At Oracle CloudWorld 2024, Oracle introduced its zettascale computing clusters, powered by NVIDIA’s Blackwell GPUs, marking a significant leap in cloud AI computing. Oracle Cloud Infrastructure (OCI) now accepts orders for the clusters, which could become the largest AI supercomputer in the cloud, featuring up to 131,072 GPUs. The system delivers an impressive 2.4 zettaFLOPS, tripling the GPU count of the Frontier supercomputer.

Mahesh Thiagarajan, Oracle’s Executive VP, emphasized that OCI’s AI infrastructure provides unmatched support for enterprise workloads. OCI’s distributed cloud enables businesses to deploy AI while ensuring data sovereignty, crucial for regulatory compliance. NVIDIA’s Blackwell GPUs, combined with OCI infrastructure, position Oracle as a major player in the high-performance AI market.

In addition to the Blackwell-powered clusters, OCI offers support for NVIDIA H100 and upcoming H200 Tensor Core GPUs, giving businesses flexibility in configuration. The infrastructure meets the growing demands of AI-driven innovation across industries like healthcare, finance, and research.

Oracle’s zettascale clusters include advanced GPU-to-GPU communication technology, NVLink, providing high-speed bandwidth for large-scale AI workloads. Oracle’s offering outperforms traditional hyperscaler systems by sixfold.

Global organizations like Zoom and WideLabs leverage OCI for AI advancements. Zoom utilizes OCI’s AI inference capabilities for its AI Companion, assisting users with tasks like summarizing meetings and drafting emails. WideLabs, on the other hand, uses OCI to power large language models in healthcare, ensuring compliance with national data regulations.

As enterprises increasingly demand powerful AI tools, Oracle’s zettascale clusters offer the performance, scalability, and security necessary for future innovation. Oracle is set to drive AI-driven advancements across industries with unmatched computing capabilities.