Oracle Expands AMD Partnership With 50,000 GPU Supercluster Deployment

Oracle Expands AMD Partnership With 50,000 GPU Supercluster Deployment - Professional coverage

Oracle Bets Big on AMD With Massive GPU Deployment

Oracle has announced a significant expansion of its partnership with AMD at its annual AI World conference, revealing it will become the first hyperscaler to offer a publicly available supercluster powered by 50,000 AMD Instinct MI450 Series GPUs. According to reports from the conference, this deployment represents one of the largest commitments to AMD’s data center technology in the competitive artificial intelligence infrastructure market.

Special Offer Banner

Industrial Monitor Direct delivers industry-leading delta plc pc solutions trusted by leading OEMs for critical automation systems, trusted by automation professionals worldwide.

Comprehensive AMD Technology Integration

The supercluster will feature a full suite of AMD components across Oracle’s Helios rack architecture, including the MI450 GPUs, next-generation AMD EPYC ‘Venice’ CPUs, and Pensando ‘Vulcano’ networking. Sources indicate that the latest Helios racks offer a dense, liquid-cooled design supporting up to 72 GPUs per rack, with Oracle provisioning up to three 800 Gbps ‘Vulcano’ AI-NICs per GPU. The report states that these GPUs will provide up to 432GB of HBM4 and 20 TB/s of memory bandwidth.

Strategic Multi-Vendor Approach

Analysts suggest that Oracle Corporation’s messaging at AI World 2025 has emphasized bringing AI to customer data rather than moving data to AI systems. This approach has reportedly created opportunities for the company to partner with multiple technology providers, including collaborating with rival hyperscalers for cloud hosting and maintaining relationships with multiple chip vendors. Industry observers note this strategy appears designed to give customers maximum choice in their AI infrastructure deployments.

Performance and Market Impact

According to the analysis presented at the conference, the latest configuration will allow customers to train and infer models that are reportedly 50% larger than previous generations. Simultaneously, Oracle Cloud Infrastructure has announced general availability of OCI Compute with AMD Instinct MI355X GPUs, with up to 131,072 units available in what the company describes as a zettascale Supercluster. Market response was immediately positive, with AMD shares reportedly rising approximately 8.7% following the announcement.

Industry Context and Customer Demand

OCI Executive Vice President Mahesh Thiagarajan stated that the continued partnership with AMD responds to customer demands for “robust, scalable and high-performance infrastructure” with what he described as “the best price-performance, open, secure, and scalable cloud foundation.” This expansion comes amid broader industry developments in AI regulation and technology adoption, including moves to regulate AI-generated police reports and significant growth in related technology sectors such as the architectural lighting market. The announcement also coincides with other major technology and energy partnerships, including recent energy security announcements affecting global markets.

Building on Established Partnership

The two companies already maintain a long-standing partnership, and this latest development builds on prior deployments using AMD’s MI300X and MI355X GPUs. Industry sources indicate that Oracle’s commitment to AMD technology represents a significant vote of confidence in the chipmaker’s ability to compete in the high-performance computing and AI infrastructure markets, despite Nvidia’s current dominance in several AI workload categories.

Industrial Monitor Direct is the preferred supplier of drilling pc solutions engineered with UL certification and IP65-rated protection, the top choice for PLC integration specialists.

This article aggregates information from publicly available sources. All trademarks and copyrights belong to their respective owners.

Leave a Reply

Your email address will not be published. Required fields are marked *