In the golden light of late afternoon on a smallholder farm outside Kisumu, Achieng Odhiambo scrolls through an app on her battered smartphone. The maize leaves show the telltale rust she has battled for years. The app — built with models trained on thousands of African crop images and speaking back in her Luo dialect — flags the problem, suggests the precise spot to treat, and factors in local weather forecasts pulled from satellite data. She smiles. This season might actually break even.
Two thousand kilometres away in a generator-powered co-working space in Lagos, 27-year-old software engineer Tunde Bakare fine-tunes an AI model that predicts flash floods for informal market traders. His team works in short bursts between power cuts. “We are not waiting for perfect infrastructure,” he says. “We are building around what exists.”
These are not isolated miracles. They are early skirmishes in the most consequential technological contest of the next decade: the race to embed artificial intelligence across Africa. The stakes are enormous — economic transformation for 1.4 billion people, or a new chapter of extraction dressed in code.
The economic promise, quantified and tempered
Recent projections from the African Development Bank suggest that strategic, inclusive deployment of AI could add up to $1 trillion in additional GDP by 2035 — roughly one-third of the continent’s current economic output. Agriculture alone employs more than 60 percent of Africa’s labour force, and AI tools that detect pests, optimise planting, or connect farmers directly to buyers are already moving from pilots to scale. Africa will have the world’s largest working-age population by 2030; if AI raises productivity fast enough, it could absorb millions of young people into higher-value work rather than informal survival.
When the grid fails the algorithm
Reality arrives with the next power outage. Africa hosts roughly 200 commercial data centres delivering about 500 megawatts of IT load — around 1 percent of global capacity. In May 2026, Kenya suspended a flagship $1 billion data-centre project backed by Microsoft and G42 — not for lack of capital, but because the facility would have consumed roughly one-third of the country’s entire installed electricity capacity. Successful deployments therefore lean heavily on edge computing, offline-first design, and models compressed to run on modest hardware. The companies winning ground treat unreliable infrastructure not as a bug but as the defining constraint.
The talent that builds — and often leaves
Walk into any thriving tech hub in Nairobi, Lagos, or Kigali and you will meet engineers whose ingenuity was forged by scarcity. Yet the continent still faces a severe shortage of advanced AI researchers, and many of the brightest move abroad for stable power, higher pay, and better research environments. This brain drain is not new, but AI accelerates the stakes. Retention strategies — equity, meaningful local impact, competitive local R&D centres — are now table stakes for any serious player.
Western companies at the gate: the smart playbook
The Western AI firms making real progress are not parachuting in with generic foundation models. They are embedding. In Kenya, Penda Health partnered with OpenAI to build a clinician copilot for primary-care doctors, designed explicitly for low-resource settings. Rising Academies, across West and East Africa, uses Anthropic’s Claude models to power lightweight tutoring tools that function on shared devices and spotty connections. The pattern is consistent: deep local partnerships, context-specific data, mobile-first interfaces, and a willingness to co-create rather than simply deploy.
Sovereignty, colonialism, and the real fight
The most charged conversation on the continent right now is not about whether AI will arrive — it is about who will control it. Critics speak openly of “digital colonialism”: African data extracted to train foreign models, value captured offshore, and algorithmic decisions made without meaningful local oversight. The counter-narrative is equally powerful: Africa has the chance to leapfrog not just old infrastructure but old power structures, through sovereign compute initiatives and local foundation models trained on African languages and realities. The question is whether these efforts can scale before foreign platforms lock in dominance.
What the next chapter requires
For Western AI companies serious about the long game, the requirements are clear: treat infrastructure constraints as design parameters; invest in local talent and retain it through real ownership; co-create data and models rather than harvest them; align with emerging sovereignty goals; and measure success in local value creation, not just revenue extracted.
The farmers checking their phones in Kisumu fields and the engineers debugging between blackouts are not waiting for permission. They are already building. The companies that meet them where they are, with humility and long-term commitment, will shape the next era of the continent’s story. The rest will become footnotes in someone else’s success narrative. The code is being written right now — on cracked screens, in generator-lit rooms, and in the quiet determination of people who have waited long enough for the future to arrive on their own terms.
Culture & Clarity helps leaders build trust, psychological safety and ethical AI culture. Explore our services or book a consultation.