ARBOR EdgeX-6000 — AMD EPYC Embedded 8004 Edge AI HPC System with Quad Dual-Slot GPUs

Specifications

Overview

The ARBOR EdgeX-6000 is a GPU-dense edge AI HPC system built around a single AMD EPYC Embedded 8004 Series processor running to 64 cores and 128 threads. Six channels of DDR5-4800 ECC memory carry up to 576 GB — 96 GB per DIMM — which is the memory class normally reserved for rack servers, and it is what allows large models and genomic or imaging datasets to be held resident at the edge.

PCIe expansion is the reason this system exists: it accepts up to four dual-slot GPUs, or eight single-slot accelerator cards in the same chassis. Front-accessible E1.S NVMe bays keep high-throughput storage serviceable from the aisle, and 10GbE links the unit to the rest of the site. The combination supports both inference and training at the edge rather than inference only.

ARBOR positions the EdgeX-6000 at genomic sequencing and medical imaging in particular — workloads where data volume and privacy rules make shipping raw data to a public cloud impractical. The system won an Embedded Computing Design Best-in-Show award at embedded world 2026 in the HPC category.

Key Benefits

64 cores / 128 threads of EPYC Embedded 8004 plus 6-channel DDR5-4800 ECC to 576 GB. Four dual-slot GPUs (or eight single-slot cards) in one edge chassis. Front-serviceable E1.S NVMe and 10GbE networking. Award-winning platform (embedded world 2026 Best-in-Show, HPC). Sourced through QS Compute with volume pricing.

Applications

Genomic sequencing and bioinformatics, medical imaging and clinical AI, on-premises model training and fine-tuning, industrial AI R&D labs, and defense or privacy-constrained sites where data cannot leave the facility.

Related Products