The race for AI dominance has intensified as cloud provider Oracle unveiled Zettascale10, an AI supercomputer which the company claims is the largest ever operating in cloud-based AI computing. The system connects hundreds of thousands of NVIDIA GPUs across multiple data centers, achieving 16 zettaFLOPS at optimal performance.
Oracle Cloud Infrastructure (OCI) Zettascale10 powers OpenAI’s Stargate supercluster in Abilene, Texas, which has been under development since 2024.
In a press release Oracle said that each deployment operates within large, gigawatt-scale data center campuses optimized for high-density GPU placement within a two-kilometer radius, minimizing latency and maximizing performance for large-scale AI model training.
Mahesh Thiagarajan, EVP, Oracle Cloud Infrastructure said, “With OCI Zettascale10, we’re fusing OCI’s groundbreaking Acceleron RoCE network architecture with next-generation NVIDIA AI infrastructure to deliver multi-gigawatt AI capacity at unmatched scale.”
Peter Hoeschele, VP, Infrastructure and Industrial Compute, OpenAI said, “The OCI Zettascale10 network and cluster fabric was developed and deployed first at the flagship Stargate site in Abilene.”
Ian Buck, VP, Hyperscale, NVIDIA said, “NVIDIA and Oracle are combining technologies to deliver AI infrastructure integrated into OCI Zettascale10, providing the computer foundation that helps organizations move to industrialized AI.”
Oracle plans to make multi-gigawatt OCI Zettascale10 deployments available to multiple enterprise customers. With 800,000 NVIDIA GPUs to support the high bandwidth required for advanced AI workloads.