Oracle has announced the launch of the world's first Zettascale cloud computing cluster. This cluster is equipped with an impressive 131,072 NVIDIA Blackwell GPUs, capable of delivering a peak performance of 2.4 ZettaFLOPS. This powerful AI infrastructure enables businesses to handle large-scale AI workloads with greater flexibility and sovereignty.
The combination of Oracle Cloud Infrastructure (OCI) and NVIDIA's advanced GPU architecture provides a scalable AI computing solution. This is crucial for enterprises and researchers worldwide. Ian Buck, NVIDIA's Vice President, stated that NVIDIA's full-stack AI computing platform combined with Oracle's cloud services will deliver AI computing capabilities on an unprecedented scale.
This new development not only supports advanced AI research and development but also ensures regional data sovereignty, which is particularly important for industries like healthcare and collaboration platforms such as Zoom and WideLabs. OCI's GPU count is six times that of other cloud service providers like AWS, Azure, and Google Cloud. In comparison, AWS UltraClusters can accommodate only 20,000 GPUs, while OCI offers three times that computing power, demonstrating Oracle's significant advantage in this field.
Oracle also claims to be the first company to offer a Zettascale AI supercomputer, achieving a performance of 2.4 ZettaFLOPS, while competitors are currently at the Exascale level. Additionally, OCI supports multiple GPUs, including NVIDIA H200, B200, and GB200, with a wide range of architectures available to handle different scales of AI workloads.
It is worth noting that OCI will be the first to provide NVIDIA GB200 Grace Blackwell superchips, which are four times faster in training and 30 times faster in inference compared to H100 GPUs. This is significant for AI models requiring real-time inference and training of large-scale multimodal language models.
Oracle's AI supercomputer not only leads in scale but also provides flexible, scalable, and highly secure AI infrastructure, making it unmatched by other cloud service providers. This positions Oracle as the preferred choice for businesses training large-scale AI models or leveraging advanced computing technologies.
OpenAI recently partnered with OCI to expand its capacity to run ChatGPT, utilizing Oracle's AI infrastructure to meet the growing demand for generative AI services. Meanwhile, Elon Musk's AI startup, xAI, is also building its own data center, planning to use 100,000 NVIDIA chips, indicating his emphasis on speed and control over infrastructure.
Furthermore, Oracle has taken a significant step in completing its multi-cloud strategy by recently partnering with AWS to launch Oracle Database@AWS services, further enhancing the flexibility and choice of its cloud infrastructure.
Key Points:
🚀 Oracle launches the world's first Zettascale AI supercomputer cluster, featuring 131,072 NVIDIA Blackwell GPUs, delivering 2.4 ZettaFLOPS performance.
🌍 OCI leads in GPU count, supporting multiple GPU architectures to meet diverse AI needs.
🤝 Oracle partners with OpenAI to expand AI infrastructure and collaborates with AWS to launch Oracle Database@AWS services.