31 reports in this theme, newest first.
By 2087, most of humanity will live in sub-Saharan Africa or South Asia. By the 2090s, the median person would live at a mean annual temperature of 26–28 °C, up from 22 °C, and the worst-off quarter beyond 28.5–29.4 °C, with one to more than two months of dangerous heat a year, in air already six times dirtier than the WHO guideline. A low-emissions, high-development path keeps unprecedented heat to 0.6 billion people instead of 3.8 billion.
By the 2090s, 4.3 to 9.3 billion people will live with frequent heat, extreme rain or long dry spells, up from 3.4 billion. The exposed population ages fast: heat exposure among people 65 and over rises nearly tenfold, to 1.3 billion. If schooling keeps rising and protecting as it has, harm falls by 70–84% in three pathways, but under SSP3 protection only holds harm level, and heat harm rises 39%.
Droughts have barely become more frequent, but they have become hot. Half of warm-season drought months in 1995–2024 were also extremely hot, up from three in ten in 1951–80 and seven in ten in 2021–24. Dry-to-wet whiplash is up 30%, and hot-dry years produce the great smoke and dust events. By the 2090s, 0.34 billion people would live with two or more compound hazards on a low-emissions path, 5.2 billion on a high one.
1.66 billion people live where two or more climate hazards overlap, but only 158 million of them also have limited means to adapt. They are concentrated in the heat-and-drought belt from Senegal to Pakistan and on cyclone coasts from Madagascar to Haiti. By 2050 that overlap nearly triples to 455 million, led by Nigeria. Halving poverty and strengthening coping would cut it to 263 million, more than lower emissions alone (356 million), though not in the most fragile states.
The world adds 1.43 billion people by 2050, 69% of them in cities and 57% in sub-Saharan Africa. People living with 30 or more days a year above a 35 °C heat index more than double, from 1.9 to 4.1 billion. Warming drives 1.74 billion of that rise and population growth 0.44 billion, and growth lands disproportionately in hot places. City dwellers in 100-year flood zones, 507 million today, are heading for 630–690 million.
Each 0.1 rise in the Human Development Index goes with about 63% fewer people hit by climate disasters, 62% fewer near-poor and 25% less death from communicable disease. Within countries over time, the share affected by climate disasters falls 40% per 0.1. Yet human development gains in 2019–2023 were half those of the 2000s, the gap between low- and very-high-HDI countries has stopped closing, and development has not damped growth volatility or pandemic mortality. Those tail risks need their own institutions.
After a severe climate shock, GDP per person in low- and lower-middle-income countries falls by 1.7%, sinks to about 5% below its previous path within four years and is still 3.7% lower after eight. Upper-middle-income countries recover within four years, and earthquakes leave no lasting scar. Low-income countries now suffer climate shocks in a quarter of all years, only one in five of these has any recorded damage, and the income gap between the median rich and poor country has grown from 18 to 25 times since 1990.
Half the world's school-age children — 962 million in 35 countries — live where most 10-year-olds cannot read a simple text and climate hazard is high. Children's exposure to dangerous heat doubles by mid-century, with sub-Saharan Africa's share tripling to a quarter, yet only one in five primary schools in low-income countries has electricity to run a fan. Because schooling is among the strongest predictors of how well people cope with disasters, every lost school day today weakens tomorrow's resilience.
Since 1975 the world's urban centres have added 2.1 billion people, and growth has leaned into the floodplain: the urban population in 100-year river flood zones tripled to 507 million while cities as a whole grew 2.3 times, with one new resident in six settling in a flood zone. Six in ten new city dwellers settled in cities exposed to at least one major hazard, and 746 million now live where two or more compound. Most of the growth still to come is unbuilt — the cheapest moment for safer settlement decisions.
The water polycrisis has two faces. In 35 countries with 4.1 billion people, from Egypt to India and China, water is physically scarce: withdrawals exceed 40% of supply, agriculture takes three-quarters and aquifers are falling. In 47 countries with 1.6 billion people, mostly in Africa, water is plentiful but nearly a third of people lack basic water, unsafe water kills at six times the rate of well-served countries and grids hang on hydropower. They meet in Sudan, Afghanistan and Yemen — and each needs a different fix.
Humanity's centre of gravity is drifting 1,700 km towards the hot belt from West Africa to South Asia just as that belt crosses danger-level heat. By the 2090s, 2.2 billion people live with a month or more of it in a middle-of-the-road world and 5.0 billion in a fragmented, high-emission one — nearly half of them people not yet born in places not yet dangerous. Low emissions and education-led development each cut that exposure by half or more, and because half of it sits on 0.5% of inhabited land, protection can be targeted.
Thirty-six countries with 3.0 billion people combine high climate hazard, high pollution exposure and weak coping capacity — across the Sahel, West and Central Africa and South Asia. The opportunity is concentrated too: the 60 countries breathing 25 µg/m³ or more emit 74% of the world's coal CO₂ and half its greenhouse gases, so in a handful of airsheds the same coal and methane cuts that slow warming would clean the air for billions.
Disasters kill far fewer people than they did — 6 per million a year in the 2010s against 99 in the 1960s — because capacity rose faster than threat. Within countries, each extra year of adult schooling has cut deaths per 1,000 people affected by about a third. Projected to 2100, that capacity keeps winning in most futures, but not in a fragmented, high-warming SSP3 world, where population and hazard outpace schooling and expected deaths rise in two-thirds of countries.
No site analysed reaches the 35 °C wet-bulb ceiling before about 4.4 °C of global warming; the binding line is lower. Healthy young adults lose the ability to cool themselves in humid heat near 30.5 °C, a level 11 of 161 sites reached for at least an hour in 2024. Under current policies, about 2.8 °C, the cities touching it would hold 161 million people, up from 9 million today — around the Gulf, across South Asia and in eastern China.
Heat cut work capacity enough in 2024 to put about $490 billion of output at risk, a conservative estimate from hourly heat stress in 149 cities. The Lancet Countdown's broader estimate is $1.09 trillion. Either figure is of the same order as all climate damages in the DICE model at 1.5 °C. The loss is regressive (3% of hours in lower-middle-income countries against 0.2% in rich ones) and convex: one extra degree adds 73%, two nearly triple it. Moving work out of the afternoon recovers 20–30%.
Where coping capacity is strong, recorded deaths from floods, storms, droughts, landslides and wildfires stay near 0.6 per million people a year whether hazard is low or high. Where it is weak they rise to 8.7 per million in the most exposed third — a 14-fold gap. About 14,000 recorded deaths a year are avoidable at that benchmark, and the true toll is larger because heat deaths are counted almost only in rich countries.
Very hot days are deadlier when the air is dirty: across 620 cities, a very hot day raised deaths by 12.8% when particle pollution was high against 5.3% when it was low. And the two hazards travel together — in 2024 ozone was higher on hot days in 131 of 149 large cities. About 2.6 billion people live in countries that are both hot and polluted, three-quarters of the population of lower-middle-income countries against 4% in high-income ones.
Humanity will live about 739 billion person-years between 2025 and 2100. On current trends, 180 billion of them will be spent under at least two of three hazards: polluted air, a hot climate and a high-risk country. Counting exposure cumulatively shows that clean air and resilience with peace each avoid more of those years this century than the gap between warming pathways, though mitigation decides how many people end up near the edge of the human climate niche.
Even if no hazard worsened, the share of humanity living in today's high-risk countries would rise from a quarter to 43% by 2100, because population is growing fastest where risk is highest. About half of the 9.5 billion babies due between now and 2100 will be born in those countries. Combining UN population projections with the INFORM risk map shows who is exposed, where and in which generation, and how far reducing risk could offset demography.
A quarter of humanity, 2 billion people, lives in countries at high or very high risk of humanitarian crisis, and all 16 of the highest-risk countries score 8.4 or more out of 10 for conflict hazard. Mapping 191 countries on the INFORM Risk Index shows that hazard is not destiny. Coping infrastructure improved in 154 countries over 2017–2026 while institutions stalled, and matching each country's income peers would cut the population at high risk by more than half.
Extreme wildfires have more than doubled in two decades, methane rose at record pace in 2020–21 as warming wetlands released more, and ozone already cuts global soybean yields by about an eighth. Mapping five pollution–climate feedback loops shows where each can be cut most cheaply, from leaking gas wells to cooking stoves.
Disasters, heat, pollution, violence and the climate damage locked in by each year's emissions already cost between 1.2 and 23.5 per cent of world output a year, depending on what is counted. Over 2025–2100 that adds up to hundreds of trillions to several quadrillion dollars, and prevention would cost a fraction of it.
Food, oil, markets, policy, conflict and disasters rarely go wrong in the same year: their year-to-year correlation averages 0.12. In the 2020s they did, with an average of 3.5 of six acute indicators in their worst quartile each year against 0.7 in the 2010s, on a chronic floor of record temperatures, conflict and displacement.
Seventy-three countries are both highly vulnerable to shocks and poorly prepared to adapt, including 23 of the 24 low-income economies. Mapping the pressures converging on the world system against the adaptive capacity of 190 countries, and the financial buffers between them, shows where shocks become crises — with a country explorer built on the ND-GAIN index.
The case for investing in adaptation and mitigation together is physical and well established. The obstacle is not the argument but the cash flow: an avoided loss has no revenue line, so private capital funds about nine per cent of adaptation. This report sets out five instruments that convert avoided losses into something financeable, and how to use them without splitting the asset.
Green industrial strategy tends to offer every sector the same package. But an aluminium smelter is stopped by the price of electricity, an electronics plant by a four-hour outage, and a garment factory by a heatwave or a flood. Sorting industries by their binding constraint produces a different — and cheaper — sequence of investments.
Climate policy treats adaptation and mitigation as parallel agendas competing for finance. In economies building both capital stocks at once, they are better understood as complements in production — with a minority of cases where they genuinely compete. This report sets out where the dependencies run, where the conflicts are real, and how to appraise an asset that discharges both duties.
Cheap adaptation technologies with strong measured returns are distributed and then abandoned. The usual explanations blame the money or the country, but neither accounts for two technologies diffusing at wildly different rates through the same channel. Five attributes of the product explain most of the variation.
Adaptation is counted as a stock of assets, but what protects people is a service that keeps working, and a service runs on a chain. Reliability multiplies rather than averages, so where an investment sits in the network matters more than what it costs — and polycrisis is the case where the links fail together.
Poor countries only began catching up with rich ones around 2000, and the resulting convergence runs at one to two per cent of the income gap a year. Climate hazard already destroys one to two per cent of output a year in low- and middle-income countries. Adaptation, read against convergence, is not insurance but growth-retention policy.
The resilience gap is usually described as a shortage of money. It is better understood as two gaps wearing one name — hard adaptation blocked at a capital gate, soft adaptation blocked at a delivery gate — and neither closes while both are treated as the first.