Dell AI Factory With NVIDIA Unveils Next-Gen Supercomputing Server
Dell Technologies has unveiled a major addition to the Dell AI Factory with NVIDIA. The company introduced the new PowerEdge XE8812 server at ISC in Hamburg, Germany, on June 22, 2026. As a result, the launch marks a fresh leap for high-performance computing and AI infrastructure worldwide.
The PowerEdge XE8812 runs on NVIDIA Vera Rubin NVL4 architecture. Therefore, it scales up to 144 GPUs per rack. It sits inside the Dell PowerRack 9100, which follows OCP standards. Consequently, this setup pushes density and efficiency further than before.
Moreover, the server is fanless and uses direct liquid cooling. This design suits institutions running molecular and multi-physics simulations. In addition, it supports demanding AI workloads at scale.
Compared to the prior generation, the new platform brings major upgrades. For instance, cores grow from 144 to 176. Likewise, host memory and GPU memory both increase. As a result, organizations can run large models entirely in-memory. This approach avoids slow staging and swapping processes, which often hurt performance.
Furthermore, the rack design follows the open ORv3 standard. This choice improves efficiency and supports modular deployment. Meanwhile, tools like iDRAC and OpenManage Enterprise simplify monitoring. They also use real-time telemetry to catch problems early.
Additionally, Dell PowerRack offers a faster path to production. It arrives pre-validated and factory-integrated. With Dell ProDeploy support, teams can launch workloads in just over six hours, according to a Dell-commissioned report.
The momentum behind the Dell AI Factory with NVIDIA continues to grow globally. More than 5,000 customers already use this platform, based on Dell’s analysis of order data. For example, Dell and NVIDIA are helping build Doudna, a new U.S. Department of Energy supercomputer at Lawrence Berkeley National Laboratory. This system uses NVIDIA Quantum-X800 InfiniBand networking for large-scale AI training.
Similarly, in France, InstaDeep is scaling its Kyber supercomputing cluster through the Dell AI Factory with NVIDIA. Kyber delivers about 0.5 exaFLOPs of FP16 performance. It supports AI training and complex chip design tasks.
In the UK, the Wellcome Sanger Institute uses Dell PowerEdge XE-Series servers to decode DNA. The institute now completes one genome every seven hours. It also manages over 100 petabytes of data on-site.
Meanwhile, in Australia, Monash University built MAVERIC with Dell, NVIDIA, and CDC Data Centres. This supercomputer supports research in cancer detection and climate science.
Arun Narayanan of Dell Technologies said the new server matches the ambition of today’s biggest research efforts. Chris Marriott of NVIDIA added that the partnership raises the bar for AI and scientific computing infrastructure.
The Dell PowerEdge XE8812 will become globally available early next year.
