Center for Strategic Studies of Eurasia

    Commentary · Sep 11, 2026

    What the 2025 military spending figures do not tell you

    SIPRI’s 2025 release is the eleventh consecutive annual rise in world military spending. Most South Caucasus figures quoted from it are not in it; the database has them, with caveats.

    Security & DefenceSouth CaucasusEurope & Russia
    A scenic aerial view of Tbilisi featuring the historic Narikala Fortress at sunset.
    ArtHouse Studio / Pexels
    Contents

    Key takeaways

    • World military expenditure reached $2,887 billion in 2025, up 2.9 per cent in real terms and the eleventh consecutive annual rise.
    • European spending rose 14 per cent to $864 billion; US spending fell 7.5 per cent to $954 billion.
    • The military burden, spending as a share of GDP, reached 2.5 per cent globally, the highest since 2009.
    • In SIPRI’s database Azerbaijan spent $4,939 million in 2025 (6.5 per cent of GDP) and Armenia $1,725 million (6.1 per cent); Armenia’s figure excludes military pensions.

    SIPRI published its military expenditure figures for 2025 on 27 April 2026. The headline numbers are unambiguous and worth having in front of anyone writing about security in this region.

    The figures

    World military expenditure reached $2,887 billion in 2025, an increase of 2.9 per cent in real terms over 2024 and the eleventh consecutive year of growth. The global military burden, spending as a share of GDP, stood at 2.5 per cent, its highest level since 2009.

    The distribution moved more than the total. European spending rose 14 per cent to $864 billion, making Europe the main driver of the global increase. Spending in Asia and Oceania rose 8.1 per cent. United States spending fell 7.5 per cent to $954 billion. The three largest spenders, the United States, China and Russia, together accounted for $1,480 billion, or 51 per cent of the world total.

    The 2.9 per cent rise was the smallest since 2021, and one country held it down: excluding the United States, world spending grew by 9.2 per cent. SIPRI attributes most of the US fall to the absence in 2025 of the supplemental appropriations that had funded military aid to Ukraine. Its programme director, Nan Tian, called the decline “likely to be short-lived”, noting that spending approved by Congress for 2026 exceeds $1 trillion.

    What the release does and does not contain

    Coverage of the release has carried country-level figures for Armenia, Azerbaijan and Georgia, generally as a share of GDP and generally without saying which year or which series they come from. Some appear to be 2025 outturn, some 2026 budget appropriations, and some are secondary estimates from aggregator sites rather than from SIPRI at all.

    The release itself says less than that coverage implies, but not nothing. The fact sheet’s map of military burden labels Azerbaijan at 6.5 per cent of GDP and Armenia at 6.1 per cent, sixth and seventh in the world, and its text records that Azerbaijan’s burden rose by 1.4 percentage points in 2025, the second-largest increase anywhere after Ukraine. Georgia does not appear, and nor do money figures for any of the three. Those are in the database.

    We are not reproducing the circulating figures. This is not caution for its own sake. In this region a defence-spending ratio is rarely quoted as an economic statistic; it is quoted as evidence about somebody’s intentions. A figure carrying that weight has to be traceable to a series, a year and a definition. The numbers below come from the April 2026 edition of the database, and we say which series each belongs to.

    What the database shows for the three states

    In current US dollars, SIPRI puts 2025 military expenditure at $4,939 million for Azerbaijan, $1,725 million for Armenia and $658 million for Georgia. As a share of GDP the figures are 6.5, 6.1 and 1.7 per cent. As a share of general government spending they are 17.7, 19.9 and 5.8 per cent. On our calculation from the constant 2024 dollar series, real spending rose in 2025 by about 24 per cent in Azerbaijan, 16 per cent in Armenia and 18 per cent in Georgia. Over the decade from 2016 the real increase was roughly 146 per cent for Armenia, 117 per cent for Azerbaijan and 52 per cent for Georgia.

    The country notes matter as much as the numbers. Armenia’s and Georgia’s figures exclude military pensions, and SIPRI estimates that Armenia’s would be 15 to 20 per cent higher with them. Georgia’s figures are the adopted budget, not actual expenditure. The three states are therefore not measured identically even inside one database. Adjusted for pensions on SIPRI’s own estimate, Armenia’s burden would be above Azerbaijan’s, not below it.

    Why the 2026 budgets complicate the picture

    The two largest spenders are now moving in opposite directions. The draft 2026 budget Armenia’s government approved in September 2025 cut defence appropriations by about 15 per cent, to 563 billion drams from 664.7 billion. Azerbaijan’s draft 2026 budget allocated 8.71 billion manats to defence and national security, about 3.8 per cent more in nominal terms than the 8.40 billion SIPRI records for 2025. Georgia’s 2026 draft raised the defence ministry’s budget by 160 million lari to 1.9 billion, mainly to pay for a 10 per cent salary rise.

    Yerevan’s own account of the cut is worth having alongside the number. In August 2025 Prime Minister Nikol Pashinyan said the 2026 budget would probably bring “no substantial increase on defense expenditures or no increase at all”, and Finance Minister Vahe Hovannisian said the government meant to lower defence’s share of public spending; critics, Azatutyun noted, argued that Baku still had preconditions for signing the peace treaty. In parliament that September Pashinyan called 2025 the exception, a “sharp and disproportionate increase of 115 billion drams”, part of which went to repaying future debts and obligations, including in defence, some originally due in 2030 or 2035. If so, part of Armenia’s 2025 figure reflects early repayment rather than a larger force, and the 2026 cut partly reverses a one-off.

    Scope changes the Azerbaijani comparison most. The 8.71 billion manat function includes 3.03 billion for the defence forces, 0.48 billion for national security, 0.54 billion for border services and 4.66 billion for other defence and security activities, of which 4.5 billion is earmarked for special defence-purpose projects. A comparison built on the defence-forces line alone would show roughly a third of the spending that one built on the whole function shows. SIPRI’s 2025 figure is on the whole-function scale. Its country note says it has found no separate special defence projects allocation since 2016; the 2026 draft names one inside the function. Whether that is new money or a relabelled line is not yet clear.

    How to use the database instead

    The database is open and gives, for each country and year, spending in constant and current dollars, in local currency, as a share of GDP and of government spending, and per capita. Its country notes matter more than usual here: what counts as military expenditure differs between states, and paramilitary and border forces are treated differently in different national accounts.

    Three rules make the numbers usable. Cite the series, not the ratio alone. Distinguish an approved budget from outturn, because in this region the gap between them has run wide in both directions. Compare in constant prices, because nominal growth in a currency that has moved against the dollar tells you about the currency.

    The case for the ratio, and its limits

    There is a fair objection. SIPRI itself calls the share of GDP “the simplest measure of the relative economic cost of the military”, and it travels across currencies and years better than any dollar figure. Azerbaijan and Armenia ranking sixth and seventh in the world by burden is a real finding: the rest of the top ten are Ukraine, Algeria, Israel, Russia, Saudi Arabia, Oman, Kuwait and Jordan.

    The objection holds as far as it goes. The ratio measures cost, not capability or intent. It moves with GDP as much as with spending. And definitions elsewhere are drifting apart. NATO’s new target lets 1.5 per cent of GDP go to loosely defined defence- and security-related spending, and SIPRI notes that NATO’s 2025 estimate for Canada was $5.0 billion above its own, with the difference unexplained. As more governments report to targets, the distance between a headline percentage and a comparable one will widen.

    The relevant finding for the neighbourhood

    The 14 per cent European increase is the number that changes conditions here, and it does so indirectly. It tightens the market for the equipment, ammunition and training capacity that smaller states in the South Caucasus and Central Asia buy, and it competes for the attention of the same suppliers. Procurement timelines lengthen for everyone outside the priority queue. That effect will show up in delivery schedules long before it shows up in anyone’s budget line.

    SIPRI’s arms-transfer data shows the size of the queue. Imports of major arms by European states more than trebled between 2016–20 and 2021–25, and at the end of 2025 twelve European states had 466 F-35 combat aircraft on order or preselected. For buyers the size of those in the South Caucasus, the binding constraint over the next few years is as likely to be a delivery slot as a budget. The next data points worth having are Armenia’s and Azerbaijan’s 2026 outturn, and SIPRI’s April 2027 edition, which will replace the 2025 figures quoted here.

    Sources

    Related publications