Microsoft’s compute allocation reveals a deliberate three-layer strategy that prioritizes first-party products over raw cloud metrics.
Layer 1: OpenAI (Shrinking %)
- Fairwater Clusters: 300MW+ GPU buildings
- $250B commitment, dedicated facilities
- But: OpenAI diversifying to AWS, Oracle, Stargate
- Trend: Declining relative share
Layer 2: Microsoft Internal (Priority)
- M365 Copilot: 15M paid seats
- GitHub Copilot: 4.7M subs (+75%)
- $30/user/month premium pricing
- Trend: Deliberate GPU prioritization
Layer 3: Third-Party (Growth Frontier)
- Anthropic, Nebius anchor tenants
- 1,500+ multi-model customers
- 80% of F500 on Azure Foundry
- Trend: Rapid expansion
The Strategic Logic
CFO Amy Hood’s admission is revealing: Azure growth could be higher if GPUs weren’t prioritized for Copilot. Microsoft deliberately trades Azure metrics for higher-margin first-party AI revenue.
Long-Term Implications
This prioritization suggests Microsoft sees Copilot products — not raw Azure compute — as its primary AI monetization path. Third-party enterprise becomes the growth hedge as OpenAI share declines.
This is part of a comprehensive analysis. Read the full analysis: Microsoft’s Frontier AI Dilemma on The Business Engineer.









