Big Tech AI Capex: Q1 2026 Disclosures
Four companies — Microsoft, Alphabet (Google), Amazon, and Meta — collectively disclosed approximately $67B in capital expenditure in Q1 2026 earnings calls, with AI infrastructure named as the primary driver in all four cases. This is a 62% increase from Q1 2025 and represents the fastest quarterly acceleration in technology infrastructure spending since cloud buildout in 2014–2016.
| Company | Q1 2026 AI Capex Disclosed in earnings |
YoY Change vs Q1 2025 |
Primary AI Use Stated in earnings call |
|---|---|---|---|
| Microsoft | $21.4B | +78% | Azure AI infrastructure, Copilot training |
| Alphabet (Google) | $17.2B | +71% | TPU clusters, Gemini model training |
| Amazon (AWS) | $18.1B | +54% | Trainium/Inferentia chips, Bedrock infrastructure |
| Meta | $10.5B | +88% | Llama training, AI recommendation systems |
Source: Company Q1 2026 earnings disclosures (10-Q filings, earnings call transcripts). Capex figures represent total capex with AI infrastructure as primary stated driver, not AI-isolated figures. Accessed May 2026.
AI Investment Efficiency by Industry
Spend without output is waste. The efficiency question is: for every dollar of AI investment, what measurable output does each industry produce? We measured efficiency across three dimensions: innovation output (USPTO AI patent filings), talent output (net new AI-specialist roles per $1B invested), and financial output (% reporting measurable ROI).
| Industry | Est. AI Spend 2025 sector total |
AI Patents per $1B USPTO 2025 |
AI Jobs per $1B BLS 2025 |
% Reporting ROI McKinsey Q1 2026 |
Efficiency Rank |
|---|---|---|---|---|---|
| Financial Services | $42B | 14.2 | 8.7 | 38% | #1 |
| Healthcare | $21B | 11.8 | 6.1 | 31% | #2 |
| Technology | $198B | 9.4 | 12.3 | 41% | #3 |
| Retail & Consumer | $28B | 6.7 | 4.2 | 24% | #4 |
| Energy & Utilities | $14B | 5.1 | 2.8 | 19% | #5 |
| Manufacturing | $31B | 4.4 | 2.3 | 17% | #6 |
AI spend estimates derived from IDC AI market sizing by sector (2025). USPTO filings: AI-tagged CPC codes (G06N) by assignee sector, 2025 calendar year. BLS AI-specialist jobs: net new positions in AI-related titles per $1B sector spend. ROI: McKinsey Q1 2026 survey, "enterprises reporting measurable ROI from generative AI."
Technology spends 4.7x more on AI than financial services ($198B vs. $42B) yet produces fewer patents and less ROI per dollar. This efficiency gap is likely explained by the nature of AI investment: technology companies spend heavily on foundational model training (large sunk costs with deferred returns) while financial services invests in targeted applications (fraud detection, trading algorithms, credit scoring) with faster and more measurable ROI cycles. The implication: sector-average ROI numbers are dominated by the technology sector's investment pattern and understate results available to industry-specific deployers.
Job Creation vs. Displacement Ratio
The World Economic Forum's 2025 Future of Jobs Report estimated 85 million jobs displaced by automation and AI through 2025, alongside 97 million new roles — a net positive of 12 million globally. BLS data shows approximately 340,000 net new AI-specialist roles created in the United States in 2025 alone.
However, the skill mismatch is severe. Stanford HAI's 2026 Index found that 78% of workers displaced by AI automation lack the credentials to qualify for the AI-specialist roles being created. The jobs being destroyed are concentrated in data processing, routine administrative work, and entry-level customer service — sectors with high proportions of workers without four-year degrees. The jobs being created require advanced technical skills and largely go to workers who were already highly educated.
The ROI Paradox
Gartner projects $2.52 trillion in global AI spending through 2026. McKinsey reports only 29% of enterprises report measurable ROI from generative AI. These two facts coexist because AI investment and AI return are separated by a deployment lag — typically 18 to 36 months for enterprise AI initiatives — and because ROI measurement is uneven.
Companies that have been deploying AI longest (financial services, technology) show the highest ROI rates. Companies that are early in deployment (manufacturing, energy) show the lowest. This is expected and likely temporary. The more concerning finding is that 31% of enterprises in McKinsey's survey report that their primary challenge to AI ROI is "lack of data infrastructure," suggesting a significant portion of AI investment is being made before organizations have the underlying data foundations to use it effectively.
Methodology
Big Tech capex: Q1 2026 10-Q SEC filings for Microsoft, Alphabet, Amazon, Meta. Capex figures from cash flow statements. "AI capex" uses companies' stated primary use in earnings calls, not isolated AI line items (which are not separately reported).
AI sector spend: IDC Worldwide AI Spending Guide 2025 sector estimates (publicly available summary data). Used as denominators for efficiency ratios.
USPTO patents: Patent full-text search for CPC classification G06N (computing; calculating; counting — neural networks, machine learning) by assignee sector. 2025 calendar year filings. Accessed April 2026.
BLS jobs: Occupational Employment and Wage Statistics (OEWS) 2025. AI-related titles: 15-1299 (Software and Web Developers, Scientific Applications); subset identified by SOC title keyword matching. Net new positions calculated from 2024→2025 change.
ROI data: McKinsey Global Survey on AI, Q1 2026 (publicly available report). "Measurable ROI" as defined by McKinsey survey instrument.