Enterprises are seeking robust computing power to enhance their AI workloads and speed up data processing, ultimately improving returns on AI training investments and user experiences. At the Oracle CloudWorld conference, Oracle Cloud Infrastructure (OCI) unveiled its first zettascale OCI Supercluster, powered by NVIDIA's Blackwell platform. This infrastructure supports the training and deployment of advanced AI models utilizing over 100, 000 of NVIDIA’s latest GPUs. The OCI Superclusters offer flexibility, enabling customers to select from various NVIDIA GPUs for deployment in on-premises, public, and sovereign clouds. These systems, available in the first half of next year, can scale to 131, 072 Blackwell GPUs, achieving a remarkable 2. 4 zettaflops of peak AI compute in the cloud. Oracle also revealed the NVIDIA GB200 NVL72 bare-metal instances, which facilitate large-scale training and real-time inference for generative AI applications. These instances support a 72-GPU NVIDIA NVLink domain functioning as a single massive GPU.

Additionally, OCI will provide NVIDIA HGX H200 to connect multiple GPUs for real-time inference at scale. NVIDIA L40S GPU-accelerated instances will now be generally available for midrange AI workloads, while Oracle’s edge solutions support scalable AI in remote areas, including smaller deployments with up to three NVIDIA L4 GPUs. Reka, a foundation model startup, is utilizing NVIDIA-powered OCI Superclusters for developing advanced multimodal AI models that enhance enterprise functionalities. Furthermore, Oracle’s Autonomous Database now supports NVIDIA GPUs, enabling accelerated data processing for machine learning. Demonstrations at the conference featured how NVIDIA's accelerated computing can enhance generative AI capabilities within Oracle environments. NVIDIA and Oracle are also collaborating on sovereign AI infrastructure to meet data residency needs. Initiatives include Brazil’s Wide Labs, which created a large language model for Brazilian Portuguese using OCI and NVIDIA technology, as well as Nomura Research Institute in Japan, which leverages OCI to enhance financial AI services. Enterprises can quickly integrate AI by deploying NVIDIA software alongside OCI’s cloud solutions, promoting automation and complex task handling. Interested parties can learn more and register for Oracle CloudWorld 2024 to explore these advancements in AI and data processing.