Dell Technologies, the global leader in AI infrastructure, announced advancements to the Dell AI Factory at Dell Technologies World 2025, focusing on simplifying and accelerating AI deployments for organizations of all sizes. Key highlights include:
Cost-Effective AI Infrastructure
The Dell AI Factory offers up to 62% cost savings for on-premises LLM inferencing compared to public cloud solutions, addressing challenges like data quality, security, and high costs. Over 3,000 global customers across industries are leveraging Dell’s comprehensive AI portfolio, spanning client devices, data centers, edge, and cloud environments.
Innovative Cooling Solutions
The Dell PowerCool Enclosed Rear Door Heat Exchanger (eRDHx) reduces cooling energy costs by up to 60%, supports warmer water temperatures (32–36°C), and enables 16% more rack capacity without increased power use. It offers air cooling up to 80 kW per rack for dense AI and HPC workloads, with advanced leak detection and thermal monitoring.
High-Performance Servers
New Dell PowerEdge XE9785 and XE9785L servers, powered by AMD Instinct™ MI350 GPUs with 288 GB HBM3e memory, deliver up to 35x better inferencing performance in liquid- or air-cooled configurations, enhancing efficiency.
AI Data Platform Enhancements
Dell Project Lightning: The world’s fastest parallel file system, offering up to 2x greater throughput for faster AI training.
Dell Data Lakehouse: Simplifies AI workflows for use cases like recommendation engines and semantic search by creating AI-ready datasets.
Expanded AI Ecosystem: Dell collaborates with partners like Cohere, Google, Meta, Glean, and Mistral AI to deliver tailored, on-premises AI solutions, including autonomous workflows, enterprise search, and secure AI applications. The Dell AI Factory now supports AMD and Intel platforms, with optimized storage networking and software stacks.
Security and Resilience: New Dell AI Security and Resilience Services ensure full-stack protection for AI infrastructure, data, and models.
Leave a Reply
You must be logged in to post a comment.