IBM Cloud will integrate AMD Instinct MI300X accelerators into its platform to enhance generative AI applications and high-performance computing.
IBM and AMD have announced a strategic collaboration that will enable the deployment of the advanced AMD Instinct MI300X accelerators as a service on the IBM Cloud platform. This new service, expected in the first half of 2025, aims to optimize performance and energy efficiency in generative artificial intelligence (AI) models and high-performance computing (HPC) applications, thereby solidifying both companies as leaders in enterprise AI-based solutions.
Technology Serving Artificial Intelligence
The AMD Instinct MI300X accelerator is designed to handle the demands of the largest and most complex AI models. With a memory capacity of 192 GB of HBM3 (high bandwidthBandwidth is the maximum transfer capacity of…), these accelerators can support large-scale model inferences and adjustments, reducing the need for multiple GPUs and, consequently, associated costs.
Alan Peacock, General Manager of IBM Cloud, stated: “AMD and IBM Cloud share a common vision: to empower businesses with the necessary tools to achieve their AI goals. With this collaboration, our enterprise clients will have a powerful option to scale their AI deployments, optimizing both cost and performance.”
Philip Guido, Executive Vice President of AMD, emphasized: “As companies adopt larger AI models and more complex datasets, it’s essential to have accelerators that can flexibly and efficiently handle intensive workloads. Our collaboration with IBM will enable the execution of generative AI models at scale without compromising cost or performance.”
Key Benefits of the Service
The integration of the MI300X accelerators into IBM Cloud will bring a number of benefits designed to meet the needs of businesses across all sectors, including highly regulated ones:
- Support for large model inferences: The MI300X accelerators provide the capability to efficiently manage advanced and complex models, reducing the number of GPUs needed.
- Enhanced performance and optimized security: Through IBM Cloud KubernetesKubernetes (commonly referred to in English as “K8s”) … Services and Red Hat OpenShift AI, users will be able to run AI applications with greater performance and ensure security in hybrid cloud environments.
- Advanced infrastructure for generative AI: This service will integrate the accelerators with IBM’s watsonx AI platform, providing businesses with additional tools to scale their workloads in hybrid environments.
- Cost and energy consumption reduction: The high energy efficiency of the MI300X contributes to minimizing the economic and environmental impact of large-scale operations.
Impact on the Business Industry
The deployment of these accelerators on IBM Cloud marks a significant milestone in the evolution of cloud computing. This service will allow businesses to leverage advanced capabilities of generative AI and HPC to transform processes and optimize operations in key areas such as finance, healthcare, manufacturing, and technology.
Additionally, IBM Cloud ensures that this new solution will provide its security and regulatory compliance capabilities, which are fundamental for clients in sensitive sectors.
via: IBM