Oracle to offer Google TPUs on OCI, a first outside Google's own cloud

What Oracle and Google announced

The announcement was made at Oracle AI World in Las Vegas on October 27, 2026. Under the plan, Google's TPUs will join OCI's AI compute portfolio. Oracle says this gives customers the flexibility to match demanding workloads with infrastructure designed for near-linear scalability and highly optimized economics.

OCI will provide the power, cooling and datacenter networking needed to deploy Google TPUs at scale within its infrastructure. Google describes the TPUs as custom accelerators co-designed with open software, built to power the entire AI lifecycle, from frontier training and large-scale inference to the multi-step reasoning data demands of agentic AI.

Mahesh Thiagarajan, president of Oracle Cloud Infrastructure, said that "different AI workloads have different compute needs." He added that bringing TPUs to OCI gives customers access to multiple leading AI compute architectures in one cloud.

What the TPUs are meant to do

The companies frame the move around the growing cost and performance pressure of AI. As models grow and production inference demands increase, the performance and economics of every training run and inference request become more important, according to the release.

Google says its TPU architecture is designed to maximize AI processing from each watt of electricity, with fast access to model data and rapid communication between chips. The TPUs can support large-model training and post-training, including dense and mixture-of-experts (MoE) models. They can also handle production inference for reasoning, agentic and long-context applications.

Customers will be able to choose Google TPUs within OCI, optimized for their workloads, performance requirements and cost objectives.

Amin Vahdat, senior vice president and chief technologist for AI and Infrastructure at Google, said that "the next generation of AI will require a diverse range of compute." He said bringing TPUs to OCI will give more organizations access to the technology.

Eighth-generation chips and timing

The release draws on what it calls more than a decade of custom AI silicon development at Google. Google's eighth-generation TPU introduces two purpose-built architectures: TPU 8t for training and TPU 8i for inference.

According to the release, TPU 8t cuts frontier model development cycles from months to weeks and delivers nearly 3x the compute performance per pod over the previous generation. TPU 8i is designed to handle the iterative work of many specialized agents. These performance claims come from the companies and are not independently verified in the announcement.

Google TPUs on OCI will be available in calendar year 2027, ramping over a multi-year period to meet customer needs. The release gives no pricing, regions or details on which TPU generations will be offered first.

What it means for customers

For organizations already running on OCI, the announcement points to a wider choice of AI hardware within a single cloud. Oracle's stated aim is to let customers pick infrastructure by the performance, throughput and economics of each workload, rather than committing to one architecture.

For Google, the deal extends access to its TPUs beyond its own cloud to a second hyperscale provider. Because availability starts in 2027 and ramps over several years, customers will have to wait before they can use the chips on OCI.