Overview
- Satya Nadella said Microsoft currently has AI GPUs sitting in inventory because it lacks electricity and finished “warm shells” to plug them into.
- He asserted the near‑term constraint has shifted from chip supply to siting facilities close to available power for rapid buildouts.
- Sam Altman predicted excess computing capacity could emerge within two to six years, describing today’s scarcity as temporary.
- Nadella outlined Microsoft’s twin focus on a “Token Factory” for infrastructure and an “Agent Factory” for the software ecosystem.
- Reporting from the conversation highlighted the need for major grid expansion, including accounts of OpenAI calling for 100 gigawatts of new power generation annually.