Oracle and Google announced on October 27 that Google’s Tensor Processing Units will be offered through Oracle Cloud Infrastructure. The companies describe OCI as the first hyperscale cloud platform outside Google scheduled to offer the accelerators.
A future addition to OCI
Availability is planned for calendar 2027, followed by a multiyear capacity ramp. OCI will supply power, cooling and data-centre networking. The announcement therefore expands Oracle’s roadmap; it does not mean customers can provision the service today.
Training and inference are different workloads
The release describes support for model training, post-training and production inference. It also highlights Google’s eighth-generation TPU family, with 8t designed for training and 8i for inference. The aim is to give customers more architectural options within OCI.
Commercial details remain to be established
The announcement provides no price list, regional schedule or independent OCI workload measurements. Those details will be needed to compare the eventual service with other accelerators. The confirmed change is a planned new deployment channel for Google’s custom silicon.
Sources
Oracle and Google joint announcement, October 27
Feature image: AI-generated editorial illustration for Guava Insights. Conceptual artwork, not an actual product photograph, interface or facility.
