AI Models

AI capex vs. the US defense budget

USD billions, 2026.

← Back to AI Capex vs. US Defense Spending

About this data

The five largest US cloud/AI providers have collectively committed to roughly $660–690B of capex in 2026 — a ~77% jump on 2025 — against a US national-defense budget of about $1 trillion. So AI infrastructure spending is now on the order of two-thirds of defense spending and rising far faster; the widely-repeated claim that tech will out-spend the military is a forward-looking projection, not yet a 2026 fact.

AI capex vs. the US defense budget

USD billions, 2026. Combined hyperscaler AI capex is now roughly two-thirds of US national-defense spending — and closing fast.

View data & sources →

Data table

AI capex vs. the US defense budget — capex_vs_defense data table (AI Capex vs. US Defense Spending)
label series usd_bn source_ref value_basis
Big Tech AI capex 2026 (5 hyperscalers) capex_vs_defense 675 futurum-capex Microsoft, Alphabet, Amazon, Meta and Oracle committed ~$660–690B capex in 2026 (midpoint ~$675B), +77% YoY from ~$410B
US defense budget, FY2026 capex_vs_defense 1000 armscontrol-defense US FY2026 national defense ~$1T ($838.7B defense appropriations + $150B via reconciliation); topline cited near $1.05T

Methodology & sources

Last updated: Jul 17, 2026

Methodology

Two source-backed charts: 2026 AI capex by company ($B) and combined hyperscaler AI capex versus the US FY2026 defense budget ($B). Per-company figures (Amazon ~$200B, Alphabet ~$185B, Meta ~$125B, Microsoft ~$120B) trace to the cited capex analysis. The aggregate ~$675B is the midpoint of the $660–690B range for the five largest US cloud/AI providers (incl. Oracle), up ~77% YoY from ~$410B. Defense ~$1T reflects $838.7B appropriations + $150B via reconciliation (topline cited near $1.05T). CAVEAT: company capex is largely but not exclusively AI. The popular claim that tech will out-spend the military is a forward projection — in 2026 defense is still the larger number. Re-verified 2026-07-17.

Sources

Comparisons are informative, not definitive. See each source for definitions and limits.

Continue researching

How relevant was this information?