When Britain’s artificial intelligence minister, Kanishka Narayan, addressed an audience at Imperial College London on October 12, his argument reached beyond the familiar promises and perils of intelligent machines. He argued that governments need the technical capacity to understand, protect and shape societies increasingly dependent on AI. Legislation alone cannot supply that capacity. Neither can confidence in technologies controlled elsewhere.
For Africa, the implications are uncomfortable. If Britain, with its universities, scientific institutions and financial resources, worries about technological dependence, what should African countries conclude about their own position?
My answer is that we must become more ambitious and more discriminating at the same time. Africa needs an AI development strategy grounded in what it can build, what it should buy, what it must protect and where it can compete. Attendance at international summits is useful. But it is no substitute for productive capability.
The major powers are pursuing overlapping ambitions from different positions. The United States combines frontier laboratories, capital markets, cloud platforms and a formidable semiconductor-design ecosystem. Stanford’s 2026 AI Index puts American private AI investment in 2025 at approximately $286bn, against about $12bn in China. These figures measure private investment, however; they do not capture the full weight of Chinese state support.
China’s achievements also expose the weakness of equating expenditure with capability. Stanford reports that the performance gap between leading American and Chinese models has narrowed sharply, with the lead changing on important comparisons. China’s strengths extend into research, industrial robotics and applying AI to production. Africa should study both America’s capacity to finance invention and China’s capacity to connect technology with industrial development, without assuming that either political model is a template.
The European Union is seeking to combine safeguards with productive scale. Its AI Continent Action Plan supports computing infrastructure, skills, data access and adoption. In April 2026, the European Commission reported a network of 19 AI factories. Europe’s experience raises a useful question: can regulation support a competitive technological economy when paired with investment, accessible infrastructure and industrial demand?
Britain brings scientific depth and an institutional emphasis on AI evaluation. Narayan’s speech added a sharper argument for domestic capability and resilience. His reasoning deserves African attention: a state cannot effectively oversee systems it lacks the expertise to understand.
These are differences of emphasis, not sealed compartments. Each power is pursuing investment, security, talent and influence. Africa must enter that contest with its own priorities.
Where, then, does the continent stand?
I would describe Africa as being in an uneven phase of adoption and adaptation, with pockets of original research and entrepreneurship, but limited capacity to shape the global technological frontier. That is an analytical assessment, not a ranking of 54 interchangeable economies. Countries, industries and institutions vary enormously.
Nor is Africa starting from nothing. The African Union endorsed its Continental Artificial Intelligence Strategy in July 2024. African researchers are building language resources and applications that imported systems have neglected. Masakhane, the African language-technology research community, demonstrates the value of collaboration organised around African needs. Its work challenges the assumption that Africans should supply data while others perform the research.
This need for institutional capability informs the work we are undertaking at GenAI Learning Concepts Ltd in Nigeria. Through AI readiness assessments, executive education and sector-specific adoption programmes, we are helping boards and management teams understand how artificial intelligence can improve their organisations and what responsible implementation requires. Our focus spans financial services, capital markets, insurance, manufacturing and other sectors, addressing business strategy, workforce preparedness, data governance, cybersecurity and effective oversight.
Yet a continental strategy does not automatically become a funded laboratory, a reliable electricity supply or a commercially sustainable enterprise. Our central difficulty is converting intent into institutions and institutions into results.
The first requirement is electricity. An AI policy that ignores power availability, reliability, stability, and cost is incomplete. Computing facilities require dependable energy, connectivity, maintenance and, depending on their design, substantial cooling resources. Governments should assess proposed facilities against local grid capacity and community needs. Expensive infrastructure that strains essential services or stands underused is a liability. Nigeria’s power capacity still hovers around 6,000 MW, barely enough to power a Tier 3 data centre.
Regional cooperation could make investment more economical. African countries should develop shared research-computing facilities where demand, power and connectivity justify them, with transparent access for universities and businesses. Development finance can help reduce early investment risk, but should follow credible demand assessments and independent procurement. The running costs matter as much as the opening ceremony.
A second requirement is a more practical understanding of sovereignty. No African country needs to manufacture every chip or reproduce every foundation model to exercise meaningful technological agency. It does need competent buyers, enforceable contracts, secure data arrangements, access to evaluation and the ability to change suppliers.
Buying foreign technology need not surrender autonomy. Dependence becomes dangerous when the buyer cannot inspect performance, negotiate terms, protect sensitive information or leave. A data centre on African soil is useful, but its location alone says little about who controls the software, equipment, contracts and access.
Africa should therefore work with American, Chinese, European and British partners while negotiating for specific benefits: research participation, technical training, access to computing, support for local enterprises and credible continuity arrangements. Partnerships should be judged by capabilities transferred and value retained. Strive Masiyiwa’s Cassava Technologies and its partnership with NVIDIA offer a promising avenue for expanding Africa’s access to the computing infrastructure that AI requires. If translated into accessible and competitively priced services, investment in GPU-powered data centres across African countries could give local researchers, start-ups and established businesses greater capacity to develop and deploy solutions suited to the continent’s needs.
Our youthful population offers an extraordinary opportunity, but demographic optimism can become an excuse for educational complacency. The African Union has highlighted that roughly 60 per cent of the continent’s population is under 25. Youth can support a large workforce of researchers, engineers, entrepreneurs and informed users. It cannot do so without investment.
Being comfortable with a smartphone is different from understanding statistics, assessing an algorithm or securing a database. The foundation remains literacy, numeracy, scientific reasoning and good teaching.
Above that foundation, we need vocational training, university research, industry apprenticeships and opportunities for experienced professionals to learn new methods.
The necessary workforce extends far beyond programmers. Africa will need nurses who can assess AI-assisted recommendations, agricultural advisers who understand model limitations, lawyers who can negotiate technology contracts and directors who can question automated decisions. Girls and young women must have equal access to these opportunities. Rural inclusion must be designed into delivery, rather than added to a launch speech.
Training also needs a labour market. Governments and companies should fund paid apprenticeships tied to real projects and publish evidence of employment and productivity outcomes. Counting certificates is easier than creating careers, but considerably less useful. AI may also automate parts of outsourcing and entry-level work. Young Africans need adaptable skills and routes into higher-value responsibilities, rather than preparation for tasks already becoming cheaper to automate.
Can Africa leapfrog?
Yes, in particular activities. The World Bank’s work on “small AI” identifies opportunities for affordable, specialised systems that can operate with fewer resources than frontier models. African businesses and public services should test these opportunities against clearly defined problems.
A voice-based agricultural service could make useful advice accessible in a farmer’s language. A teacher could use a locally evaluated assistant to prepare differentiated lessons. A clinic could improve stock forecasting; a small manufacturer could identify defects; a financial institution could strengthen fraud detection. Each application should be assessed against the existing process, including cost, accuracy, accessibility and the consequences of failure.
The promise is to extend scarce expertise and improve decisions. A clinical assistant still needs an accountable health professional, and a useful diagnosis still needs a functioning referral system. Leapfrogging can shorten a technological journey. It cannot abolish the need for institutions.
Language and data offer another opening. Systems that misunderstand African names, livelihoods and languages will serve Africans poorly. Universities, public agencies and businesses should collaborate on representative datasets and independent evaluations, with consent, privacy and appropriate compensation. Useful public information can be made accessible without exposing personal records. Local knowledge should generate local scientific and commercial value.
Government procurement could create early markets for capable African firms. Contracts should be awarded competitively, divided into manageable projects where appropriate, and linked to measurable improvements. Publish the results: time saved, errors reduced, service access widened and total costs incurred. Stop projects that fail. Public procurement should reward delivery rather than proximity to power.
Safety must develop alongside adoption. I have consistently advocated Responsible Human-in-the-Loop oversight. This requires people with competence, authority, time and information to challenge a system. A human who merely approves a machine’s recommendation supplies little protection. High-impact applications need audit trails, clear responsibility, incident reporting and accessible routes of appeal.
Ultimately, this programme requires political renewal.
Africa needs a new crop of leaders able to connect technological change with education, energy, enterprise and public accountability. More young people and women should enter decision-making. But age alone guarantees neither competence nor integrity. The qualities that matter are curiosity, respect for evidence, financial discipline and a willingness to build institutions that survive their founders.
Such leaders would protect research budgets, appoint technically capable officials, publish procurement terms and accept independent scrutiny. They would recognise that an AI system can help detect irregularities but cannot supply the political will to investigate them. They would measure progress through functioning services and expanding opportunity.
I would urge African governments to begin with a disciplined three-year programme: establish an honest readiness baseline, select a few valuable applications, finance the necessary infrastructure and skills, and publish annual results. Continental cooperation should reduce duplication and help successful enterprises operate across borders.
Britain’s anxiety about technological dependence should concentrate African minds. We have talent, unmet demand and the possibility of choosing more efficient routes to development. What remains uncertain is whether our institutions will allow those advantages to flourish.
Africa’s place in the age of AI will be shaped by work its people can do, the enterprises they can build and the decisions they retain the power to make. That is the future our leaders should be organising now.
Iroche is the founder and CEO of GenAI Learning Concepts Ltd, Lagos