Oracle Expands AI Compute Options: Adds NVIDIA, Google TPU and Cerebras
At the Oracle AI World conference, Oracle announced the expansion of AI compute on its cloud platform, adding NVIDIA Vera Rubin NVL72 systems and Vera CPUs, and plans to offer Google TPU and Cerebras inference technology on OCI starting in 2027. Oracle will also partner with Intel to launch OCI X14 based on Diamond Rapids, and expand AMD EPYC Venice instances and SambaNova SN50 inference systems. In addition, Oracle and OpenAI are expanding their collaboration to connect Oracle applications and enterprise data to ChatGPT, and plan to deploy the Jalapeño inference accelerator jointly developed by OpenAI and Broadcom.
Body
At its Oracle AI World conference in Las Vegas, Oracle announced plans to expand its AI computing infrastructure, adding NVIDIA’s latest GPUs, Google’s TPUs, and Cerebras’ inference technology to its cloud platform.
Oracle plans to scale its NVIDIA Vera Rubin infrastructure to multi-gigawatt capacity. Oracle now offers its Vera Rubin NVL72 systems and Vera CPUs on Oracle Cloud Infrastructure (OCI), and the expansion is intended to support both AI training and inference.
CPUs will also handle tasks such as running code, retrieving data and executing tools between AI model calls.
Oracle and Google will bring Google’s Tensor Processing Units (TPUs) to OCI from 2027, with capacity expected to increase over several years.
The partnership will allow Oracle customers to use Google’s custom chips for training and running AI models without moving to Google’s cloud.
Oracle is also partnering with Cerebras Systems to offer its AI inference technology on OCI in 2027. Cerebras uses a large processor that combines computing and memory on a single wafer, an approach designed to speed up AI responses.
The service will target developers and businesses building applications that require frequent model calls, including AI agents.
The company also announced a multiyear collaboration with Intel to expand Xeon-based computing on OCI. The partnership includes planned OCI X14 offerings powered by Intel’s next-generation Diamond Rapids server processors, targeting enterprise applications and AI workloads. The new offerings do not yet have a firm availability date.
The company is also expanding its AMD infrastructure portfolio. Upcoming OCI Compute E7 instances will use AMD’s next-generation EPYC processors, codenamed Venice. Oracle also plans to bring SambaNova’s SN50 systems to OCI for AI inference.
Beyond computing infrastructure, Oracle and OpenAI expanded their collaboration to connect Oracle applications and enterprise data with ChatGPT.
The companies announced new AI products, including Oracle AI Workspace, App Composer and AI Security Mesh.
Oracle also plans to deploy OpenAI’s Jalapeño inference accelerators, developed with Broadcom, to expand its computing capacity.
Oracle will also offer Snowflake’s AI Data Cloud as a native service on OCI. Customers can procure and use Snowflake through Oracle’s cloud platform, bringing its analytics and AI capabilities closer to data stored in Oracle’s infrastructure.
The announcements broaden Oracle’s options for customers building and running AI applications, combining different chip architectures with enterprise software and data services.