Stratechery Microsoft and Software Survival
341 snips
Feb 3, 2026 A tour of which tech giants felt most threatened by the AI era and how that fear shifted over time. A close look at Microsoft's bets: Azure GPU strategy, OpenAI ties, Copilots and the per-seat licensing risk. Discussion of how AI-written code reshapes software economics and why SaaS vendors may gain or feel pressure. Consideration of GPU scarcity, capacity allocation, and what survival requires.
AI Snips
Chapters
Transcript
Episode notes
OpenAI Dominates Azure Obligations
- Microsoft reported OpenAI represents 45% of Azure's remaining performance obligations, which markets now view as a liability.
- This concentration, plus slower Azure growth and big GPU spend, helped trigger a dramatic market sell-off.
Code Is A Natural Fit For AI
- AI-generated code matches probabilistic inputs with deterministic, testable outputs, making it especially effective for software development.
- Thompson believes most software will eventually be written by AI, though architects will remain important for longer.
Sell The Whole Product, Not Just Code
- Don't assume companies will build their own software just because AI can write code; they value hired products for non-code responsibilities.
- Thompson advises software vendors to emphasize support, compliance, integration, and maintenance as enduring value.