Most of humanity now lives where births fall short of replacement

The most consequential dataset in the world is filed by clerks. Birth registers, aggregated by the UN's population division, now say this: 131 countries and territories, 55% of all of them, have fertility below the roughly 2.1 births per woman that holds a low-mortality population steady, a benchmark that sits higher where child mortality is, and those countries hold 68% of the world's people¹. The planet's average has fallen to 2.25, one full child fewer than a generation ago, and the UN's central projection has it crossing below replacement in the late 2040s¹.

Nothing else covered here, not chips, not cobalt, not the price of a battery pack, operates on the scale of this variable. Fertility compounds silently for decades and then arrives all at once as school closures, labour shortages, pension arithmetic, and the relative weight of nations. It is also the variable most reliably mis-narrated, oscillating between overpopulation panic and birth-dearth panic without pausing at the data. This piece pauses at the data.

What the numbers actually say

Three findings organise the 2024 revision¹. First, the peak is now visible: world population, 8.2 billion in 2024, is projected to crest around 10.3 billion in the mid-2080s and then decline, a forecast revised earlier and lower with each revision. Second, the peak has already happened for a fifth of humanity: dozens of countries, China foremost, are past their maximum population and shrinking. Third, the decline is general, not cultural: fertility has fallen on every continent, in every major religion's heartlands, under pro-natalist governments and indifferent ones alike, which is the strongest hint about causes that the aggregate data offers.

The spread around the average is where the century's geography gets written. East Asia holds the floor: South Korea's rate touched 0.72 in 2023 before rising to 0.75 in 2024, its first increase in nine years, celebrated nationally as a triumph at roughly a third of replacement². China, Japan, and much of southern and eastern Europe cluster between 1.0 and 1.4. Sub-Saharan Africa holds the ceiling, averaging above 4 and supplying an ever-larger share of the world's births; the UN projects most population growth from here to the peak happens in a handful of African and Asian countries¹. North Africa and the Middle East sit in between and falling, many countries near or below replacement already.

The timing collision with technology is worth a paragraph, because the two big slow variables of the century are converging from opposite directions. The rich world's workforces begin their steepest declines in exactly the decades when AI systems promise to substitute for scarce labour. Whether that is providential timing or a race lost by one generation of pensioners is the sort of question that will look obvious only afterwards. What is already observable is the direction of investment: the countries automating fastest, Korea and Japan lead the world in robot density, are precisely the ones whose registers emptied first. Demography is driving the machine adoption curve at least as hard as wages are.

Why it falls everywhere

No single cause survives contact with the data, and the strongest correlates are the boring ones: education, urbanisation, and opportunity cost. Fertility falls where girls stay in school, where families move to cities that price space per square metre, and where a woman's earning years compete with childbearing years, which is to say it falls wherever development succeeds. Contraception is the mechanism, not the motive. The transition that took France a century and Korea a generation now completes in decades everywhere it starts, and it starts everywhere.

What the correlates cannot yet explain is the second drop: the fall from moderate sub-replacement, the 1.8 of a Sweden or a France in the 2000s, toward the 1.2 floor that spread through the 2010s and 2020s across rich countries with very different policies. Housing costs, later partnership formation, and what demographers dryly call the incompatibility of intensive parenting norms with everything else are the current suspects. The honest position is that rich-world fertility fell below every model's floor, and the models are being rebuilt around the fact.

What money has and has not bought

Governments have tried to buy births for decades, and the results are now legible. Cash bonuses, tax credits, and parental leave shift the timing of births more than the total; the most generous systems, in the Nordics and France, long held fertility higher than their neighbours' but have converged downward with everyone else this decade. Korea has spent staggering sums across two decades for the world's lowest rate, and its 2024 uptick tracked a marriage rebound delayed by the pandemic more than any new subsidy². The defensible reading is narrow: policy can support the children people already want, and nothing yet devised makes people want children at replacement scale against the grain of modern urban life.

Migration is the rich world's actual working policy, whatever its politics. Every developed country with a stable workforce imports it, and the competition for young migrants is quietly becoming a structural feature of international relations, with points systems, diaspora courtship, and talent visas as its instruments. The corollary lands on the sending countries: the remittance economies of North Africa and South Asia are exporting the one commodity whose global supply is now shrinking, working-age people, and the terms of that trade will move in the exporters' favour for the rest of the century.

Closer to home, the register reads double-edged. The Maghreb and the wider Arab world completed one of history's fastest fertility declines, from among the world's highest rates two generations ago to near replacement in much of the region today, which means the youth bulge that dominated every analysis of the region's politics is a closing window rather than a permanent condition. The next two decades are the demographic sweet spot, many workers, fewer dependents, exactly the configuration East Asia converted into miracles and other regions squandered. Morocco's industrial build-out is, among other things, a race against that window's closing.

The arithmetic nobody has repealed

A population's age structure is destiny with a lag. The workers of 2050 are already born; no policy changes their number, only their location and their productivity. That is why the fertility register is the quietest leading indicator in economics: today's birth counts are the 2040s' labour forces, the 2050s' pension ratios, and the 2060s' great-power weights, published decades in advance and priced by almost nobody. Japan is the pioneer, managing three decades of aging with automation, participation, and calm; China is attempting the same transition at a fraction of Japan's income and several times its speed; Europe splits the difference with migration it struggles to politically metabolise.

Both kinds of panic can be retired. The overpopulation alarms of the last century missed the transition already under way; the extinction alarms of this one extrapolate exponential decline through a century of unknowns, ignoring that a peak at 10.3 billion still adds two billion people, overwhelmingly in countries least equipped for them, before any global decline begins¹. Both panics skip the actual policy problem, which is neither too many births nor too few, but the speed of the switch between regimes, faster than any welfare state has yet rebalanced.

India is the pivot case the forecasts lean on hardest. The world's most populous country slipped below replacement at the national level in the early 2020s, a generation earlier than most projections once assumed, while still holding the century's largest cohort of young workers. Its demographic dividend is real and already ticking: the question its planners state openly is whether the country grows rich, or at least industrial, before it grows old, the race China is now visibly losing from ahead. Every multinational supply-chain map and every long-term growth forecast quietly assumes India wins it. That assumption needs regular checking.

What to watch

Four registers, annually. Korea's rate, because the world's hardest case is the test of whether any floor exists². India's, already below replacement at the national level, because the country the century expects to supply workers and growth is one generation behind the East Asian curve. Sub-Saharan Africa's pace of decline, because whether the continent's transition takes thirty years or sixty is the largest open question in the world economy. And the UN's own revisions, because each one has moved the peak closer and lower¹, and the direction of revision is itself the finding: the century keeps surprising the demographers on the same side. The clerks will file the answer, one register at a time. Almost everything else moves faster and matters less. The births of 2026 are the entire labour supply of 2050, already counted, already located, and already setting the boundaries inside which every other variable will do its work, from the demand for phosphate to the supply of soldiers.