Artificial Intelligence

Oracle to Offer Google TPUs on OCI in Landmark Cloud Partnership

Advertisement
Advertise on GlobeWire Daily

· GlobeWire Daily Editorial Desk · Source: www.techradar.com

2 min read

Key takeawaysOracle has announced a new partnership to make Google Cloud's Tensor Processing Units (TPUs) available through its Oracle Cloud Infrastructure (OCI) platform. This development marks the first time a hyperscaler other than Google will offer Google's custom AI accelerators to cloud customers.
Oracle x Google Cloud

Image: TechRadar · Photo source · Resized and letterboxed to 16:9; no AI generation

Oracle Expands AI Hardware Options with Google TPUs

Oracle has announced a new partnership to make Google Cloud's Tensor Processing Units (TPUs) available through its Oracle Cloud Infrastructure (OCI) platform. This development marks the first time a hyperscaler other than Google will offer Google's custom AI accelerators to cloud customers.

The collaboration was announced at Oracle's annual AI World conference. According to the companies, the move is designed to give enterprise customers greater flexibility in selecting the hardware they use to train and run artificial intelligence models, providing an alternative to widely used graphics processing units (GPUs).

Understanding the Hardware

Google's TPUs are custom-designed application-specific integrated circuits optimized specifically for machine learning workloads. They are divided into two primary functions:

  • Model Training: Building and teaching new AI models.
  • AI Inference: Running trained models to make predictions or generate outputs.

Oracle plans to introduce Google's eighth-generation hardware through this integration, specifically the TPU 8t for training and the TPU 8i for inference. Both units were initially launched by Google Cloud earlier this year.

Industry Context and Practical Implications

While Nvidia continues to dominate the AI accelerator market, major cloud providers are increasingly investing in custom silicon to meet diverse computing demands. Industry executives highlighted the shifting landscape:

  • Compute Diversity: OCI President Mahesh Thiagarajan noted that different AI workloads require varying compute resources.
  • Future Infrastructure: Google AI & Infrastructure SVP Amin Vahdat stated that future generations of AI will demand a diverse range of compute options.

Verification and Next Steps

  • Availability: The integration is scheduled to become available to OCI customers in 2027.
  • Verification: Interested users can review the official announcements released by Oracle at the AI World conference or check updates directly through the Oracle Cloud Infrastructure platform.
  • Current Status: Confirmed details currently include the upcoming inclusion of TPU 8t and 8i chips on OCI; further implementation timelines and pricing remain subject to future updates.
Share this article
Read the original source: 'The next generation of AI will require a diverse range of compute': Oracle deal means Google TPUs are coming to a rival cloud for the first time

AI-assisted reporting; source claims are attributed and have not been independently verified.

About the author

GlobeWire Daily Editorial Desk — AI-assisted news based on attributed published sources.

More to read