EPICFURY Power Index · 2026

Compare military resources, with the formula showing.

A model ranking within a sample of 50 countries, using public resource estimates. Adjust the weights to explore how the ranking changes. Scores describe this dataset and formula; they do not predict combat outcomes.

15 comparable metrics selected. Scores are rounded for display; ranks use full scores. Equal full scores share a rank. Excluded because observations are missing in the comparison pool: Military satellites.

Metric definitions and source limits
Active personnel
Published active personnel estimate, thousands; separate from reserves and paramilitary.
Reserves
Published reserve personnel estimate, thousands; national definitions differ.
Defence budget estimate
GFP published defence-budget estimate in USD billions. Estimates may use different fiscal bases; these are not SIPRI expenditure figures or audited actual spending.
Military aircraft
Total fixed-wing and rotorcraft inventory in the source; excludes UAVs. Overlaps fighters and helicopters, so these categories are not independent.
Fighters / interceptors
Source category: fighters/interceptors, not every multirole aircraft.
Tanks (all source classes)
Source tanks category includes main battle, medium and light tanks.
Artillery (towed+SP)
Sum of published towed and self-propelled artillery stocks; excludes MLRS.
Submarines
All source submarine classes including small/midget craft; hull counts do not indicate equivalent capability.
Aircraft carriers
Source aircraft-carrier category only; helicopter/assault carriers are separate and excluded. Listed hulls do not imply readiness.
Destroyers + frigates
Sum of published destroyer and frigate hulls; source classifications vary.
Estimated nuclear stockpile
SIPRI estimated national military stockpile, January 2026; excludes retired warheads and foreign-hosted warheads. Not a GFP input.
Military satellites
Comparable military-satellite counts are not established; null means unknown, and this metric is excluded for the entire sample.
Population
Population estimate, millions, published in the source review; not necessarily a 2026 census.
Oil production
Source oil-production estimate scaled to thousands of barrels per day; observation year may predate the 2026 review.

Head-to-head matchups

Compare countries →

Comparisons use the weights selected above.

🇮🇳 vs 🇵🇰 India – Pakistan🇮🇳 vs 🇨🇳 India – China🇺🇸 vs 🇨🇳 United States – China🇺🇸 vs 🇷🇺 United States – Russia🇮🇱 vs 🇮🇷 Israel – Iran🇷🇺 vs 🇺🇦 Russia – Ukraine🇨🇳 vs 🇹🇼 China – Taiwan🇮🇳 vs 🇧🇩 India – Bangladesh🇵🇰 vs 🇮🇷 Pakistan – Iran🇰🇵 vs 🇰🇷 North Korea – South Korea🇹🇷 vs 🇬🇷 Türkiye – Greece🇨🇳 vs 🇯🇵 China – Japan

Methodology. Each metric is square-root-compressed, min-max normalised across the 50 sampled countries, and combined as a weighted sum; the sample leader is rescaled to 100. Only metrics observed for the whole sample enter the score. Unknown is distinct from zero. Aircraft totals can include fighters and helicopters, so their weights deliberately overlap. Budgets and inventories may use different source years and definitions; they do not measure readiness, alliances or training. Snapshot: 2026.

Related: Air Defence Shield of India · WARROOM II

About the model ranking

Which country ranks highest in this model?

United States leads under the default weights in this 50-country sample. This is a model of selected resource quantities, not a global military effectiveness ranking or a prediction of conflict outcomes. Changing the weights can change the ordering.

Is this the Global Firepower ranking?

No. Epicfury uses its own published formula and adjustable weights. A source may contribute observations, but its original ranking and methodology are separate. Our rank applies only to countries and comparable metrics in this sample.

How is the Firepower Index calculated?

Eligible metrics are square-root compressed, min-max normalised across the sample, multiplied by their selected weights and summed; the sample leader is rescaled to 100. Metrics missing observations for any sampled country are excluded, rather than treating unknown as zero. Equal scores share a rank; all-zero weights produce no ranking. The dataset currently defines 16 metrics.

Can I compare any two countries?

Yes. Select any two different countries from the 50-country dataset. The comparison retains your selected weights and shows raw observations, source years where available and missing-data warnings. It does not establish which country would win a conflict.