Dell introduced PowerEdge XE9685L and XE7740 servers for AI infrastructure

The PowerEdge XE9685L model in the 4U form factor is designed to install two AMD EPYC Turin processors. Liquid cooling is used. 12 PCIe 5.0 expansion card slots are available. There is talk about the possibility of using NVIDIA HGX H200 or B200 accelerators.

Dell says the PowerEdge XE9685L offers the industry’s highest GPU density with support for up to 96 NVIDIA accelerators per rack. The new product is suitable for organizations solving large-scale computing problems, such as creating large AI models, running complex simulations, or performing genomic sequencing. The server design provides optimal thermal performance under high workloads, and the presence of a liquid-liquid system improves energy efficiency.

Image source: Dell

The second model, PowerEdge XE7740, also has a 4U size, but uses air cooling. It is possible to install two Intel Xeon 6 processors based on powerful P-core cores (Granite Rapids). Customers will be able to choose configurations with eight double-wide AI accelerators, including Intel Gaudi 3 and NVIDIA H200 NVL, as well as 16 single-wide accelerators, such as NVIDIA L4. The server is suitable for various use cases, such as fine-tuning generative AI models, inference, data analytics, etc. The machine is designed to effectively balance cost, performance and scalability.

Dell is also preparing to release a new PowerEdge XE server based on the NVIDIA GB200 Grace Blackwell NVL4. There is talk of support for up to 144 GPUs per 50OU rack (Dell IR7000).