Dell has announced a major expansion to its AI Factory during the SC24 high-performance computing conference in Atlanta.
The Dell AI Factory is a comprehensive framework designed by Dell to help organizations accelerate their adoption and deployment of artificial intelligence (AI) at scale. It offers a wide range of AI-focused products, solutions, and services, tailored to support AI workloads across various environments, including on-premises data centers, cloud platforms, edge locations, and even AI-enabled PCs.
The announced update includes new high-density servers and extended services aimed at simplifying how enterprises scale AI across their operations. By enhancing its Integrated Rack Scalable Systems (IRSS) program and deepening partnerships, Dell is positioning itself as a comprehensive provider for AI deployment.
Next-Level Server Integration in AI Factory
The newly introduced PowerEdge XE9685L and XE7740 servers will be part of the IRSS framework, which provides plug-and-play rack-scale systems. This program supports both air and liquid cooling configurations, enabling businesses to choose based on their workload needs.
The 4U liquid-cooled PowerEdge XE9685L comes with dual 5th Generation AMD EPYC processors and either NVIDIA HGX H200 or B200 platforms. This setup supports up to 96 GPUs per rack, making it one of the highest-density options available for machine learning and high-performance computing tasks.
Narayanan, Dell’s SVP of server and networking products, highlighted that these units are configured at the factory for a seamless customer experience. It is designed so that clients can simply receive and start using the rack with minimal setup.
The PowerEdge XE7740, also a 4U model but air-cooled, utilizes Intel’s Xeon 6 processors with P-Cores and can hold up to eight double-wide or 16 single-wide accelerators. This design flexibility allows businesses to customize their servers according to specific AI inferencing or model-tuning needs.
NVIDIA Partnership and New Chip Support
A significant part of this expansion revolves around Dell’s continued collaboration with NVIDIA. The AI Factory now integrates NVIDIA HGX H200 and H100NVL Tensor Core GPUs, achieving up to 1.9 times the performance of the previous H100 units.
These improvements enhance complex data processing and AI-driven query capabilities, including retrieval-augmented generation workflows.
Looking ahead, Dell plans to incorporate the NVIDIA GB200 Grace Blackwell NVL4 Superchip in its future server models for the Integrated Rack 7000 system. This server is built for intensive AI workloads and can hold up to 144 GPUs in a 50OU rack, maximizing both space and performance.
Strengthening Data Capabilities with Apache Spark
An essential update to Dell’s Data Lakehouse involves integrating Apache Spark, which enables distributed data processing for better scalability. Apache Spark processes data in memory, speeding up complex analytics tasks and enhancing efficiency. This upgrade helps organizations manage vast amounts of data while maintaining a unified approach to data processing and analysis.
To support developers, Dell launched Validated Designs for AI PCs, which are open-source guides that facilitate building AI applications. These modular designs are customizable, enabling integration of large language models, visual tools, and speech processing capabilities. This flexibility ensures that businesses can deploy AI solutions across various platforms, aligning with diverse hardware specifications.
Enhanced Professional Services for AI Integration
Dell is addressing skill gaps in AI adoption through expanded professional services. The company now offers support for sustainable data centers, AI-optimized networking, and comprehensive data management. New services include implementing ServiceNow’s Now Assist, which applies generative AI to streamline workflow productivity through automated content creation.
The PowerEdge XE9685L is set to launch globally in Q1 2025, followed by the XE7740 in Q2 2025. The updated Data Lakehouse features, along with Dell’s Validated Designs for AI PCs, are available now. Other services are being rolled out selectively in various regions.
