Core Numbers
2026 Big Tech AI capital expenditure reaches historic levels:
| Company | 2026 Capex | YoY Change | Q1 Cloud Growth | Key Moves |
|---|---|---|---|---|
| Amazon | $200B | Major increase | AWS +28% (fastest in 15 quarters) | Bedrock multi-cloud |
| Microsoft | $190B | Doubled | Azure steady growth | Copilot Agent 365 GA |
| $175-190B | Major increase | Cloud +63% | Gemini full rollout | |
| Meta | $125-145B | Up $10B | - | Llama ecosystem + AI metaverse |
| Total | ~$725B | +77% | - | - |
Compared to $410B in 2025, a 77% increase. Q1 alone spent $130B.
Where Is the Money Going?
1. Data Center Construction (Largest Share)
Four giants’ data center construction is reshaping the US power grid:
- US data center grid transformer crisis continues
- By 2030, US data center power investment expected to triple
- Amazon, Microsoft building hyperscale AI data centers across multiple states
2. AI Chip Procurement
| Chip Type | Main Buyers | Scale |
|---|---|---|
| NVIDIA GPU | All four | Hundreds of thousands |
| Google TPU | Google internal | Self-developed |
| AWS Trainium | Amazon internal | Reduce NVIDIA dependency |
| AMD MI Series | Multiple | Cost-effective alternative |
3. Agent Infrastructure Investment
Q1 2026 earnings show all four giants listing AI Agents as core growth drivers.
AWS 28% Growth: Fastest in 15 Quarters
Drivers:
- AI workload migration: More enterprise AI inference/training moving to cloud
- Bedrock multi-model strategy: Customers can use Claude, GPT, Llama simultaneously
- Agent infrastructure: Bedrock Agents attracting enterprise AI Agent builds
Google Cloud 63% Growth But Supply Constrained
Google Cloud’s 63% growth is impressive but officially supply constrained, meaning:
- Demand exceeds capacity
- GPU/TPU resources tight
- May delay some customers’ AI projects
Meta’s $10B Increase Strategy
Additional $10B mainly used for:
- Llama model training: Maintain open-source competitiveness
- AI social features: AI-generated content and interaction
- Metaverse + AI: AI-driven virtual reality experiences
ROI Analysis
| Metric | Amazon | Microsoft | Meta | |
|---|---|---|---|---|
| Cloud revenue growth | 28% | Steady | 63% | N/A |
| AI revenue visibility | High (Bedrock) | High (Copilot) | Medium-High | Medium |
| Margin pressure | Medium | Medium | Medium-High | High |
| Long-term potential | High | High | High | Medium |
Recommendations
| Role | Action |
|---|---|
| Enterprise CTO | Lock in long-term cloud contracts during expansion for discounts |
| AI startups | Leverage cloud vendor free quotas and developer programs |
| Investors | Focus on AI infrastructure supply chain (chips, power, cooling) |
| Job seekers | AI infrastructure operations and MLOps in high demand |
$725B means AI is no longer a “track” but an infrastructure arms race.