Oracle And AMD Expand Partnership To Deliver Next-Generation Ai Superclusters

Oracle and AMD will deepen their collaboration with plans to launch a large-scale Ai supercluster from the third quarter of 2026, with Oracle becoming the first hyperscaler to publicly offer infrastructure powered by 50,000 AMD Instinct MI450 Series GPUs.

The companies announced the agreement in Sydney on October 15, highlighting how demand for high-performance Ai compute continues to grow as next-generation models surpass the limits of existing clusters.

Building On Long-Term Collaboration

The announcement extends Oracle and AMD’s partnership, which began with Oracle Cloud Infrastructure (OCI) adoption of AMD Instinct MI300X GPU-powered shapes in 2024. This was followed by the release of AMD Instinct MI355X GPUs for OCI compute earlier this year.

Mahesh Thiagarajan, Executive Vice President of Oracle Cloud Infrastructure, said the extended partnership would provide the scale required for emerging Ai workloads.

“Our customers are building some of the world’s most ambitious Ai applications, and that requires robust, scalable, and high-performance infrastructure,” he said.

“By bringing together the latest AMD processor innovations with OCI’s secure, flexible platform and advanced networking powered by Oracle Acceleron, customers can push the boundaries with confidence. Through our decade-long collaboration with AMD we’re continuing to deliver a scalable cloud foundation to meet customer needs for this next era of Ai.”

AMD Technology At The Core

The planned superclusters will use AMD’s new “Helios” rack design. This includes Instinct MI450 GPUs, next-generation EPYC CPUs codenamed Venice, and Pensando networking technology codenamed Vulcano.

Forrest Norrod, Executive Vice President and General Manager of AMD’s Data Center Solutions Business Group, said the collaboration is focused on enabling customers to advance their Ai initiatives.

“With our AMD Instinct GPUs, EPYC CPUs, and advanced AMD Pensando networking, Oracle customers gain powerful new capabilities for training, fine-tuning, and deploying the next generation of Ai,” he said.

“Together, AMD and Oracle are accelerating Ai with open, optimized, and secure systems built for massive data centers.”

Technical Capabilities

Each AMD Instinct MI450 GPU will offer up to 432 GB of HBM4 and 20 TB/s of memory bandwidth, enabling models 50 per cent larger than previous generations to run entirely in memory.

The Helios rack design will be liquid-cooled and support 72 GPUs per rack, with dense scale-up connectivity and Ethernet-based networking aligned with Ultra Ethernet Consortium standards.

Other features include:

• Next-generation EPYC CPUs providing confidential computing and security features
  
  
• Pensando DPU-based converged networking to accelerate data ingestion and improve security
  
  
• Support for up to three 800 Gbps Pensando Vulcano AI-NICs per GPU for high-speed connectivity
  
  
• UALink protocol for direct, hardware-coherent networking and memory sharing across GPUs
  
  
• Full support for the open-source ROCm software stack to maintain vendor flexibility
  
  
• Advanced partitioning and virtualization for multi-tenant use and efficient GPU allocation

OCI Supercluster Roadmap

To complement the launch of the MI450-based clusters, Oracle announced the general availability of OCI compute instances powered by AMD Instinct MI355X GPUs. These will operate in OCI’s zettascale supercluster environment, capable of scaling to 131,072 GPUs.

The companies said the expanded infrastructure is intended to support customers training large language models, running generative Ai workloads, and advancing high-performance computing use cases.

Oracle confirmed the first deployment of the MI450 supercluster is scheduled for Q3 2026, with expansion in 2027 and beyond.

Last Updated on October 15, 2025 by Nick Ross