The Polycrisis Apocalypse Humanity Is Facing

A polycrisis differs from a bad decade because its shocks are causally linked: drought cuts hydropower, power cuts curtail cold chains, food prices rise, budgets cannot subsidise, unrest follows, capital flees, and debt service rises just as revenue falls (Lawrence and colleagues, 2024). The decisive variable, as Joseph Tainter argued for past collapses, is the ratio of stress to repair capacity. That is why HH-2026-34 defines an SRR not by hazard alone but by hazard binding on low capacity: a wealthy city with heat is not an SRR if its grid, hospitals and treasury hold; a poor city with the same heat is.

The tiers are cumulative. Pollution damages lungs, hearts and developing brains everywhere it is high. Functional heat takes working months and school days long before it kills. Missing electricity and clean cooking remove the means of coping. Dangerous heat is the tail that becomes deadly for the old, the sick and those working outdoors. On the moderate definition, 4.6 billion people live with tier A, 1.9 billion with tier C and 1.2 billion with tier B today; 2.5 billion live in the 80 countries with low capacity.

What turns exposure into a Polycrisis Apocalypse is persistence. Past regressions, the Late Bronze Age collapse or the fall of the western Roman Empire, lasted centuries but took place in a stable climate. This one would not. The IPCC finds that many changes from past and future emissions are irreversible for centuries to millennia, and that warming stabilises only when net carbon dioxide emissions reach zero. The heat of the 2090s is a floor for the centuries that follow.

The red cells are measured Survivability Risk Regions: places where at least two tiers bind and national capacity is low. Today, on the moderate definition, they cover Nigeria and much of West Africa, the Sahel, Sudan, the Horn, DR Congo and Uganda; in Asia, Pakistan, Afghanistan, Bangladesh, Myanmar and Nepal; and parts of Central America. Orange cells have two or more tiers but adequate capacity: India, the Arabian peninsula and parts of mainland Southeast Asia today, joined by southern China by the 2090s. Their exposure is real, but their institutions currently hold.

Switch to the 2090s and the belt widens. Months of hazardous heat spread across West and Central Africa, the Sahel and South Asia, so that in countries already short of capacity, the second tier binds almost everywhere. Under SSP3-7.0, with more warming and faster population growth, the red belt is broader still, reaching deeper into East Africa and Southeast Asia. Because national conditions are held at today's levels, the map shows what warming alone does to today's capacity, not a forecast of institutions.

The definition matters as much as the climate. The strict definition captures the core of the belt: places with very heavy pollution or very low access and very weak institutions. The broad definition pulls in India, much of Latin America and Southeast Asia, where pollution exceeds 15 µg/m³ and governance is middling. HH-2026-34's scenario value of 35% today sits between the moderate and broad maps, which is the honest summary: the SRR is real, but its size is a matter of where one draws the line.

On the central, moderate definition, 1.24 billion people, 16% of humanity, live in a measured Survivability Risk Region today. The strict definition, which requires very heavy pollution or very low access and very weak institutions, gives 0.74 billion (10%); the broad one, which counts any pollution above 15 µg/m³ and middling governance, gives 3.54 billion (47%). HH-2026-34's scenario figure, 2.9 billion (35%), lies between the moderate and broad measurements.

Warming alone, applied to today's institutions and pollution, roughly doubles the measured population. By the 2090s under SSP2-4.5, the moderate SRR holds 2.79 billion people (30%); under SSP3-7.0, 4.45 billion (37%). On the broad definition the 2090s figures are 5.65 billion (61%) and 8.58 billion (71%), the scale of the scenario's weak path. A majority of humanity in SRRs by the 2050s, as the scenario's weak path has it, requires both the broad definition and continued warming; on the moderate measurement, it would also require institutions and pollution to get worse, not just stay the same.

The measurement is conservative in one way and generous in another. It ignores the sub-national pockets of poverty, conflict and failed services inside countries rated as adequate, and it holds today's institutions fixed, though they may erode. But it also ignores the improvement that most developing countries have achieved in electricity, clean cooking and poverty over the past two decades. The simulator in section 9 and the atlas in section 10 let readers set their own thresholds.

On the moderate definition, sub-Saharan Africa holds 51% of SRR residents today, the Middle East and North Africa region, which in the World Bank's current grouping includes Pakistan and Afghanistan, 26%, and South Asia 16%. Pakistan (210 million people in SRRs), Nigeria (202 million), Bangladesh (166 million) and DR Congo (84 million) lead, followed by Myanmar, Uganda, Sudan and Afghanistan. These countries combine heavy pollution or low access with weak capacity, and, increasingly, with heat.

By the 2090s under SSP2-4.5, Africans make up 60% of SRR residents on the moderate definition and 70% on the strict one. Nigeria alone would have 557 million people in SRRs, Pakistan 329 million, DR Congo 183 million, Bangladesh 176 million and Uganda 147 million; the Philippines (130 million), Afghanistan, Niger, Yemen and Sudan follow. Two forces drive the shift: Africa's population grows fastest, and warming adds the heat tier across the West and Central African belt where capacity is already thin.

None of this is about who people are. Fertility remains highest, and fiscal capacity lowest, in the same places that face the most heat and pollution; that is the whole of the arithmetic. Africa's general government revenue is a fraction of the OECD's as a share of GDP, and its countries pay more to borrow. Where the growth is, the buffers have so far lagged. Where buffers have been built, from Mauritius and Botswana to Morocco's solar build-out and Kenya's geothermal grid, the same geography produces very different outcomes.

Africa's share of humanity was about 9% in 1950 and is 19% today; on the UN's medium projection it is 37.5% in 2100, as population peaks elsewhere and continues to grow in Africa. Beyond 2100 there are no official projections. The H Heuristics Africa Ascendant scenario, in which Africa's long-run fertility settles just above replacement and the rest of the world's at replacement, puts the share at about 42% in 2200, 48% in 2300 and 52% in 2500. That scenario is acutely sensitive to small differences in long-run fertility.

Combining that demography with measured SRR rates tests HH-2026-34's proposition that the most common human circumstance becomes "an African living in an SRR." On the broad definition it holds: by 2500, Africans in SRRs are 46% of humanity under SSP2-4.5 (49% under SSP3-7.0), against 25% for SRR residents elsewhere, close to the scenario's 48%. On the moderate definition, Africans in SRRs reach 34–37% by 2500, and people living outside Africa and outside any SRR remain the larger group. On every definition, Africans make up most SRR residents after 2100: 71–86% on the strict and moderate definitions, 53–66% on the broad, against 62% by 2500 in HH-2026-34.

So the defensible statement is narrower and more useful than the proposition. The exposed population of the coming centuries will be increasingly African because Africa's population is growing in the hottest, most polluted and least-buffered regions. Whether its median member lives in a risk region depends on decisions that have not yet been made: how fast capacity, clean energy and clean air spread, and how much warming the world allows. Demography sets who is exposed. It does not decide whether exposure becomes catastrophe.

The world's nuclear arsenal has shrunk by more than four fifths since its 1986 peak of about 70,000 warheads, but the decline has stalled. SIPRI counted 12,241 warheads in January 2025, about 9,614 of them in military stockpiles, with every nuclear-armed state modernising and arms-control agreements lapsing. The Bulletin of the Atomic Scientists set its Doomsday Clock at 85 seconds to midnight in January 2026, the closest in its history.

The consequence that matters most for survivability is not blast but climate. Xia and colleagues modelled the soot from burning cities and its effect on crops, fisheries and livestock: a war between India and Pakistan could leave more than 2 billion people dead from famine, and a war between the United States and Russia more than 5 billion. The countries least able to absorb the food shock are net food importers with thin reserves, which is to say the SRR belt.

Below the nuclear threshold, the cost of violence is falling. Cheap drones, commercial satellite imagery and encrypted communications let armed groups contest territory and infrastructure where state forces are underpaid and under-equipped. World military spending reached a record $2,887 billion in 2025. In a buffered region, a war is a catastrophe from which societies recover; in an SRR, with thin food reserves, a fragile grid and contested authority, it can tip a region from stress into post-state dissolution.

The Uppsala Conflict Data Program recorded 65 active state-based armed conflicts in 2025, the most in any year since its series begins in 1946, up from 31 in 2010. The five deadliest years for battle deaths since 1989 are the five most recent: 2022, with about 285,000 deaths, driven by Ethiopia and Ukraine, then 2021, 2024, 2025 and 2023. The long post-Cold War decline in war has reversed.

Displacement has followed. People forcibly displaced under UNHCR's mandate rose from 57 million in 2015 to 113 million in 2024, before easing to 109 million in 2025, most of them displaced within their own countries. Sudan, the Sahel, DR Congo, Myanmar and Haiti are among the largest crises, and all lie in or near the SRR belt. Conflict destroys infrastructure, diverts budgets and pushes people into more exposed places, and above all removes the state's capacity to provide the services that keep the other tiers at bay.

Turmoil in wealthy countries works more slowly but in the same direction. Ageing, fiscal strain, polarisation and a retreat from multilateralism reduce the concessional finance, technology transfer and diplomatic effort that a polycrisis requires. HH-2026-34 notes that official development assistance fell by about 23% in real terms in 2025. A world whose external buffers are thinning at the moment its exposed population is growing is the setting for what the report calls a planetary Haiti.

Haiti is the clearest contemporary case of the polycrisis at national scale, and the easiest to misread. Its revolution of 1791–1804 abolished slavery and created the first Black republic. In 1825 France, with warships offshore, demanded an indemnity for the "loss" of the colony, repaid with loans over more than a century, so that a large share of public revenue went to foreign creditors rather than roads, schools or health. The United States occupied the country from 1915 to 1934; the Duvalier dictatorship (1957–1986) added three decades of predation. Deforestation, driven by poverty and charcoal, magnified every storm and flood.

The data show the cost of the loop. Haiti's real income per person peaked at about $2,150 in 1980 and has fallen to about $1,110, a loss of nearly half; the Dominican Republic, starting lower, grew to about $9,290. The 2010 earthquake killed an estimated 100,000 or more people and pulled life expectancy down to 46 that year; cholera followed. After the assassination of the president in 2021 and a second earthquake, armed groups expanded their control of the capital. The International Organization for Migration counted 1,412,199 internally displaced people in September 2025.

HH-2026-34 draws four lessons that this dashboard endorses. Collapse is capacity, not culture: each link in the loop can be traced to a policy or an external action. Compounding beats size: a series of individually survivable shocks did the damage. The state's monopoly on force is the final buffer, and recovery must start with security. And outside help cannot substitute for internal capacity. A planetary Haiti would arise not from who the majority of humanity are, but if the same loop of debt, environmental loss, disaster and capture ran in many states at once, with no external buffer left to hold it.

The defining feature of the Polycrisis Apocalypse is its duration. Carbon dioxide emitted now stays in the atmosphere–ocean system for tens of thousands of years; the IPCC projects sea-level rise of 2 to 6 metres over the next two millennia at 2 °C of warming and 19 to 22 metres at 5 °C. Topsoil, once lost, regenerates over centuries. HH-2026-34 argues that the right measure for such a world is not a headcount at a date but the sum of person-years lived in unsafe conditions.

The dashboard reproduces the report's three pathways from its published anchors. On buffered convergence (S1), cumulative SRR exposure reaches about 0.8 trillion person-years by 3000 CE. On unsafe stabilisation (S2), where population declines slowly and 85% of people live in SRRs, it reaches 4.3 trillion by 3000 CE, more than all human life ever lived, and about 21 trillion by 12,000 CE. On the endgame path (S3), with population falling towards extinction, it reaches 2.7 trillion by 3000 CE and levels off near 9 trillion. Extinction, terrible as it is, caps the total; unsafe stabilisation does not.

These are thought experiments with a disciplined structure, not forecasts. No demographic or climate model has skill over tens of thousands of years, and whether a reduced, impoverished humanity would go extinct at all is deeply uncertain. Their value is to show what matters: the conclusion that prolonged unsafe stabilisation is the worst outcome holds as long as a few hundred million people persist in high-exposure conditions for a few thousand years. The simulator lets you change the plateau, the population and the starting point, including the measured values from section 3.

Build capacity where the population is growing. The measured SRR is defined by capacity as much as hazard: a country that crosses from low to adequate capacity leaves the SRR even if its heat does not change. Revenue mobilisation, debt restructuring on faster timelines, concessional and green finance, civil-service and security-sector reform, and digital public infrastructure are the levers. The demographic dividend of a young Africa is real if investment in schooling, health, power and jobs keeps pace; it becomes a burden only if it does not.

Cut the tiers directly. Clean electricity is the single most leveraged intervention in the belt because it addresses pollution, heat (through cooling), services and the fiscal load of imported fuel at once. Morocco's solar build-out, Kenya's geothermal-led grid and Ethiopia's hydropower show that deployment at scale does not require wealth first. Clean cooking is among the cheapest health interventions available. Heat-labour protocols, cooling refuges and school calendars that follow the heat can be put in place within years.

Prevent the discontinuities and preserve the means to rebuild. Nuclear arms control, bio-risk governance and the cyber-resilience of grids and finance reduce the chance of a break during the Danger Window. Security and governance support, funded before cascades rather than after, keeps the Haiti loop from starting. And the cheapest insurance against the long tail is knowledge: robust, distributed archives of scientific, medical and engineering knowledge, seed banks and the training of people who can use them. Above all, cutting emissions now sets how hot the floor of the coming centuries will be.

Heat: World Bank CCKP CMIP6 multi-model median climatologies (0.25°), 1995–2014 and 2040–59 and 2080–99 under SSP1-2.6 to SSP5-8.5: days with heat index above 35 °C (tier B at 60+ days) and above 41 °C (tier D at 30+ days); mean annual temperature for unprecedented heat (≥ 29 °C). Population: SSP-consistent gridded projections (Jones and O'Neill) for 2020 and period midpoints. Countries assigned with rasterised world-atlas boundaries. National tiers: SatPM PM2.5 (via Our World in Data, 2024); electricity and clean-cooking access (WDI); poverty at $4.20 a day (2021 PPP, latest since 2010); WGI government effectiveness and political stability.

Definitions: strict / moderate / broad thresholds for PM2.5 (35 / 25 / 15 µg/m³), electricity (90 / 95 / 99%), clean cooking (50 / 70 / 90%), poverty (40 / 25 / 10%), government effectiveness (−0.75 / −0.5 / −0.25) and political stability (−1.0 / −0.75 / −0.5). SRR = two or more tiers and low capacity. National conditions held at today's levels in projections. Beyond 2100: 2090s SRR rates for Africa and elsewhere held constant and applied to the Africa Ascendant demography (42% / 48% / 52% African in 2200 / 2300 / 2500). Long tail: HH-2026-34 Appendix A anchors, monotone cubic interpolation.

Caveats: national tiers ignore sub-national variation; heat thresholds are summaries of heat stress, not mortality; institutions and pollution are held fixed, though they may improve or erode; the SRR is a definition, and its size is highly sensitive to it; the long-run figures are scenario arithmetic. Conflict, displacement, weapons and Haiti figures are as published by UCDP, UNHCR, FAS/OWID, SIPRI, the Bulletin of the Atomic Scientists, IOM and the World Bank. The analysis concerns where and how people live, and makes no claim about any group's character or capacity.