Oracle Cloud to Offer Google TPUs From 2027, the First Hyperscale Cloud Outside Google

What happened

The division of labor is Oracle supplying the facilities and Google supplying the chips: OCI provides the power, cooling and datacenter networking to deploy TPUs inside its own facilities; TPUs cover large-model training, post-training and production inference, supporting dense and mixture-of-experts models for reasoning, agentic and long-context applications, and customers can choose TPUs within OCI matched to workload, performance and cost objectives. Oracle Cloud Infrastructure president Mahesh Thiagarajan said different AI workloads have different compute needs, and bringing Google TPUs to OCI gives customers access to multiple leading AI compute architectures in one cloud.

The hardware is Google's eighth-generation TPU, in two variants: TPU 8t for training and TPU 8i for inference. Oracle and Google said TPU 8t delivers nearly 3x the compute performance per pod over the previous generation and can cut frontier model development cycles from months to weeks, while TPU 8i targets the iterative work of many specialized agents; the companies said TPUs will be available on OCI in calendar year 2027, ramping over a multi-year period, without disclosing quantities for 2027. Amin Vahdat, Google's senior vice president and chief technologist for AI and infrastructure, said the next generation of AI will require a diverse range of compute, and bringing TPUs to OCI gives more organizations access to the technology.

On the same day Oracle also announced bringing ChatGPT and Codex into its enterprise applications, including Oracle AI Workspace, which connects OpenAI frontier models with Oracle applications, enterprise data and cloud services; CNBC reported that OCI's accelerator lineup also includes OpenAI's in-house Jalapeño chip and Cerebras hardware. Oracle also announced on the same day that NVIDIA Vera Rubin systems are generally available on OCI, with the NVIDIA side of the partnership continuing.

Key facts

1. Announcement: Oracle and Google announced on Oct 27 at Oracle AI World in Las Vegas that Google TPUs will be available through OCI.
2. First outside hyperscale cloud: Oracle said this is the first time Google TPUs reach a hyperscale cloud platform outside Google.
3. Timing: planned for availability on OCI in calendar year 2027, ramping over a multi-year period, with no quantities for 2027 disclosed.
4. Eighth-generation pair: TPU 8t for training, officially described as nearly 3x the compute performance per pod over the previous generation; TPU 8i for inference and multi-agent iterative work.
5. Division of labor: OCI provides power, cooling and datacenter networking; TPUs support training, post-training and production inference, covering dense and mixture-of-experts models.
6. Same-day moves: Oracle announced bringing ChatGPT and Codex into enterprise applications; CNBC reported OCI's accelerator lineup also includes OpenAI's in-house Jalapeño chip and Cerebras hardware; NVIDIA Vera Rubin systems became generally available on OCI the same day.