Composite model — major great-power war by 2050
~45%
Cumulative · range 35–55%
Modal high-risk window
2027–30
Peak annual hazard ≈ 2028
Dalio "hot world war" est.
~50%
Within ~10yr (up from 35%)
Doomsday Clock 2026
85s
Closest to midnight ever
Full nuclear "extinction" risk
~0.1%
/century (Ord) — very low
Read the headline carefully. "Major great-power war" (direct conflict between major powers, e.g. over Taiwan) is plausibly a coin-flip-ish risk over 25 years. A full global thermonuclear WW3 is much rarer (~10–18% by mid-century on these models). Human extinction from nuclear war is rarer still (~0.1%/century). These are three very different questions stacked inside "WW3."
The Model — When?
The model takes a historical base rate of system-level great-power war (~1%/year — roughly one per century: Napoleonic → WWI → WWII), then applies a "Big-Cycle multiplier" (3–6×) for today's elevated indicators: late-stage debt, polarisation, deglobalisation, militarisation and an active US–China rivalry. That produces a hazard rate that rises into a 2027–2030 danger window, then tapers as the cycle resolves.
Cumulative Probability Over Time
Annual Hazard Rate (% per year)
Why 2027–2030? Three independent clocks point there: (1) Dalio's debt + cycle timing flags ~2028; (2) the much-discussed "Davidson window" for Taiwan invasion capability is 2027; (3) US ICBM/New START arms-control lapsed in Feb 2026, removing guardrails. The model treats this as the peak — not a guarantee anything happens.
Ray Dalio's Big-Cycle Gauges — Current Readings
Dalio's framework scores five driving forces. Today, the US-centred system reads "red" on most — the same configuration seen in the 1930s. He places the US in Stage 5, transitioning toward Stage 6 of the six-stage internal-order cycle.
Debt / financial cycle (US ~85% through)
85
External conflict (US–China: 4 of 5 war types active)
80
Internal conflict (wealth gap + polarisation)
75
Deglobalisation / mercantilism
70
Reserve-currency erosion (USD <60% of FX reserves)
60
1New order
2Build systems
3Peace & prosperity
4Excess & debt
5Bad finances & conflict ◄ US now
6Civil war / war
With 60–80% of Dalio's internal "red flags" present, his framework implies roughly a 1-in-6 chance of severe internal conflict this decade — a separate risk that can amplify external war.
How Forecasters Compare
The spread is huge — and that spread is the honest answer. Pessimists (Dalio) and crowd markets (Metaculus) cluster at 30–50%; disciplined superforecasters go far lower for nuclear specifically. The truth is genuinely unknown.
| Source | Question | Probability | Horizon |
|---|
| Ray Dalio | Hot world war incl. major powers | ~50% | ~10 yr |
| Metaculus (crowd) | World War Three before 2050 | 35% | by 2050 |
| Metaculus | ≥1 nuclear detonation in war | 35% | by 2050 |
| Good Judgment | Taiwan full blockade | 29% | by 2030 |
| Good Judgment | Taiwan full invasion | 20% | by 2030 |
| Metaculus | Global thermonuclear war | ~11% | by 2070 |
| Domain experts (ONN) | Nuclear catastrophe (10M+ deaths) | 5% | by 2045 |
| Superforecasters (ONN) | Nuclear catastrophe (10M+ deaths) | 1% | by 2045 |
| Toby Ord (The Precipice) | Existential catastrophe from nuclear war | 0.1% | /century |
Why The Risk Is Rising — The Drivers You Named
You're right that mercantilism and extreme politics are surging. The hard data confirms it — every structural driver of great-power war is flashing.
Trade restrictions (2025)
2,500+
≈5× the 2015 level (Global Trade Alert)
Global military spending
$2.89T
11th straight yr up · 2.5% of GDP
Countries autocratising
44
41% of world population (V-Dem)
Trade-to-GDP ratio
56.8%
Down from 62.8% peak (2022)
US average tariff
~90yr
Highest in roughly 90 years
Global Military Spending ($ Trillion)
Democracy in Retreat — Countries Autocratising
Where It Hurts Most
| # | Region / Zone | Why | Impact |
|---|
| 1 | Taiwan Strait / Asia-Pacific | 90% of advanced + 99% of AI chips; world's busiest lane; US–China direct clash | $10.6T · 9.6% global GDP |
| 2 | Eastern Europe (Russia–Ukraine–NATO) | Active war in year 4; nuclear doctrine; Article 5 trigger risk | Nuclear escalation |
| 3 | Middle East (Iran–Israel) | Strait of Hormuz (34% of seaborne crude); energy & food price shock | Global oil spike |
| 4 | US ICBM belt + major cities | Silo fields in Montana / N. Dakota / Wyoming (~450 warheads); NYC, LA, DC | Primary nuclear targets |
| 5 | Global financial hubs | NYC & London (targets); Hong Kong & Shanghai (Taiwan spillover) | System paralysis |
| 6 | Korean Peninsula | DPRK nuclear-test resumption; Seoul in artillery range | Regional nuclear |
Northern Hemisphere industrial/military cores absorb the direct blows; even a "regional" nuclear exchange injects soot that cuts Northern Hemisphere crop yields catastrophically (Penn State 2025: up to −80% in a full exchange).
Where Is Safer
The Xia et al. 2022 Nature Food nuclear-winter study and the Global Peace Index converge on the same answer: remote, neutral, food-self-sufficient, Southern-Hemisphere nations. These "island refuges" could sustain >2,200 kcal/person/day even in a severe (150 Tg soot) scenario.
SAFEST · 01
🇳🇿 New Zealand
Isolation + food surplus; GPI #3. Top academic "island refuge."
SAFEST · 02
🇮🇸 Iceland
GPI #1; geothermal energy autonomy; remote, neutral.
SAFEST · 03
🇦🇺 Australia / Tasmania
Southern Hemisphere; major food exporter; island refuge.
SAFEST · 04
🇦🇷 Argentina
Southern-cone breadbasket; beef + grain self-sufficiency.
SAFEST · 05
🇨🇭 Switzerland
Armed neutrality; fallout shelters for ~100% of population.
SAFEST · 06
🇮🇪 Ireland
GPI #2; neutral; Atlantic isolation, no strategic targets.
SAFEST · 07
🇨🇱 Chile
Southern Hemisphere; agriculture + renewable energy.
SAFEST · 08
🇫🇯 Fiji
Remote Pacific; negligible strategic value; fishing + forests.
The common thread: distance from NATO/Russia/China target maps, ability to feed yourself without global trade, political neutrality, and Southern-Hemisphere geography (circumpolar winds limit fallout crossing the equator). Iceland and New Zealand score on every axis.
Bottom Line
So — is WW3 inevitable, and when? The honest synthesis: a major great-power war is a serious, elevated risk (~35–55% over 25 years), with the danger most concentrated in 2027–2030 around Taiwan. But "inevitable" overstates it — most of probability mass is still on no direct great-power war, and a full nuclear WW3 remains a low-double-digit tail risk, with civilisation-ending outcomes lower still. The structural drivers you identified (mercantilism +400% trade restrictions, record militarisation, 44 nations autocratising) are real and rising — which is exactly why the model's near-term hazard sits 4–6× above the historical baseline. The deterrents (nuclear MAD, economic interdependence, the sheer cost) are also real, which is why catastrophe is not the base case.
Methodology & sources: Composite model = historical base rate of system-level great-power war (~1%/yr) × Big-Cycle multiplier (3–6×) derived from current indicator readings, calibrated against published forecasts. Frameworks & data: Ray Dalio, Principles for Dealing with the Changing World Order + 2024–26 LinkedIn/Fortune essays (Great Powers Index, six-stage cycle, Stage 5→6, ~50% hot-war estimate); Metaculus prediction markets (WW3 by 2050 = 35%, nuclear detonation by 2050 = 35%, thermonuclear by 2070 ≈ 11%); Good Judgment superforecasters (Taiwan blockade 29% / invasion 20% by 2030); Open Nuclear Network expert vs superforecaster survey (5% vs 1% by 2045); Toby Ord, The Precipice (0.1%/century extinction); Bulletin of the Atomic Scientists Doomsday Clock 2026 (85 seconds); SIPRI military expenditure 2025 ($2.887T); Global Trade Alert (2,500+ restrictions); V-Dem Democracy Report 2026 (44 autocratising); Xia et al. 2022 Nature Food + Penn State 2025 nuclear-winter crop models; Global Peace Index 2025; CFR Conflict Tracker / Preventive Priorities 2026; St. Louis Fed & Bloomberg Economics Taiwan cost ($10.6T). This is analytical scenario modelling for educational purposes — not a prediction, and not safety, financial, or relocation advice.
