted are sound. Connectivity, devices, and computing resources are unreliable or insufficient, making sustained experimentation difficult or impossible. Questions around ethics, privacy, and academic integrity also remain unresolved, leaving staff and students hesitant to use these tools for work that is formally assessed. Time is another less visible constraint. Faculty have limited time to rethink how they teach and conduct research alongside existing responsibilities.

Capacity sits underneath all of it. Students are often moving faster than their institutions, and faculty development has not kept pace, which says less about individual willingness than about how little structured training exists. What is needed is institutional-wide support: practical skills for students, support for faculty applying these tools to teaching and research, and guidance for senior leaders making investment decisions about rapidly changing technologies. 

Where African universities have made visible progress, they have generally started with people rather than platforms. As highlighted by Prof. Chika Yinka-Banjo, Professor of AI and Robotics at the University of Lagos, the capability behind the University’s AI and Robotics Lab and the UNDP-supported innovation pod that followed was built over several years, starting with young learners and extending to students, doctoral researchers, and eventually faculty. Technology investment does not create institutional readiness on its own.

Infrastructure, Talent, and the Case for African-Led Research

Infrastructure is often treated as an operational constraint. In the AI context, it is becoming a structural one. Institutions without reliable access to computing power cannot realistically build or test models using their own data. This reinforces dependence on systems trained on other people’s languages, datasets, and knowledge systems,  leaving the imbalance unchanged.

Infrastructure also affects where talent works. A university may train researchers and students, but without the computing resources and research environment to support their work, it risks losing them to institutions that can provide them. This is a challenge that is widely recognized, and there are efforts to address this gap through initiatives such as Masakhane by building African language datasets, models, and tools, alongside the local research and technical capacity needed to support more inclusive and contextually relevant AI systems. Dependable infrastructure, alongside the funding to sustain it, is therefore one of the more important tools available to help retain AI talent on the continent.

The answer is not infrastructure alone. African universities need greater capacity to conduct research on AI and shape the systems being developed. This includes work on African languages, datasets, knowledge systems, and solutions built for local conditions rather than adapted from elsewhere.  It also means stronger connections between universities so that there can be exchange in research, expertise, and infrastructure, and to build on existing work instead of starting from scratch each time. Outreach and training matter, but they need to sit alongside stronger research capacity.

Universities also have a role well beyond their own campuses, and some are already working this way. At Mulungushi University in Zambia, that has meant hosting the UNESCO International Centre for Education Innovation, bringing more than 18 universities onto a shared training platform, convened around 100 participants to draft a common AI-in-education policy, and made generative AI compulsory for incoming students so that its uses and limits are learned early. A similar approach is emerging in health education. Through the African Forum for Research and Education in Health (AFREhealth), comparable training on prompting and ethics has reached health professional educators across the continent, supported by a readiness survey jointly developed with DG and IREX that surfaced many of the same gaps higher education is now naming. Health systems have already had to confront questions about how professionals should use AI, and there is learning worth borrowing from this area.

These examples point to a wider shift: universities are not simply recipients of national AI strategies. They can be places where those strategies are tested, adapted, and translated into skills, research, and public value.

The funding question is beyond technology

If the goal is to move from experimentation to institutional capability, investment has to work on three fronts at once. Infrastructure without capacity produces facilities that go underused; capacity without infrastructure produces researchers who may leave; governance without either regulates activity the institution is not yet equipped to carry out.

This suggests a different investment question for funders and governments. Rather than simply focusing on how to increase access to AI tools, they need to consider what combination of investments allows universities to build and sustain capacity. That includes connectivity, computing and data environments; faculty development, researchers and students, and technical expertise; and the policies, incentives and institutional systems for responsible use. 

Two conditions will determine whether these investments translate into practice. The first is incentives. Access to computing power becomes research capacity only when scientists have reason and support to use it. This is why targeted funding for researchers applying AI to health, agriculture, and climate challenges tends to go further than generic capacity support. The second is decision guidance for university leadership. Senior leaders are being asked to make choices about tools, infrastructure, policies, and partnerships in an environment where evidence is still developing. Helping leaders understand their options and sequence investments may be one of the least expensive gaps in the system to close and one of the most consequential.

There is also a category of investment individual universities cannot make efficiently on their own. Shared infrastructure, common curricula, joint research collaboration, and pooled procurement can change what is affordable to an institution. Regional networks such as ARUA can play an important role in coordinating these investments and creating opportunities for universities to learn from one another. Funders, governments, universities, and networks achieve considerably more here acting together than separately.

The same applies to governance. Most institutional AI policies are often framed primarily on what students may and may not do. A more useful approach needs to cover teaching and research, data privacy, intellectual property, and ethics and expectations for faculty as well as students. It should also consider how universities engage with national AI institutions and with the communities they serve.

From Adoption to Institutionalization

Taken together, these examples suggest that the AI readiness conversation in Africa is moving beyond access to tools. The emerging agenda is about whether institutions have the people, infrastructure, research ecosystems, and governance arrangements to use AI effectively and responsibly. This all leads to a different way of thinking about AI readiness. The challenge for African universities is not simply to increase AI adoption. It is to turn scattered individual use into coordinated institutional practice, supported by the capabilities, infrastructure, governance, and partnerships needed to sustain it. We have come to think about this as a shift from adoption to institutionalization.

That framing starts from what is already true on campuses rather than assuming universities are waiting to begin. It also recognizes that universities are at genuinely different stages.  Some are experimenting, others are developing policies, and others are building research capacity or shared infrastructure. The useful question therefore is not whether a university is “AI ready” but where the institution is now, what it is trying to achieve, and what the realistic next step looks like.

For some institutions, that next step may be a structured assessment, a way of measuring institutional maturity over time rather than at a single point. For others, it may be faculty development, a governance framework, a sandbox environment to test implementation, improved access to computing, or a partnership that makes shared infrastructure possible.  Our work with IREX, ARUA, universities and partners across the continent is increasingly focused on understanding what that transition requires in practice: how institutions can build from experimentation already happening on their campuses towards responsible, sustained use of AI that strengthens teaching, research and public value.

African universities are not waiting to begin. They are already experimenting, learning and building. The opportunity now is to create the institutional conditions that allow that experimentation to become capability. That is the shift that we see from ambition to adoption, and ultimately from adoption to institutionalization.

***

This piece draws on insights shared by Prof. Chika Yinka-Banjo, Prof. Hewan D. Degu, Dr. Brian Halubanza, Vincent Nkundimana, and Loise Ochanda during the regional discussion hosted by Development Gateway, IREX, and ARUA on September 9, 2026.