Particle.news
Download on the App Store

Microsoft Will Fold OpenAI Chip Designs Into Its Own Silicon Through 2030

Satya Nadella says the designs will be industrialised under Microsoft IP to power Fairwater datacentres.

Overview

  • Nadella confirmed multi‑year access to OpenAI’s chip and hardware research through 2030, with continued access to OpenAI models through 2032 under the revised pact.
  • Microsoft will integrate OpenAI’s system‑level hardware innovations into its in‑house chip programs, first industrialising them and then extending them under Microsoft’s intellectual property.
  • OpenAI is co‑developing specialised AI processors and networking hardware with Broadcom, signaling a push beyond software into core compute components.
  • Microsoft is rolling out Fairwater datacentres, with an Atlanta site featuring NVIDIA GB200 NVL72 rack‑scale systems, a high‑throughput rack architecture, and advanced liquid cooling that uses almost no water.
  • The expanded hardware alignment is intended to reduce reliance on external suppliers and sharpen Microsoft’s competition with Google and Amazon in cloud‑scale AI.