Oracle Offers First Zettascale Cloud Computing Cluster

Core Insights - Oracle has announced the launch of the largest AI supercomputer in the cloud, featuring up to 131,072 NVIDIA Blackwell GPUs, capable of delivering 2.4 zettaFLOPS of peak performance [1][2][3] - The OCI Supercluster offers more than three times the GPU capacity of the Frontier supercomputer and over six times that of other hyperscalers, highlighting its significant scale and performance capabilities [2][3] - The new supercomputer is designed to support demanding AI workloads, providing customers with flexibility in deploying cloud and AI services while ensuring data sovereignty [1][4] Oracle Cloud Infrastructure (OCI) Superclusters - OCI Superclusters can be configured with NVIDIA H100 or H200 Tensor Core GPUs, with H100 configurations scaling up to 16,384 GPUs and H200 configurations scaling up to 65,536 GPUs [3] - The performance metrics for H100 GPUs reach up to 65 ExaFLOPS, while H200 GPUs can achieve up to 260 ExaFLOPS, showcasing the advanced capabilities of OCI Superclusters [3] - The infrastructure includes ultra-low latency networking options and high-performance computing storage, enhancing the overall efficiency of AI workloads [2][3] Customer Use Cases - WideLabs, an AI startup in Brazil, is utilizing OCI to train one of the largest Portuguese language models, Amazonia IA, while ensuring compliance with local data sovereignty regulations [4][5] - Zoom is leveraging OCI's capabilities for its AI personal assistant, Zoom AI Companion, which enhances user productivity by providing features like email drafting and meeting summarization [6] - Both companies benefit from OCI's high-performance infrastructure and data sovereignty features, which are critical for their AI initiatives [5][6]