Complementing the Xeon 6 processors, Intel’s Gaudi 3 AI accelerators are specifically engineered for large-scale generative AI tasks. The Gaudi 3 is equipped with 64 Tensor Processor Cores (TPCs) and eight Matrix Multiplication Engines (MMEs), which facilitate accelerated deep neural network computations. It includes 128 gigabytes of HBM2e memory, ensuring efficient data handling for both training and inference processes. Additionally, the accelerator offers 24 200 Gigabit Ethernet ports to support scalable networking needs. The Gaudi 3 is fully compatible with the PyTorch framework and supports advanced models from Hugging Face, enabling seamless integration with existing AI development tools. In collaboration with IBM, Intel is deploying Gaudi 3 accelerators as a service on IBM Cloud, aiming to reduce the total cost of ownership (TCO) and enhance the scalability of AI deployments for enterprises.

Intel’s introduction of the Xeon 6 CPUs and Gaudi 3 AI accelerators underscores its commitment to optimizing TCO and performance per watt through its established x86 infrastructure and extensive open ecosystem. Currently, 73% of GPU-accelerated servers utilize Intel Xeon as the host CPU, highlighting its widespread adoption in the industry. Partnerships with original equipment manufacturers (OEMs) such as Dell Technologies and Supermicro enable the development of co-engineered systems tailored to specific customer requirements, facilitating effective AI implementations. Intel also addresses the challenges of transitioning generative AI solutions from prototypes to production-ready systems through collaborative engineering efforts and the Open Platform Enterprise AI (OPEA) framework. This framework integrates scalable retrieval-augmented generation (RAG) systems optimized for Xeon and Gaudi accelerators, allowing customers to easily integrate applications from platforms like Kubernetes and Red Hat OpenShift AI. Additionally, Intel’s Tiber portfolio provides comprehensive solutions that tackle issues related to access, cost, complexity, security, efficiency, and scalability across AI, cloud, and edge environments. The Intel Tiber Developer Cloud offers preview systems of Xeon 6 for technical evaluation and testing, while select customers receive early access to Gaudi 3 accelerators to validate AI model deployments ahead of large-scale production rollouts. New service offerings, such as the SeekrFlow AI platform, incorporate the latest Intel Gaudi software and tools, including support for Xeon 6 processors, to facilitate the development of reliable and efficient AI applications.