NVIDIA and Oracle Unveil Zettascale AI Supercomputing for Enterprises

Oracle and NVIDIA join forces to revolutionize enterprise AI: Oracle Cloud Infrastructure (OCI) and NVIDIA have announced a significant collaboration to enhance AI and data processing capabilities for enterprises, with plans to deploy cutting-edge GPU technology and infrastructure solutions.

Zettascale supercomputing comes to the cloud: OCI is set to introduce the first zettascale OCI Supercluster, powered by NVIDIA’s Blackwell platform, offering unprecedented computational power for AI workloads.

• The new system will utilize over 100,000 of NVIDIA’s latest-generation GPUs, providing up to 2.4 zettaflops of peak AI compute.
• Deployment options include on-premises, public cloud, and sovereign cloud environments, offering flexibility for diverse enterprise needs.
• Availability is expected in the first half of next year, marking a significant leap in cloud-based AI computing capabilities.

Advanced GPU instances for AI applications: Oracle is expanding its offerings with new GPU-accelerated instances to cater to various AI workloads and requirements.

• NVIDIA GB200 NVL72 liquid-cooled bare-metal instances are being previewed, designed for large-scale training and real-time inference of trillion-parameter models.
• OCI will offer NVIDIA HGX H200 instances, connecting eight H200 Tensor Core GPUs in a single bare-metal instance, scalable up to 65,536 GPUs.
• NVIDIA L40S GPU-accelerated instances are now generally available for midrange AI workloads, NVIDIA Omniverse, and visualization tasks.

Edge AI solutions for diverse environments: Oracle is enhancing its edge offerings with NVIDIA GPU acceleration, enabling AI deployment in remote and disconnected locations.

• Smaller-scale deployments using Oracle’s Roving Edge Device v2 will support up to three NVIDIA L4 Tensor Core GPUs, bringing powerful AI capabilities to the edge.
• This expansion allows for scalable AI solutions from single-node to multi-rack configurations, catering to various enterprise needs.

Accelerating database workloads with GPU support: Oracle Autonomous Database is integrating NVIDIA GPU support to enhance data processing capabilities.

• Oracle Machine Learning notebooks will leverage NVIDIA GPUs to accelerate workloads within Oracle Autonomous Database.
• Demonstrations at Oracle CloudWorld showcase how NVIDIA GPUs can accelerate vector embeddings, graph index generation, and inference for text generation and translation.

Sovereign AI infrastructure worldwide: NVIDIA and Oracle are collaborating to address data residency needs for governments and enterprises globally.

• Brazilian startup Wide Labs trained the Amazonia IA language model using NVIDIA H100 GPUs in OCI’s Brazilian data centers.
• Nomura Research Institute in Japan is utilizing OCI’s Alloy infrastructure with NVIDIA GPUs to enhance its financial AI platform.
• Zoom will use NVIDIA GPUs in OCI’s Saudi Arabian data centers to support compliance with local data requirements.

Enterprise-ready AI solutions: The collaboration between NVIDIA and Oracle aims to accelerate task automation and AI adoption for enterprises.

• NVIDIA software solutions like NIM microservices and cuOpt are being integrated with OCI’s cloud solutions.
• The NVIDIA AI Enterprise software platform, available on the Oracle Cloud Marketplace, includes various tools to support AI development and deployment.

Looking ahead: Implications for enterprise AI: This partnership between Oracle and NVIDIA represents a significant step forward in making advanced AI capabilities more accessible to enterprises of all sizes.

• The introduction of zettascale computing in the cloud could dramatically accelerate AI research and development across industries.
• Improved sovereign AI infrastructure may lead to increased adoption of AI technologies in regions with strict data residency requirements.
• As these technologies become more widely available, we may see a surge in AI-driven innovation and transformation across various sectors.