Oracle and AMD have announced a strategic partnership to create a massive AI compute cluster powered by 50,000 AMD MI450 GPUs. The initiative, planned for rollout in the third quarter of next year, will be hosted at Oracle’s data centers and is expected to reach a total capacity of approximately 200 megawatts.
High-Performance Infrastructure for AI
The new cluster will leverage AMD’s Helios server systems, combining MI450 GPUs with AMD CPUs, designed to compete directly with Nvidia’s upcoming Vera Rubin lineup. The collaboration reflects the surging demand for AI computing resources and represents a significant investment in next-generation data center infrastructure.
Strategic Partnership Through 2027
Oracle and AMD plan to continue their collaboration through 2027 and beyond, emphasizing the long-term need for AI infrastructure. While financial details remain undisclosed, the partnership strengthens both companies’ positions amid intensifying competition in AI chips and cloud platforms.
Market and Stock Impact
Following the announcement, AMD shares rose over 2% in pre-market trading, while Oracle stock experienced a modest decline of around 1%. Previously, AMD secured a multi-year agreement with OpenAI, expected to generate up to $100 billion in revenue, highlighting its growing influence in AI hardware.
Oracle has also committed to providing $300 billion in cloud capacity over five years, reflecting the skyrocketing demand for AI resources. According to Mahesh Thiagarajan, EVP of Oracle Cloud Infrastructure, the current demand exceeds available capacity, forcing many clients to wait. The new cluster with AMD aims to alleviate some of this shortage, particularly in public cloud segments.
Competing in the AI Era
The AMD Helios systems, integrating high-performance GPUs and CPUs, will position Oracle’s data centers to compete with Nvidia’s offerings next year. This strategic move underscores the escalating race for AI compute power, critical for both cloud providers and AI-driven enterprises.
