In 2025 the world spent $2,887 billion on its armed forces. That is the highest figure SIPRI has ever recorded, the eleventh consecutive annual increase, and it works out at $352 for every person alive.

Below is what that money bought, counted as carefully as open sources allow: aircraft, ships, submarines, carriers and people, country by country.

Two warnings before the tables. First, every number here is an estimate from a named source with a stated date — they are not interchangeable, and mixing editions produces nonsense. Second, and more importantly, the definitions do more work than the numbers. A “destroyer” is whatever a navy decides to call a destroyer. Whether a country has 291 warships or 461 depends entirely on which rulebook you open. We have flagged those traps as we go, because they are where almost every viral military comparison falls apart.

Part One: The Money

Spending is the cleanest of all these measures, because SIPRI has been doing it consistently since 1949 and publishes its methodology. It is also the best single predictor of everything else on this page, with a lag of about a decade.

<iframe src="https://migflug.com/afterburner/wp-content/uploads/sites/4/2026/08/military-spending-share-2005-vs-2025-pie.html" style="width:100%;border:0" height="700" scrolling="no"></iframe>

Burden, Not Just Bulk

Absolute spending flatters large economies. Military burden — spending as a share of GDP — shows how much a country is actually straining.

Ukraine has now spent four consecutive years with the highest military burden on earth — 40% of GDP and 63% of all government spending. Russia’s military share of government spending reached 20% in 2025, the highest SIPRI has ever recorded for the country.

The Long View: 1, 5, 10 and 20 Years

A single year tells you very little. Trends tell you what a country has decided about the next decade.

That last line is the whole story of the century so far. Chinese and Russian spending have both grown roughly 350% in twenty years. American spending has grown 8.5%. The United States still spends more than the next seven countries combined, but the gap has been closing for two decades, and 2025 was the first year in some time that Washington’s spending actually fell.

NATO Versus Everyone

Alliance arithmetic changes the picture completely. NATO’s 32 members between them spent $1,581 billion in 2025 — 55% of all military spending on earth.

European NATO spending rose faster in 2025 than in any year since 1953. Germany passed 2% of GDP for the first time since 1990; Spain for the first time since 1994, with a 50% single-year jump. Poland now carries the heaviest burden in the alliance at 4.5%.

Part Two: Airpower

There are 52,231 active military aircraft in the world, spread across 161 nations. The number fell by 411 last year, mostly because the entire Syrian air force — 414 aircraft — was deleted from the count.

That tanker figure is the one professionals notice. Combat aircraft can be bought; the ability to move them across an ocean and keep them there cannot. Three quarters of the world’s aerial refuelling capacity belongs to one country, and it is the single clearest measure of who can actually fight far from home.

Fighters: Who Actually Has Them

Total fleet size is a blunt instrument. It counts trainers, transports and helicopters alongside the aircraft that would actually contest an air battle. Strip it back to combat aircraft and the order changes immediately — China moves to the top.

The column that matters is the third one. Roughly 1,300 of the world’s 52,231 military aircraft are fifth-generation stealth fighters — about 2.5% of the total — and they are distributed more unequally than anything else on this page.

A definitional note first, because this label is abused more than any other in aviation. We count the F-22, le F-35, le J-20, le J-35 and — with reservations explained below — the Su-57. Aircraft still in development, such as Turkey’s KAAN and South Korea’s KF-21, are not counted. Nor are 4.5-generation types like the F-15EX, Rafale or late-tranche Typhon, which are in several respects more capable than the Su-57 despite the generational label.

The Line That Puts It in Perspective

Read the table again and note where Russia sits. The Netherlands — a country of 18 million people with a single fighter type — fields more fifth-generation aircraft than the Russian Federation.

Russia’s difficulty is not design ambition but production. Sanctions have kept some of the Su-57’s advertised systems out of the aircraft entirely, and serial deliveries have run at single digits a year. Algeria is expected to become the first export customer, with a reported order for twelve.

The opposite problem exists in the two categories nobody exports. The F-22 line closed in 2011 after 187 aircraft, and the United States has never sold one abroad. China applies the same ban to the J-20. Both countries keep their best air-superiority aircraft for themselves and export the multirole one — the F-35 for America, and probably the J-35 for China, which is openly touted for export.

Which leaves one asymmetry that matters more than any individual airframe. The F-35 has been ordered by around twenty countries. Every one of those fleets is interoperable, shares a common logistics chain, and can pass targeting data to the others. China’s stealth fleet is larger than anyone’s but Washington’s, and it flies alone.

The counting caveat applies here too, as it does everywhere on this page. FlightGlobal’s directory records 177 active F-22s and 883 F-35s worldwide, because it excludes aircraft in storage or undergoing upgrade. Lockheed Martin has delivered more than 1,300. Both numbers are correct; they answer different questions.

The Chinese Number That Jumped

China’s combat fleet rose 14.6% in a single edition, from 1,583 to 1,814. Almost none of that was new construction. It was a counting revision.

China is now building somewhere between 220 and 280 new fighters a year, on independent estimates. And the total above still excludes the Shenyang J-35 entirely — so the real figure is higher than 3,529.

There is a matching caution on the Russian side, in the opposite direction. FlightGlobal counts 1,559 Russian combat aircraft. Justin Bronk of RUSI, counting only genuinely serviceable modern types and discounting stored MiG-29, Su-24 et Su-25 airframes, arrives at roughly 545 to 560 fighters plus the bombers — a gap of about 2.4 times. Both numbers are defensible. They are answering different questions.

Bombers: A Club of Three

Only three countries operate strategic bombers. Everyone else’s long-range strike is done with fighters, cruise missiles or not at all.

The Russian caveat matters. Ukraine’s June 2025 drone raid on Russian airbases reportedly destroyed around a fifth of Moscow’s operational strategic bombers on the ground — aircraft that no longer exist but which still appear in inventories.

Part Three: Navies

This is where counting becomes genuinely treacherous, and where most published comparisons are simply wrong.

Aircraft Carriers: Six Countries, Twenty Ships

A fourth Chinese carrier is under construction at Dalian. Satellite imagery from May 2026 shows a hull about 286 metres long with two large compartments carrying the hallmarks of shielded reactor containment.

The Pentagon’s stated expectation, published in December 2025: the PLAN aims to produce six carriers by 2035, for a total of nine.

The carriers currently under construction around the world, including China’s Type 004 and France’s nuclear-powered replacement for the Charles de Gaulle.

Every aircraft carrier in service in 2026, country by country.

Warships: The 291-or-461 Problem

Ask how many ships the US Navy has and the honest answer is another question: counted how?

The US counting rule is a Navy instruction, not a law. It admits commissioned warships and certain support vessels, and excludes patrol craft, unmanned vessels, tugs, barges, hospital ships and most sealift. On that rule the US Navy had 293 battle-force ships in October 2025. Count every hull of every kind and the same navy is 461.

The same trap applies to the famous China comparison. The Pentagon’s last published figure had the PLAN at over 370 battle-force ships against the US Navy’s 291 — and contrary to a widespread claim, that Chinese figure explicitly excludes around 60 missile-armed patrol catamarans. It is not padded with speedboats.

Worth noting for anyone quoting the Pentagon: the December 2025 edition of the China Military Power Report dropped the battle-force number entirely. The last published figure dates from the 2024 edition.

Why “Destroyer” Means Nothing

The column above is close to meaningless without this warning, so here it is plainly. Ship classes are national branding decisions, not engineering categories.

China’s Type 055 displaces 12,000 to 13,000 tonnes. The Pentagon calls it a cruiser; China calls it a destroyer. China’s “cruiser” count in American tables jumped from zero to eight purely because of US naming convention.

Germany has not rated a ship as a destroyer since 2003 — as deliberate policy, to symbolically underline military restraint — despite operating 7,000-tonne frigates. France and Italy operate the identical Horizon-class hull with the identical missile system: France calls it a frigate, Italy calls it a destroyer. Britain’s Type 26 frigate outweighs its Type 45 destroyer. Italy’s 6,000-tonne, 16-cell PPA is officially an offshore patrol vessel; Indonesia bought two and commissioned them as frigates. Iran calls 1,500-tonne ships destroyers; US naval intelligence classifies them as corvettes.

Compare hull counts across navies without reading the definitions and you are comparing marketing departments.

Tonnage, Missile Cells and the Better Questions

Because hull counts are so unreliable, analysts prefer tonnage. Here honesty requires an admission: no current aggregate tonnage figure for either navy exists from an authoritative public source. The comparison everyone repeats rests on a 2018 Brookings estimate — the US Navy at roughly twice China’s tonnage, a gap of about three million tons — and a 2024 CSIS study. Anyone quoting a 2026 tonnage figure is inventing it.

What CSIS did establish in 2024 is more interesting than the raw number.

Better metrics exist. Vertical launch cells — the actual magazine depth of a fleet — put the US Navy at around 9,900 against the PLAN’s 4,200. That still sounds comfortable until you see the trend: in 2004 the ratio was 222 to one.

That last line is the one that keeps American planners awake. The ONI slide put Chinese shipyard capacity at over 23 million tons against under 100,000 for the United States — a ratio of more than 230 to one. The US Navy confirmed the slide was genuine while cautioning it was drawn from public sources and was never meant as a deep analysis of Chinese commercial shipbuilding.

Carrier size compared — the United States, China and the United Kingdom side by side.

Part Four: People

Manpower has its own version of the definitions problem: authorised strength, funded billets and actual serving headcount are three different numbers, and countries quote whichever flatters them.

Military Spending Per Soldier

Put the two halves of this article together — the money and the manpower — and a third number falls out: what a country actually spends per person in uniform. It is one of the more revealing figures in defence economics, because it separates mass from capability.

There is a complication, and it is the reason this comparison is usually done badly. Converting budgets at market exchange rates assumes a dollar buys the same amount of defence everywhere. It does not. A Russian conscript, an Indian mechanic and an American staff sergeant are paid in different currencies at wildly different local costs, while a fighter jet is priced globally in dollars.

The economist Peter Robertson built a measure for exactly this: military purchasing power parity, which prices personnel, operations and equipment separately rather than applying a single blunt GDP-PPP factor. Both views are below.

The headline is the American one. The United States spends about $742,000 per active service member per year — and because the US is the base country for the index, that figure is the same on either measure. Nobody else is close. Even after purchasing-power adjustment more than doubles Russia’s effective budget, Russia still spends barely half as much per soldier as America does.

The adjustment matters enormously for everyone else, though. At market rates India appears to spend $67,000 per soldier; in military-PPP terms it is $245,000, a 3.7-fold uplift and the largest in the table. China moves from $165,000 to $301,000, Indonesia from $35,000 to $118,000. These are not accounting tricks — they are an attempt to answer what the money actually buys at home.

Read the two columns together and the shape of each force appears. China fields the largest army on earth and spends less per head than Russia does. The United States fields a force barely two thirds the size of China’s and spends two and a half times as much on each person in it. That difference — not the headline budget, and not the headcount — is most of what people mean when they talk about a quality gap.

One honest caveat. Six of the fifteen largest armed forces cannot appear here at all: North Korea, Pakistan, Iran, Vietnam, Egypt and Thailand publish too little for a military-PPP estimate to be built. Their absence is itself informative.

Demographics are quietly becoming the binding constraint. South Korea has gone from 563,000 active personnel in 2019 to about 450,000 in 2025, and projects 290,000 by 2030 — not through policy, but through birth rates. India loses around 60,000 people a year to attrition. The US Navy is roughly 20,000 sailors short at sea.

What None of These Numbers Tell You

Every table above measures inventory. War is decided by things inventories cannot capture, and it is worth being explicit about the gaps.

Readiness. FlightGlobal counts aircraft not recorded as in storage or undergoing upgrade. That is an accounting definition, not a statement that they can fly tomorrow. Of Russia’s nine Akula-class attack submarines, five are in refit and one is in reserve — leaving perhaps three available from a paper fleet of nine.

Munitions. Nobody publishes credible stockpile figures. A fleet of aircraft with two weeks of precision weapons is a two-week air force, and several NATO members discovered exactly this after 2022.

Logistics and training. Tankers, transports, spare parts and flying hours decide whether a force can be used. The tanker column is the closest proxy on this page, and it is brutally lopsided.

The counting problem cuts both ways. The most instructive example of 2026 came from the Gulf. After the campaign against Iran, US Central Command reported that a fleet which had eleven ships in the Gulf of Oman had none. Yet analysts assessed that 80 to 90% of Iran’s small boats and minelaying craft survived — and those are the vessels that actually threaten shipping in the Strait of Hormuz. The ships that appeared in the tables were destroyed. The ships that never appeared in any table are the ones that still matter.

The Shape of It

Strip out the detail and four things are happening at once.

The United States retains an enormous, and in some categories overwhelming, lead — a quarter of the world’s military aircraft, three quarters of its tankers, eleven carriers, two thirds of the world’s nuclear submarines and a third of all military spending. That lead is real and it is not close to disappearing.

China is closing specific gaps very quickly, and has already passed the US on hull count, on warships over 1,000 tonnes, on fleet age, on auxiliaries and on shipbuilding capacity by a factor of over two hundred. Its 31st consecutive annual spending increase is the longest unbroken run in SIPRI’s entire database.

Europe is rearming faster than at any point since the early 1950s, from a very low base, and mostly by writing cheques rather than by fielding new units — the equipment those budgets buy will not arrive for years.

And the constraint everyone is running into is not money. It is people, shipyards and production lines. Japan cannot fill its ranks. Korea is shrinking by demography. The United States can fund carriers faster than it can build them. In that world, the twenty-year spending trend at the top of this article is a better guide to 2040 than any inventory table below it.

Sources: SIPRI, Trends in World Military Expenditure 2025 (April 2026) and the SIPRI Military Expenditure Database; FlightGlobal World Air Forces 2026 with data partner Cirium; US Congressional Research Service reports RL32665 and RL33153 by Ronald O’Rourke; US Department of Defense, Military and Security Developments Involving the People’s Republic of China (2024 and 2025 editions); CSIS, Unpacking China’s Naval Buildup (2024) and Hidden Reach (2026); RUSI; IISS Military Balance; US Naval Institute News; Naval News; NATO; national defence ministries.

A note on method: every figure carries its source and year. Where sources disagree, both are given rather than averaged. Where no authoritative figure exists — current fleet tonnage, most national drone inventories — we have said so instead of estimating.