# Meta's $145B AI spending dwarfs most military budgets Tech giant's 2026 capex surpasses all but three nations' defense spending - **Source:** [Fortune](https://fortune.com) - **Published:** 2026-09-24 - **Topic:** [Business](https://brifing.si/c/business) - **Original article:** - **On Brifing:** https://brifing.si/s/990efafe-1083-4d33-b886-64ec1c69d1e4 ## Unprecedented tech spending scale Meta plans $145 billion capital expenditure in 2026, a 101% increase from $72.2 billion in 2025. This exceeds military budgets of all nations except the U.S., China, and Russia. The spending surge reflects Meta's aggressive infrastructure buildout for AI systems, including data centers and cloud capacity. Zuckerberg confirmed AI will consume significant compute resources for model training and product development. ## Historical spending trajectory Meta's capex has grown exponentially from $28.1 billion in 2023 to $72.22 billion in 2025, consistently surpassing its own forecasts. Property/equipment spending followed similar growth from $27.05 billion to $69.69 billion. The acceleration coincides with Meta's strategic pivot to AI, requiring massive infrastructure investments. SEC filings show actual expenditures repeatedly exceeded initial projections by billions. ## Enterprise AI monetization strategy Zuckerberg outlined multiple revenue streams from AI investments: APIs, business agents, direct compute sales, and services for large customers. These complement existing ad business improvements from AI. The company expects AI to simultaneously enhance current products while creating new product categories. Personal agents are positioned as foundational for future revenue growth beyond social media. ## Industry-wide spending arms race Meta joins Alphabet, Amazon, and Microsoft in projected $730 billion collective 2026 AI spending, up from $600 billion estimates. This reflects intense competition for AI infrastructure supremacy. Meta's $169 billion total expense projection includes talent acquisition and compensation costs. The spending surge comes despite uncertain ROI timelines, indicating priority on market positioning over short-term profitability. ## Infrastructure buildout specifics Meta is constructing massive data centers while securing third-party computing capacity to meet AI demands. The infrastructure supports both internal AI model training and external customer services. Compute allocation will balance core business needs with new enterprise offerings. Physical infrastructure represents the tangible manifestation of Meta's AI ambitions at unprecedented scale. ## Strategic timing considerations Zuckerberg called 2026 pivotal for achieving 'superintelligence' capabilities. The accelerated spending suggests Meta believes current AI development pace requires immediate, massive investment to maintain competitiveness. The company is betting that establishing infrastructure leadership now will determine long-term industry positioning. This aligns with broader tech sector urgency around AI infrastructure. ## Comparative military spending context Meta's $145B capex surpasses Germany's $113B and India's $92B military budgets. Only the U.S. ($954B), China ($336B), and Russia ($190B) spend more on defense. This comparison underscores how tech giants now rival nation-states in resource allocation. The data comes from Stockholm International Peace Research Institute's 2025 military expenditure rankings. ## Core business transformation AI investments already enhance Meta's apps, ad targeting, and product development velocity. The company views AI as simultaneously optimizing current operations while enabling entirely new business lines. This dual approach aims to future-proof Meta's revenue streams beyond social media advertising. Zuckerberg's 'superintelligence' framing suggests aspirations beyond commercial applications into fundamental technological capabilities. ## Sources covering this story - https://fortune.com/2026/09/24/meta-145-billion-capex-military-budgets/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._