Listen to some commentators, mostly web journalists, and AI is a magic solution that will transform Africa and let it leapfrog. Yet AI will not build roads, bridges, hospitals, schools, or water and power networks.
AI does not replace good governance, the effective organisation of society or social peace. Without that groundwork, “Organize to not agonize”, it risks mainly amplifying existing inequalities.
The Novissi lesson
Take Novissi, Togo’s mobile cash-transfer programme. During COVID-19 it used machine learning on satellite and mobile-phone data to help target support to the poorest. It is a genuine innovation.
But sending virtual money to a rural citizen is not enough if the roads are impassable, the markets flooded and farming unproductive. Money that cannot reach a market, a clinic or a school changes little.
Even the richest economies are judged on their foundations
The American Society of Civil Engineers grades US infrastructure in its 2025 Infrastructure Report Card: an overall C, with a US$3.6 trillion investment gap over the next ten years. If the world’s largest economy still has to pour concrete, lay pipes and modernise its grid, no algorithm will do that work for Africa.
Optimise, not replace
AI can optimise development: better targeting, better planning, better maintenance, better decisions. It cannot replace its foundations. Without effective governance, social stability and massive investment in infrastructure, we will go nowhere.
Two news stories, a few weeks apart, tell the same story. In France, SoftBank has pledged up to €75 billion for AI data centres. In Kenya, a flagship US$1 billion Microsoft–G42 data centre has stalled for lack of electricity. What separates them is not algorithms, chips or talent. It is energy.
France: electricity as a competitive advantage
At the Choose France summit on 1 June 2026, France announced about €93 billion (around US$108 billion) in foreign investment pledges.
The largest came from SoftBank: €45 billion for a first phase of 3.1 GW by 2031, potentially rising to €75 billion for 5 GW of AI data-centre capacity in northern France, widely described as Europe’s largest AI infrastructure pledge.
Why France? Because it offers what is becoming rare in the age of generative AI: abundant, stable and low-carbon electricity, largely supported by its fleet of 57 nuclear reactors and a growing power surplus.
As hyperscale AI data centres multiply, electricity is becoming the new strategic currency of competitiveness.
GM
Kenya: when the power maths catches up
In 2024, Microsoft and the UAE’s G42 announced a US$1 billion data centre in Kenya, to be powered by geothermal energy from Olkaria.
Kenya produces most of its electricity from clean sources (commonly cited at 80–90%, mainly geothermal). But its grid has little surplus baseload.
At full build-out the campus would need about 1,000 MW, roughly one third of the country’s grid-connected installed capacity.
In May 2026, President William Ruto explained: “To switch on that one data centre, we would need to shut off power for half the country.” The project is on hold; the government says it has not withdrawn.
Kenya has the vision, the talent and the clean energy. What it lacks is surplus: grid stability, affordable electricity, peak-demand management and long-term generation capacity. That is the real gap between AI’s computing appetite and the power systems of developing nations.
Africa is starting to think nuclear
Rwanda has explored partnerships for small modular reactors (SMRs) as part of its long-term energy strategy.
Ghana continues to advance its plans for a nuclear power programme.
South Africa, Africa’s nuclear pioneer, is reassessing its expansion capacity.
Kenya has bold digital and AI ambitions, but still has to solve power reliability and scale for energy-intensive infrastructure such as hyperscale data centres.
We need an Estates General of Africa’s energy potential
Africa needs a continental Estates General of its energy potential: a complete, clear-eyed and strategic map of our resources. Hydropower, solar, gas, oil, nuclear, geothermal, wind, biomass, green hydrogen and regional interconnections.
The goal is not only to list what Africa has, but to decide how to turn that potential into real power: for industrialisation, security, jobs, digital transformation, data centres, modern agriculture and economic sovereignty.
A federal Africa cannot depend forever on fragmented models, weak national grids, scattered external financing or isolated energy policies. We need a continental blueprint for energy sovereignty, built on pooled resources, long-term planning, regional energy corridors, integrated electricity markets and a shared vision of energy as infrastructure of power.
Africa must stop treating energy as a simple matter of electricity production and see it as the foundation of its industrial and technological renaissance. Without abundant, stable, affordable and well-managed energy, there will be no digital sovereignty, no agricultural transformation, no local value chains and no lasting economic emergence.
This Estates General must bring together states, regional institutions, engineers, financiers, universities, the private sector, the diaspora, young innovators and strategic partners to build a common African energy doctrine.
Peace is an energy policy
This is precisely why our generation must commit to stabilising the Great Lakes region and the Sahel, to fostering peace, to pooling our resources and to building a genuine African energy strategy. The future of AI, industrialisation and our economic sovereignty will depend on our ability to turn our wealth into a shared vision and lasting prosperity.
Congo 🇨🇩 and Niger 🇳🇪: our energy lungs.
No power, no AI
The next AI powers may not simply be those with the best engineers, but those able to supply reliable, scalable and affordable energy to run AI infrastructure at scale. For Africa, AI readiness must go beyond digital skills and regulation. It must include energy strategy, grid modernisation, data-centre infrastructure and bold long-term investment.
No power, no AI.
GM
About the author: Akuété Giana MATHEY-APOSSAN is a project and portfolio manager and the founder of HOPE FOR AFRIKA. The goal: to lead large international infrastructure projects in AI, data centres, transport and energy.
What astonishes me is the upheaval under way. AI breaks every rule and works around every safeguard: intellectual-property rules ignored, mass data collection, privacy violations, and much more.
A USD 1.5 billion settlement… for whom?
Anthropic reportedly downloaded around 7 million books to develop its AI models.
Only around 500,000 titles are covered by the USD 1.5 billion settlement.
Gross compensation is estimated at about USD 3,000 per eligible title, an amount that may be shared between authors and publishers.
To be eligible, a book had, among other things, to: come from LibGen or PiLiMi; have an ISBN or ASIN; have been registered in time with the US Copyright Office.
Result: around 6.5 million downloaded titles are not covered by this settlement.
African authors, left out
Many foreign authors, African authors in particular, did not know these American formalities or had no reason to register their works in the United States.
Their works may have contributed to the value created by AI, yet they will not automatically benefit from the reparations.
In theory they can bring a separate claim, but very few have the means to sue a major American technology company.
Conclusion: the same disregard for everything that is not American continues
The works were used worldwide, but the reparation remains American, restrictive and hard to access. AI thus reproduces inequality not only in its outputs, but also in how damages are distributed.
Every AI answer is paid for in electricity, water and computing power. Africa is short on all three.
One email, one bottle of water
Asking a model like GPT-4 to write one 100-word email uses about 519 ml of water, a little more than a bottle, and 0.14 kWh of electricity, enough to light 14 LED bulbs for an hour. Once a week for a year, that becomes 27 litres of water and 7.5 kWh. Now multiply by a whole country.
Electricity: we cannot power what we do not have
Training one major AI model can use as much electricity as 1,000 homes consume in a year, and about half of it goes to cooling, not thinking. Meanwhile, roughly 600 million Africans still live without electricity. Every megawatt sent to a data centre is a megawatt not sent to a clinic, a school or a home.
Water: cooling machines while people wait for water
AI chips run so hot that data centres evaporate large volumes of water to cool them. In regions already facing drought and water stress, spending clean water to cool servers is a choice we cannot afford.
Compute: the price keeps rising
Building a frontier model has gone from roughly $1,000 (GPT-1, 2018) to millions (GPT-3) to close to $100 million (GPT-4). One high-end AI chip costs $25,000 to $40,000, and large models need 20,000 to 30,000 of them. Add top researchers earning $500,000 to $1 million a year. At that price, Africa becomes a customer and a data source, not a builder.
So what do we do?
We stop copying a model built for rich grids and full reservoirs. Africa should invest first in reliable, renewable power; in smaller, efficient models; in shared regional computing; and above all in people. Programming, data analysis, mathematics and statistics are the new Office suite. We must master them to use AI wisely, not just consume it.
— Akuété Giana MATHEY-APOSSAN, CEO & Founder, HOPE FOR AFRIKA · PMI Authorized Trainer · PfMP | PgMP | PMP | CPMAI | RMP
Figures on cost, electricity and water come from IST 692 Responsible AI course material (Syracuse University, Week 1). Email estimates come from a 2024 Washington Post / UC Riverside analysis for US data centres and vary by location. Electricity access figure: IEA, approximate.
An illustrated story: Papi Koffi answers his grandchildren’s questions about how much electricity artificial intelligence really needs.
1. Dangote Refinery ≈ 435 MW
Kossi: Papi, how much power does the Dangote Refinery need?
Africa’s biggest refinery makes fuel for millions. Its own power plant gives about 435 MW. — Papi Koffi
2. One AI plant ≈ 4½ Dangote refineries
Ama: And a giant AI plant?
One of the world’s biggest, in the USA, is reported to need about 2,000 MW. — Papi Koffi
3. One AI plant ≈ 1/3 of Nigeria’s electricity
Esi: How much electricity does all of Nigeria have?
Our national grid’s record is about 5,800 MW, shared by more than 200 million people. — Papi Koffi
4. One house in three goes dark
Kossi: What would that mean for us?
Plug that AI plant into our grid, and it is like switching off 1 house in every 3 across Nigeria. — Papi Koffi
Power for AI is power taken from homes, schools and clinics.
5. So should Africa forget AI?
No! First we build more power from sun, wind and water. Then smaller, smarter AI. And young people who master maths, data and code. — Papi Koffi
Build the power. Build the skills. Then build the AI.
Where do these numbers come from?
Dangote Refinery power plant: about 435 MW (Nairametrics, 2023). Giant AI plant: xAI Colossus, USA, reported at about 2,000 MW in early 2026 (industry reports; not officially confirmed). Nigeria’s grid record: about 5,802 MW on 4 March 2025 (Transmission Company of Nigeria). These are capacity figures, not exact daily use. AI plants are growing fast, so numbers change quickly.
— Akuété Giana MATHEY-APOSSAN, CEO & Founder, HOPE FOR AFRIKA · PMI Authorized Trainer · PfMP | PgMP | PMP | CPMAI | RMP
AI’s foundations are simple to understand. Its progress comes from the convergence of four developments:
📊 More data: sensors, transactions, research and digital activities provide information from which AI can learn.
⚡ More computing power: GPUs, supercomputers, cloud infrastructure and edge devices enable faster processing at greater scale.
🔗 More connected systems: digitalization and interoperability allow sectors and organizations to exchange and use information.
🧮 Better algorithms: mathematics and statistics, applied through languages such as Python and R, help turn data into patterns, predictions and useful insights.
Together, these developments enable AI to analyze volumes of information that humans could not practically process unaided.
Speed is not accuracy
But speed does not guarantee accuracy. A pattern does not prove causation. Human judgment and accountability remain essential.
🚀 Is this the path toward superintelligence?
Possibly. These advances support increasingly capable AI, but superintelligence — intelligence exceeding human capabilities broadly — remains a hypothetical frontier, not an established achievement.
Super Intelligence Area 😎 — policy watch. On 29 September 2026 the White House issued the executive order “Inaugurating the Era of Super Intelligence”, which directs U.S. federal agencies to use the terms “Super Intelligence” and “SI” instead of “Artificial Intelligence” and “AI” in official communications. A change of vocabulary is not a change of capability: the questions of accuracy, causation and accountability above still apply.
AI expands what we can process. Its value depends on how wisely we use it.
— GM · Akuété Giana MATHEY-APOSSAN, CEO & Founder, HOPE FOR AFRIKA · PMI Authorized Trainer · PfMP | PgMP | PMP | CPMAI | RMP
Artificial intelligence is moving from pilots to everyday operations — in banks, telecoms, tax administrations and public services. With it come new questions: who is accountable for an AI system’s decisions, how is data protected, how are risks such as bias or errors managed? ISO/IEC 42001, published in December 2023, gives organisations a structured answer.
What ISO/IEC 42001 is
ISO/IEC 42001 specifies requirements for an AI management system (AIMS): the policies, roles, processes and controls an organisation uses to develop, provide or use AI responsibly. Like ISO/IEC 27001 for information security, it follows the common ISO management-system structure, so it can be integrated with existing systems.
Who should care
Organisations that develop AI solutions for clients.
Organisations that buy and deploy AI tools — chatbots, scoring models, document automation.
Public institutions that use AI in services to citizens.
Organisations already certified to ISO/IEC 27001 that want to extend their governance to AI.
Five steps to get started
Inventory the AI systems you develop or use, including tools bought from vendors.
Assign accountability: who owns each system and its risks?
Assess risks and impacts — on people, data, operations and reputation.
Define policies and controls proportionate to those risks.
Train the people who build, buy and run AI systems.
Projects and AI governance go together
Many AI failures are project failures: unclear objectives, poor data, no plan for adoption. Combining ISO/IEC 42001 with an AI-specific delivery method such as PMI-CPMAI™ helps teams deliver AI that is both useful and well governed.
HOPE FOR AFRIKA offers ISO/IEC 42001 Lead Implementer and Lead Auditor programmes with PECB, and PMI-CPMAI™ training. See the Academy.
