Oracle expands AI chip options on OCI

Oracle and NVIDIA plan to scale OCI Superclusters equipped with NVIDIA Vera Rubin NVL72 systems to multi-gigawatt capacity.[1] Oracle says the NVL72 systems and NVIDIA Vera CPUs are generally available on OCI, and that the expanded infrastructure is intended for AI training, inference and agentic applications.[1] Oracle also says customers can scale CPU and GPU resources independently to match workloads and reduce idle capacity.[1] For x86-based applications, Oracle plans to introduce the NVIDIA HGX Rubin NVL8, an eight-GPU option.[1]

Oracle says its planned HGX Rubin NVL8 platform will include two 64-core Intel Xeon 6 CPUs, 4 TB of DDR5 memory and 6.4 Tbps of RDMA networking.[1] Oracle also says OpenAI has trained and served GPT-6 Astra on an OCI Supercluster in Abilene with more than 100,000 NVIDIA Blackwell GPUs.[1] That example concerns Blackwell infrastructure; it is distinct from the Vera Rubin systems announced as generally available.[1]

Google TPUs will be offered on a hyperscale cloud platform outside Google for the first time, with Oracle saying OCI availability is planned for calendar year 2027 and capacity will ramp over multiple years.[3] Oracle says the TPUs are designed for training, inference and agentic AI workloads.[3] Google’s eighth-generation portfolio includes TPU 8t for training and TPU 8i for inference.[3] Google says TPU 8t delivers nearly three times the compute performance per pod of the previous generation and cuts frontier-model development from months to weeks.[3] That performance and development-time statement is Google’s claim.[3]

Cerebras Systems and Oracle announced plans to make Cerebras AI inference available on OCI for AI labs, enterprises and developers.[4] The service is planned for calendar year 2027.[4] Cerebras says its Wafer-Scale Engine integrates compute and memory on a single wafer-scale processor.[4] Oracle described the addition as another ultrafast and deterministic option in OCI’s compute portfolio.[4] Oracle also says its existing AI infrastructure includes Dedicated AI Clusters in OCI Enterprise AI, and that bringing Google TPUs to OCI gives customers access to multiple leading AI compute architectures in one cloud.[4][3]