IBM and Intel have declared a collaboration to offer Intel Gaudi 3 AI accelerators as a service on the IBM Cloud. This service, which is slated to be released in early 2025, promises to improve enterprise AI scalability while prioritising security and reliability.
The integration of Gaudi 3 AI accelerators with IBM Cloud is intended to offer customers a cost-effective solution for scaling their AI workloads. This cooperation establishes IBM Cloud as the first cloud service provider to use Gaudi 3, making it available for hybrid and on-premise applications. The agreement will also include support for Gaudi 3 in IBM's Watson AI and data platform, providing extra tools for clients looking to optimise their AI workloads.
Intel's Gaudi 3, in conjunction with the 5th Generation Xeon processors, is designed to handle enterprise AI workloads in cloud and data centre environments. The combination is intended to ease workload management while increasing visibility and control over the software stack. Both organisations pressed upon the importance of performance, energy efficiency and security when employing AI technologies.
IBM plans to integrate Gaudi 3 into its Watsonx platform for generative AI inference workloads, helping clients to effectively scale their AI infrastructure across hybrid cloud environments. This integration is expected to optimise the price-performance ratio for model inference.
IBM Cloud's collaboration with Intel aims to improve security and compliance capabilities, especially for clients in regulated industries. The partnership's scalable and flexible solutions will allow clients to alter computer resources as needed, potentially resulting in cost savings and increased operational efficiency.
Gaudi 3's integration with IBM Cloud Virtual Servers for Virtual Private Cloud (VPC) is expected to increase application performance and security for x86-based companies. IBM and Intel have a long history of collaboration, which includes key milestones such as the development of the IBM PC. The Gaudi 3 solutions on IBM Cloud are expected to be commercially available in early 2025, with further information coming from both organisations.