Sign in Subscribe
Data centers

Oracle Unveils Largest AI Supercomputer in Cloud, Delivers 10X Peak Performance with ZettaFLOPS

Strategic Insights

— 1 min read

Oracle announced the largest AI supercomputer in the cloud, OCI Zettascale10, designed to deliver up to 16 zettaFLOPS of peak performance. Built on Oracle Acceleron RoCE networking architecture with NVIDIA AI infrastructure, OCI Zettascale10 will provide multi-gigawatt AI workload capacity and scale, supporting the deployment of the first Zettascale cloud computing cluster, introduced at the Oracle AI World conference.

Key Takeaways:

  • OCI Zettascale10 is the largest AI supercomputer in the cloud, delivering up to 16 zettaFLOPS of peak performance.
  • Built on Oracle Acceleron RoCE networking architecture with NVIDIA AI infrastructure, OCI Zettascale10 provides multi-gigawatt AI workload capacity and scale.
  • OCI Zettascale10 is designed to deliver breakthrough scale, extremely low GPU-GPU latency, industry-leading price-performance, and improved cluster utilization.
  • The architecture is being deployed with OpenAI at the Stargate site in Abilene, Texas, supporting the first Zettascale cloud computing cluster.
  • OCI Zettascale10 will be powered by NVIDIA AI infrastructure, with up to 800,000 NVIDIA GPUs, delivering predictable performance and strong cost efficiency.
  • Oracle Acceleron RoCE networking delivers scale, reliability, and efficiency for AI on OCI Zettascale10, with features including wide, shallow, resilient fabric, higher reliability, consistent performance, power-efficient optics, and operational flexibility.
  • OCI is now taking orders for OCI Zettascale10, which will be available in the second half of next calendar year.

Statistics:

  • Up to 16 zettaFLOPS of peak performance delivered by OCI Zettascale10.
  • Up to 800,000 NVIDIA GPUs in a single OCI Zettascale10 cluster.

*OCI Zettascale10 delivers multi-gigawatt AI workload capacity and scale.

  • Predictable performance and strong cost efficiency delivered by Oracle Acceleron RoCE networking.
  • 2-kilometer radius data center campuses that are hyper-optimized for density to offer the best GPU-GPU latency for large-scale AI training workloads.

Sources:

  • Oracle AI World
  • Oracle Corporation press release, "Oracle Unveils Largest AI Supercomputer in Cloud, Delivers 10X Peak Performance with ZettaFLOPS"
  • Oracle Corporation press release, "Oracle Acceleron RoCE Networking Architecture Delivers Scale, Reliability, and Efficiency for AI on OCI Zettascale10"
  • NVIDIA Corporation press release, "NVIDIA and Oracle Bring Together Cloud and AI Infrastructure to Deliver AI at Extraordinary Scale"

Read more