LAS VEGAS, United States — Oracle and Google have announced a collaboration to make Google’s Tensor Processing Units (TPUs) available through Oracle Cloud Infrastructure, expanding the range of computing technologies available to organizations developing and deploying artificial intelligence applications. The announcement was made on October 27, 2026, during Oracle AI World in Las Vegas, where the companies outlined plans to introduce Google’s specialized AI processors into Oracle’s cloud computing portfolio. Under the arrangement, Oracle Cloud Infrastructure will become the first hyperscale cloud platform outside Google to offer TPU technology, with availability expected during 2027.
Oracle Expands Its AI Computing Infrastructure
Tensor Processing Units are specialized processors developed by Google to accelerate machine learning and artificial intelligence workloads. Unlike general-purpose processors, TPUs are designed to handle the computational requirements of AI model training and inference, making them relevant to organizations developing increasingly complex AI systems. Oracle plans to incorporate these processors into its cloud infrastructure portfolio, allowing customers to select computing technologies according to their workload requirements.
Oracle will provide the data center infrastructure required to deploy Google TPUs, including electricity, cooling systems, and networking capabilities. The collaboration is intended to give enterprise customers greater flexibility when choosing infrastructure for applications with different performance, processing capacity, and operating cost requirements. By expanding the range of available computing architectures, Oracle aims to support organizations developing AI systems that require substantial computing resources.
Google Introduces Eighth-Generation TPU Technology
Google’s eighth-generation TPU portfolio includes two architectures designed for different stages of artificial intelligence computing. TPU 8t focuses on model training, while TPU 8i is designed for inference workloads, which involve using trained models to generate responses, predictions, and other outputs. These specialized processors are intended to support demanding applications involving large language models, reasoning systems, and autonomous AI agents.
According to Google’s announced specifications, TPU 8t provides nearly three times the computing performance per pod compared with the previous generation. The company says the architecture can help accelerate advanced model development while improving the efficiency of large-scale computing operations. These performance figures represent company-reported capabilities, and actual results will depend on the applications, configurations, and deployment conditions involved.
Partnership Responds to Growing Enterprise AI Demand
Demand for artificial intelligence computing infrastructure has increased as businesses deploy more sophisticated models and integrate AI capabilities into everyday operations. Organizations require computing resources not only to train new systems but also to operate existing applications at scale. Inference workloads can become particularly demanding when AI services support large numbers of users or perform complex reasoning tasks.
Specialized processors offer one approach to improving performance and managing infrastructure costs. Oracle’s partnership with Google expands the computing options available through its cloud platform, while Google gains another distribution channel for its TPU technology beyond its own hyperscale cloud services. The collaboration also demonstrates how major technology companies can cooperate on infrastructure while continuing to compete in other areas of the cloud computing market.
Google TPUs Expected on Oracle Cloud in 2027
The companies expect Google TPUs to become available through Oracle Cloud Infrastructure during calendar year 2027, with deployment capacity increasing over a multiyear period. The announcement does not mean the processors are immediately available to all Oracle customers, and organizations evaluating the service will require additional information about pricing, regional availability, supported configurations, and integration requirements.
The partnership represents another development in the competition to provide infrastructure for advanced artificial intelligence applications. Its long-term commercial impact will depend on enterprise adoption and the performance of Google TPU technology within Oracle’s cloud environment. For organizations investing in AI, the agreement promises another option for accessing specialized computing resources as demand for high-performance infrastructure continues to grow.
This website uses cookies.