The Century of Africa

2025: one in five

About 1.55 billion people live in Africa today, one of every five humans alive. This is not a story about a future continent. Nigeria already has more people than Russia, and Ethiopia has more than Japan. Neither fact has reached most mental maps of the world.

Countries shaded by population in 2025. The rest of the world sits on the same scale, dimmed so Africa reads first.

2100: closer to two in five

By the end of the century Africa reaches 3.8 billion, better than a third of everyone alive. Europe finishes smaller than it started. Asia peaks around 2055 and turns down. China ends the century with fewer than half the people it has today. Africa is the only region still climbing when the clock runs out, and getting there means housing, powering and watering an extra 2.3 billion people.

Regional totals, 1950 to 2100. The shaded band around Africa is the UN's own low to high range, which reaches from 2.7 to 5.3 billion by 2100.

The multiplication map

Same countries, different question. Not how many people, but how many times over. Angola and DR Congo land near four times their current size, with Tanzania just behind. That multiple is the one a water utility, a power grid or a road network has to plan against. The blue countries have the opposite problem: Italy, Spain and Japan all finish the century well below where they stand now.

Why this is not a guess

Projections like these are not predictions about how people will behave. They are mostly arithmetic on an age pyramid. The map now shows the median age of every country. In Niger it is 15.6 years, in DR Congo 15.8, in Somalia 15.6. In Japan it is 49.8 and in Italy 48.2.

Women in Niger have 5.8 children on average today, and in DR Congo 5.9, against 1.5 in Germany. Those numbers are falling almost everywhere and the UN expects them to keep falling. The growth arrives regardless, because the women who will have children in 2050 have already been born. Demographers call it momentum, and it is why birth rates are not the lever here. The lever is what gets built before the people arrive.

The overtaking years

Rankings are how most of us carry a sense of national weight, so here is the moment they flip. Each line is the year an African country passes a Western one and never falls back behind it. Every seat, quota and funding formula weighted by population is reading a table that expires on these dates.

Read off the UN's annual series. A pass only counts if the lead holds all the way to 2100.

1975: six cities

Six African urban areas held a million people. Cairo was the only large one, at 7.3 million. Lagos had 2.5 million, Kinshasa barely one, Dar es Salaam fewer than 300,000. Tokyo alone held 24 million that year, more than double Cairo, Lagos and Kinshasa put together.

2025: the arrival

Eighty-eight African urban areas now clear a million people. Cairo is at 25 million, Lagos 13, Kinshasa 11, every one of them larger than Paris or London.

Where the circles start to touch

Follow the coast east from Abidjan and the gaps close. Seven cities of a million or more sit inside 811 km, from Abidjan through Kumasi, Accra, Lomé and Cotonou to Lagos. They hold 35 million people today and reach 47 million by 2050.

They also sit in five countries, and four of the seven are national capitals. The strip is becoming one continuous urban region and nothing about the way it is run matches that: each city plans its water, its power and its roads alone.

2050: past the West

Cairo reaches 32 million by mid century and Luanda 20. New York sits about where it has sat for fifty years, near 13 million, and Tokyo is falling. Africa will hold 129 cities above a million people. What has to be built is not one megacity. It is a hundred and twenty-nine of them, and most are names that have never run in a Western headline.

2100: cities with no precedent

The UN's city projections stop at 2050. Two researchers, Daniel Hoornweg and Kevin Pope, carried the curves to the end of the century and landed on numbers with no parallel in history: Lagos near 88 million, Kinshasa 84, Dar es Salaam 74. Lagos on its own would be larger than Germany. Kinshasa would just about match it. No city has ever been run at that size, so nobody knows yet what running one takes.

The hollow rings are those 2100 figures. They are drawn hollow because they are one team's projection rather than a count, and they sit at the high end of what other researchers estimate.

What the map inherited

African railways were mostly laid to carry copper, cocoa and oil from the interior to the nearest port, not to carry people between cities. The lines run seaward, and they rarely meet each other. Kinshasa and Lubumbashi are the two largest cities in DR Congo, about 1,570 km apart, with no railway between them. The gap this story is about does not begin in 2050. It was designed in.

Who is building, and why

The headline railways of the past decade were financed and built by China: Mombasa to Nairobi, Addis Ababa to the port of Djibouti, Lagos to Ibadan. Beijing has read the same population tables as everyone else and moved first. In 2023 the United States and the European Union answered with the Lobito Corridor, an old line rebuilt from the Angolan coast into the copper belt. So building is happening. Whether it is happening at the rate the arrivals demand is the open question, and whoever lays the track tends to set the trade routes for a century.

Where the traffic wants to go

Transport planners estimate future flow with a gravity model: two cities pull on each other in proportion to their populations, and against the square of the distance between them. Run it across the thirty largest African cities of 2050 and the strongest links draw themselves. The West African coast. The Nile. A ladder up through the Great Lakes. Kinshasa reaching for Luanda.

A formula drawing lines is easy to dismiss, so here is the test. Of the 36 links it demands, 24 already have a mapped primary road along at least four fifths of the route, and the median link is 90 per cent covered. The strongest link it finds, by a factor of six, is Cairo to Alexandria, which Egypt built long ago. A model whose top answer is the corridor that exists is not drawing noise.

Coverage counts a mapped motorway, trunk or primary road from OpenStreetMap within about 24 km of the route. It measures whether a road exists, not whether it is paved, open or big enough. Terrain, borders and money are not in the model.

The long way round

Take Lagos and Luanda. Straight across the Gulf of Guinea they are 2,046 km apart, and that is the distance the gravity score is calculated on. Nobody drives it. The road follows the coast through Cameroon, Gabon, Congo and DR Congo: 2,421 km, a fifth further, across five borders.

Drawn straight, this link was 98.5 per cent sea, and twelve of these thirty-six were lines through open water. That is not just an ugly map: the coverage test samples points along each route, so it was counting roads in the Atlantic. Lagos to Luanda scored 8 per cent that way. Along the road that exists it is 88 per cent served.

Where the road runs out

With the routes on land, the badly served links are no longer the ones that crossed water. They are interior and cross-border: Dar es Salaam to Kampala at 42 per cent, Luanda to Kinshasa at 48, Cairo to Khartoum at 48, Addis Ababa to Khartoum and Kinshasa to Yaoundé at 52.

Luanda and Kinshasa are the sharpest of them. They are 549 km apart, close enough that the model ranks the pair fifth by demand out of all thirty-six, and less than half the way between them has a mapped primary road. That is the gap between where the traffic wants to go and where it can.

The lights are already on

Satellites have photographed the earth at night for decades. This is 2020, and the lit ground follows the same lines the model drew: the coastal strip, the Nile, the eastern highlands. These corridors are not a forecast waiting to begin. They are lit tonight.

Harmonized DMSP and VIIRS nighttime lights, 2020 (Li et al.), with the modeled corridors and the built railways still faintly drawn underneath.

What the lights leave out

A map of light is also a map of who has power, so here is that map directly: the share of people in each country with electricity at home. Across Africa 623 million people live without it, and 454 million without a basic supply of drinking water.

DR Congo is the sharpest version. It is building toward a capital of 84 million people, and today 78% of Congolese have no electricity and two in three have no safe water.

World Bank open data, most recent year available for each country, generally 2022 or 2023.

Building fast, and still losing

Six hundred million people without power reads like failure. It is not. The share of Africans with power at home went from 37.7% to 58.5% between 2000 and 2022, while the number without it rose from 510 million to 600 million. Water runs the same way: 52.8% to 70.1% access, and the number without still rose.

The building is real and the arithmetic is outrunning it, but a race is not a verdict. The map shades by what each country gained since 2000: Kenya went from 15% to 76% at 54 million people, Rwanda from 6% to 51%. The arrival of the people is arithmetic. The arrival of the power is a decision.

Percentage points gained, 2000 to 2022, counting only countries reporting in each year against UN population for that year. The 623 million on the previous map is this same shortfall measured against today's larger population, which is why it runs ahead of the 2022 figure here.

Two capitals, one river

The Congo River widens here into a lake called the Malebo Pool. Kinshasa sits on the south bank, capital of the Democratic Republic of the Congo. Brazzaville sits on the north, capital of the Republic of the Congo. No two capital cities on earth are closer together. There is still no bridge.

1975: the city on the plan

A million people, and a plan. The sand-colored ground is everything built by then, and the colonial town is still legible inside it: a dense strip along the water, a grid running inland, then farmland. This is the last decade in which the city and its street map are the same size.

Built ground mapped from the satellite archive by the European Commission's Global Human Settlement Layer, in squares about 92 meters across.

1990 to 2000: growth without a boom

Mobutu falls, two wars follow, the economy comes apart. The city triples anyway, to 3 million by 1990 and 5.2 million by 2000. This is where the two rates come apart: people keep arriving on the demographic schedule, and building follows the budget of a collapsing state. Almost none of the new ground is planned. Neighborhoods assemble themselves house by house and push south into the hills.

2010 to 2020: the megacity

Ten million people. New building runs east along the highway toward N'Djili airport and climbs slopes nobody surveyed for streets. Then look at the thin gray lines. The road network carrying those ten million is largely the one drawn for a city of one million. Ten times the people on much the same streets, which is the whole argument in a single frame.

2030 and past it

The satellite record carries built ground to 2030, when the UN expects 11.6 million residents. The century-end projection is 84 million, one city holding about as many people as Germany does today, standing on the ground you have just watched fill in. Nothing about that number is under anyone's control. What gets built on it still is.

It grew up, not out

Built ground across Kinshasa and Brazzaville grew by about half since 1975. The number of people living on that same 1975 ground went from 2.3 million to 14 million. The city did not sprawl so much as thicken, because thickening is what a city does when nothing is built to let it spread. The brightest patches below carry more than 60,000 people per square kilometer, over five times the density of New York City, and almost none of it was drawn by a planner.

Population per cell from the European Commission's GHS-POP dataset, 1975 and 2025, on the same grid as the built-ground maps. The 2025 layer is modelled rather than counted: summed inside Kinshasa's real commune boundaries it runs about a fifth above the UN's figure for the city, in 1975 and today alike, so read these totals as carrying that much slack.

Where the people actually went

Measure it in rings out from the old river front and the pattern is hard to miss. Built ground grew by half. The city as a whole added fourteen million people, and 83% of them are standing on ground that was already built in 1975. Half the city still lives within ten kilometres of the colonial core. Those streets, drains and pipes were sized for the city of 1975 and are carrying most of the city of today.

Each row is a five kilometre ring. Dark is 1975, amber is today. The city did spread, but the spreading is the small half of the story.

The flat ground ran out

There is a reason the packing happened. The old city sits on a shelf of flat land by the river, at a median slope of 2 degrees, and only about one part in forty of it is steep. That land is finished. Every decade since has had to build on worse ground: the growth added by 1990 came in at twice the slope of the old city, and roughly one part in eleven of it was properly steep.

On bare tropical hillsides with no drainage, heavy rain does not run off, it digs. Kinshasa loses houses and streets to erosion gullies most wet seasons, and the map of where that happens is largely the map of where the city had nowhere flat left to go. Drainage is buildable. It was not built.

Slope from the Copernicus 30 m elevation model, averaged to the built-up grid. The 2030 layer is the JRC's projection and covers a small area, so its slope figures move around more than the rest.

Streets built for a tenth of the city

Chapter by chapter this story has shown the ground filling in. Here is what did not. Counting every street inside the built-up area, the network grew about 37% between the 1975 footprint and the 2030 one. Over the same stretch the city went from roughly one million people to more than eleven.

Per person, that is a fall from about nine metres of street to one. Nothing collapsed and nothing was demolished. The city simply arrived, and the streets did not. The distance between those two clauses is the distance between a projection and a plan.

Streets from OpenStreetMap, measured inside each epoch's built footprint. OSM maps the network as it is now, so the early years are generous: some of those streets were not there yet, which makes the real decline steeper than the line.

Is Kinshasa just badly run?

It is the fair objection, and it is testable. The same measurement run over nine African cities: every street OpenStreetMap knows, inside the same satellite footprint, against the same UN population.

Kinshasa comes last, so it is not a typical case. It is the sharp end of a pattern that holds everywhere it was measured. Every city with a 1975 figure has less street per resident today than it had then, and the median across the nine is 1.82 metres. Cairo fell least, from 2.50 to 1.82, which is what fifty years of money and a state that builds looks like.

Streets from OpenStreetMap inside JRC built-up footprints, against UN city population. OSM has no history, so every 1975 figure is an upper bound and every fall is at least as steep as shown. Luanda has no UN 1975 series, so it appears with today's figure only.

One road to the sea

Zoom out and the last problem shows up. Kinshasa sits hundreds of kilometres inland on a river that ocean ships cannot climb, because the Livingstone Falls sit between the city and the estuary. Almost everything imported for it lands at Matadi and finishes the journey on the two lines drawn here: one national road, and one railway.

That is the whole connection between a city heading for tens of millions of people and the rest of the world. Chapter 3 asked which corridors a growing continent will need. Here is the answer from the ground: one road, one railway, and a projection of 84 million people at the end of them.

The part that is still open

The people in this story arrive whether or not anyone is ready. Four chapters of maps have shown how little of that is in doubt. Everything else here is a decision nobody has taken yet. The map holds more than four chapters can show, so take it apart: shade the countries by any measure, run the cities back and forth through time, and click anything for its numbers.

Explore the map →