699556061cffd280693973-af1e6e404e163ca23c624dbd6f871243ecf33e7c_ASUS_ESC8000A_E13P_front.png

699556061d5f0577232310-651ac020c1cbcb869a106a58949248c645d4d01d_ASUS_ESC8000A_E13P_back.png

699556061dbb3388676875-85484bfbb544cda7ac5984bc4a1aae08fa0a76b1_ASUS_ESC8000A_E13P_sideopen.png

SKU: SCS-ESC8000A-E13P

ASUS ESC8000A-E13P

4U dual-socket GPU server, 8 NVMe hot-swap drives, 8 PCIe 5.0 x16 dual-slot GPU slots and NVIDIA MGX compatible

Manufacturer:

Specifications

Dual-socket AMD EPYC 9005 processors (SP5, up to 500W TDP, 192 Zen 5c cores)
24 DDR5 slots (up to 6TB, 6400 MT/s, 12 channels per CPU)
8x 2.5" NVMe hot-swap + 2x M.2 PCIe 5.0 x4
8x PCIe 5.0 x16 dual-slot GPUs + 5x PCIe 5.0 x16 (NIC/Storage) + 1x PCIe 5.0 x8
Dual 10GbE (Intel X710-AT2) + dedicated Mgmt LAN (AST2600 BMC)
3+1 redundant 3200W 80PLUS Titanium power supplies

This ASUS server targets extreme GPU workloads in AI training/inference, High Performance Computing (HPC) and enterprise AI applications. Primary markets include generative AI, Large Language Models (LLM), scientific simulations and data-intensive analytics environments. The GPU-optimized design supports up to eight dual-slot GPUs (e.g. NVIDIA H200/H100/RTX PRO 6000 Blackwell) with PCIe 5.0 x16 connectivity and optional NVLink bridge support in NVIDIA MGX architecture. Storage includes eight 2.5" NVMe hot-swap bays plus two M.2 PCIe 5.0 slots. Networking ranges from dual 10GbE (Intel X710-AT2) to dedicated 400GbE PCIe cards, complemented by ASMB11-iKVM remote management. Cooling with 5 system fans is designed for dual EPYC 9005 (up to 500W TDP) and GPU TDP up to 600W.