An opinion piece
Every few months, a new AI model drops from Silicon Valley, a new frontier lab secures a billion-dollar cheque, and the world’s tech press celebrates yet another leap forward. The question that hangs quietly over the African continent — often unasked but deeply felt — is this: where do we stand in all of this? Are we watching from the sidelines, or are we in the game?
The answer, it turns out, is more complicated and more hopeful than the dominant narrative gives us credit for.
The Case for “Just a Spectator”
Let us be honest first. The numbers are sobering. As of mid-2025, 159 African AI startups had collectively raised a total of $803 million over five years. That sounds impressive until you hold it next to the global picture: in 2024 alone, global private investment in AI hit between $100 and $130 billion. Africa’s entire five-year total represents less than what a single major US AI company raises in one funding round.
The infrastructure gap is equally stark. Only about 5% of African AI innovators have reliable access to advanced computing resources, and the continent faces an estimated 7 million GPU hours of unmet demand for model training over the next three years. Power instability, limited data centres, and the chronic underrepresentation of African languages in global datasets are not minor inconveniences — they are structural barriers that slow the pace of foundational AI work.
Brain drain compounds everything. Talented engineers and researchers, trained at African universities, continue to migrate abroad in search of better-resourced environments. This is not a moral failure — it is a rational response to an uneven global system. But it depletes the very talent pool that could accelerate homegrown innovation.
The Case for “We Are Actually Building”
And yet. Walk through the evidence more carefully, and a very different story begins to emerge.
The conversation on the continent has shifted — quietly but meaningfully — from adoption to ownership. African startups are no longer simply integrating GPT APIs into existing apps and calling it AI. A new generation of builders is working at the foundational level: training models on local language data, designing AI-ready hardware infrastructure, and engineering the compute pipelines needed to power it all.
Take the semiconductor space. Cairo-based InfiniLink — a startup developing silicon-photonics chiplets specifically designed to cut energy consumption inside AI data centres — raised $10 million in early 2025. That is not an application layer play. That is deep infrastructure work, the kind of thing that historically only emerged from labs in the United States, Taiwan, or South Korea. The startup was subsequently acquired by GlobalFoundries, a US-based semiconductor manufacturer — validation that African deep-tech can reach the global stage.
Cape Town’s Cerebrium, founded in 2021, has built what may be Africa’s most significant serverless AI infrastructure platform. Its GPU optimisation tools and custom runtimes allow engineers to deploy machine learning models with dramatically reduced cold start times, directly reducing African developers’ dependence on expensive global cloud providers. An $8.5 million seed round in mid-2025 — backed by Google’s Gradient Ventures and Y Combinator — signalled to the international investment community that Africa can produce not just AI users, but AI infrastructure builders.
Then there is Tunisia’s InstaDeep, perhaps the continent’s most globally recognised AI success story. Founded in 2014 in Tunis, it grew to a $680 million valuation before being acquired by BioNTech. What made InstaDeep notable was not the exit valuation but what preceded it: a serious investment in compute infrastructure in 2018, years before the broader AI boom, that catalysed a cluster of deep-tech startups in its wake. Tunisia now has just nine funded AI startups — but those nine companies have collectively raised $244 million, averaging over $27 million each, demonstrating a strategic focus on building globally competitive companies rather than spreading thin.
The Distinctly African Approach
What strikes me most is not the amount of money being raised or the number of startups being founded. It is the kind of problems being solved, and the solutions being designed in response.
African AI builders are not simply copying Western models and localising them. They are confronting challenges that most Western AI labs have never had to think about: how to build AI systems that work reliably in low-bandwidth environments; how to train language models that understand Hausa, Swahili, Amharic, and hundreds of other languages that global LLMs largely ignore; how to make AI-powered healthcare diagnostics accessible in places where specialist physicians are scarce; how to deploy agricultural AI tools via SMS because smartphones are not ubiquitous.
Ethiopia’s Gebeya Dala, launched in 2025, is a perfect illustration. It is an AI app builder specifically designed for the African context — mobile-first, capable of generating full-stack code from plain-language prompts in local languages including Hausa, Swahili, and Arabic, and optimised for low-data environments with mobile money payment integration built in. You would not build that product in San Francisco. You build it when you understand the actual constraints of your users.
Kenya’s Shamba Records uses AI to give over 50,000 smallholder farmers smart credit access, market connections, and climate-resilient agricultural guidance. Rwanda’s Smartel Agri Tech deploys AI-powered, solar-driven devices to warn farmers of crop disease outbreaks — all via SMS. Ghana’s SmartSkin Africa uses AI to deliver personalised skincare guidance specifically calibrated for African skin types, a market the global beauty-tech industry has historically neglected.
These are not niche applications. They are solutions to problems affecting hundreds of millions of people, built by people who live those realities.
Where I Stand
My honest view: Africa is neither a pure spectator nor an equal participant in the AI boom. We occupy a third category that the Western tech press does not have a good vocabulary for — a continent that is simultaneously constrained by infrastructure deficits and energised by a generation of builders who are turning those very constraints into design principles.
The geographic concentration is a legitimate concern. Kenya, Tunisia, Egypt, South Africa, and Nigeria account for the overwhelming majority of Africa’s AI activity. Forty-four countries are largely absent from the conversation. This is not an African AI boom — it is a five-country AI boom on a continent of 54 nations. That distinction matters enormously for policy and investment priorities.
The window to shape Africa’s role in AI is also narrowing. The foundational models, the compute infrastructure, the data pipelines that will define the next decade of AI development — these are being built now. An Africa that waits another five years risks being locked into a permanent consumer relationship with technology developed elsewhere, trained on data that does not reflect African realities, optimised for problems that are not African problems.
But I am genuinely encouraged by what is emerging. The shift from consumerism to infrastructure building is real. The funding, while modest globally, is growing. Governments in Nigeria, Kenya, Ghana, and others have announced AI roadmaps. New data centres — including Nairobi’s IXAfrica in partnership with Safaricom — are coming online. The World Economic Forum estimates that investment in green, AI-ready compute infrastructure could unlock $1.5 trillion in economic value for Africa by 2030.
Africa is not watching the AI boom from the sidelines. But we are not yet building the game-defining technology either. We are somewhere in the middle — scrappy, imaginative, solving real problems with limited tools, and slowly, steadily building the foundation for something larger. That is not a story of failure. That is a story worth paying attention to.
The question is whether the rest of the world — and our own governments and investors — will pay attention before the window closes.
Sources: TechCabal, StartupList Africa, World Economic Forum, African Exponent, Google for Startups Africa, African Business Magazine — all 2025.
