takeoff.watch

US military spending reaches 10% of GDP

Will SIPRI report US military expenditure at 10% of GDP or more for any year?

Chance by Q4 2031

1%

Cumulative probability

0%25%50%75%100%
  • Latest forecast

Forecast history

Forecast made on 11 Oct 2026

Latest of 4 forecasts · Methodology v2.1

Model reasoning

Forecast made 11 Oct 2026 · Methodology v2.1

The research institute's latest figure puts American military spending at about 3.1% of the economy. Reaching 10% would take three to four times today's spending, last seen in the 1950s around the Korean War. Even the largest budget now proposed would be under 5%. The small remaining chance, about 1% by 2031, comes almost entirely from a long war with China; such a war breaking out would most change this forecast.

Where things stand

  • The latest figure. SIPRI's April 2026 database (workbook v1.2) puts US military spending in 2025 at 3.12% of GDP. That is $954bn, down 7.5% in real terms, and down from 3.43% in 2024.

  • The historical record. I downloaded SIPRI's United States share-of-GDP series. It was last at or above 10% in 1958 (10.07%). The Vietnam War peak was 9.42% in 1967, and since 1995 the series has stayed below 5%.

  • The only jumps of this size were war mobilisations.

    • Korean War: 5.08% in 1950, 10.17% in 1951 and 13.86% in 1952, a 2.7-fold rise in two years.
    • World War II was larger still.

  • How large a rise is needed. Dividing $954bn by 3.123% gives an implied GDP of about $30.6tn. I assumed 4.5% nominal growth. On that basis the 10% threshold is about:

    • $3.3tn in fiscal 2027;
    • $3.8tn in fiscal 2030.

    That is 3.5 to 4 times the 2025 level. Even the Pentagon's full $1.5tn request for fiscal 2027 would be about 4.5% of GDP.

  • The budget is not moving that way. The $1.5tn request faces an "uphill battle" in Congress (Military.com, 12 Aug 2026). Fiscal 2027 began on a stopgap funding bill at fiscal 2026 levels, running through 11 December (Breaking Defense, 4 Sep 2026).

  • The Iran war is too small to matter. It has cost about $38bn through July 2026, plus $2–3bn a month (Roll Call, quoting the Congressional Budget Office, 15 Sep 2026). That is around 0.1% of GDP.

Release calendar

SIPRI updates the database once a year in late April (27 April 2026; 28 April 2025). That gives these chances:

SIPRI release Fiscal year reported Chance of 10%
April 2027 2026 Effectively none (about 3.3%)
April 2028 2027 Small
April 2029 2028 Small
April 2030 2029 Small
April 2031 2030 Small

Fiscal 2031 data would not appear until April 2032, after the window closes. So the risk is concentrated in the second-quarter releases from 2028 to 2031. I left a sliver of risk in other quarters for wartime special estimates or corrections.

Reference class and model

Reference class. The US has gone from about 3% of GDP to 10% only through great-power war mobilisation.

Main path to YES: a long US–China war.

  • Metaculus put the chance of a US–China war before 2035 at 12% (as of 27 May 2026, cited by vandermerwe.com, July 2026). That is about 1.5% a year.
  • Polymarket prices a Chinese invasion of Taiwan by the end of 2027 at about 10.5%, and by the end of 2026 at about 2%. I discount these for the usual overpricing of long shots.
  • Near-term risk looks lower than that average. The People's Liberation Army purges limit readiness, Trump and Xi held a summit on 22–25 September 2026, and Beijing has an interest in Taiwan's 2028 election.

My scenario model. These numbers are my judgement, not data.

  • War hazard: 1% a year until early 2028, then 1.5% a year.
  • Chance a war is protracted: 35%.
  • If protracted, chance spending reaches 10% by the second full fiscal year after the war starts: 55%.
  • Short war followed by heavy rearmament: about 3%.

That gives about 0.67% through the April 2031 release, weighted towards fiscal 2029 and 2030. A wartime fall in GDP would help push the ratio over 10%.

Other paths add about 0.3–0.5%: a full NATO–Russia war, a very large defence-run AI programme, a SIPRI data error, or an out-of-cycle publication.

Total: about 1.2% by the end of 2031.

Strongest case on each side

For higher. The world is volatile: an active US war with Iran, China's 2027 military readiness target, Russia–NATO tensions, and a 10.5% market price on an invasion by 2027. A wartime GDP collapse would also shrink the denominator.

For lower. Peacetime politics are pushing the other way. Congress is balking at $1.5tn and federal debt is above 100% of GDP. Even a US–China war might be short or might escalate to nuclear use. Payments lag appropriations, and outside World War II the US has never tripled its military share within two years.

On balance I stay near 1%.

Sources

Resolution criteria

Opened 16 Sep 2026Not yet resolved

Question. By the last day of each quarter, has SIPRI reported United States military expenditure at 10% of GDP or more for any year?

Definitions.

  • SIPRI: the Stockholm International Peace Research Institute.
  • Military expenditure as a share of GDP: SIPRI’s United States series in the SIPRI Military Expenditure Database, expressed as a percentage of GDP, using SIPRI’s own expenditure definition and GDP denominator, exactly as SIPRI publishes it. SIPRI states that all of its figures for the United States are for the US financial year (1 October of the previous year to 30 September of the stated year) rather than the calendar year; the series counts on that basis, and no conversion to calendar years is required or applied. This is not the share of government expenditure or of worldwide military expenditure.
  • Eligible year: a year in SIPRI’s series, as SIPRI labels it, whose underlying period has ended and ended after the opening date, under the Observation period convention. SIPRI’s 2026 figure for the United States covers 1 October 2025 to 30 September 2026, so the earliest eligible year is 2026.
  • Reported value: SIPRI’s numerical estimate for a completed eligible year, including a provisional or estimated figure. Use the most precise value SIPRI publishes in the relevant release, including the underlying downloadable data, without further rounding. If only a rounded percentage is published, use that value as printed; 10% or 10.0% qualifies. A range qualifies only if its lower bound is at least 10%.

Resolves YES if, within the Resolution window and by the quarter-end, SIPRI publishes a qualifying reported value of at least 10% for at least one eligible year. A revision that first raises an eligible year to the threshold counts on the revision’s publication date. A subsequent downward revision does not undo a YES, under the Resolution window convention. No attribution to AI is required.

Resolves NO if no qualifying publication exists by the quarter-end. Exclude forecasts, budget proposals, appropriations, spending targets, ratios from any source other than SIPRI (including US government fiscal-year ratios), and figures solely for periods ending before the opening date. Figures published by NATO, US government agencies, the World Bank or other third parties do not independently qualify, even if they cite SIPRI. Do not calculate a substitute ratio from expenditure and GDP series.

Resolving source. SIPRI’s annually updated Military Expenditure Database, specifically the United States military-expenditure-as-a-share-of-GDP series in its downloadable workbook, or a SIPRI-authored publication explicitly reporting that series, such as the annual Trends in World Military Expenditure fact sheet or SIPRI Yearbook. The current database covers 1949–2025; the April 2026 fact sheet reports the series in Table 1, “Spending as a share of GDP (%)”. If simultaneous SIPRI sources differ, the database value takes precedence. SIPRI publications only, a narrower source requirement than Public confirmation (A2). Apply Resolution date (A3), the Resolution window and the Observation period conventions.

Download all forecasts (JSON)