Illustrative product image. Confirm the exact supplied model and configuration.
Image 1 of 3
Use the full model, manufacturer part number and revision. Your quotation will list the condition and included components.
Build around your workload
HPE ProLiant Compute XD685
Model ProLiant Compute XD685
HPE ProLiant Compute XD685. Gpu: 8 x NVIDIA HGX B200 or B300 SXM6 GPUs in configuration-specific NVL8 builds; Form factor: 5U direct-liquid-cooled rack server. Ask for current pricing and help choosing the right configuration.
Form factor
5U direct-liquid-cooled rack server
Gpu
8 x NVIDIA HGX B200 or B300 SXM6 GPUs in configuration-specific NVL8 builds
Manufacturer model / P/N
ProLiant Compute XD685
Processor platform
2 x 5th Gen AMD EPYC 9005
Transaction requirements
For B300 / GB300 configurations
B300 & GB300 — In-stock transactions only
Direct paying buyers only.
Proof of funds and an in-person meeting at our office are required.
Check the installed modules, supported type and population rules.
Use this checklist to describe the change. The host system and full configuration determine which parts will fit.
03
YOUR QUOTE REQUEST
Tell us what you need
Add the quantity, options and installation details you want us to quote.
Confirm the exact manufacturer part number, host compatibility and included components. Optional software is quoted separately only when requested.
04
SUPPORT & REPLACEMENT
Plan for support and future replacement
Check support dates, replacement options and whether continued use still makes sense.
Support dates
We do not have confirmed sales and support dates for this part number.
Send us the full ProLiant Compute XD685 order code, software release and support details before planning long-term use.
Ask us to check the current sales and support status
Choosing a replacement
Products with similar names or connectors may still be incompatible. Compare the full specifications, installed system and software requirements.
Compare related products These products may be related, but we have not confirmed them as replacements.What should I compare?
Physical fit, connectors and power
Capacity and required performance
Software, firmware and feature support
Licensing, service coverage and data and configuration migration
USING HARDWARE RESPONSIBLY
Maintain, redeploy or replace
Consider repairability, remaining support and actual operating demand. Extending service life is useful only when the equipment still meets your security and reliability requirements.
MaintainCheck parts, firmware and service options.
RedeployMatch a suitable role and securely erase any retained data.
Retire responsiblyUse documented data removal and an appropriate electronics recycler.
Compare operating energy
Use measured average input power for the same workload. PSU ratings and processor TDP are different from total system power.
Energy difference = (current W − proposed W) × hours × units ÷ 1,000. Excludes cooling, embodied emissions and purchase costs.
CONNECTION DETAILS
Network and storage connections
Choose the host adapters around the workload and the storage protocol. A server NIC, Fibre Channel HBA and SAS HBA serve different connections.
CONNECTS TOUsers & application traffic
→
ProLiant Compute XD685GPU Platform
→
USED FORNetwork and storage services
What to check before installation
Qualified adapters, slots, drivers and optics
Network paths, storage protocol and operating system
Browse the categories above for related equipment, then send us the full configuration so we can check the fit.
PRODUCT DETAILS
Features and applications
Manufacturer model / P/N
ProLiant Compute XD685
Form factor
5U direct-liquid-cooled rack server
Processor platform
2 x 5th Gen AMD EPYC 9005
Accelerator subsystem
8 x NVIDIA HGX B200 or B300 SXM6 GPUs in configuration-specific NVL8 builds
System memory
DDR5 RDIMM platform; capacity and speed depend on CPU and BOM
Local storage
NVMe data and M.2 boot options selected by configuration
Network and I/O fabric
High-speed GPU fabric and north-south adapters configured to order
Power architecture
Redundant high-wattage hot-plug PSU subsystem; exact set by configuration
Cooling
Direct liquid cooling
Confirm option codes, firmware, region, lifecycle and the supplied configuration with your quotation. Transactions follow the direct-buyer and in-person meeting policy above.
HPE / order paths
Part numbers and ordering options
Keep the HPE order references separate
Option P/N
Requested upgrade or option-kit identity
Spare P/N
Service-replacement identifier from the installed item
System SKU
Base or configured system to which the requirement belongs
Manufacturer
HPE
Model
ProLiant Compute XD685
Manufacturer P/N
Confirmed in the quotation
Server configuration
Selection and installation requirements
model receiving and deployment photography
HPE ProLiant Compute XD685: three physical views
Three model-bound views are provided for physical review. Final labels, option population and accessories follow the allocated unit and written BOM.
Match the modelConfirm model, GPU generation, cooling, drive layout and power input before order approval.Power-state ruleLEDs are shown lit only where power cables are visibly connected; disconnected equipment remains unlit.Approve before dispatchRequest allocated-unit labels, serials, physical condition and export-packing photographs.
Form factor5U direct-liquid-cooled rack serverProcessor platform2 x 5th Gen AMD EPYC 9005Accelerator subsystem8 x NVIDIA HGX B200 or B300 SXM6 GPUs in configuration-specific NVL8 buildsCoolingDirect liquid cooling
02 / ARCHITECTURE
Compute and fabric architecture
A 5U AMD EPYC AI server with manufacturer-documented liquid-cooled HGX B200 and B300 configurations.
Foundation-model training and fine-tuning
Large-scale inference and reasoning
RAG and multimodal pipelines
HPC, simulation and scientific computing
B200 / B300
NVIDIA SXM accelerator modules. These names alone do not identify a complete server.
HGX B200 / B300
An eight-GPU NVL8 baseboard/platform integrated into a named NVIDIA or OEM system.
HPE ProLiant Compute XD685
The complete named system on this page, with its own chassis, CPU, cooling, power and orderable configuration.
Cisco UCS B200
An unrelated Cisco blade-server model. It is not an NVIDIA B200 GPU system.
01
Rack and handling
Confirm rack width/depth, rail type, service clearance, floor loading and a rated material lift.
02
Power
Validate voltage, connectors, branch capacity, redundancy and sustained/peak rack power from the final BOM.
03
Cooling
Match air or liquid cooling, CDU and facility-water requirements, hose routing and leak detection.