Oracle unveils the world’s first Zettascale AI supercomputing cluster

Oracle today announced the launch of world’s first Zettascale Artificial Intelligence (AI) cloud computing clusters - where Zettascale computing in AI is different from Zettascale in actual HPC land -  beating the Japanese government and Fujitsu to the punch, which in this case is  powered by NVIDIA Blackwell GPUs offering up to 131,072 GPUs and delivering 2.4 zettaFLOPS of peak performance.

Oracle’s robust AI infrastructure, powered by NVIDIA GPUs, enables businesses to handle large-scale AI workloads with enhanced flexibility and sovereignty.

“We have one of the broadest AI infrastructure offerings and are supporting customers that are running some of the most demanding AI workloads in the cloud,” said Mahesh Thiagarajan, EVP of Oracle Cloud Infrastructure.

The Future of AI and Generative AI, by keynote Matthew Griffin

By combining NVIDIA’s advanced GPU architecture with Oracle’s cloud infrastructure, they’re offering scalable AI computing solutions critical for businesses and researchers globally.

“NVIDIA’s full-stack AI computing platform on Oracle’s broadly distributed cloud will deliver AI compute capabilities at unprecedented scale,” said Ian Buck, VP of Hyperscale and HPC, NVIDIA.

This new development from Oracle supports advanced AI research and development while ensuring regional data sovereignty, a crucial factor for industries like healthcare and collaboration platforms such as Zoom and WideLabs among others. Oracle Cloud Infrastructure (OCI) offers up to 131,072 NVIDIA B200 GPUs, which is six times more than other cloud hyperscalers like AWS, Azure, and Google Cloud.

For example, while AWS UltraClusters can host up to 20,000 GPUs, OCI offers more than three times that amount, enabling unprecedented levels of computational power.

Oracle also claimed that it is first to offer a zettascale AI supercomputer, delivering 2.4 zettaFLOPs while competitors have only reached exascale levels.

Moreover, OCI supports a wide range of GPUs, including NVIDIA H200, B200, and GB200 GPUs, with general availability across multiple architectures (Hopper, Blackwell) and the ability to run AI workloads of varying scales.

OCI will be the first to offer NVIDIA GB200 Grace Blackwell Superchips, which are designed for 4x faster training and 30x faster inference compared to H100 GPUs. This is important for AI model requiring real-time inference and multimodal LLM training.

That explains why OpenAI recently partnered with OCI to expand its capacity for running ChatGPT, leveraging its AI infrastructure to meet the rising demand for generative AI services, alongside its collaboration with Microsoft Azure.

“Leaders like OpenAI are choosing OCI because it is the world’s fastest and most cost-effective AI infrastructure,” said Larry Ellison, Oracle Chairman and CTO. Now with its supercluster scaling up to 131,072 NVIDIA B200 GPUs, there is no stopping Oracle and OpenAI.

“OCI will extend Azure’s platform and enable OpenAI to continue to scale,” said OpenAI’s chief Sam Altman.

Despite the happy news though Elon Musk is not smiling after Oracle and Elon Musk’s AI startup xAI recently ended talks on a potential $10 billion cloud computing deal, with xAI opting to build its own data center in Memphis, Tennessee.

At the time, Musk emphasised the need for speed and control over its own infrastructure. “Our fundamental competitiveness depends on being faster than any other AI company. This is the only way to catch up,” he added.

xAI is constructing its own AI data center with between 100,000 and 300,000 NVIDIA B200 chips. It claimed that it will be the world’s most powerful AI training cluster, marking a significant shift in strategy from cloud reliance to full infrastructure ownership.

After Microsoft Azure and Google Cloud, Oracle has finally partnered with AWS to launch Oracle Database@AWS, enabling customers to access Oracle Autonomous Database and Exadata services on AWS infrastructure.

“To meet this demand and give customers the choice and flexibility they want, Amazon and Oracle are seamlessly connecting AWS services with the very latest Oracle Database technology, including the Oracle Autonomous Database,” said Larry Ellison, Oracle Chairman and CTO.

The partnership offers zero-ETL integration and unified support, helping businesses modernise and improve outcomes.

“This partnership provides customers with a unified experience for migrating and managing their Oracle databases,” said AWS chief Matt Garman.