Oracle Launches Major AI Supercluster Partnership with AMD

LAS VEGAS, Nevada — Oracle has announced a significant expansion of its partnership with AMD, aiming to be the first hyperscaler to launch a publicly available AI supercluster equipped with 50,000 AMD Instinct MI450 Series GPUs by the third quarter of 2026. This move, revealed during Oracle AI World, marks a strategic commitment to compete against NVIDIA in the AI infrastructure sector.

Oracle CEO Safra Catz emphasized the importance of this development, highlighting Oracle’s plans to enhance its AI capabilities for large clients. These clients are reportedly tied to a larger contract worth over $30 billion annually. Forrest Norrod, executive vice president at AMD, added, “AMD and Oracle continue to set the pace for AI innovation in the cloud,” referring to the transformative potential of their collaboration.

The partnership comes as major players in the AI market, including OpenAI, diversify their partnerships. OpenAI has established deals with both AMD and NVIDIA, suggesting a broader industry trend towards reducing dependency on a single vendor. Mahesh Thiagarajan, executive vice president at Oracle Cloud Infrastructure, stated, “Our customers are building some of the world’s most ambitious AI applications, and that requires robust, scalable infrastructure.”

Oracle’s AMD Instinct MI450 GPUs will feature up to 432 GB of HBM4 memory and a bandwidth of 20 TB/s, enabling customers to work with significantly larger AI models. This latest AI infrastructure announcement also includes the launch of OCI Zettascale10, touted as the largest AI supercomputer in the cloud, integrating up to 800,000 NVIDIA GPUs.

Alongside these initiatives, Oracle has debuted OCI Dedicated Region25, which will lower the barriers to entry for on-premises clouds from a previous 10–12 rack requirement to just three racks, addressing growing concerns around data sovereignty. Scott Twaddle, senior vice president at OCI, remarked, “Organizations want the freedom to run AI and cloud services where they deliver the most value.”

This shift is particularly timely, as data sovereignty regulations intensify worldwide, pushing organizations to rethink their cloud strategies. The new three-rack configuration allows businesses to maintain compliance with data regulations while still tapping into cloud efficiencies. Kazushi Koga, of Fujitsu, expressed that this solution would enhance their ability to deploy services swiftly while ensuring scalability.

Oracle also introduced Multicloud Universal Credits, simplifying procurement processes for customers using Oracle services across multiple cloud platforms, including AWS and Azure. This new licensing model aims to reduce administrative burdens and improve flexibility for organizations managing several cloud environments.

As Oracle builds its AI infrastructure, the company is addressing the fundamental operational challenges faced by enterprises in harnessing the full potential of AI technologies. This comprehensive approach signals Oracle’s commitment to meeting the needs of its clients through enhanced performance and compliance in a rapidly evolving landscape.