The collaboration between tech giants NVIDIA and Oracle promises to revolutionize the landscape of artificial intelligence (AI) and data processing for companies around the world. In an announcement made during the Oracle CloudWorld conference, both companies revealed ambitious plans to take high-performance computing to new levels.
Cloud Supercomputing: The OCI Supercluster
Oracle Cloud Infrastructure (OCI) unveiled its first OCI Supercluster at zetta scale, powered by the NVIDIA Blackwell platform. This cutting-edge system will enable companies to train and deploy next-generation AI models using over 100,000 state-of-the-art NVIDIA GPUs.
The OCI Superclusters offer customers a wide range of NVIDIA GPU options and the flexibility to deploy them in different environments:
- On-premises
- Public cloud
- Sovereign cloud
Systems based on Blackwell are expected to be available in the first half of next year, with the ability to scale up to 131,072 Blackwell GPUs. These systems will use NVIDIA ConnectX-7 NICs for RoCEv2 or NVIDIA Quantum-2 InfiniBand networks, providing an astonishing AI computing capacity of 2.4 zettaflops in the cloud.
New Instances and Capabilities
Oracle also teased information about new instances and capabilities:
- NVIDIA GB200 NVL72 liquid-cooled bare-metal instances for generative AI applications.
- NVIDIA HGX H200 instances, connecting eight NVIDIA H200 Tensor Core GPUs in a single bare-metal instance.
- General availability of NVIDIA L40S GPU-accelerated instances for mid-range AI workloads, NVIDIA Omniverse, and visualization.
Sovereign AI: A Global Approach
NVIDIA and Oracle are collaborating to offer sovereign AI infrastructure worldwide, addressing the data residency needs of governments and businesses. Some examples of this collaboration include:
- Wide Labs in Brazil: trained and deployed Amazonia AI, one of the first large language models for Brazilian Portuguese.
- Nomura Research Institute in Japan: uses OCI’s Alloy infrastructure with NVIDIA GPUs to enhance its financial AI platform.
- Zoom: will use NVIDIA GPUs in OCI data centers in Saudi Arabia to meet local data requirements.
- RSS-Hydro: demonstrates how their flood mapping platform, built on NVIDIA Omniverse and powered by L40S GPUs in OCI, can simulate flooding impacts in the Kumamoto region, Japan.
Enterprise-Ready AI
Companies can accelerate task automation in OCI by deploying NVIDIA software such as NIM microservices and NVIDIA cuOpt with OCI’s scalable cloud solutions. These solutions enable companies to quickly adopt generative AI and build agentic workflows for complex tasks like code generation and route optimization.
The NVIDIA AI Enterprise software platform, including cuOpt, NIM, RAPIDS, and more, is available on the Oracle Cloud Marketplace.
This collaboration between NVIDIA and Oracle promises to transform how companies approach AI and data processing, offering scalable, high-performance solutions for a wide range of business needs.