Educating a Continent for the Price of a Phone

 

Africa's youth are its greatest asset, but only if they are schooled. The way we have always tried to get there — more buildings, more teachers, more textbooks — cannot move fast enough or cost little enough to close the gap. There is a cheaper, faster way. The catch is that it has to be led as public infrastructure, by Africa's own institutions, rather than left to the market.

Africa is about to hold the youngest workforce on the planet. By the middle of this century, close to one in three of the world's young people will be African. That fact is usually sold as a gift, the famous "demographic dividend." It is worth being honest about the condition attached to it. A young population pays out only if those young people can read, reason and earn. If they cannot, the same demographics turn into a bill nobody can settle.

So the real question is not whether Africa has the human potential. It plainly does. The question is how you get a functioning education to hundreds of millions of children, most of them poor, many of them rural, fast enough to matter and cheaply enough to be possible. The argument here is that the answer already exists, sitting in pockets and on market stalls across the continent, and that the whole cost of reaching a child for a year now comes to roughly the price of a low-end smartphone. The technology already exists. What is missing is a short list of hard problems still to be solved, and a decision about who takes responsibility for solving them.

The gap we are actually trying to close

Start with the size of the thing. Around 98 million children and adolescents in sub-Saharan Africa are out of school, and this is the only region on earth where that number is still climbing rather than falling. The gap widens as children age: roughly 28 million of primary-school age, 22 million of lower-secondary age, and 36 million of upper-secondary age are outside a classroom altogether.

Enrolment, in any case, flatters the truth. A child can be on a school register and learning almost nothing. In low-income countries, nine out of ten ten-year-olds cannot read and understand a simple passage of text. Nigeria alone, the continent's largest country, has more than 8.7 million children out of school. Add the structural gaps behind those figures — four in five schools with no electricity, single teachers running multi-grade classes, ten pupils sharing one textbook — and the scale of the shortfall becomes clear.

None of this is for want of trying. African governments spend a larger share of their budgets on education than most rich countries do. The effort is there; what fails them is the model.

Why we cannot build our way there

The reflex is to close the gap the way it has always been closed: build more schools and hire more teachers. Run the numbers on that route and it stops looking like a plan.

Building a rural primary school runs from about 60,000 to 200,000 US dollars; a secondary school with laboratories can reach 1.5 million. Africa needs on the order of 18 million new classrooms by 2030, which is well over 100 billion dollars in construction alone before a single teacher is paid. And teachers are the larger cost by far. Salaries already swallow between 60 and 70 percent of education budgets, and the continent would need to recruit and train roughly 15 million more of them by 2030 to reach every child. Reaching the global education targets would mean low-income countries doubling what they already spend. That money does not exist.

The ambition to educate every African child was never the problem; the delivery model is. A system whose costs climb with every extra pupil, one more classroom and one more teacher at a time, cannot outrun the fastest-growing young population on earth. The architecture is the constraint, and changing it is what changes the arithmetic.

The opening that did not exist ten years ago

Here is what has shifted. Sub-Saharan Africa now has 527 million unique mobile subscribers, about 44 percent of the population, and roughly 360 million people using mobile internet. Smartphones are still a minority of the phones in people's hands, but they are the fastest-growing segment, and by 2030 they are forecast to make up the large majority of the region's mobile connections. For millions of young Africans, the phone is the first and only computer they will ever own.

The device is getting cheaper by the month. Entry-level Androids sit around 100 dollars today, but a coalition of the big operators is pushing hard toward a 30-dollar 4G handset that could put a smartphone in the hands of 50 million first-time users. Price cuts work fast in this market: when Kenya scrapped the tax on handsets, purchases jumped more than 200 percent and penetration climbed from 50 to 70 percent inside two years.

Data is the other half of the equation, and it is wildly uneven. A gigabyte can cost as little as 20 cents in Nigeria, yet more than a dollar and a half in Kenya, Cameroon and South Africa.

That unevenness matters, because a learning platform cannot assume cheap data everywhere. The way through is twofold: build the product so it works offline first and sips data when it does connect, and strike zero-rating deals with operators so educational traffic does not eat a family's airtime. Platforms like Eneza have already shown operators will do this; its lessons were billed straight off mobile airtime through Safaricom, MTN and Orange. Treat those partnerships as core business rather than an afterthought, and unlimited access to learning stops being a fantasy even on a tight data plan.

The economics are a message to finance ministers

Now the part that gives the article its title. Strip out the buildings and the salaries, and the cost of educating a child collapses.

A well-run mobile platform costs roughly 40 dollars per student per year, all in: a slice of device subsidy, a slice of subsidised data, a small amount for AI tutoring, and the bandwidth and support around it. Traditional delivery runs to around 323 dollars per student per year and rising. That is about one-eighth of the cost, for a model where each extra learner adds almost nothing, because digital content copies for free and software tutors do not draw a salary.

Scale that across 25 years and the difference stops being a budget line and becomes a development strategy. Delivering education the traditional way to Africa's growing school-age population would cost around 2 trillion dollars cumulatively by 2050. The mobile route would cost about a quarter of a trillion. The gap, roughly 1.76 trillion dollars saved, is close to the entire annual output of sub-Saharan Africa outside its two biggest economies.

And that saving is the number a treasury should be looking at. The money freed in a single year could pay for basic healthcare for hundreds of millions of children, or sink millions of clean-water boreholes, or wire up millions of homes with solar.

That is a public investment case, and it belongs in finance ministries as much as in any boardroom.

It reaches them, but does it teach them?

Cost means nothing if children do not actually learn. This is where the sceptic's guard should go up, and where the evidence is stronger than the doubts suggest.

M-Shule, running personalised lessons over plain SMS, produced measurable learning gains for under 10 dollars a student a year, with especially strong results for girls. The World Bank's AI-tutoring trial in Nigeria moved learners about a third of a standard deviation in English, the equivalent of one and a half to two years of ordinary schooling, at around 15 dollars a head. Eneza reached 5 million learners. These are not slideware demos. They are working programmes with results a ministry can audit.

The technology has grown up too. A progressive web app installs straight from a browser with no app store to navigate, loads in seconds on a poor network, and caches its lessons on the device so learning continues when the signal drops. AI tutoring can run in layers, a lightweight model on the phone for offline questions and a cloud model for harder ones when a connection appears, so the tutor is never entirely gone. Short video, a few minutes at a stretch, teaches better than hour-long lectures and costs a fraction of the bandwidth.

Why this cannot be left to the market

Everything so far might read like a business plan. That is exactly the trap to avoid.

Start with the money that would fund it. African EdTech venture capital boomed to 291 million dollars in 2021, then collapsed to 27 million by 2023 in the global funding winter.

You cannot build a 25-year universal-access system on capital that evaporates with every interest-rate cycle. Worse, venture money has to chase returns, and that pulls any product toward the people who can pay: urban, connected, exam-sitting families. The 98 million out-of-school children, poor and rural and often without a device, are precisely the segment a return-seeking market will skip. Left to itself, the best app gets built for the child who already had options.

The market also fragments. Look at how today's platforms actually compare: each is strong on one axis and weak on the rest, and none is complete.

A dozen walled gardens, each holding a slice of learners on incompatible systems, is the opposite of a system. Both problems share a root: at 40 dollars a child, mobile education is infrastructure, not a quick-exit venture. Finance and govern it like a road or a power grid: owned publicly, used by many.

Whose job is each part

So who builds it? The clearest way to answer is to break the work into a value chain and ask which layer belongs to whom. Three tiers do most of the work.

The rails: Africa's own institutions

Some parts only public bodies can supply, and they are best supplied once, at continental scale, rather than rebuilt country by country. Patient, counter-cyclical financing, drawn from regional development banks rather than thin secretariat budgets. Standards and interoperability, so platforms and devices can talk to one another. An open content commons, built once per language region and free for every country to adapt, instead of fifty ministries paying for the same mathematics course. And certification, the hardest problem of all.

Here the groundwork is already being laid. The African Union adopted a new Continental Education Strategy, CESA 26-35, in February 2025, and launched a Decade of Accelerated Action for education and skills running to 2034. More to the point, the African Continental Qualifications Framework, now in implementation and designed to work across all 55 states and the eight regional economic communities, exists precisely to make credentials portable across borders. It is anchored in the African free-trade area's protocols on services and the free movement of people. The African Union Commission's own estimate is that weak credential recognition costs the continent somewhere between a tenth and a sixth of the economic value that skilled-labour mobility could create. A mobile-learning system does not need to invent certification. It needs to plug into this framework and into national examination bodies such as WAEC, Umalusi and KNEC. And the exam machinery itself is going digital: Nigeria has begun migrating its WAEC examinations to computer-based testing, a phased rollout that started with objective papers in late 2025. That is a state function, and increasingly a continental one.

Pooling changes the bargaining, too. A single small country asking operators to zero-rate learning traffic, or haggling over the price of a 30-dollar handset, has little leverage. A bloc buying on behalf of 400 million learners has a great deal. Setting standards, pooling procurement and negotiating as one is exactly what the African Union and the regional communities — ECOWAS, SADC, the EAC and the rest — are built to do.

The adopters: national governments

The rails are continental; the trains are national. Each country still has to fit the content to its own curriculum and languages, connect it to the examinations its children actually sit, and run the last mile: the community hubs, the human mentors, the tie-in with existing schools. Mobile learning works best beside people, not instead of them, and the people are a national responsibility.

The builders: an invited private sector

None of this shuts the private sector out. It invites the private sector into the parts of the value chain where competition drives things forward, on open standards the public side has set. Devices, edge connectivity, the delivery platforms themselves and the AI tutoring underneath them are the layers where private firms will iterate faster than any ministry. CESA 26-35 says as much, recognising that public, private and faith-based providers all have a part to play. What the market should not do is set the mission, decide who gets served, or own the rails everyone else depends on.

The part that could still go wrong

Two honest risks sit over all of this.

The first is capacity. The African Union and the regional bodies have a long record of slow delivery, overlapping mandates and dependence on donor money. If they cannot execute, "hand it to the continent" fails the same way "hand it to weak states" does. The answer is the division of labour above. Regional bodies do the things done by treaty and framework, the standards, the financing and the certification, which is what they are built for, while delivery capacity sits with the competitive layer rather than with a secretariat in Addis Ababa. And the money has to come from institutions that can actually carry it, the African Development Bank and its regional peers, blended with member financing, not from the AU's own thin budget.

The second risk is capture. A continental programme is a very large budget, and large budgets in this field have a way of being captured: single-vendor mega-tenders, funds earmarked for transformation that quietly leak, enforcement that never quite arrives. The defences have to be designed in from the start. Funding that follows the learner rather than the vendor. Content kept open, so nobody can rent-seek on it. No single-supplier lock-in. Independent audit with real teeth. Get that wrong and a good idea becomes the next procurement scandal.

The decision, not the technology

The pieces have rarely lined up this well. The infrastructure is in place, the teaching has evidence behind it, and the continental scaffolding is being built as we speak: a new education strategy, a qualifications framework, a free-trade area that needs portable skills to work at all. What is missing is the decision to treat mobile education as public infrastructure and to lead it as Africans, through African institutions, rather than wait for a market to serve children it was never going to reach.

Educating a continent has always sounded like the most expensive ambition in the world. It turns out it might cost about the price of a phone. The money is manageable and the technology is ready. What the continent still has to find is the will to treat this as a public good, and to build it together.

---

The back story

Africa has three great endowments, to my mind: its land, its minerals and its people. The first two have been fought over for centuries. The third is the one I keep returning to, because it is the endowment that compounds, and because it is the one we are closest to squandering. That belief sits underneath this piece.

My interest in education is homegrown, in the most literal sense. My mother started a primary school in Nigeria in 1985. Forty years on it is still standing, with a secondary school that came later beside it. I have watched at close range exactly how hard it is to build a school and keep it alive: the approvals, the land, the buildings and everything that goes wrong with them, finding and retaining good teachers, the salaries, the fees families struggle to find, the constant arithmetic of doing more with less. Nobody who has seen that from the inside thinks school-building fails for want of will.

My own schooling took the long way round. Two primary schools in Nigeria, then four high schools across Nigeria, Zimbabwe and South Africa. Those four schools also put me through three different examination systems: WAEC in Nigeria, Cambridge GCSE in Zimbabwe, and SAFCERT in South Africa, the body that later became Umalusi. Two things stayed with me. The first was the sheer unevenness of access, quality and investment, sometimes between countries, sometimes between one school and the school down the road. The second was how little the systems spoke to one another. Crossing a border meant reopening the question of what my education was worth. Years later, reading about the African Continental Qualifications Framework, I recognised the problem straight away. I had lived it at fifteen.

I have also been fortunate to work on projects going into primary and high schools in South Africa. Every one of them reinforced the same conclusion. Solving access to quality education is hard, and the way we are going about it does not scale.

The device idea first occurred to me around 2011. Poor matric results were a running national argument in South Africa at the time, and textbooks were in the news for all the wrong reasons. Then I came across Khan Academy, which had built a platform that could teach a student mathematics from basic arithmetic through to calculus. My first technology idea came out of that: a tablet carrying etextbooks, past papers and learning guides. I called it Edutab, which tells you everything you need to know about my naming instincts. I was thrilled by it. I was convinced something like it would stop textbook failures from ever happening again, and I wrote my first blog post saying so. I laugh at myself now, mostly at the confidence that a system would embrace an idea simply because the idea was reasonable.

A lot has happened since. Innovators have delivered educational tablets and laptops. Phones have become cheap enough to carry the load on their own. Reaching learners through a device is no longer a novel thought, it is widely accepted. What has not happened is the step that matters: treating it as the scalable answer to a problem we cannot build our way out of.

So the hypothesis I brought to this piece is the one I have carried since 2011, that a child can be reached through a device, and that it is cheaper and faster than the alternative. The evidence bore that out, and then corrected me on the harder half. The technology was never the blocker. The 2011 version of me thought the difficult part was the tablet. The difficult part is the architecture: who finances it, who certifies it, who owns the rails, and whether the thing gets built for the children who can pay or the ones who cannot.

One disclosure, for anyone who has read this far. I am involved in building a piece of this. It is called myNext, an AI agent for African youth that includes AI tutoring and subject learning material. It is a source of real excitement for me and the culmination of a long journey to do something for Africa's youth.

A note on scope: this piece argues at the level of systems and cost structures across the continent, not about any single country, school, teacher or family.

Research: the underlying research was done with Kimi Deep Research, NotebookLM and Google Gemini.  Claude (Anthropic) with live web search was used to verify load-bearing figures and current facts.

Writing: Claude (Anthropic) for structuring, drafting and revision.

Charts: built in matplotlib from the underlying research data and rebuilt to improve layout using Claude (Anthropic).

Images: the hero image was generated with Nano Banana Pro (Gemini 3 Pro Image) via kie.ai, selected after a four-model comparison against GPT Image 2, Flux-2 and Ideogram V3, run through a self-hosted n8n workflow. The accompanying LinkedIn carousel was produced the same way. Garbled text corrections were done using GPT Image 2 through a separate n8n workflow

Video: none used.