Oracle unveiled plans to build the first zettascale cloud computing cluster, powered by NVIDIA Blackwell GPUs. This new AI supercomputer, available via Oracle Cloud Infrastructure (OCI), will be the largest of its kind in the cloud, offering up to 131,072 NVIDIA Blackwell GPUs. Oracle is now accepting orders for the cluster, which is designed to support demanding AI workloads across various industries.
The OCI Supercluster delivers up to 2.4 zettaFLOPS of peak performance, more than three times the capacity of the Frontier supercomputer and six times that of other hyperscale cloud providers. The clusters are equipped with NVIDIA H100, H200, or Blackwell GPUs, enabling scalability up to 65,536 GPUs with 260 ExaFLOPS of performance. These clusters offer advanced networking options such as NVIDIA Quantum-2 InfiniBand and NVLink Switch, ensuring ultra-low latency and high-speed data transfer between GPUs.
At Oracle CloudWorld 2024 in Las Vegas, Oracle said customers like WideLabs and Zoom are already utilizing OCI’s infrastructure for AI projects, benefiting from its robust performance and strong data sovereignty controls. Ian Buck, VP of Hyperscale and HPC at NVIDIA, highlighted the importance of such large-scale computing clusters for accelerating AI research, development, and deployment efforts worldwide.
- Oracle introduced the first zettascale cloud computing cluster.
- OCI Supercluster offers up to 131,072 NVIDIA Blackwell GPUs.
- The system delivers 2.4 zettaFLOPS of peak performance.
- OCI supports NVIDIA H100, H200, and Blackwell GPUs.
- Customers like WideLabs and Zoom are leveraging OCI for AI projects.
- OCI offers high-speed networking with NVIDIA Quantum-2 InfiniBand.
“As businesses, researchers, and nations race to innovate using AI, access to powerful computing clusters and AI software is critical,” said Ian Buck, vice president of Hyperscale and High Performance Computing at NVIDIA.