What could happen, when, and how to prepare without surrendering to panic.
Evidence review: October 8, 2026. Version 1.1. Corrections and link checks: October 8, 2026.
A crisis rarely arrives alone. A failed harvest can become a food-price shock. A food-price shock can become a health emergency, displacement, or political unrest. Add a power outage or a badly timed war, and suddenly the problems are helping one another. They have formed a committee. Unfortunately, it is a very effective committee.
This guide organizes the Collapse Update research around the Universe Institute’s 15-crisis framework. Its purpose is to help readers see the evidence, distinguish different kinds of warnings, and make proportionate decisions about preparation, adaptation, resilience, and possible relocation. [P1]
The central finding: credible sources publish serious risks and some specific tipping-point windows. They do not establish a single date for worldwide societal collapse, a single death total for the polycrisis, or a year when everyone must migrate. Missing numbers mean missing estimates—not missing danger!
Choose your reading route
- Start here: five things to know
- The risk timeline: dates, threats, and mortality
- How to read death estimates
- The British actuarial report: the billion-death claim
- Mortality evidence: the full ledger
- All 15 crises: what the sources actually date
- Ecosystem collapse: onset versus completion
- Climate, AMOC, and World3: where forecasts differ
- How risks amplify one another
- What can change the dates and consequences
- Use this information without panic
- What governments should do immediately
- Frequently asked questions
- Glossary
- Method and credibility checks
- Linked bibliography
- Related pages and next steps
Ten-minute route: read the five points, the timeline, the actuarial correction, and the preparation section. The remaining sections let you audit the details. Every bracketed alphanumeric code opens its supporting bibliography entry.
Start here: five things to know
- Danger does not require a collapse deadline. A hospital without medicine or a town without reliable water can face a life-threatening emergency while the rest of the world continues functioning.
- Dates describe different things. A warming threshold, a potential ecosystem transition, a supply gap, and an intelligence planning horizon cannot be treated as interchangeable.
- Deaths are measured differently. Annual deaths, cumulative deaths, deaths associated with a risk, and extra deaths caused by that risk answer different questions.
- Worst cases matter, but labels matter too. A scenario can justify prevention without being the most likely future.
- Action changes outcomes. Emissions cuts, safer water, functioning health services, reliable infrastructure, and effective cooperation can reduce losses. Preparation is a tool for protecting life, not a hobby in collecting frightening headlines.
Read the timeline as a map of evidence and planning horizons. It is not a countdown clock. The arrows point toward things to watch and actions to take; they do not point toward an appointment with the apocalypse.
The Polycrisis Risk & Survival Timeline: a preparedness horizon
For your wall: use this chart to schedule risk reviews and resilience investments. Each line pairs a particular warning with its mortality evidence. Do not add the death numbers. Estimates spanning 2025–2050 or 2020–2100 remain whole-period estimates; their position here does not allocate them to one decade. Dates are milestones within the bands, not claims that risks begin only then.
| Date range / trigger | What the sources place in this horizon | Mortality estimates and their limits | Practical planning use — editorial guidance |
|---|---|---|---|
| Already observed, through 2026 | Coral central thermal threshold crossed [T6]; novel-entities boundary exceeded in the 2022 assessment [T4]; Saudi groundwater access threshold reported crossed [T3]; democratic backsliding documented for 2025 [T13]. | These observations do not supply a future global death count. Current damage is not a reason to invent one. | Check existing local exposure. Protect health, drinking water, cooling, finances, and social connections. |
| 2026–2030 | AI/cyber vulnerability outlook to 2027 [I11]; geoeconomic survey horizon to 2028 [T19]; debt at 100% of GDP in 2029 [T11]; PJM adequacy risk from 2029 [T18]. By 2030: older 40% water demand/supply gap projection [I6], 520,940 potentially uninsurable Australian homes [T10], and 435 million extreme poor in fragile economies [T12]. ASIO and AI capability outlooks run to 2030 [I10] [T14]. WMO: repeated annual 1.5°C exceedances are likely in 2026–2030 [T22]. | Aid-cut models, original study windows ending by 2030: 9.4m or 22.6m additional deaths under different all-donor scenarios [M10]; separate USAID model 14.1m [M9]. HIV models: 0.77–2.93m in 2025–2030 [M11]; updated UNAIDS 4.2m in 2025–2029 [M12]. These estimates overlap. No defensible global death toll is assigned to the water, debt, cyber, grid, poverty, or insurance dates. | Build a household plan; secure medicine access and records; check local flood/fire/heat exposure and insurance; support services that prevent deaths. |
| 2030–2039 | Long-term 1.5°C best estimate: first half of 2030s [T15]. Migration hotspots from 2030 [T9]; U.S. NIE water/migration conflict risk grows especially after 2030 [I7]. Potential ecosystem onset from 2030 (detail below) [I12]. Copper gap 25% in 2035 [T2]. Early-2030s food catastrophe appears in an alternative intelligence scenario [I8]; World3 welfare/output turning points roughly around 2030 are conditional [T21]. High-emissions ecological disruption can begin in tropical oceans before 2030 [T7]. | WHO: approximately 250,000 extra deaths per year during 2030–2050, selected climate-health causes only [M1]. None of the other milestones here comes with a supported global mortality total. The corrected AMOC model interval starts at 2037, but supplies no death toll [T16]. | Invest in heat-safe homes, resilient water and food systems, health capacity, and backup infrastructure. Set measurable relocation review triggers. |
| 2040–2050 | 2040 remains an intelligence outlook horizon, not a collapse year [I1] [I2] [I3] [I4] [I5]. Habitability-related displacement becomes more important mainly after 2040 [I2]. By 2050: up to 216m internal climate migrants [T9]; 10–47% of Amazon forests exposed to potentially transition-triggering disturbance [T8]; forest/high-latitude ecological disruption in a high-emissions scenario [T7]. Possible rainforest/mangrove collapse onset from 2050 [I12]. 2021 U.S. NIE placed 2°C around mid-century; 2026 Parasol Lost warns before 2050 without action [I7] [A3]. | 14.5m cumulative additional deaths by 2050, WEF/Oliver Wyman [M2]; 529,000 extra deaths in 2050 from modeled food/diet changes [M3]. AMR: 39.1m attributable / 169m associated during 2025–2050; in 2050 alone 1.91m / 8.22m respectively [M4]. Older AMR scenario: 10m/year in 2050, different scope and assumptions [M16]. WHO estimate continues [M1]. Not additive. | Review long-lived investments against several scenarios. Plan services and housing for people staying and for people moving. Maintain prevention and adaptation. |
| 2051–2070 | Corrected statistical AMOC central estimate 2065, within 2037–2109; disputed, conditional [T16] [T17]. OECD baseline plastic leakage roughly doubles by 2060 [T5]. Earlier ecosystem transitions may still be unfolding [I12]. | Older OECD outdoor-air-pollution model: 6–9m premature deaths/year by 2060 [M5]. A newer OECD PM2.5 baseline projects declines through 2050, showing the importance of model vintage and pollutant scope [M6]. No credible global death count accompanies the AMOC date or plastics projection. | Keep adaptation flexible. Update models before using old projections to justify new infrastructure or a move. |
| 2071–2109 | Actuarial conditional GDP warning: 50% loss in 2070–2090 [A4]. UN population peak around 10.3bn in mid-2080s [T1]. UNEP 2025 century pathways: 2.8°C under current policies / 2.3–2.5°C with full pledges [T23]. Corrected AMOC interval extends to 2109 [T16]. | Bressler: 83m cumulative temperature-related excess deaths over 2020–2100 in baseline; about 9m on its modeled optimal path; annual burden rises above 4m at 4°C late in its baseline [M7]. Pozzer et al.: roughly 30m deaths/year late-century, total non-optimal-temperature plus PM2.5 burden under SSP2-4.5, not all extra climate deaths [M8]. | These horizons guide durable land-use, infrastructure, emissions, and institutional decisions. They do not establish a personal survival probability. |
| Event-triggered or undated: relevant at any time | Nuclear war and pandemics are not assigned calendar start dates [M13] [M14]. DDIS windows are relative to military/political conditions: roughly six months, two years, five years; near-term limited attacks assessed separately [I9]. Pollinator-loss scenarios have no dated onset [M15]. The actuarial 2°C/3°C severity matrix is illustrative [A1]. | Nuclear famine scenarios: >2bn / >5bn deaths in selected India–Pakistan / U.S.–Russia cases, concentrated in the first few post-war years [M13]. Influenza model: 720,000 annualized expected deaths; 1918-type scenario 21–33m at historical calibration [M14]. Pollination losses: 700,000 or 1.42m additional annual deaths for 50% or complete service loss [M15]. Actuarial >2bn / >4bn bands are not mortality forecasts [A1]. | Use official emergency instructions, readiness, diplomacy, health surveillance, and ecological protection. Do not translate a hypothetical trigger into a scheduled disaster. |
Risk profile: the near term combines service interruption, funding choices, conflict, and information risks; the 2030s add more dated climate/ecosystem milestones; mid-century assessments bring greater migration, health, and ecological stresses; later-century outcomes diverge widely by policy. This is an editorial synthesis of the sources above, not a calibrated probability rating. A later date does not make a risk more certain or a present emergency less urgent.
Return to contents | Return to the risk timeline
Before reading a death number, ask four questions
| Question | What to check | Example |
|---|---|---|
| How long? | One year, an entire period, or years after an event? | 250,000 per year differs from 14.5 million accumulated by 2050. [M1] [M2] |
| Compared with what? | Total burden or additional deaths versus a no-change/reference world? | The late-century atmospheric total includes existing pollution and hot/cold temperature burdens. [M8] |
| Which cause? | Attributable to the risk or associated with an infection/exposure? | AMR-associated deaths include AMR-attributable deaths. Adding 169m to 39.1m double counts. [M4] |
| Under which assumptions? | A model forecast, conditional scenario, expected annual loss, or illustrative severity band? | The actuaries’ billion-death bands and the nuclear-famine model have different evidentiary meanings. [A1] [M13] |
Why there is no grand total: the same person can be exposed to hunger, heat, pollution, resistant infection, and interrupted medical care. Studies use different populations, baselines, assumptions, and time windows. Their combined effects could be worse than any single model, but a valid combined estimate requires a joint model that handles overlapping causes and interactions. A calculator cannot do that by pressing “plus” with unusual confidence.

Four panels distinguish annual and cumulative deaths, show 39.1 million attributable deaths inside 169 million associated deaths over 2025–2050, explain overlap among studies, and identify the actuarial bands as illustrative.
The British actuarial report: what the billion-death claim really means
The billion-death claim discussed in this YouTube video concerns numbers that appear in the January 2025 IFoA–University of Exeter report, Planetary Solvency: Finding our balance with nature. IFoA is a professional actuarial body with expertise relevant to insurance and pensions; this is not a mortality forecast agreed by the entire British insurance industry. The interpretation below is based on the report itself. [V1] [A1]
Figure 12 on printed page 32 contains a severity matrix. The qualification immediately above it says the matrix ranks potential impacts and is “not a prediction or central scenario.” That distinction changes how the figures can honestly be reported. [A1]
| Report band | Illustrative climate descriptor | Illustrative human and economic impact | How to use it |
|---|---|---|---|
| Catastrophic | 2°C or more by 2050 | More than 2 billion deaths; GDP losses of at least 25%. [A1] | A high-consequence stress-test band. Not proof that reaching 2°C causes 2bn deaths, or that those deaths occur by 2050. |
| Extreme | 3°C or more by 2050 | More than 4 billion deaths; GDP losses of at least 50%. [A1] | A more severe band. Not a calculated causal relationship or an established cumulative mortality forecast. |
The table also associates these bands with ecosystem disruption, migration, conflict, and recurring mass mortality. But grouping impacts in one row does not demonstrate that a specified temperature mechanically causes the listed number of deaths. The report does not provide a validated global mortality model deriving those totals, nor a separate accumulation period for the deaths. The temperature column’s 2050 marker must not be silently turned into a death deadline. [A1]
A defensible sentence: “The 2025 actuarial report illustrates catastrophic and extreme severity bands with more than two billion and four billion deaths, respectively; it explicitly does not present those bands as forecasts.” [A1]
The substantive risk-management argument remains valuable: financial planning should consider potentially severe, interconnected losses, including difficult-to-model outcomes. Its conditional GDP warning for 2070–2090 is a separate claim, not a statistical confirmation of the billion-death bands. [A4]
The newer reports: January 2026’s Parasol Lost warns that reduced aerosol cooling can accelerate warming, with 2°C before 2050 without action. April 2026’s Tipping into the wild unknown focuses on food-system fragility and its financial consequences. It calls for sustainable land use, pollinator protection, and stronger supply chains. Neither announcement converts the 2025 illustrative mortality bands into validated forecasts. [A3] [A2]
This is a reason to improve both preparedness and evidence. The smoke alarm is useful. It does not become more useful if we announce that it has counted everyone who will die.
Return to contents | Return to the risk timeline
Mortality evidence: the full ledger
The following are the principal quantitative mortality forecasts and scenarios identified in this review, including older estimates that need qualification. “Global” here sometimes means a model across many affected countries, not every country. This is a broad public-source review, not a claim that every credible forecast in every language has been found.
| Mortality pathway and source | Time measure | Published estimate | Conditions and interpretation |
|---|---|---|---|
| Climate-related illness — WHO | 2030–2050; annual | Approximately 250,000 extra deaths per year. [M1] | Selected undernutrition, malaria, diarrhoea, and heat pathways only. Older assessment; excludes many direct and cascading effects. Not the entire climate toll. |
| Climate-related health shocks — WEF / Oliver Wyman | Cumulative by 2050 | 14.5 million additional deaths. [M2] | Scenario-based assessment of six climate-related hazards under SSP2-4.5. Floods and droughts dominate its modeled toll. Not 14.5m every year or a universally agreed estimate. |
| Food and diet changes — Springmann et al. | In 2050; annual burden | 529,000 extra deaths; 95% CI 314,000–736,000. [M3] | Climate-mediated dietary/bodyweight changes. Includes chronic-disease pathways; not just starvation. Emissions mitigation changes the outcome. |
| Bacterial AMR — GBD / GRAM | 2025–2050 cumulative; 2050 annual | Cumulative: 39.1m attributable (UI 33–46m), 169m associated (145–196m). In 2050: 1.91m attributable, 8.22m associated. [M4] | Reference scenario. Associated is the wider category and includes attributable; do not add them. Better prevention, care and drug access can avert many deaths. |
| Older resistance scenario — O’Neill review | 2050; annual | 10 million deaths per year under uncontrolled resistance. [M16] | An older, broader commissioned scenario, including diseases beyond the newer bacterial model. Retain as historical scenario context, not a replacement for M4 or a second independent toll. |
| Outdoor air pollution — older OECD model | 2060; annual | 6–9 million premature deaths per year. [M5] | Older policy/model assumptions and particulate matter plus ozone scope. The 2025 triple-crisis outlook instead projects PM2.5 mortality declines through 2050 [M6]. These are different model vintages and horizons; do not present the older figure as inevitable. |
| Temperature-related climate mortality — Bressler | 2020–2100 cumulative; late-century annual | Central baseline 83m excess deaths; modeled optimal path about 9m, a 74m reduction. Baseline reaches about 4°C in 2096–2100, with over 4m extra deaths/year at 4°C. [M7] | Integrated economic–mortality model, sensitive to warming and adaptation assumptions. Not an all-pathway polycrisis toll. No defensible decade allocation is made here. |
| Temperature plus particulate pollution — Pozzer et al. | Late century; annual | Approximately 30m deaths/year under SSP2-4.5; abstract uncertainty range 12–53m. [M8] | Total modeled attributable burden from non-optimal heat/cold and PM2.5, not exclusively additional climate deaths. Population, age structure and exposure assumptions matter. |
| USAID cuts — Cavalcanti et al. | Original 2025–2030 forecast window; cumulative | About 14.1m additional deaths (UI 8.48–19.66m), including 4.54m children under five. [M9] | Conditional on modeled withdrawal of services versus continued funding. Observational historical associations inform the model; replacement financing and policy changes can alter it. |
| All-donor aid cuts — Ferreira da Silva et al. | Forecast to 2030; cumulative | Mild: 9.4m excess deaths (UI 6.2–12.6m), including 2.5m under-five. Severe: 22.6m (16.3–29.3m), including 5.4m under-five. [M10] | 93 LMICs; alternative scenarios, not two tolls to add. Overlaps the USAID/HIV projections. These are model totals, not a count of deaths remaining after October 2026. |
| HIV funding cuts — ten Brink et al. | 2025–2030; cumulative | 0.77–2.93m additional HIV deaths. [M11] | Range across different funding/mitigation scenarios; not a confidence interval. LMIC extrapolation from country models. Overlaps wider aid-loss estimates. |
| PEPFAR program discontinuation — UNAIDS | 2025–2029; cumulative | Updated model 4.2m additional AIDS-related deaths. Earlier rapid estimate 6.3m was revised. [M12] | 55 supported countries; permanent cessation assumption. Scenario, not a count of deaths already occurring. Changed services and substitute resources change outcomes. |
| Nuclear war — Xia et al. | Years after a war; no start date | Selected India–Pakistan scenario >2bn deaths; largest U.S.–Russia scenario >5bn. [M13] | Famine modeled following soot-driven climate disruption. Severe food deficit occurs in the first two years; largest calorie-production reduction approaches 90% in years 3–4. Assumes disrupted trade and specified food responses. Neither forecasts a war date nor establishes inevitable extinction. |
| Influenza pandemic risk — Fan et al. | No calendar date; annualized expectation | About 720,000 expected deaths/year; a 1918-type event modeled at 21–33m deaths in 2015-era calibration. [M14] | A probability-weighted risk measure, not 720,000 guaranteed annual deaths. Historical calibration and uncertainty make it unsuitable as a precise present-day pandemic forecast. |
| Pollinator loss — Smith et al. | Undated hypothetical service loss | 50% pollination-service loss: 700,000 extra deaths/year. Complete loss: 1.42m (1.38–1.48m). [M15] | Nutrition/disease consequence model. Does not predict complete pollinator loss by any date. Percent service loss is not a percent of insect species extinct. |
The human meaning: a smaller estimate is not a small tragedy. Hundreds of thousands of preventable deaths justify substantial action. Conversely, an enormous number needs especially clear assumptions. Credibility comes from the method and label, not the number of zeros.
Return to contents | Return to the risk timeline
All 15 crises: what has a date, and what does not
This table retains the original project’s crisis order. Intelligence and government-security findings appear first in each row, followed by scientific or institutional findings. “No dated forecast found” means no substantiated public forecast emerged in this review; it says nothing about classified material. No intelligence assessment located supplies a complete 15-crisis global-collapse calendar.
| Crisis | Intelligence / government-security evidence | Other dated evidence | What remains unproven or unquantified |
|---|---|---|---|
| 1. Population pressure and uneven demographic growth | NIC outlook through 2040: expanding urban demand can strain services. [I1] | UN population peak around 10.3bn in mid-2080s. [T1] | Demographic turning point, not a collapse threshold. No attributable global death forecast for population pressure alone found. |
| 2. Overshoot, overconsumption, and resource depletion | 2012 water assessment: demand about 40% above then-current sustainable supply by 2030 under its external model. [I6] | IEA 25% copper gap in 2035 [T2]; specific groundwater tipping points [T3]; World3 overshoot scenarios [T21]. | Supply gaps are not total exhaustion. Food/diet mortality [M3] is one related pathway, not a death count for all overshoot. |
| 3. Pollution of air, water, soil, and food | NIC environmental pressures through 2040. [I2] | Novel-entities boundary exceeded by 2022 [T4]; baseline plastic leakage roughly doubles by 2060 [T5]. | Boundary breach is not a dated irreversible collapse. Air-pollution forecasts differ [M5] [M6]; combined atmospheric burden [M8] is not wholly extra climate mortality. |
| 4. Biodiversity loss and ecological breakdown | UK security assessment: possible ecosystem collapse onset from 2030 / 2050, depending on system. [I12] | Coral threshold [T6]; high-emissions disruption before 2030/by 2050 [T7]; Amazon disturbance exposure by 2050 [T8]. | Specific ecosystem transitions, not dated global ecological collapse. Pollination mortality scenarios have no calendar date. [M15] |
| 5. Migration and displacement | NIE water/migration conflict especially after 2030 [I7]; some permanent uninhabitability mainly after 2040 [I2]. | Groundswell: hotspots from 2030, up to 216m internal climate migrants across six regions by 2050. [T9] | People moving are not deaths. No supported global migration-driven mortality total or migration-system collapse date found. |
| 6. Crime, conflict, terrorism, and war | DDIS conditional capability windows: six months/local, two years/regional, five years/large European war. [I9] | Nuclear-war food-collapse scenario has event-relative consequences, not a war start year. [M13] | No dated global/nuclear war forecast. Crime and terrorism cannot inherit a nuclear scenario’s mortality number. |
| 7. Economic fragility and financial instability | NIC debt/financial constraints through 2040. [I3] | Debt reaches 100% of GDP in 2029 [T11]; regional uninsurability by 2030 [T10]; actuarial GDP risk in 2070–2090 [A4]. | No universal debt-collapse threshold or dated global financial collapse found. Illustrative actuarial mortality is not a forecast. [A1] |
| 8. Political instability and government failure | NIC anticipates multiple political breakdown risks over its 2021–2040 horizon. [I4] | World Bank fragile-economy vulnerabilities to 2030. [T12] | No universal government-collapse date or global death total found. Aid-service loss supplies a specific conditional mortality pathway, not a death forecast for political failure as a whole. [M9] [M10] |
| 9. Authoritarianism, executive overreach, and nationalism | NIC assesses vulnerabilities in democracies and authoritarian states through 2040. [I4] | V-Dem 2026 documents deterioration during 2025. [T13] | Trend evidence, not a dated worldwide political tipping point. No valid separate global mortality forecast found. |
| 10. Pandemics, conflict-related disease, and AMR | Intelligence outlooks identify continuing health vulnerability; no date for the next pandemic established. [I1] [I7] | AMR forecasts to 2050 [M4] [M16]; pandemic expected-loss model [M14]; HIV service-loss models [M11] [M12]. | A rising burden does not imply all antibiotics stop working in one year. Pandemic timing remains uncertain. |
| 11. Inequality, poverty, and social fracture | NIC societal fragmentation outlook through 2040. [I5] | 435m extreme poor in fragile/conflict-affected economies by 2030. [T12] | No universal social-collapse threshold found. Aid-cut mortality illustrates a preventable pathway through poverty and service loss. [M9] [M10] |
| 12. AI manipulation, surveillance, and psychological distortion | ASIO security outlook to 2030 highlights AI-enabled deception. [I10] | International AI Safety Report 2026 includes capability scenarios to 2030. [T14] | No established date or quantified worldwide mortality forecast for this crisis found. AI can amplify failures without a credible death calendar. |
| 13. Accelerating climate change | 2021 NIE: about 1.5°C around 2030 and 2°C around mid-century on its trajectory. [I7] | IPCC long-term 1.5°C in early 2030s [T15]; corrected AMOC 2037–2109 / central 2065, disputed [T16] [T17]; WMO near-term outlook [T22]. | Warming thresholds are not societal-collapse dates. Multiple mortality models cover selected pathways. [M1] [M2] [M3] [M7] [M8] |
| 14. Infrastructure fragility and cyber disruption | NCSC conditional AI/cyber vulnerability to 2027. [I11] | NERC PJM high resource-adequacy risk from 2029. [T18] | Regional shortfall and attack risk, not predicted worldwide grid collapse. No defensible global infrastructure-death total found. |
| 15. Geoeconomic fragmentation and supply-chain division | NIC outlook through 2040: greater economic fragmentation. [I3] | WEF geoeconomic confrontation leads survey horizon 2026–2028 [T19]; IMF long-run fragmentation scenarios have no collapse year [T20]. | Risk rankings are not probabilities. Health-aid cuts can be deadly, but their mortality estimates cannot be attributed to all trade fragmentation. |
Return to contents | Return to the risk timeline
Ecosystem collapse: starting is not finishing
The UK assessment is a reasonable worst-case security exercise, publicly issued by Defra using intelligence methods. Overall confidence in exact timing is low. The following are possible onset dates and transition durations, not completion deadlines or global death forecasts. [I12]
| Ecosystem | Potential onset | Confidence | Transition duration |
|---|---|---|---|
| Amazon rainforest | From 2050 | Low | 50–100 years |
| Congo Basin rainforest | From 2050 | Low | Uncertain |
| Southeast Asian coral reefs | From 2030 | Medium | 10 years |
| Southeast Asian mangroves | From 2050 | Medium | Uncertain |
| Himalayas | From 2030 | Medium | 50–1,000 years |
| Russian boreal forests | From 2030 | Low | 40–100 years |
| Canadian boreal forests | From 2030 | Low | 40–100 years |
Three examples distinguish possible onset from transition duration: Southeast Asian coral reefs from 2030 over about 10 years; Amazon rainforest from 2050 over 50–100 years; Himalayas from 2030 over 50–1,000 years.
Do not turn “from 2030” into “nothing happens until 2030,” or into “everything is gone in 2030.” Harm can precede a threshold; a transition may unfold afterward. These distinctions help planners prepare earlier without pretending to know a precise final date.
Climate Time-Compression Watch: When Several Systems Fail Together
Climate risk is also a race between growing hazards and our ability to cope. A community might withstand a heatwave, a power outage, or a food-price shock separately, yet struggle when all three arrive together. Interacting emergencies can shorten the practical time available to adapt—even without a sudden jump in the global temperature average. [T15]
The Universe Institute proposes earlier warming dates. The chart below presents its conditional stress-test scenarios, without assigned probabilities. These scenarios warrant examination, but their greater accuracy has not been established. Keep this exploratory schedule separate from the evidence baseline in the main risk timeline; it should not be used to re-date all 15 crises or assign death totals without additional supporting models. [F1]
The Climate Time-Compression Watch Chart
These are Universe Institute scenario dates—not consensus forecasts, guaranteed deadlines, or predicted mortality dates. Temperatures represent increases above the 1850–1900 average.
| Global warming level | Institute low scenario | Institute medium scenario | Institute high scenario | Source |
|---|---|---|---|---|
| 2°C | About 2030 | About 2028 | About 2027 | [F1] |
| 3°C | About 2044 | About 2035 | About 2032 | [F1] |
| 4°C | About 2059 | About 2043 | About 2037 | [F1] |
| 5°C | About 2074 | About 2050 | About 2042 | [F1] |
| 6°C | About 2088 | About 2058 | About 2047 | [F1] |
| 7°C | Unlikely by 2100 | About 2065 | About 2053 | [F1] |
This chart follows F1’s detailed threshold schedule: the low scenario places 3°C around 2044 and 4°C around 2059. Rounded prose elsewhere in F1 gives approximately 2040 and 2050 instead. These discrepancies should remain visible; neither set of scenario dates should be treated as a precise, validated deadline. [F1]
Dots show low, medium, and high scenario crossing dates for 2°C through 7°C. The dates are conditional stress tests, with no assigned probabilities. The accessible table above contains the values.
The calculation uses annual CO₂-equivalent concentration growth of 1%, 2%, or 3%; equilibrium climate sensitivity of 4.5°C; assumed 2025 warming of 1.7°C; no carbon removal; and additional warming allowances for feedbacks. Concentration growth and fossil-fuel-use growth are different quantities. [F1]
Why these assumptions matter: equilibrium climate sensitivity describes the eventual warming response to doubled carbon dioxide. It does not, by itself, establish how quickly that warming occurs. Ocean heat uptake and changes in greenhouse gases and aerosols affect the response over decades. Therefore, using an equilibrium relationship to assign calendar-year temperatures requires testing against a model of the time-dependent response. IPCC AR6 assesses a best estimate of 3°C, a likely range of 2.5–4°C, and a very likely range of 2–5°C. A 4.5°C assumption is a higher-sensitivity case within the broader assessed range, rather than an established replacement for the central estimate. [F4]
An important reality check: WMO’s March 2026 assessment puts observed 2025 warming at approximately 1.43°C above the pre-industrial average. Its May 2026 outlook assigns less than a 1% chance to even one year exceeding 2°C during 2026–2030. Paris Agreement thresholds refer to sustained warming, typically assessed over about 20 years. These observations and definitions need to be reconciled before treating the accelerated schedule as a reliable temperature forecast. A scenario can help test preparedness without being the most likely future. [F5] [T22]
An important note on the Universe Institute climate change temperature calculations: unlike other think tanks, the Universe Institute assigns reasonable forecast-modeling values to climate change system and subsystem variables, which other climate change forecasts almost always exclude or underestimate. A full list of these important climate change forecasting variables, and how they affect future climate modeling, is in another Universe Institute paper on this page.
The temperature forecast and timetables on this page will be extremely valuable as the conditions, circumstances, and variables described here occur sooner than government and media climate change studies predict. To better understand why the Universe Institute also includes in its models climate change variables not found in other climate change modeling, it is important to understand something called the highly illustrated Climaeddon Feedback Scenario.
If you find global temperature rising significantly faster than previous government and media predictions, remember this Universe Institute forecast page and remember this Climaeddon Feedback Scenario page.
Four panels distinguish annual from sustained warming, summarize WMO’s 2026–2030 probabilities, and explain why eventual climate sensitivity does not establish warming speed.
Why Interacting Crises Can Whipsaw a Community
Peer-reviewed polycrisis research describes how shared pressures, domino effects, and feedback loops can connect crises across systems. Some connections amplify harm; others help stabilize systems. This supports examining linked vulnerabilities without treating every interaction as a guaranteed cascade. [F7]
The Climageddon Feedback Loop is Job One for Humanity’s name for a systems-risk framework. Its updated explanation distinguishes established mechanisms from hypotheses and precautionary estimates. It does not claim that scientists have detected one universal master feedback loop. [F2]
There is credible scientific reason to investigate these interactions. Armstrong McKay and colleagues’ 2022 assessment identified several tipping elements potentially vulnerable around 1.5–2°C. Wunderling and colleagues’ 2024 review found that many assessed interactions are destabilizing, while others stabilize systems or remain uncertain. Interactions can alter tipping thresholds and enable cascades, although their strength and timing remain difficult to establish. Much of the research still relies on conceptual models. This supports investigating dangerous combinations; it does not independently validate the dates in the Institute’s chart. [F6] [F3]
Consider an illustrative cascade. Heat and drought reduce harvests while raising electricity demand. A strained grid fails, disrupting cooling, water pumping, refrigeration, and medical care. Food losses and rising prices deepen household hardship. Repeated losses strain insurers and public budgets, leaving less capacity for the next emergency. Physical hazards and institutional weakness then reinforce one another. This illustrates dependencies, rather than predicting a fixed sequence everywhere. [T15] [A2] [F7]
Heat and drought converge on a vulnerable electricity system. A power failure can disrupt water, cooling, food, and care, reducing recovery capacity. Protecting essential services can interrupt the cascade.
Multiple related systems in crisis can whipsaw and interact with each other, making consequences far worse than a single-hazard forecast suggests—and arriving far sooner than a single-hazard preparation plan anticipates.
Three ideas help explain this danger:
- Cumulative damage: repeated disruptions exhaust savings, supplies, infrastructure, and recovery capacity.
- Nonlinear responses: a modest additional stress can cause a disproportionately large loss when a system approaches its limits.
- Synergistic effects: one disruption strengthens another—for example, an electricity failure makes extreme heat more dangerous by disabling cooling.
The magnitude and timing depend on the connections involved, local exposure, and remaining capacity. Some interactions amplify change; others delay or counteract it. A reinforcing feedback does not automatically mean unlimited warming or inevitable global collapse. [F2] [F3]
Earth’s systems do not take turns politely. Unfortunately, there is no global ticket dispenser saying, “Drought, please wait until the hospital has finished dealing with the blackout.”
Review this list of the primary and secondary consequences of climate change to get a better idea of how related systems and subsystems can whipsaw and interact as temperatures rise. The severity, scale, duration, and reversibility of potential damage should influence how closely a risk is investigated, prevented, managed, and prepared for. UNEP’s September 2026 report, Limiting Overshoot, describes an optimistic scenario in which warming peaks at about 1.8°C above pre-industrial levels; other scenarios peak higher. António Guterres, the United Nations Secretary-General, called for any overshoot above 1.5°C to be as small and short as possible. The 1.8°C figure is a scenario result, not a universal threshold at which all consequences become permanent. Separately, the IPCC finds that sea-level rise will continue for centuries to millennia because of deep-ocean warming and ice-sheet melting. Long-lasting, potentially irreversible harm therefore warrants proportionate scrutiny even when its precise timing is uncertain. [F8] [T15]
Use the Warning to Gain Options
Plan across a range of futures. Keep assessed climate projections as the evidence baseline, then test essential services against faster or more severe conditions. Can water, cooling, food, medicines, communications, and evacuation still function if two or three dependencies fail together? Where the answer is no, identify a practical improvement now. Backup plans work best when they do not all depend on the same grid, supplier, road, or bank account.
Use local decision triggers. Watch recurring unsafe indoor heat, unreliable water, evacuation frequency, insurance availability, and access to essential care. Assess possible relocation destinations using current local evidence, affordability, employment, legal access, and social support. A global temperature date alone cannot tell you when to move or which destination will remain livable.
The timetable and consequences can change. Emissions and methane reductions, aerosol changes, carbon-sink behavior, natural variability, and feedback strength affect warming and hazards. Poverty, exposure, infrastructure, health care, and governance affect the losses those hazards cause. Uncertainty about timing is a reason to build flexibility and monitor evidence. It is also a reason to preserve the actions that can improve outcomes. [F4] [T15]
Governments should prepare for combined failures immediately. Assess heat, drought, flood, power, food, and health risks together. Protect critical-service continuity, publish understandable warning indicators, support vulnerable households, and plan equitable adaptation and relocation where necessary. Pair these measures with rapid emissions reductions and ecosystem protection. Measure progress through reliable services and avoided harm. [T15] [A2]
Three places where the evidence needs extra care
1. Warming averages: one hot year is not the same measure as long-term warming
WMO’s 2026 outlook puts the chance of at least one 2026–2030 year above 1.5°C at 91%, and the five-year mean above it at 75%. It puts a single year above 2°C in that period below 1%. Paris thresholds use sustained long-term warming, commonly assessed over about 20 years. [T22]
The IPCC’s early-2030s 1.5°C estimate is a long-term threshold estimate. The source page’s roughly 2°C-around-2030 scenario must therefore remain an Institute scenario, not be relabeled the WMO or IPCC forecast. Different emissions paths, aerosol changes, variability and definitions can change timings; they do not erase these differences in evidence. [P1] [T15] [T22] [A3]
2. AMOC: a corrected estimate and genuine disagreement
AMOC is a large Atlantic circulation system, not simply another name for the Gulf Stream. The 2023 Ditlevsen–Ditlevsen estimate was corrected in August 2025 after code errors. The corrected central date is 2065, with a 2037–2109 interval, replacing 2057 and 2025–2095. Its 95% interval is conditional on the statistical model and assumptions; it does not capture every form of uncertainty. [T16]
Baker and colleagues’ 2025 analysis across 34 climate models finds twenty-first-century collapse unlikely. Both results belong in the evidence record. Neither supports a single consensus collapse year, and neither supplies a global death toll. Plan for the consequences of substantial circulation change while following the evolving literature. [T17]
3. World3: overshoot scenarios are not a dated mortality forecast
Herrington’s comparison found two World3 scenarios fitting observed data most closely. They pointed to several welfare, food and industrial indicators stopping growth roughly around 2030 and declining thereafter, but only one portrayed collapse; the other showed more limited decline. These are conditional system-behavior scenarios, not evidence that worldwide collapse is established for 2030. There is no validated global death count attached to this finding. [T21]
Return to contents | Return to the risk timeline
Why these risks can become more than a list
A polycrisis is a network problem. Looking at one crisis at a time can miss shared dependencies and the ways losses spread. The following is an explanatory synthesis, not a model that predicts probabilities or deaths. Food-system and national-security assessments describe these types of connections. [A2] [I7] [I12]
| Stage | How the problem can spread | Protective intervention — editorial synthesis |
|---|---|---|
| Physical or ecological shock | Heat, drought, flooding or lost pollination strains harvests and water. | Protect watersheds and habitats; diversify crops; reduce emissions; improve water efficiency. |
| Service and price shock | Food becomes less affordable; power, transport or medical supply failures make recovery harder. | Protect essential imports and logistics; backup power for care; targeted income/food support. |
| Human and institutional stress | Illness, job loss, debt, displacement or public mistrust reduce the capacity to respond. | Maintain care and public health; transparent information; fair access to assistance; functioning local institutions. |
| Feedback into the crisis | Conflict or political failure can damage services, block trade, and weaken environmental protection. | Diplomacy; rights protections; accountable emergency powers; coordinated reconstruction and restoration. |
For example, a heatwave combined with a blackout may be more dangerous than either alone. A food shock during an aid cut can undermine nutrition and treatment together. These are interaction mechanisms, not permission to multiply unrelated risk percentages or add separate death forecasts.
What can change the timeframes and consequences?
- Emissions and aerosols: long-term greenhouse-gas trajectories, methane cuts and changes in cooling pollution affect warming. Cleaner air should be paired with rapid climate mitigation; the aerosol issue is not an argument for keeping people exposed to toxic pollution. [T15] [A3]
- Adaptation and access: cooling, safe water, vaccination, infection control, affordable treatment and food support can change mortality even at a given warming level. [M1] [M3] [M4] [M10]
- Funding and institutions: restored or substitute aid, reliable procurement, local capacity, public trust and competent government can change service-loss outcomes. [M9] [M10] [M11] [M12]
- Ecological protection: land use, forest protection, fire management and pollinator conservation affect transition risk. [T8] [M15] [A2]
- Trade and conflict: military choices and export restrictions can turn a regional shock into an international one; diplomacy and resilient logistics can limit transmission. [I9] [T20] [M13]
- Population and vulnerability: aging, poverty, urbanization and uneven service access change both exposure and the number of people harmed. [I1] [M4] [M8]
- Models and new observations: revised methods, corrections and changing baseline policies can shift forecasts. AMOC corrections, UNAIDS revisions and different OECD pollution outlooks illustrate why every estimate needs a date and a label. [T16] [M12] [M5] [M6]
Some preventive actions influence whether a threshold is crossed; others reduce the damage after it is crossed. They are complementary. Preparation does not replace prevention, and prevention does not remove the need for preparation.
Why Long-Lasting, Lower-Probability Risks Require Greater Precaution
The precautionary principle means that when credible evidence points to serious or irreversible harm, incomplete scientific certainty should not become an excuse to postpone effective prevention. This approach is recognized in the United Nations’ Rio Declaration. The European Commission adds an essential safeguard: precautions should match the threat's seriousness, weigh the benefits and harms of acting or not acting, and be reviewed as evidence improves.
A low probability of catastrophic harm is not the same as a small risk. As potential damage becomes more severe, more frequent, more widespread, and longer-lasting, even less-likely outcomes deserve closer investigation and stronger safeguards. Repeated disruptions can exhaust recovery capacity; irreversible losses can remove options permanently. When consequences could persist for decades, centuries, or millennia, today’s decisions also impose risks on generations who cannot consent to them. That justifies greater scrutiny, earlier monitoring, and proportionate prevention—not treating every frightening possibility as a certainty. [R1] [R2]
Sea-level rise illustrates the stakes. The Intergovernmental Panel on Climate Change concludes that sea-level rise will continue for centuries to millennia because of ocean warming and ice-sheet melting. Uncertainty about the fastest possible ice loss does not remove the need to reduce emissions, protect essential infrastructure, and avoid creating new coastal vulnerabilities. Waiting for complete certainty can mean waiting until the affordable options have disappeared—the ocean does not offer a thirty-day return policy. [T15] The practical lesson is to investigate plausible severe outcomes, preserve options, and choose protections that remain useful across several futures. Precaution does not require eliminating every risk at any cost. It requires taking credible, potentially enduring harm seriously enough to act before prevention becomes impossible.
How to use the timeline logically and calmly
Use global evidence to decide what to investigate locally. Do not treat a world mortality scenario as the probability that you or your family will die. Personal risk depends on location, housing, health, income, services, and the ability to respond. The steps below combine FEMA’s basic preparedness approach with editorial planning suggestions for longer-term resilience. [S1]
- Start with your actual hazards. Identify local flood, wildfire, heat, water and outage risks, official alerts, evacuation routes and assistance needs. Make a contact plan and an accessible document backup.
- Protect essential daily functions. Build practical emergency supplies, food and water capacity, lighting, communications, and a plan for medicines or medical equipment. Discuss treatment continuity with clinicians. Follow local evacuation orders and official incident instructions.
- Strengthen connections. Make agreements with neighbors, friends and family about checking on vulnerable people, transport, communication and temporary shelter. Reliability is often a social resource before it is a product to purchase.
- Set review triggers. Examples: repeated hazardous indoor heat, worsening water reliability, unaffordable insurance, recurrent evacuations, or declining access to essential care. Record the trigger, the evidence to check, and the action it would prompt.
- Assess relocation before an emergency forces it. Compare more than temperature: reliable water, flood/fire exposure, housing cost, employment, care access, power, legal residence, and social support. Investigate candidate places with current local evidence. No place is guaranteed safe; a distant destination can create new vulnerabilities.
- Review periodically. A six-month review, and another after a major local event, is a reasonable editorial starting point. Update assumptions rather than endlessly refreshing frightening headlines. Your nervous system is not an emergency-management agency.
| Useful decision | Misleading shortcut |
|---|---|
| “Our home overheats during outages; let’s improve cooling, backup plans, and assistance.” | “A global death estimate tells us our exact survival odds.” |
| “A possible 2030 ecosystem onset means protect dependencies and monitor conditions now.” | “There is no reason to prepare before 2030.” |
| “Compare relocation options using local evidence and agreed triggers.” | “A planetary average identifies one guaranteed safe country.” |
| “Maintain emergency readiness while supporting prevention and public services.” | “Buying supplies means the wider causes no longer matter.” |
A one-page household exercise: list your three largest dependencies; identify the interruption that would matter most; choose one affordable action for each; assign a review date. A modest plan you can carry out beats a dramatic plan that lives exclusively in a browser tab.
Optional research prompt: “Using primary sources, assess my location’s heat, flood, wildfire, water, insurance, health-care and grid risks. Distinguish observed conditions, projections and scenarios. State dates and uncertainty. Help me define measurable review and relocation triggers without inventing safety guarantees.” Avoid sharing sensitive household information unnecessarily.
Return to contents | Return to the risk timeline
What governments should start doing immediately
The following priorities are an editorial policy synthesis from the evidence, not a claim that every cited institution endorses every recommendation. The aim is to prevent deaths, maintain essential functions, and reduce the chance that one crisis disables the response to another.
- Coordinate risks across ministries. Give one accountable national process responsibility for identifying shared food, water, energy, health, finance and infrastructure dependencies. Publish assumptions and regularly update stress tests that include simultaneous shocks. [A1] [A2] [I7]
- Keep life-saving services operating. Secure medicines, vaccination, disease surveillance, infection control, safe water and sanitation. Protect humanitarian support and arrange funded transitions before withdrawing programs; otherwise a budget decision can become a mortality event. [M4] [M9] [M10] [M11] [M12]
- Reduce both warming and immediate exposure. Accelerate greenhouse-gas reductions while funding heat protection, resilient hospitals, early warnings and safe housing. Pair cleaner air with mitigation, rather than treating aerosol cooling as a reason to retain harmful pollution. [T15] [A3] [M1]
- Protect food, water and ecosystems together. Restore watersheds, soils and habitats; protect pollinators; diversify crops, supply sources and routes. Test food-system resilience against concurrent harvest and trade shocks. [A2] [T8] [M15]
- Harden critical infrastructure. Maintain grids, water networks and transport; secure digital systems; provide backup capacity for hospitals, cooling and communications. Update adequacy plans as demand, retirements and supply projects change. [I11] [T18]
- Protect people with the least capacity to absorb losses. Target support to children, older people, disabled people, low-income households and exposed workers. Maintain affordable food, care and housing, and monitor who is excluded from assistance. [M4] [M10] [T12]
- Make migration safer and more orderly. Plan housing, work and services in receiving areas; support adaptation where viable and voluntary movement where needed. Avoid confusing migration projections with mortality forecasts or treating displaced people as the cause of the crisis. [T9]
- Prevent conflict and nuclear escalation. Maintain diplomacy, crisis communication, arms-risk reduction and cooperation over essential resources. No national supply stockpile can substitute for preventing a nuclear-famine scenario. [I9] [M13]
- Protect accountable institutions and reliable information. Preserve oversight of emergency powers, transparent procurement and independent scrutiny. Build defenses against AI deception and cyber disruption while maintaining civil rights. [I10] [I11] [T13] [T14]
- Measure avoided harm. Publish service-continuity, water, heat, health, infrastructure and ecological indicators. Compare preventive investments with response costs; update policies when evidence changes. Do not substitute an impressive dashboard for a functioning response. [A1] [A2]
The most useful national plan prepares for several plausible futures, protects citizens during present emergencies, and works to prevent worse ones. We do not need perfect foresight before repairing obvious vulnerabilities. We need competent action before another committee of crises gets together.
Return to contents | Return to the risk timeline
Frequently asked questions
Is worldwide collapse predicted for 2030?
No substantiated universal date was found. Some sources identify specific potential onsets or alternative scenarios around 2030; others merely use it as a planning milestone. [I8] [I12] [T21]
Did British insurers predict two billion deaths at 2°C?
The IFoA–Exeter report gives >2bn and >4bn as illustrative severity bands, not modeled forecasts. The matrix itself says so. [A1]
Why include a number that is not a forecast?
Because it is widely circulated. Its proper label prevents a hypothetical impact band being mistaken for a measured temperature–mortality relationship. [A1]
Can I add all the death estimates?
No. Causes overlap and time measures differ. AMR-associated includes attributable deaths; aid and HIV models overlap; climate-health models may cover the same pathways. [M4] [M10] [M11]
Does “no death estimate” mean a risk is harmless?
No. It means this review found no defensible quantitative global estimate. Governance, cyber, migration and financial failures can still endanger lives.
Are the intelligence reports predictions?
Some contain forecasts, some planning outlooks, and some alternative futures. The NIC food-catastrophe story is one scenario; DDIS readiness windows are conditional capabilities. [I8] [I9]
Does an ecosystem onset date mean total loss in that year?
No. Starting a transition and completing it are different. Stated transition durations can extend far beyond the onset window. [I12]
Is an AMOC collapse in 2065 scientifically settled?
No. That is one corrected statistical model’s central estimate. A 34-model study finds this-century collapse unlikely. [T16] [T17]
Do 216 million migrants mean 216 million deaths or refugees?
Neither. Groundswell models internal climate migration across six regions in a pessimistic scenario by 2050. [T9]
Why are the mortality estimates so different?
They estimate different populations, causes, periods and counterfactuals. The WHO selected-cause estimate, an atmospheric total burden and an illustrative actuarial band are not three interchangeable estimates of the same quantity. [M1] [M8] [A1]
Does the aid-cut forecast describe deaths still to come after today?
No. Published totals cover the studies’ original forecast windows. This page does not subtract modeled or observed deaths to calculate a remaining toll. [M9] [M10]
Can policy still make a difference?
Yes. The studies are conditional and several explicitly examine prevention or mitigation. A modeled avoidable burden is a reason to protect services and reduce exposure. [M3] [M4] [M7] [M10]
Should I move immediately?
A global timeline cannot answer that. Immediate threats require official instructions. Longer-term relocation should follow current local evidence, practical feasibility and personal dependencies. [S1]
How should I update this guide?
Check the original sources, revisions and local indicators periodically. Record new evidence, assumptions and dates. Keep disagreements and uncertainty visible.
-
Glossary: Plain-Language Guide to the Abbreviations and Terms
You do not need to memorize these terms. Use this glossary whenever an unfamiliar word or abbreviation interrupts your reading.Abbreviations, Organizations, and Model Names:
AI — Artificial intelligence: Computer systems that perform tasks such as recognizing patterns, generating text, or supporting decisions. Their answers can contain mistakes.
AIDS — Acquired immunodeficiency syndrome: The most advanced stage of infection with HIV, when damage to the immune system leaves a person vulnerable to serious illnesses.
AMOC — Atlantic Meridional Overturning Circulation: A large Atlantic Ocean circulation system that carries warm water northward near the surface and colder water southward at depth. It helps redistribute heat and influences weather. It is more extensive than the Gulf Stream alone.
AMR — Antimicrobial resistance: When microbes become harder to treat because medicines that previously controlled them become less effective. Bacterial AMR concerns bacteria; resistance can also affect other microbes.
AR6 — Sixth Assessment Report: The sixth major assessment of climate science produced by the Intergovernmental Panel on Climate Change. It consists of several reports, rather than one document.
ASIO — Australian Security Intelligence Organisation: Australia’s domestic security intelligence agency.
CFL — Climageddon Feedback Loop: Job One for Humanity’s name for a framework examining how climate processes and human vulnerabilities might interact and amplify harm. It is not the scientific name of one established physical feedback loop.
CI — Confidence interval: A statistical range calculated around an estimate. A 95% confidence interval comes from a method designed to capture the true value in about 95 out of 100 repeated applications, under its assumptions. It does not cover every possible source of error.
CO₂ — Carbon dioxide: A heat-trapping gas released by activities such as burning coal, oil, and natural gas. It is also exchanged naturally among the atmosphere, oceans, plants, and soils.
CO₂-equivalent, or CO₂e: A way to express the combined warming influence of different greenhouse gases in comparable carbon-dioxide units. The calculation depends on the purpose and time period being considered.
DDIS — Danish Defence Intelligence Service: Denmark’s foreign and military intelligence service.
Defra — Department for Environment, Food & Rural Affairs: A United Kingdom government department responsible for environmental, food, agricultural, and related matters.
DESA — Department of Economic and Social Affairs: A United Nations department whose work includes population estimates and projections.
DIA — Defense Intelligence Agency: A United States agency that provides military intelligence.
DICE — Dynamic Integrated Climate-Economy model: A computer model linking economic activity, emissions, climate change, and economic damage.
DICE-EMR — Dynamic Integrated Climate-Economy Model with an Endogenous Mortality Response: A version of DICE that also estimates how temperature changes affect deaths. “Endogenous” means that the death response is calculated within the model. [M7]
DOI — Digital Object Identifier: A permanent identifying code for a research paper or other publication, helping readers locate it even if its website address changes.
ECS — Equilibrium climate sensitivity: The eventual change in global average temperature following a doubling of atmospheric carbon dioxide, after the climate has largely adjusted. It does not, by itself, tell us when that warming will occur.
FEMA — Federal Emergency Management Agency: A United States agency supporting disaster preparedness, response, and recovery.
GBD — Global Burden of Disease: A research program estimating deaths, illnesses, and disability from different causes across countries.
GDP — Gross domestic product: The value of goods and services produced within an economy during a period. It does not measure every aspect of health, fairness, environmental quality, or well-being.
GRAM — Global Research on Antimicrobial Resistance: A research project estimating the worldwide health effects of infections that resist treatment. [M4]
HIV — Human immunodeficiency virus: A virus that damages the immune system. Effective treatment can control it and prevent progression to AIDS.
IEA — International Energy Agency: An international organization that analyzes energy systems, markets, policies, and related resources.
IFoA — Institute and Faculty of Actuaries: A professional body for specialists who assess financial risks, including risks affecting insurance and pensions.
IMF — International Monetary Fund: An international organization that analyzes economies and provides financial assistance to member countries.
IPCC — Intergovernmental Panel on Climate Change: The United Nations body that assesses published climate research. It evaluates existing scientific evidence rather than conducting all the research itself.
LMIC or LMICs — Low- and middle-income country or countries: Countries grouped by income levels. The label does not mean that every person in those countries has the same income or vulnerability.
MI5 — Security Service: The United Kingdom’s domestic security intelligence agency. “MI5” is its familiar historical name.
MI6 — Secret Intelligence Service: The United Kingdom’s foreign intelligence agency. “MI6” is its familiar historical name.
NCSC — National Cyber Security Centre: A United Kingdom organization that helps protect computer systems and essential services against digital threats.
NDC — Nationally Determined Contribution: A country’s climate-action commitment under the Paris Agreement. A commitment is not proof that its measures have been implemented.
NERC — North American Electric Reliability Corporation: An organization that develops electricity reliability standards and assesses risks to the large interconnected electricity system.
NIC — National Intelligence Council: A United States intelligence body that produces assessments of major international issues and possible future developments.
NIE — National Intelligence Estimate: An assessment expressing the United States intelligence community’s judgments about an important issue. It can include uncertainty and disagreement.
ODA — Official development assistance: Government-provided assistance intended to support development and welfare in eligible countries.
OECD — Organisation for Economic Co-operation and Development: An international organization that researches economic, environmental, social, and public-policy issues.
PDF — Portable Document Format: A common file format that preserves a document’s layout.
PEPFAR — President’s Emergency Plan for AIDS Relief: A United States program supporting the prevention and treatment of HIV and AIDS internationally.
PJM — PJM Interconnection: An organization coordinating electricity transmission and wholesale electricity markets across parts of the eastern and central United States. Its name comes from Pennsylvania, New Jersey, and Maryland, its original core states. [T18]
PM2.5 — Fine particulate matter: Airborne particles about 2.5 micrometres or less in diameter—far smaller than the width of a human hair. They can enter deep into the lungs and harm health.
RCP — Representative Concentration Pathway: A research scenario describing a possible future level of heat-trapping influence from greenhouse gases and other factors.
RCP8.5: A pathway reaching a high level of heat-trapping influence by 2100. It is used to study high-emissions outcomes; it is not a guaranteed future.
SSP — Shared Socioeconomic Pathway: A research storyline describing possible changes in population, development, technology, and society. It can be paired with a climate pathway.
SSP2-4.5: A scenario combining a middle-of-the-road development storyline with a specified level of heat-trapping influence by 2100. Its name does not mean a 4.5°C temperature increase.
UI — Uncertainty interval: A range showing uncertainty around an estimate, according to the study’s methods. It may include several sources of uncertainty, but not necessarily every possible error or unknown.
UK — United Kingdom: England, Scotland, Wales, and Northern Ireland.
UN — United Nations: An international organization through which member countries cooperate on issues including peace, development, health, and the environment.
UNAIDS — Joint United Nations Programme on HIV/AIDS: The United Nations program coordinating international efforts against HIV and AIDS.
UNEP — United Nations Environment Programme: The United Nations body focused on environmental issues.
UNU — United Nations University: A United Nations research and education institution studying major global challenges.
US or U.S. — United States: The United States of America.
USAID — United States Agency for International Development: The United States agency identified in the cited studies as providing international development and humanitarian assistance.
V-Dem — Varieties of Democracy: A research project measuring different features of democracy and political institutions.
WEF — World Economic Forum: An international organization that publishes reports and convenes discussions involving business, government, and other groups.
WHO — World Health Organization: The United Nations agency concerned with international public health.
WMO — World Meteorological Organization: The United Nations agency coordinating international work on weather, climate, and water.
World3: A computer model used in Limits to Growth research to explore interactions among population, industry, food, resources, and pollution. Its scenarios are not fixed appointments with global collapse.
Climate, Nature, and Interacting Systems
Adaptation: Changes that help people or ecosystems cope with conditions already occurring or expected—for example, safer buildings, heat protection, and more reliable water supplies.
Aerosols: Tiny particles or droplets in the air. Some cool the climate by reflecting sunlight or changing clouds; others contribute to warming. Many also harm health.
Aerosol masking or aerosol cooling: The cooling influence of some air pollution that partly offsets greenhouse-gas warming. Removing that pollution can reveal more warming, making simultaneous greenhouse-gas reductions important.
Atmosphere: The layer of gases surrounding Earth.
Biodiversity: The variety of living organisms, their differences, and the ecosystems they form.
Boreal forest: The broad belt of northern forests, found in places such as Canada and Russia.
Carbon removal: Taking carbon dioxide out of the atmosphere and storing it. Different methods vary in cost, effectiveness, durability, and environmental effects.
Carbon sink: Something that absorbs more carbon than it releases over a period, such as certain forests or parts of the ocean.
Cascade or cascading risk: A disruption that spreads through connected systems. For example, a power failure can interrupt water pumping and medical care.
Climate feedback: A process through which a climate change causes effects that then influence the original change.
Climate sensitivity: A measure of how much the climate responds to a change in its heat-trapping conditions. Different measures describe different timescales.
Climate threshold: A level at which an important change or risk is assessed. A threshold is not automatically a universal collapse point.
Climate time compression: In this article, the possibility that worsening or overlapping threats leave less practical time to prepare and recover. It is a planning phrase, not a universally defined scientific measurement.
Compound hazards or compound risks: Hazards occurring together or in a sequence that makes their combined effects more dangerous.
Concentration: The amount of a substance present in something else—for example, carbon dioxide in the atmosphere.
Critical services or critical infrastructure: Services and physical systems people depend on, including electricity, water, transport, communications, and health care.
Cumulative damage: Harm that builds up through repeated events, reducing the ability to withstand the next disruption.
Dependency: Something a person or system needs to function. A hospital, for example, depends on electricity, staff, medicines, water, and transport.
Domino effect: A sequence in which one disruption triggers another, as falling dominoes knock over the next pieces. The chain can stop when systems have enough protection or spare capacity.
Ecological disruption: A disturbance to living systems that changes their species, relationships, or ability to function.
Ecosystem: Living organisms and their physical surroundings interacting in a particular place or region.
Ecosystem collapse: A major loss or transformation of an ecosystem’s defining features and functions. It does not necessarily mean that every organism disappears.
Ecosystem services: Benefits people receive from nature, including pollination, water regulation, food production, and carbon storage.
Emissions: Substances released into the environment. In climate discussions, this usually means greenhouse gases released into the atmosphere.
Equilibrium: A condition in which a system has largely adjusted to a change. Climate equilibrium can take much longer to approach than the short periods used in household planning.
Feedback loop: A chain of effects that circles back to influence its starting point.
Global average temperature: An average across Earth’s land and ocean surfaces. It is not the temperature at every location, and regional warming can differ considerably.
Global warming: The long-term rise in Earth’s average surface temperature.
Greenhouse gas: A gas that traps some of the heat Earth would otherwise release to space.
Habitability: How suitable a place is for people to live, considering conditions such as heat, water, food, housing, and functioning services.
Heat-trapping influence: The effect of gases and other factors on how much heat Earth retains.
Internal climate migration: Movement within a country influenced by climate-related changes. It is not the same as crossing an international border.
Irreversibility: A change that cannot readily be undone on the timescale being discussed. Something effectively irreversible over a human lifetime might change over much longer periods.
Long-term or sustained warming: Warming measured over an extended period to reduce the influence of unusually hot or cold years. A single hot year is a different measurement.
Mangroves: Trees and shrubs growing along certain tropical and subtropical coasts. They provide habitat and help protect shorelines.
Methane: A greenhouse gas released from sources including fossil-fuel operations, livestock, waste, and wetlands.
Millennium; millennia: A millennium is one thousand years; millennia means multiple periods of one thousand years.
Mitigation: Action to reduce the cause or severity of a risk. Climate mitigation includes reducing greenhouse-gas emissions and increasing durable carbon removal.
Nonlinear response: A response that does not grow in simple proportion to the original change. A small additional pressure can sometimes produce a much larger consequence.
Novel entities: Human-created substances or materials, such as certain synthetic chemicals and plastics, that can introduce environmental risks.
Ocean heat uptake: The ocean’s absorption of heat. It affects the rate of surface warming and contributes to changes such as sea-level rise.
Overshoot: Exceeding a limit. Ecological overshoot means demands exceed nature’s ability to regenerate resources or absorb waste. Temperature overshoot means warming temporarily exceeds a target.
Paris Agreement: An international climate agreement whose temperature goals include holding warming well below 2°C and pursuing efforts to limit it to 1.5°C above pre-industrial levels.
Particulate matter: Small solid particles or liquid droplets suspended in air.
Permafrost: Ground that remains frozen for at least two consecutive years. Thawing can release carbon dioxide and methane from previously frozen material.
Planetary boundary: An assessed limit on human pressure affecting an important Earth-system process. Crossing a boundary increases concern; it does not automatically establish an immediate or irreversible collapse.
Planetary solvency: In the actuarial reports, the ability of Earth’s life-support systems to continue supporting human well-being and functioning societies.
Pollination: The transfer of pollen that enables many plants to produce seeds and fruit.
Pollinator: An animal, such as a bee, butterfly, bird, or bat, that helps transfer pollen.
Positive or reinforcing feedback: A process that strengthens an initial change. “Positive” means amplifying, not beneficial.
Negative or stabilizing feedback: A process that opposes or reduces an initial change. “Negative” does not mean harmful.
Pre-industrial baseline: A reference period before most modern industrial warming. Climate reports commonly use the 1850–1900 average as a practical reference.
Resilience: The ability to maintain essential functions, recover from disruption, and adapt to changing conditions.
Runaway warming: A phrase that needs qualification. In some organizational writing, it means warming that society has failed to control. It should not automatically be interpreted as a physically unstoppable, Venus-like runaway greenhouse.
Subsystem: A smaller connected part of a larger system—for example, ocean circulation within the climate system.
Synergy or synergistic effects: Interactions in which combined effects become stronger or different than the effects considered separately.
System: Connected parts that influence one another, such as an electricity network, an ecosystem, or a health service.
System collapse: A major loss of essential functions. A collapse can be local or regional without being worldwide.
Threat multiplier: Something that makes other dangers more severe or harder to manage.
Tipping element: A major component of a system that may undergo substantial change after crossing a threshold.
Tipping point: A threshold beyond which a change can become self-reinforcing or difficult to reverse. Timing, reversibility, and consequences vary among systems.
Transition duration: How long a change takes to unfold after it begins. Crossing a threshold and completing a transition are different events.
Transient climate response: Warming that develops while climate conditions are changing, before the system has fully adjusted. It differs from the eventual equilibrium response.
Whipsaw: Here, repeated or rapidly changing disruptions that strain a community’s ability to respond—for example, drought followed by damaging floods.
Death Estimates, Forecasts, and Risk Assessment
Actuary or actuarial: An actuary uses mathematics and statistics to assess financial risks. “Actuarial” describes that work.
Annual mortality: Deaths occurring during one year.
Annualized expected deaths: A probability-weighted average of possible deaths expressed per year. It does not mean that this number of deaths will occur every year.
Associated deaths: Deaths involving a specified condition or exposure, using a study’s definition. For antimicrobial resistance, this is a broader category than deaths attributable specifically to resistance.
Attributable deaths: Deaths estimated to result from a particular risk compared with a stated alternative situation.
Avoidable or averted deaths: Deaths a model estimates could be prevented under a different policy, treatment, or exposure scenario.
Baseline: The reference conditions or scenario used for comparison. A baseline is not necessarily the future that will occur.
Burden or health burden: The amount of death, illness, or disability associated with a problem.
Calibration: Adjusting a model using observations or other evidence so that its starting conditions or behavior better match what is known.
Causal relationship: A relationship in which one factor contributes to producing another, rather than merely occurring alongside it.
Central estimate: A model’s main representative estimate. It is not a guarantee or a complete description of uncertainty.
Conditional capability window: A period in which an organization could gain a particular ability if specified conditions hold. Military capability does not prove an intention to attack.
Conditional scenario: A possible future based on stated assumptions.
Confidence: The strength of support for a finding, considering the evidence and agreement among sources. “Low confidence” is not the same as a low probability that an event will happen.
Consensus: Broad agreement among relevant experts based on available evidence. It does not mean complete unanimity or certainty.
Counterfactual: The alternative situation used for comparison—for example, continued funding compared with funding cuts.
Cumulative mortality: Deaths added across a specified period, rather than deaths in a single year.
Double counting: Counting the same deaths or people more than once when combining estimates.
Event-relative timeframe: Time measured from an event, such as “two years after a war,” rather than from a known calendar date.
Excess or additional deaths: Deaths above the number expected under a reference situation.
Exposure: Contact with a hazard—for example, living in a flood-prone area or breathing polluted air.
Extrapolation: Extending an estimate beyond the places, periods, or conditions directly studied. Its reliability depends on the assumptions used.
Famine: An extreme shortage of food affecting a population and threatening survival. Formal humanitarian classifications have specific criteria.
Forecast: An estimate of future conditions based on evidence and assumptions.
Forecast horizon or planning horizon: The period a report examines. Its endpoint is not automatically a predicted disaster date.
Fragile or conflict-affected economy: An economy operating amid serious institutional weakness, conflict, or related instability.
Geoeconomic confrontation: The use of economic tools, such as trade restrictions or financial measures, in political or strategic rivalry.
Geoeconomic fragmentation: The division or weakening of economic connections among countries because of rivalry, restrictions, or security concerns.
Governance: How decisions are made, implemented, and held accountable.
Hazard: Something with the potential to cause harm, such as extreme heat, flooding, or an infectious disease.
Hypothesis: A proposed explanation that can be examined and tested against evidence.
Illustrative impact band or severity band: A hypothetical category showing how serious an outcome could be. It is not necessarily a forecast.
Indicator: A measurement used to track conditions or change—for example, water reliability or hospital capacity.
Integrated assessment model: A model connecting several areas, such as climate, economics, and health, to examine their combined effects.
Intelligence assessment: An evaluation produced by an intelligence organization. It can contain judgments, scenarios, and uncertainty rather than definite predictions.
Likelihood or probability: How likely an outcome is to occur. This is different from confidence in the evidence supporting a finding.
Model: A simplified representation of how something works, used to explore relationships or estimate outcomes.
Model assumptions: Choices about starting conditions, relationships, policies, or future behavior that shape a model’s results.
Model validation: Checking how well a model represents relevant real-world evidence. Validation does not guarantee that every future prediction will be correct.
Model vintage: When a model or estimate was produced. Older estimates may use different data, policies, and assumptions.
Mortality: Deaths in a population. A mortality count gives the number of deaths; a mortality rate relates deaths to the population and period studied.
Non-optimal temperature: Temperature above or below the level associated with the lowest death rate in a particular study. It includes harmful cold as well as heat.
Observational study: Research examining what occurred without researchers assigning the conditions. Observed relationships do not automatically establish cause and effect.
Onset: The beginning of a change or disruption.
Optimal modeled pathway: The pathway a model identifies as best under its chosen goals and assumptions. It is not a universally agreed definition of the best future.
Pandemic: An infectious-disease outbreak spreading widely across countries or continents.
Pathway or trajectory: A possible course of change over time.
Peer-reviewed research: Research examined by other specialists before publication. Peer review improves scrutiny but does not guarantee that every finding is correct.
Polycrisis: Several connected crises whose interactions can amplify their combined consequences.
Precautionary planning: Preparing for a potentially serious risk despite uncertainty about its exact timing or likelihood.
Premature deaths: Deaths occurring earlier than they would have under a comparison situation.
Primary source: The original research, report, or official evidence behind a claim.
Projection: An estimate of what could happen under specified assumptions.
Reasonable worst-case scenario: A severe but plausible scenario used to test preparedness. It is not necessarily the most likely outcome or the worst imaginable outcome.
Risk: The possibility of harm, considering hazards, exposure, and vulnerability.
Risk profile: A description of the types of risks affecting a place, population, or period.
Risk ranking: An ordering of risks by a survey or assessment. Rank alone does not tell us their numerical probabilities.
Scenario: A possible future constructed from specified assumptions. A scenario is not automatically a prediction.
Scenario range: The spread of results across different scenarios. It differs from a statistical confidence interval.
Service continuity: Keeping essential services functioning during disruption.
Shock: A disruption that puts a system under pressure.
Soot: Dark particles produced by burning. Large amounts entering the upper atmosphere can reduce sunlight reaching Earth’s surface.
Stress test: An examination of how a system would perform under difficult conditions.
Supply chain: The connected organizations, materials, transport routes, and processes that deliver a product or service.
Surveillance: Monitoring. Disease surveillance tracks health threats; digital surveillance monitors people or activities.
Systems-risk framework: An approach examining connected risks and how failures may spread among systems.
Undernutrition: Insufficient energy or nutrients to maintain health and normal growth.
Uncertainty: Limits on what is known about an estimate, its assumptions, or future conditions.
Vulnerability: How susceptible someone or something is to harm, including limitations on the ability to prepare, respond, or recover.
Numbers, Symbols, and Reference Labels
°C — Degrees Celsius: A temperature unit. In this article, “2°C of warming” means an increase above the stated reference average—not a local outdoor temperature of 2°C.
% — Percent: Parts out of 100. A 25% increase means one-quarter more than the comparison amount.
m — Million: One million is 1,000,000. In this article’s death and population figures, “14.1m” means 14.1 million.
bn — Billion: One billion is 1,000,000,000—one thousand million.
> — Greater than: “>2bn” means more than two billion.
≥ — Greater than or equal to: “≥25%” means at least 25%.
~ or approximately: About. The number is not intended as exact.
A range, such as 2037–2109: The two endpoints of an interval. The meaning depends on the source; it is not a guarantee that an event will happen within those years.
Per year: A yearly amount. It must not be confused with a total accumulated across several years.
Et al.: A shortened Latin expression meaning “and others.” It indicates that a publication has additional authors.
- Method, coverage, and credibility checks
This review followed the 15 categories in the Universe Institute’s framework [P1], checked the original actuarial PDF, revisited dated intelligence/security and scientific assessments, and searched primary institutional and peer-reviewed mortality sources. It prioritizes global or multinational mortality estimates relevant to the crisis framework. Individual-country projections and observed historical death counts are not presented as global forecasts.
The scope is broad but not exhaustive: no public-source search can guarantee every credible prediction worldwide, including unpublished, classified, paywalled or untranslated material. Several source websites restrict automated full-text retrieval; this review used original abstracts, institutional releases or accessible original PDFs where possible. The actuarial interpretation is based on the original report, rather than treating a video’s characterization as the supporting evidence.
- Corrected: actuarial illustrative bands were not promoted to forecasts. [A1]
- Corrected: AMOC values use the 2025 correction; conflicting model evidence remains visible. [T16] [T17]
- Corrected: UNAIDS’ initial 6.3m calculation is distinguished from the revised 4.2m scenario. [M12]
- Qualified: older air-pollution and resistance scenarios are identified by vintage and scope. [M5] [M6] [M16]
- Excluded from numerical totals: unsupported global-collapse dates, unattributed billions, and invented mortality for unquantified crisis categories.
- Separated: scientific findings, intelligence judgments, source scenarios, and this page’s editorial planning recommendations.
No source’s reputation makes a scenario certain. No uncertainty makes inaction automatically reasonable. The practical task is to examine evidence carefully, preserve options, and reduce avoidable exposure.
Return to contents | Return to the risk timeline
Linked bibliography and supporting material
Codes beginning I identify intelligence/government-security material; T identify timeline sources; M identify mortality evidence; A identify actuarial reports; P is the project framework; S is preparedness guidance; V identifies the video discussed in the actuarial-claim section; F identifies climate scenarios and interacting-systems sources; R identifies precautionary-principle sources. Each code above links here. External titles open the original source. Access reviewed October 8, 2026; availability can change.
-
- [P1] Universe Institute. The Global Polycrisis (updated August 28, 2026). Source of the 15-category framework. Its own probability judgments and dates are not automatically the findings of its cited institutions. Back to timeline
- [A1] Institute and Faculty of Actuaries & University of Exeter. Planetary Solvency: Finding our balance with nature (January 2025). Appendix I, Figure 12, printed page 32: illustrative impact matrix. Read the qualification above the table. The mortality bands are not modeled forecasts. Back to timeline
- [A2] IFoA & Anglia Ruskin University. Planetary Solvency: Tipping into the wild unknown (April 30, 2026). Newer report announcement: food-system fragility, nature loss, prices, and financial transmission. Back to timeline
- [A3] Institute and Faculty of Actuaries & University of Exeter. Parasol Lost: Recovery plan needed (January 14, 2026). Acceleration and aerosol cooling; report warns of 2°C before 2050 without action. Does not validate the illustrative billion-death bands as forecasts. Back to timeline
- [A4] IFoA. Planetary Solvency launch and summary (January 16, 2025). Conditional warning of 50% GDP loss in 2070–2090; risk framework and dashboard. Back to timeline
- [V1] YouTube video discussing the actuarial mortality claim (video ID: 2eRXo_tOafc). Context for the claim examined here; the supporting actuarial evidence is the original report in [A1]. Back to timeline
- [I1] U.S. National Intelligence Council. Global Trends 2040: Demographics and human development (2021). Urban services and demographic pressures; 2040 is an outlook horizon. Back to timeline
- [I2] U.S. NIC. Global Trends 2040: Environment (2021). Pollution, environmental disruption, displacement, and habitability risks. Back to timeline
- [I3] U.S. NIC. Global Trends 2040: Economics (2021). Debt, constraints, and fragmentation; not a dated global financial-collapse forecast. Back to timeline
- [I4] U.S. NIC. Global Trends 2040: State dynamics (2021). Institutional pressure, political breakdown, and democratic/authoritarian vulnerabilities. Back to timeline
- [I5] U.S. NIC. Global Trends 2040: Societal dynamics (2021). Fragmentation and insecurity. Back to timeline
- [I6] U.S. intelligence community. Global Water Security (2012; principally drafted by DIA). 2030 water-demand/supply gap reproduces an external model; it is an older conditional projection. Back to timeline
- [I7] U.S. intelligence community. Climate Change and International Responses Increasing Challenges to US National Security Through 2040 (2021). Climate baseline and post-2030 water/migration conflict risks; not a societal-collapse timetable. Back to timeline
- [I8] U.S. NIC. Global Trends 2040: Tragedy and Mobilization (2021). Early-2030s food catastrophe in one alternative future; illustrative scenario, not prediction. Back to timeline
- [I9] Danish Defence Intelligence Service. Assessment of the threat from Russia (September 24, 2026). Reaffirms six-month/two-year/five-year conditional capability windows; also addresses near-term limited attacks. Back to timeline
- [I10] ASIO. Annual Threat Assessment 2025, hosted by Australia’s Office of National Intelligence. Security outlook to 2030, including AI-enabled deception and trust risks. Back to timeline
- [I11] UK National Cyber Security Centre. Impact of AI on cyber threat from now to 2027 (2025). Conditional critical-infrastructure vulnerability and attacker-capability assessment. Back to timeline
- [I12] UK Defra. National security assessment on global ecosystems (published January 20, 2026; accessible version added February 2, 2026). Reasonable worst-case assessment using intelligence uncertainty methods. Public attribution is Defra; do not relabel it an independently published MI5/MI6 report. Back to timeline
- [T1] UN DESA. World Population Prospects 2024: launch summary. Mid-2080s peak around 10.3 billion; demographic projection, not catastrophe forecast. Back to timeline
- [T2] IEA. Global Critical Minerals Outlook 2026: Executive summary. Projected 25% copper supply gap in 2035 based on the project pipeline; policy and supply can change it. Back to timeline
- [T3] UN University. Interconnected Disaster Risks 2023: Risk tipping points. Groundwater access, insurance, glaciers, heat, extinctions and debris. Saudi groundwater example is a specific system threshold. Back to timeline
- [T4] Persson et al. Outside the Safe Operating Space of the Planetary Boundary for Novel Entities (2022). Boundary exceedance is not itself a dated collapse event. Back to timeline
- [T5] OECD. Global Plastics Outlook: Policy Scenarios to 2060 (2022). Baseline leakage growth; not a mortality estimate. Back to timeline
- [T6] Global Tipping Points Report 2025: Earth system tipping points. Warm-water coral central thermal threshold crossed; local timing and remaining reefs vary. Back to timeline
- [T7] Trisos, Merow & Pigot. The projected timing of abrupt ecological disruption from climate change. Nature (2020). High-emissions RCP8.5 scenario: tropical oceans before 2030, forests and higher latitudes by 2050. Back to timeline
- [T8] Flores et al. Critical transitions in the Amazon forest system. Nature (2024). 10–47% of forest potentially exposed to compounding disturbance by 2050; not whole-Amazon collapse probability. Back to timeline
- [T9] World Bank. Groundswell report release (September 13, 2021). Up to 216 million internal climate migrants in six regions by 2050 in a pessimistic scenario; hotspots from 2030. Back to timeline
- [T10] UN University. Five facts on disasters and insurability (2024). 520,940 Australian homes potentially uninsurable by 2030; regional market-function risk. Back to timeline
- [T11] IMF. Fiscal Monitor, April 2026: Fiscal Policy under Pressure. Global public debt projected at 100% of GDP in 2029; this is not a universal failure threshold. Back to timeline
- [T12] World Bank. Fragile and conflict-affected situations: report release (June 27, 2025). 435 million people in extreme poverty in these economies by 2030; no dated government-collapse forecast. Back to timeline
- [T13] V-Dem. Democracy Report 2026: release and findings. Describes democratic backsliding during 2025; not a future universal collapse date. Back to timeline
- [T14] International AI Safety Report 2026. Capability scenarios through 2030 and risks; no established global mortality/collapse timetable. Back to timeline
- [T15] IPCC. AR6 Synthesis Report, longer report (2023). Best estimate of long-term 1.5°C in first half of 2030s for most assessed pathways. Back to timeline
- [T16] Ditlevsen & Ditlevsen. Warning of a forthcoming collapse of the AMOC (2023, corrected 2025). Corrected central estimate 2065 and conditional 95% interval 2037–2109; see linked author correction of August 21, 2025, DOI 10.1038/s41467-025-63201-y. Back to timeline
- [T17] Baker et al. Continued Atlantic overturning circulation even under climate extremes. Nature (2025). 34-model analysis finds twenty-first-century collapse unlikely; disagreement matters. Back to timeline
- [T18] NERC. 2025 Long-Term Reliability Assessment (published 2026). PJM high resource-adequacy risk from 2029 under assessed assumptions; regional shortfall risk, not inevitable permanent collapse. Back to timeline
- [T19] World Economic Forum. Global Risks Report 2026: digest. Geoeconomic confrontation ranks first on two-year horizon to 2028; survey, not collapse model. Back to timeline
- [T20] IMF. Geoeconomic Fragmentation and the Future of Multilateralism (2023). Long-run losses in fragmentation scenarios, without a collapse year. Back to timeline
- [T21] Herrington. Update to limits to growth: Comparing the World3 model with empirical data (online 2020; journal issue 2021). Two closest-fitting scenarios diverge between collapse and more limited decline. Neither establishes a precise global-collapse date or death toll. Back to timeline
- [T22] WMO. New report suggests more global temperature records ahead (May 28, 2026). 2026–2030 annual and five-year probabilities; less than 1% chance of a single year above 2°C in that window. Back to timeline
- [T23] UNEP. Emissions Gap Report 2025. Current-policy century warming around 2.8°C; full NDC implementation 2.3–2.5°C. These are pathways, not certain outcomes. Back to timeline
- [M1] WHO. Climate change and health fact sheet (2023), summarizing its 2014 quantitative assessment. Approximately 250,000 extra annual deaths in 2030–2050 from selected causes only; not all climate pathways. Back to timeline
- [M2] WEF & Oliver Wyman. Quantifying the Impact of Climate Change on Human Health (2024). 14.5 million additional deaths by 2050 in the report’s modeled climate-health scenario; a cumulative estimate. Original PDF is linked on this publication page. Back to timeline
- [M3] Springmann et al. Global and regional health effects of future food production under climate change. Lancet (2016). 529,000 additional deaths in 2050; 95% CI 314,000–736,000. Food availability, dietary composition, and bodyweight pathways. Back to timeline
- [M4] GBD 2021 Antimicrobial Resistance Collaborators. Global burden of bacterial AMR 1990–2021: forecasts to 2050. Lancet (2024). 39.1 million attributable and 169 million associated deaths during 2025–2050; associated includes attributable. Annual 2050: 1.91 million attributable / 8.22 million associated. Back to timeline
- [M5] OECD. Health at a Glance 2025: Environment and health. Reiterates older projection of 6–9 million outdoor-air-pollution deaths annually by 2060. Compare newer baseline in M6. Back to timeline
- [M6] OECD. Environmental Outlook on the Triple Planetary Crisis (2025). Newer projection has PM2.5 premature deaths declining through 2050 across regions. Different scope/horizon from the older PM2.5-plus-ozone forecast. Back to timeline
- [M7] Bressler. The mortality cost of carbon. Nature Communications (2021). DICE-EMR central baseline: 83 million cumulative excess deaths over 2020–2100; modeled optimal path about 9 million. Not a total of every climate effect. Back to timeline
- [M8] Pozzer et al. Atmospheric health burden across the century and the accelerating impact of temperature compared to pollution. Nature Communications (2024). Roughly 30 million deaths/year late in the century (abstract interval 12–53 million) for non-optimal temperature plus PM2.5 under SSP2-4.5. Total attributable burden, not wholly extra anthropogenic-climate deaths. Back to timeline
- [M9] Cavalcanti et al. Evaluating two decades of USAID interventions and projecting defunding effects to 2030. Lancet (2025). Approximately 14.1 million excess deaths by 2030, UI 8.48–19.66 million, including 4.54 million under-five deaths, conditional on modeled cuts. Back to timeline
- [M10] Ferreira da Silva et al. Impact of humanitarian and development assistance and mortality consequences of defunding to 2030. Lancet Global Health (2026). 93 LMICs; mild cuts 9.4 million excess deaths (UI 6.2–12.6); severe cuts 22.6 million (16.3–29.3). Under-five subsets 2.5 million and 5.4 million, respectively. Back to timeline
- [M11] ten Brink et al. Impact of an international HIV funding crisis on infections and mortality in LMICs. Lancet HIV (2025). 0.77–2.93 million additional HIV-related deaths in 2025–2030 under alternative funding/mitigation scenarios. Scenario span, not confidence interval. Back to timeline
- [M12] UNAIDS. Estimating the potential impact of HIV response disruptions (April 2025). Updated Goals model: 4.2 million additional AIDS-related deaths in 2025–2029 if PEPFAR-supported programs permanently discontinue. Earlier rapid estimate 6.3 million is superseded in this note. Back to timeline
- [M13] Xia et al. Global food insecurity and famine from reduced production due to nuclear-war soot. Nature Food (2022). Selected India–Pakistan scenario >2 billion deaths; largest U.S.–Russia scenario >5 billion. Event-relative famine estimates, not predictions of war occurrence. Back to timeline
- [M14] Fan, Jamison & Summers. Pandemic risk: how large are the expected losses? Bulletin of the WHO (2018). Expected annualized influenza-pandemic mortality about 720,000, using 2015-era calibration; no calendar date. Discusses 21–33 million deaths in a 1918-type event. Back to timeline
- [M15] Smith et al. Effects of decreases of animal pollinators on human nutrition and global health. Lancet (2015). Hypothetical 50% service loss: 700,000 additional annual deaths; complete loss: 1.42 million (1.38–1.48 million). No calendar forecast. Back to timeline
- [M16] Review on Antimicrobial Resistance, chaired by Jim O’Neill. Background and commissioned scenario (2014–2016). Older uncontrolled-resistance scenario of 10 million annual deaths by 2050; broader scope and different method from M4, not an additional burden to add. Back to timeline
- [S1] FEMA. Are You Ready? An In-Depth Guide to Citizen Preparedness. Emergency supplies, household plans, alerts, evacuation and local hazard assessment. Back to timeline
Additional Bibliography: Climate Time-Compression Watch
[F1] Universe Institute. 2026 Climate Change Temperature and Timeframe Forecast, June 2026. Section 7 supplies the detailed scenario schedule. Its assumptions and qualifications are discussed above. Back to timeline
[F2] Job One for Humanity. The Climageddon Feedback Loop: A Systems-Risk Theory of Interacting Climate Tipping Points, Feedbacks, Nonlinear Reactions, and Human Cascades, updated August 24, 2026. Distinguishes established science, supported risks, organizational hypotheses, and precautionary planning estimates. Back to timeline
[F3] Wunderling and colleagues. Climate tipping point interactions and cascades: a review. Earth System Dynamics, 15, 41–74, 2024. DOI: 10.5194/esd-15-41-2024. Reviews interactions, possible cascades, timescales, model limitations, and uncertainty. Back to timeline
[F4] IPCC, Sixth Assessment Report, Working Group I. Chapter 7: The Earth’s Energy Budget, Climate Feedbacks, and Climate Sensitivity, 2021. See especially sections 7.3–7.5 and Box 7.1 for equilibrium and transient sensitivity, assessed ranges, aerosol forcing, and the time-dependent climate response. Back to timeline
[F5] World Meteorological Organization. State of the Global Climate 2025, published March 23, 2026. Reports observed 2025 warming of approximately 1.43°C above the 1850–1900 average. Back to timeline
[F6] Armstrong McKay and colleagues. Exceeding 1.5°C global warming could trigger multiple climate tipping points. Science, 377, eabn7950, 2022. DOI: 10.1126/science.abn7950. Assesses tipping thresholds and uncertainty; crossing a threshold and completing a transition occur on different timescales. Back to timeline
[F7] Lawrence and colleagues. Global polycrisis: The causal mechanisms of crisis entanglement. Global Sustainability, 2024. Explains shared pressures, domino effects, and inter-system feedbacks; conceptual causal framework, not a numerical global-collapse or mortality forecast. Back to timeline
[F8] United Nations Environment Programme. Limiting Overshoot: official report announcement, September 2, 2026. Describes an optimistic peak around 1.8°C and Guterres’s call to minimize the magnitude and duration of overshoot. This is not a universal irreversible-damage threshold. Back to timeline
Additional Bibliography:
[R1] United Nations. Rio Declaration on Environment and Development, Principle 15, 1992. Establishes that incomplete scientific certainty should not justify postponing cost-effective measures against threats of serious or irreversible environmental damage. Back to timeline
[R2] European Commission. Communication on the Precautionary Principle, 2000. Explains evidence-based precaution, proportionate measures, examination of action and inaction, and review as scientific knowledge improves. Back to timeline
Related pages and next steps
Continue with the Universe Institute’s full polycrisis framework. Use the sources in this guide to keep the distinction between organizational scenarios and independently published forecasts visible.
For practical climate preparation and adaptation, visit Job One for Humanity. For values, community resilience and constructive participation, visit Way of the Universe. These organizational resources are separate from the primary-source mortality evidence above.
See. Prepare. Protect. Build. We can be serious about the risks without becoming certain about what the evidence cannot yet tell us. Protecting life starts with honest information and useful action—and occasionally with closing the doom-scrolling tab long enough to do the useful action.
Return to the top | Contents | Risk timeline
Do you like this page?