Recently, Dell launched its new Dell AI Platform at its AI factory, integrating Intel's latest Gaudi3 AI accelerator. This platform aims to simplify the adoption, integration, and scaling of AI technology for enterprises. This fully validated end-to-end solution combines the powerful performance of Intel Gaudi3 hardware, the flexibility of open-source software, and Dell's extensive experience in enterprise infrastructure, providing a convenient solution path for AI transformation across various industries.
"Our collaboration with Dell brings Intel Gaudi3's powerful capabilities into an integrated solution for enterprise deployment. This platform optimizes modern AI requirements, covering a variety of applications from large language models to edge inference while offering the flexibility and openness required by enterprises," said Saurabh Kulkarni, Vice President of Intel Data Center AI Strategy and Product Management.
Source Note: Image generated by AI, licensed by MidJourney
For enterprises, deploying AI often requires balancing high performance, scalability, flexibility, and cost-effectiveness. The launch of the Dell AI Platform precisely addresses this pain point. The Gaudi3 accelerator offers a 70% better price-performance ratio compared to Nvidia H100, particularly excelling in inference tasks for the Llama380B model. Additionally, the platform's scalable architecture evolves with changing enterprise needs, ensuring IT professionals maintain flexibility while meeting growing demands.
In supporting AI workloads, the core of the Dell AI Platform is the optimized PowerEdge XE9680 server. This server features eight Intel Gaudi3 accelerators, 128GB of high-bandwidth memory (HBM), and a bandwidth of 3.7TB per second, making it ideal for processing large language models and computer vision tasks. It also comes equipped with 5th-generation Intel Xeon processors, supporting up to 64 cores, with efficient complex computing capabilities.
To accelerate developer workflows, the platform integrates pre-validated open-source software stacks, supporting popular frameworks like PyTorch and Hugging Face, and introduces Kubernetes for flexible resource scheduling and orchestration. Moreover, Dell Enterprise Hub provides a series of models optimized for Intel Gaudi3, helping developers quickly implement applications.
Dell Omnia, as an open-source tool for high-performance cluster management, can dynamically allocate resources, allowing IT teams and developers to quickly adapt and optimize ever-changing workloads. This platform delivers the perfect combination of reliability and flexibility through seamless integration with various tools and frameworks.
For more information about the features of Intel Gaudi3 AI accelerators, visit Intel booth 654 or Dell booth 833 at Dell Technologies World Conference on May 19-22, 2025.