Dell: Server for AI Workloads & AI Factory with NVIDIA

Dell revealed the Dell PowerEdge XE8812 server, which joins the Dell AI Factory with NVIDIA portfolio.

Dell positions the system as designed specifically for organisations handling the most intensive high performance computing (HPC) and AI workloads, equipped with NVIDIA Vera Rubin NVL4 architecture.

With capacity for up to 144 GPUs per rack, the XE8812 could represent what Dell describes as "a generational leap in compute".

According to a McKinsey report, compute power is becoming one of the most vital resources of this decade as the AI sector continues to expand.

Server specifications and capabilities

Dell positions its fanless, direct liquid cooled Dell PowerEdge XE8812 server as built for leading institutions managing complex HPC and AI workloads, including molecular and multi-physics simulations.

The platform transitions from NVIDIA GB200 NVL4 to NVIDIA Vera Rubin NVL4, bringing enhanced host memory, additional cores (increasing from 144 to 176), greater GPU memory and supplementary compute capabilities.

The architectural improvements enable the server to handle increasingly complex computational tasks that require substantial processing power. 

The direct liquid cooling system represents a significant engineering achievement, enabling higher density deployments whilst maintaining optimal operating temperatures.

This cooling approach eliminates the need for traditional fan-based systems, reducing noise and improving energy efficiency.

Arun Narayanan, Senior Vice President of Compute and Networking at Dell, says: "The institutions doing the world's most important research, like decoding the human genome, modelling the energy systems of the future and building the sovereign AI infrastructure that nations depend on, deserve infrastructure that matches the ambition of their work.

"The Dell PowerEdge XE8812 reflects Dell's commitment to pushing the boundaries of what's possible, giving these organisations the density, memory and open architecture they need to tackle workloads that once seemed impossible."

Addressing infrastructure demands

The volume and speed of AI and HPC workloads are surpassing what gradual infrastructure improvements can accommodate, whilst the worldwide drive for AI innovation is intensifying requirements for high-performance infrastructure across data and compute operations.

According to Gartner's latest forecast on AI spending trends, global AI spending is projected to total US$2.52tn by next year, representing a 44% increase year-over-year.

Gartner's analysis also suggests AI infrastructure will contribute US$401bn in spending during the same period. This substantial investment reflects the growing recognition amongst enterprises and research institutions that robust computing infrastructure is essential for maintaining competitive advantage.

The increasing complexity of AI models and HPC simulations requires infrastructure that can scale efficiently whilst delivering consistent performance. Traditional computing approaches are proving inadequate for the demands of modern AI and scientific computing workloads.

Chris Marriot, NVIDIA's Vice President of Enterprise Platforms, says: "The convergence of AI and HPC is redefining what organisations should expect from their infrastructure. 

"Dell and NVIDIA are raising that bar together, combining NVIDIA Vera Rubin NVL4 architecture and CUDA-X libraries with Dell's engineering and at-scale deployment expertise to provide the performance, efficiency and openness required for the world's most demanding AI and scientific computing workloads."

Portfolio expansion and deployments

The new PowerEdge XE8812 server will integrate into Dell's AI Factory with NVIDIA, a collection of products, solutions and services configured for AI workloads that currently serves over 5,000 customers worldwide. 

Working alongside NERSC, Dell and NVIDIA are developing Doudna, the forthcoming flagship US Department of Energy supercomputer.

This system will utilise Dell PowerEdge XE8812 servers featuring NVIDIA Vera Rubin NVL4 and connected through NVIDIA Quantum-X800 InfiniBand networking, supporting larger-scale HPC workloads, AI training and inference alongside data intensive workflows.

According to Dell, the system will accelerate breakthroughs from the molecular level to astronomy, reshaping science and everyday life. The company's customer base includes organisations such as French AI company InstaDeep, the Wellcome Sanger Institute in the UK (deploying it for DNA decoding) and Monash University in Australia.