Oracle and AMD have announced an expanded partnership to develop a massive AI "supercluster" powered by 50,000 AMD Instinct MI450 Series GPUs. This collaboration aims to address the growing demand for AI infrastructure, enabling Oracle's cloud customers to train, fine-tune, and deploy next-generation AI models more efficiently. The AI supercluster, which will start being deployed in Q3 2026, is designed to provide unparalleled performance and scalability for AI workloads.
The AMD Instinct MI450 Series GPUs will provide up to 432 GB of HBM4 memory and 20 TB/s of memory bandwidth, allowing customers to train and run models that are 50% larger than previous generations. This increased capacity will enable Oracle's cloud customers to tackle more complex AI tasks and achieve better results. The partnership also includes the development of next-generation AMD EPYC CPUs, codenamed "Venice," which will offer confidential computing capabilities and built-in security features to safeguard sensitive AI workloads.
The advanced networking capabilities of the AI supercluster will be powered by Oracle Acceleron and AMD Pensando "Vulcano" AI-NICs, providing lossless, high-speed, and programmable connectivity. This will enable the supercluster to handle the demanding requirements of AI workloads, including large language models and generative AI. With this expanded partnership, Oracle is positioning itself as a major player in the AI infrastructure market, offering a competitive alternative to Nvidia's dominance.
The partnership has been well-received by investors, with AMD's shares rising 3% following the announcement. While Oracle's shares slipped 1.8%, the company's focus on AI infrastructure is expected to drive growth and innovation in the industry. As the demand for AI capabilities continues to grow, Oracle and AMD's expanded partnership is poised to deliver cutting-edge solutions for cloud customers.
