AI at Full Throttle.
When compromise is off the table, this is the AI system to build around. Designed for large scale inference, massive models, and nonstop demand, it brings formidable GPU compute, fast local storage, and high-speed networking together in one 4U platform. One serious investment, with room to scale for whatever comes next.
System Requirements:
Power: 4x dedicated 20A circuits (220-240V). A redundant UPS and PDU solution are recommended.
Rackmount: 4U of space in a standard 19 inch, four post rack. The chassis depth is 800mm. (1-meter rail kit included)
Environment: Dedicated server rooms with HVAC required (system heat produced is up to 20,500 BTU/hr)
Installation: At least two people or a suitable lifting device is required for installation. Total system weight: Up to 60KG (depending on the configuration)
Availability:
Estimated Lead Time: 30-35 business days from order confirmation.
Special Orders: This configuration requires multiple special-order items that may affect lead times. Contact us for more details.
AI Inference Server – Dual AMD EPYC | 8x GPU | 4U Rackmount
AI at Full Throttle.
When compromise is off the table, this is the AI system to build around. Designed for large scale inference, massive models, and nonstop demand, it brings formidable GPU compute, fast local storage, and high-speed networking together in one 4U platform. One serious investment, with room to scale for whatever comes next.
System Requirements:
Power: 4x dedicated 20A circuits (220-240V). A redundant UPS and PDU solution are recommended.
Rackmount: 4U of space in a standard 19 inch, four post rack. The chassis depth is 800mm. (1-meter rail kit included)
Environment: Dedicated server rooms with HVAC required (system heat produced is up to 20,500 BTU/hr)
Installation: At least two people or a suitable lifting device is required for installation. Total system weight: Up to 60KG (depending on the configuration)
Availability:
Estimated Lead Time: 30-35 business days from order confirmation.
Special Orders: This configuration requires multiple special-order items that may affect lead times. Contact us for more details.
From: $212,607.00
Made in Canada
Tested for performance
AI at Full Throttle.
When compromise is off the table, this is the AI system to build around. Designed for large scale inference, massive models, and nonstop demand, it brings formidable GPU compute, fast local storage, and high-speed networking together in one 4U platform. One serious investment, with room to scale for whatever comes next.
System Requirements:
Power: 4x dedicated 20A circuits (220-240V). A redundant UPS and PDU solution are recommended.
Rackmount: 4U of space in a standard 19 inch, four post rack. The chassis depth is 800mm. (1-meter rail kit included)
Environment: Dedicated server rooms with HVAC required (system heat produced is up to 20,500 BTU/hr)
Installation: At least two people or a suitable lifting device is required for installation. Total system weight: Up to 60KG (depending on the configuration)
Availability:
Estimated Lead Time: 30-35 business days from order confirmation.
Special Orders: This configuration requires multiple special-order items that may affect lead times. Contact us for more details.
Platform: ASUS ESC8000A-E13P 4U Rackmount | NVIDIA MGX Architecture
CPUs: 2× AMD EPYC 9005 Processors (Up to 192 Cores each | 384 Cores Total)
GPU: 8× NVIDIA Server Edition Graphics Cards | Up to 141 GB HBM3e each | 1128 GB total GPU memory
AI Performance: Up to 4 PFLOPS FP4 per GPU | 5th-Gen Tensor Cores | MIG support (up to 7 instances/GPU)
System RAM: Up to 3 TB DDR5-6400 ECC Registered (24 × 128GB)
Storage: 2x M.2 Gen5 22110 SSDs and up to 8x Gen5 NVMe U.2 SSDs
Networking: 2× 10GbE LAN (Intel X710) | 1× Dedicated Management Port
Connectivity: 8× PCIe 5.0 x16 GPU slots | 5× PCIe 5.0 x16 NIC slots
Power: 3+1 Redundant 3200W 80 PLUS Titanium PSUs
Management: ASUS ASMB12-iKVM | IPMI | Redfish API | ASUS Control Center Enterprise
Form Factor: 4U Rackmount
Rack Rails: Factory-bundled sliding mounting rail kit Included
