Particle.news
Get it on Google Play
Download on the App Store
7 ARTICLES
·
4h ago

Oracle Will Put Google’s TPU 8 Chips on Its Cloud

The deal gives enterprise customers a new on‑cloud option to run Google’s training and inference accelerators instead of relying only on Nvidia hardware

Image
OpenAI logo and a keyboard in this illustration created on October 19, 2026. REUTERS/Dado Ruvic/Illustration/File Photo
Image
WIZ, Google and OpenAI logos appear in this illustration taken October 22, 2026. REUTERS/Dado Ruvic/Illustration

Overview

  • Oracle and Google announced the partnership at Oracle AI World on October 27, 2026, saying Oracle Cloud Infrastructure will offer Google’s TPU 8t for training and TPU 8i for inference beginning in 2027.
  • TPUs are specialized accelerators built to speed the matrix math at the heart of machine learning, and Google’s TPU 8t is designed for large model training while the TPU 8i is tuned for high-volume inference.
  • The companies set 2027 as the start date with a phased rollout that will scale over several years, so broad TPU capacity on OCI is not expected to appear overnight.
  • Oracle frames the move as part of a multi‑vendor hardware strategy that keeps customer data and apps on OCI while offering choice alongside Nvidia, Cerebras and SambaNova.
  • Key commercial details such as regional availability, pricing, and framework and orchestration support have not been disclosed and will determine whether TPUs on OCI become a practical alternative for enterprise AI teams.