Big Tech AI Spending Tracker
Every dollar the seven biggest tech companies pour into AI — data center capex, acquisitions, minority investments, compute contracts, and talent — from 2020 through 2026 guidance. The big four hyperscalers alone plan roughly $730 billion of 2026 capex, up 78% from ~$410B in 2025.
The four largest hyperscalers plan to spend roughly $730 billion on AI infrastructure in 2026 — up about 78% from ~$410B in 2025.
Guidance as of Q2 2026 earnings (late July 2026): Amazon raised to ~$220B, Alphabet to $195–205B, and Meta narrowed to $130–145B; Microsoft's ~$175B reflects a lease-accounting change, not a spending cut. Full breakdown in the $730B AI capex breakdown.
2026 AI Capex Guidance by Company
Company-guided capital expenditure for calendar 2026, as of Q2 2026 earnings. Amazon leads at ~$220B; Microsoft's ~$175B (including finance leases) is down from ~$190B only because of a lease-accounting change.
2026 capex guidance ranges ($B)
Four companies now out-spend most national infrastructure budgets
Source: company earnings calls and guidance through Q2 2026.
| Company | 2025 capex | 2026 guidance | Change | Primary focus |
|---|---|---|---|---|
| Amazon (AWS) | ~$128B | ~$220B | +72% | AWS data centers, Trainium chips, Bedrock |
| Microsoft (Azure) | ~$118B | ~$175B | +48% | OpenAI infrastructure, MAIA chips, Copilot scale |
| Alphabet (Google) | ~$91B | $195–205B | +120% | TPU v7, Gemini training, Google Cloud AI |
| Meta | ~$72B | $130–145B | +90% | Llama training, superintelligence labs |
| Combined big four | ~$410B | ~$730B | +78% | AI data centers, GPUs, custom silicon, power |
| Company | 2026 capex (est.) | Source |
|---|---|---|
| Amazon | ~$220B | Amazon 2026 cash capex guidance (~$220B, raised Jul 30, 2026) |
| $195–205B | Alphabet guidance, raised to $195–205B at Q2 2026 (Jul 22, 2026) | |
| Microsoft | ~$175B | Microsoft calendar-2026 capex incl. finance leases (~$175B after lease/useful-life accounting change; was ~$190B) |
| Meta | $130–145B | Meta guidance, narrowed to $130–145B at Q2 2026 (Jul 29, 2026) |
| Apple | $15–25B | Estimated AI data center buildout |
| Tesla | $12–16B | Colossus/xAI growth trajectory (est.) |
| NVIDIA | $4–6B | Fabless model; R&D-focused |
Tracked AI Spend by Company, 2020–2025
Broader than capex guidance alone: infrastructure capex attributable to AI plus AI acquisitions, minority investments, structured deals, and talent raids. Total tracked spend grew from ~$30B in 2020 to $381B in 2025 — a 12x ramp in five years.
Annual tracked AI spend, stacked by company ($B)
2025 alone (~$381B) exceeds 2020–2023 combined
Source: company filings, earnings calls, and reported deal values.
| Year | Amazon | Meta | Microsoft | Tesla | NVIDIA | Apple | Total | |
|---|---|---|---|---|---|---|---|---|
| 2020 | $5B | $7.7B | $5B | $2B | $0.8B | $8.4B | $1.3B | $30.2B |
| 2021 | $6.2B | $8B | $5B | $24.7B | $1.8B | $2.5B | $1.5B | $49.7B |
| 2022 | $15.2B | $10.5B | $10B | $5B | $1.7B | $5.3B | $1.9B | $49.6B |
| 2023 | $18.1B | $19.8B | $14B | $21.2B | $2.2B | $5.6B | $2.5B | $83.4B |
| 2024 | $39.5B | $54.4B | $27.5B | $30.4B | $3.5B | $10.2B | $5.2B | $170.6B |
| 2025 | $106.3B | $105B | $79.6B | $66B | $4.8B | $11.3B | $8B | $381.1B |
Cumulative AI Spend and Strategy, 2020–2025
$764B tracked across seven companies in six years. Amazon and Google lead on totals; every company runs a distinctly different playbook.
Cumulative tracked AI spend, 2020–2025 ($B)
Amazon edges Google for the biggest six-year AI bill
| Company | 2020–2025 total | 2025 spend | 5-yr CAGR | Strategy |
|---|---|---|---|---|
| Amazon | $205B | $105B | 69% | Cloud + Model Partnerships |
| $190B | $106.4B | 84% | Full Stack / Vertical Integration | |
| Microsoft | $149B | $66B | 101% | Partnership-First |
| Meta | $141B | $79.6B | 74% | Open Source + Infrastructure |
| NVIDIA | $43B | $11.3B | 6% | Picks & Shovels |
| Apple | $20B | $8B | 45% | On-Device / Stealth |
| Tesla | $15B | $4.8B | 43% | Specialized / Autonomous AI |
Where the Money Goes
Infrastructure — data centers, GPUs, custom silicon — dominates every year, but 2025 saw a surge in external deals: Google's $32B Wiz acquisition, Meta's $41B in Nebius and CoreWeave compute contracts, and Amazon's $50B cumulative OpenAI position.
Tracked AI spend by category ($B, stacked)
Infrastructure is ~80% of every dollar; M&A spiked in 2025
| Year | Infrastructure | M&A | Investments | Structured deals | Talent |
|---|---|---|---|---|---|
| 2020 | $20.8B | $9.3B | $0B | $0B | $0B |
| 2021 | $28B | $19.7B | $2B | $0B | $0B |
| 2022 | $42.5B | $6.8B | $0.3B | $0B | $0B |
| 2023 | $69.9B | $0.2B | $13.3B | $0B | $0B |
| 2024 | $156.8B | $1.0B | $9.5B | $3.4B | $0B |
| 2025 | $308.4B | $53.9B | $16.3B | $1B | $1.5B |
The 15 Biggest AI Deals, 2020–2025
Acquisitions, cumulative minority stakes, and multi-year compute contracts of $1B or more. Follow the two biggest recipients on our OpenAI news hub and Anthropic news hub.
| # | Deal | Buyer | Year | Value | Type |
|---|---|---|---|---|---|
| 1 | OpenAI (cumulative stake) | Amazon | 2025 | $50B | Minority Investment |
| 2 | Wiz | 2025 | $32B | Acquisition | |
| 3 | Nebius 5-yr compute | Meta | 2025 | $27B | Infrastructure Contract |
| 4 | Nuance Communications | Microsoft | 2021 | $19.7B | Acquisition |
| 5 | Scale AI (49% stake) | Meta | 2025 | $14.3B | Acquisition |
| 6 | CoreWeave 6-yr compute | Meta | 2025 | $14.2B | Infrastructure Contract |
| 7 | OpenAI (cumulative stake) | Microsoft | 2023 | $13.8B | Minority Investment |
| 8 | Anthropic (cumulative stake) | Amazon | 2024 | $8B | Minority Investment |
| 9 | Mellanox Technologies | NVIDIA | 2020 | $6.9B | Acquisition |
| 10 | Mandiant | 2022 | $5.4B | Acquisition | |
| 11 | Intersect Power | 2025 | $4.8B | Acquisition | |
| 12 | Anthropic (cumulative stake) | 2025 | $3.3B | Minority Investment | |
| 13 | Manus AI | Meta | 2025 | $2.8B | Acquisition |
| 14 | Character.AI license | 2024 | $2.7B | Structured Deal | |
| 15 | G42 minority stake | Microsoft | 2024 | $1.5B | Minority Investment |
The $500B Stargate Project
Announced in early 2025, Stargate is a joint venture between OpenAI, SoftBank, and Oracle to build a $500B AI data center network across the United States — on top of the hyperscaler budgets tracked above. SoftBank provides the majority of financing, Oracle supplies cloud infrastructure, and OpenAI is the primary compute consumer, with initial campuses in Texas. It is the single largest private infrastructure investment in AI history, deploying capital over 4–5 years. Track every announced US site, megawatt, and construction project on our AI Infrastructure Buildout Tracker.
AI Spending — Common Questions
How much are tech companies spending on AI in 2026?
The four largest hyperscalers — Amazon, Alphabet (Google), Meta, and Microsoft — plan to spend roughly $730B on capital expenditures in 2026, up about 78% from ~$410B in 2025, with the overwhelming majority going to AI data centers, Nvidia GPUs, custom silicon, and power. After Q2 2026 earnings in late July, Amazon leads at ~$220B, followed by Alphabet ($195–205B), Microsoft (~$175B, lowered from ~$190B by a lease-accounting change rather than a spending cut), and Meta ($130–145B).
Which company is spending the most on AI?
Amazon is the single largest AI infrastructure spender in 2026, with 2026 capex guided to about $220B, much of it for AWS data centers, Trainium chips, and Bedrock. Alphabet (Google) follows at $195–205B, then Microsoft at ~$175B and Meta at $130–145B.
Who benefits most from big tech AI spending?
Nvidia remains the primary beneficiary — its H200, B200, and GB200 GPUs power most AI training and inference at every major hyperscaler, capturing roughly 40% of hyperscaler capex. Other winners include TSMC (fabrication), Arista (networking), Vertiv and Eaton (power/cooling), and AI clouds like CoreWeave and Nebius — Meta alone signed $41B in compute contracts with the two of them.
How much have big tech companies spent on AI since 2020?
Our tracker counts roughly $764B in AI spending by Google, Amazon, Meta, Microsoft, Tesla, NVIDIA, and Apple from 2020 through 2025 — infrastructure capex plus AI acquisitions, minority investments, structured deals, and talent. Adding 2026 guidance (~$730B from the big four alone) pushes the cumulative total past $1.5 trillion. The $500B Stargate project (OpenAI/SoftBank/Oracle) adds further capacity on top of hyperscaler budgets.
Is big tech AI spending sustainable?
By mid-2026 the consensus has shifted toward AI capex being a permanent feature of big tech cost structures rather than a cycle. Bulls cite concrete returns — Microsoft Azure growing ~40%, Meta's AI ad targeting boosting revenue, Google Cloud growing 82% in Q2 2026. Bears note that ~$730B in a single year still outpaces demonstrable AI revenue, GPU depreciation assumptions are under scrutiny, and the arms race shows no signs of slowing.
How much is AI infrastructure spending in 2026?
Total AI capex by the four main hyperscalers reaches approximately $730B in 2026 — Amazon ~$220B, Google $195–205B, Microsoft ~$175B, Meta $130–145B — up 78% from ~$410B in 2025. Including the $500B Stargate project (OpenAI/SoftBank/Oracle), Oracle's own buildout, and other AI-focused companies, total sector AI infrastructure investment in 2026 exceeds $1 trillion.
What is Amazon spending on AI in 2026?
Amazon is projecting approximately $220B in cash capital expenditure for 2026, raised from $200B on its July 30, 2026 earnings call, the largest of any single hyperscaler. The majority goes to AWS data centers, Trainium and Inferentia custom chips, and Bedrock AI services infrastructure — on top of its cumulative $50B+ position in OpenAI and $8B in Anthropic.
Methodology
"2026 capex guidance" figures are company-guided total capital expenditures (the overwhelming majority now AI-driven), from earnings calls through Q2 2026. "Tracked AI spend 2020–2025" is our estimate of AI-attributable infrastructure capex plus disclosed AI acquisitions, cumulative minority investments, structured deals, and talent packages — so 2025 tracked spend (~$381B across seven companies) is not directly comparable to the ~$410B big-four total-capex figure. Estimates carry uncertainty ranges; sources include company filings, earnings transcripts, and reported deal values. Updated .