Updraft O5: Africa is Making the Same Mistake, Again
African tech stakeholders need to pause and re-think our position on the AI foodchain.
It has been scarcely a year since ChatGPT 4o’s live demo. There had been a good deal of fussing over the release of its voice feature, leaps ahead in emulating natural speech; with a wider emotional range, tone switching and inoffensive humor. I watched that historical demo, when the speculative walls fencing sci-fi from reality supposedly crumbled, while eating lunch. I did not care much for it as an African. Like TTS technology, AI assistants are not made with the African tongue or thought in mind. It is not as though these behemoth tech companies lack the means to build products that support indigenous African languages, TechCabal writes, but they steer clear because Africa is characterized as a ‘low-revenue market.’ In an interview with software developer Nana Ghartey he explains that this was a simple matter of priority, ‘They’ve prioritized the languages they want to focus on first.’ Ghartey is the brain behind Abena AI, a voice assistant fluent in Twi, a language spoken by 60% of Ghana’s population. He had begun his foray into artificial intelligence and voice technology when he created Kofi for his grandmother, a Fante assistant that helped her claim control of her Western-oriented smartphone, training the AI on voice recordings of his grandmother and sister.
Ghartey’s innovation - an independently funded response to an observed local need- is hailed as an inspiring feat (which it is) when in fact it only amplifies warning signs. Africa is making the same mistake it did at the start of the internet era; in letting itself be designated as a consumer. not a highly regarded one, on the food chain.
While the internet has existed since around the cold war, by 2001 less than 1% of sub-saharan Africa had internet access, according to data from the Global Policy Institute. Only one undersea fiber-optic cable connected Sub-Saharan Africa to the global internet: SAT-3, controlled by a consortium of largely foreign telecom firms. This monopoly kept bandwidth scarce and prices sky-high. By 2024, the figure had risen to 40% and the investments that made this possible had mostly come from Google and other Western tech giants (someone had to do it if the governments weren’t going to). This was not philanthropy, but strategic decisions to create a continent-wide workforce of young IT professionals trained on their products and systems, source and market in one place. The same colonial playbook; expose Africans just well enough to make for decent (and cheap) labor. The same colonial playbook; only slaves and natural resources became metaphors for other things. This is the same playbook being used against us now in the AI revolution. According to Partech Africa's 2023 report, over 70% of startup funding in Africa came from international investors. This time we don’t have the excuse of being infant nations just recovering from the rape of colonialism.
A story published in The African Exponent this March states that the ‘African AI market is projected to reach $4.92 billion by the end of 2025’, with at least 2,400 AI startups currently operating across the continent. The average African reader whose use (and knowledge) of AI application is limited to LLMs is right in musing where all that money is coming from and going to. Most of it is made up by private equity and venture capital. AI and robotics startups in Nigeria alone have received over $1 billion in seed funding from global tech giants in recent years. Server costs and IaaS fees also factor in this figure. This is not to say that African AI startups are not penetrating local audiences or turning in a profit, some are. Just last year, South-Africa based Dataprophet, a startup optimizing manufacturing with AI, reported a $6.6 million revenue. But who are their clients? In many cases, the clientele/target audience of African AI startups is wholly or mostly European/American. Such appears common sense and benign, after all where is the market for advanced AI-driven optimizations in countries still building basic public infrastructure?
The problem really rears when you consider how projects are chosen for funding in the first place. Investors (local and foreign) often chase what is validated by Silicon Valley. As a result, startups building “local-first” or “non-Western” products—say, an AI assistant in Twi or Igbo—are often seen as niche, unserious, or low-return, even if they’re solving real problems. That is how we ended up with literal thousands of AI businesses offering services like edtech that assumes widespread digital literacy and personal device ownership or neo-bank apps focused on crypto trading in countries largely operating informal economies. But there are stakeholders and investors with wallets that have to be catered to. Enter the first world markets.
If you are cynic enough to see this situation was deliberately guided to where it has gotten to, you are not far from truth. Africa imports or licenses the hardware and models its applications are built on. In some cases, any existing patents are filed abroad, where the lawyers and capital are. Even when data training is local, pretrained models like GPT, BERT are the foundation — meaning the innovation is ultimately derivative.
The question is no longer about how much we are participating in the AI revolution, but whether the systems we are building are rooted in local agency, or we are merely positioning ourselves as African extensions of foreign value chains.
Tech pundits say we are living through the 4th Industrial Revolution (an exasperating euro-centric categorization of the global timeline, by the way). The AI Revolution. Critics like Ed Zitron argue that this is all a massive exaggeration. As he points out in a frothing essay disassembling the ‘AI revolution’, ‘the Large Language Model paradigm is also yet to produce a successful, mass market product.’ In any case, AI would continue to generate excitement within investment and stock communities for the foreseeable future. This new order of things isn’t driven by hunger, or visionaries but by the Valley which in turn is driven by an insatiable profit craze. It is key that this point is clear; the technological evolution of the world lies in the hands of CEOs who have no problems selling unhelpful products if it gives more market share. African investors and developers should be obsessed with maximizing research into this emerging technology, creating truly helpful products that may become revolutionary, rather than piggybacking on hype and foreign initiative. The continent certainly does not lack challenges that could be addressed meaningfully with machine learning. Projects targeting agriculture or energy sectors should be the norm, rather than outliers. Of course, the price of such inward focus may be alienation from many sources of Western funding. This is why a true AI revolution in Africa has sail on African money; local and diaspora investors, state funding.
Artificial Intelligence and machine learning technology may define reality as we would know it for the next 20 years, whether we like it or not. The capacity to steer the course is in our hands as a collective; everybody has a role to play- investors, the state, AI entrepreneurs, end-users. We should have begun acting already; the current trajectory indicates that the world’s second largest continent might spend another 20 years playing warehouse for its counterparts elsewhere.
Culture journalist and music commentator. Ebri’s essays have appeared in The Republic, Vanguard Nigeria, Afrocritik, African Writer Magazine, and elsewhere. MAAR 2025 fellow, he was shortlisted for the Ikenga Prize in 2024.