
NVIDIA’s investment strategy creates a circular demand engine: every company they invest in needs massive compute for training and inference. That compute comes from NVIDIA.
GPU Demand by Portfolio Company
| Company | Estimated GPU Demand | Notes |
|---|---|---|
| xAI | 100,000 H100s | Largest single cluster |
| CoreWeave | ~70,000 GPUs | GPU cloud provider |
| OpenAI | ~50,000+ GPUs | Via Azure + direct |
| Anthropic | ~40,000 GPUs | Via AWS + GCP |
| Mistral | ~10,000+ GPUs | European AI leader |
The Circular Demand Engine
Every investment NVIDIA makes pays dividends twice:
- Equity appreciation — If the company succeeds, NVIDIA’s stake appreciates
- Hardware sales — The company must buy NVIDIA GPUs to operate
This is the “circular economics” pattern that makes NVIDIA’s position so powerful. They’re not just betting on AI — they’re the house that supplies the chips to every player at the table.
Why Competitors Can’t Replicate
- Capital requirements — Building this portfolio took $10B+ in investments
- Hardware lock-in — CUDA ecosystem creates switching costs
- Network effects — More customers = more software optimization = better performance
- Time advantage — Started building relationships years before competitors
This is part of a comprehensive analysis. Read the full analysis on The Business Engineer.









