8× NVIDIA A100 80GB SXM4 Microsoft ZT C2660 GPU Servers Available from Stock
Microsoft / ZT Systems C2660 GPU servers equipped with 8× NVIDIA A100 80GB SXM4 GPUs, 2TB DDR4 RAM and dual AMD EPYC processors are available through the server-parts.eu supply network, subject to stock confirmation.
NVIDIA A100 80GB GPU Servers
Special Pricing & No Lead Time
The Microsoft / ZT Systems C2660 is a powerful 6U AI server featuring 8× NVIDIA A100 80GB SXM4 GPUs, 640GB GPU memory and NVIDIA NVLink/NVSwitch technology. Ideal for AI training, machine learning, inference and high-performance computing.
Customers interested in purchasing Microsoft / ZT Systems NVIDIA A100 GPU servers are invited to contact server-parts.eu for current pricing, availability, technical specifications and delivery arrangements.
Microsoft ZT NVIDIA A100 80GB Systems C2660 – Stock Availability
The Microsoft / ZT Systems C2660 is a hyperscaler-class GPU server platform built around the NVIDIA HGX A100 8-GPU architecture.
server-parts.eu can assist customers looking to source these systems from the refurbished enterprise hardware market.
The referenced configuration has been offered from UK stock, with current quantities and shipping availability subject to confirmation.
Availability details:
Server model: Microsoft / ZT Systems C2660
GPU configuration: 8× NVIDIA A100 80GB SXM4
GPU platform: NVIDIA HGX A100 8-GPU
Total GPU memory: 640GB HBM2e
Processor: 2× AMD EPYC 7V12
System memory: 2TB DDR4 ECC
Storage: 3.84TB NVMe SSD
Networking: 8× Mellanox ConnectX-6 HDR 200Gb/s
Form factor: 6U
Stock location: UK stock referenced; current location subject to confirmation
Condition: Configured enterprise GPU servers; condition confirmed per offer
Shipping: Available upon request
Warranty: Subject to quotation
Pricing: Available upon request
Availability may change without notice.
Customers planning NVIDIA A100 GPU infrastructure projects are encouraged to
contact us for current stock information and a detailed quotation.
Microsoft ZT Systems C2660 NVIDIA A100 80GB Server Specifications
The Microsoft / ZT Systems C2660 is an AMD EPYC-based GPU computing server designed for demanding AI and HPC environments.
The NVIDIA HGX A100 platform combines eight NVIDIA A100 SXM4 GPUs with high-speed NVLink and NVSwitch connectivity.
The offered configuration includes:
Server model: Microsoft / ZT Systems C2660
Server type: 6U enterprise GPU compute server
GPU: 8× NVIDIA A100 80GB SXM4
GPU architecture: NVIDIA Ampere
GPU platform: NVIDIA HGX A100 8-GPU
GPU memory: 80GB HBM2e per GPU, 640GB total
GPU interconnect: NVIDIA NVLink 3.0 / NVSwitch
GPU power rating: Approximately 400W per GPU
Processor: 2× AMD EPYC 7V12
CPU specification: 64 cores per processor, 2.40GHz, 256MB cache, 240W
Total CPU cores: 128
System memory: 32× 64GB DDR4-2933 ECC RDIMM
Total system RAM: 2TB
Storage: 3.84TB M.2 NVMe SSD
Storage adapter: Microsoft ZTSYS Quad M.2 NVMe Adapter
Primary networking: 8× Mellanox ConnectX-6 HDR 200Gb/s InfiniBand / 200GbE adapters
Network adapter model: MCX653105A-HDAT
Network interface: QSFP56, PCIe x16
Aggregate network port bandwidth: Up to 1.6Tb/s across eight 200Gb/s interfaces
Power supplies: 6× LiteOn PS-2302-3N 3000W
Installed PSU capacity: 18kW
Input voltage: 200–240V AC, 50/60Hz
Power redundancy: Redundant PSU configuration
Rack height: 6U
Rail kit: Included in the referenced configuration
The full hardware configuration, component quantities and available accessories can be provided upon request. Final specifications should be confirmed before ordering.
NVIDIA A100 80GB SXM4 – 640GB GPU Memory per Server
One of the main advantages of the Microsoft / ZT Systems C2660 is its eight-GPU NVIDIA HGX A100 configuration.
Each server incorporates eight NVIDIA A100 80GB SXM4 GPUs, providing a combined 640GB of HBM2e GPU memory.
The NVIDIA A100 is based on the NVIDIA Ampere architecture and is designed for AI training, deep learning, inference and high-performance computing.
The SXM4 configuration supports high-speed communication between GPUs through NVIDIA NVLink and NVSwitch technology.
Key GPU features include:
8× NVIDIA A100 SXM4 GPUs
80GB HBM2e memory per GPU
640GB total installed GPU memory
NVIDIA Ampere architecture
NVIDIA NVLink 3.0
NVIDIA NVSwitch interconnect
NVIDIA HGX A100 8-GPU platform
Support for AI training and inference workloads
GPU acceleration for HPC and scientific computing
The high-speed GPU interconnect is particularly useful for workloads that require frequent communication between multiple GPUs.
For organizations looking to purchase NVIDIA A100 80GB servers, the HGX A100 SXM4 platform remains an option for GPU-intensive computing environments.
Microsoft ZT Systems C2660 NVIDIA A100 80GB – Dual AMD EPYC Processors and 2TB RAM
The Microsoft / ZT Systems C2660 combines NVIDIA GPU acceleration with AMD EPYC server processors and high-capacity ECC memory.
The referenced configuration includes two AMD EPYC 7V12 processors.
Each processor provides 64 cores, delivering 128 physical CPU cores per server.
CPU and memory configuration:
2× AMD EPYC 7V12 processors
64 CPU cores per processor
128 total CPU cores
2.40GHz base frequency
256MB cache per processor
32× 64GB DDR4-2933 ECC RDIMM
2TB total system memory
The large system memory capacity makes these servers suitable for data-intensive AI applications, dataset preparation, CPU-side processing and GPU-accelerated computing. The combination of 2TB system RAM and 640GB GPU memory provides a substantial hardware platform for organizations running complex AI and HPC workloads.
Mellanox ConnectX-6 HDR 200Gb/s Networking – NVIDIA A100 GPU Servers
The Microsoft / ZT Systems C2660 includes high-speed Mellanox networking designed for multi-node GPU infrastructure. The referenced configuration is equipped with eight Mellanox ConnectX-6 network adapters.
Networking specifications:
Adapter: Mellanox ConnectX-6
Part number: MCX653105A-HDAT
Quantity: 8 adapters per server
Network speed: 200Gb/s per adapter
Connectivity: HDR InfiniBand / 200GbE
Connector: QSFP56
PCIe interface: PCIe x16
Aggregate port bandwidth: Up to 1.6Tb/s per server
These network interfaces can support high-bandwidth communication between GPU servers, storage infrastructure and compatible data center networking equipment.
The platform is also designed to support GPU-aware networking technologies such as NVIDIA GPUDirect RDMA in compatible environments.
This is particularly important for organizations building multi-node AI training clusters and distributed GPU computing infrastructure.
server-parts.eu can also assist customers looking for:
NVIDIA Mellanox ConnectX-6 network adapters
NVIDIA HDR 200Gb/s InfiniBand switches
NVIDIA Quantum InfiniBand networking equipment
200G QSFP56 InfiniBand cables
InfiniBand DAC and optical cables
GPU cluster networking infrastructure
Refurbished enterprise switches and network components
Network equipment and cable compatibility should be confirmed for each deployment.
Microsoft ZT Systems C2660 NVIDIA A100 80GB – Power Supply and Rack Infrastructure
The Microsoft / ZT Systems C2660 is a 6U GPU server with a high-capacity redundant power supply configuration.
The referenced system includes six LiteOn 3000W power supplies.
Power and installation specifications:
Power supply manufacturer: LiteOn
Power supply model: PS-2302-3N
Power supplies: 6× 3000W
Total installed PSU capacity: 18kW
Input voltage: 200–240V AC
Input frequency: 50/60Hz
Maximum rated input current: 16A per PSU
Power architecture: Redundant PSU configuration
Rack height: 6U
Mounting rails: Included in the referenced configuration
Important: The 18kW figure represents the total installed power supply capacity, not the actual operating power consumption of the complete server.
Actual power draw depends on the GPU workload, CPU utilization, memory, storage and system configuration.
Before installation, customers should confirm the server's electrical requirements, rack compatibility, cooling capacity, power connections and data center infrastructure. Detailed power consumption, dimensions, weight and cooling requirements should be confirmed for the particular system before deployment.
Microsoft ZT Systems C2660 8× NVIDIA A100 GPU Servers for AI Infrastructure
The Microsoft / ZT Systems C2660 is designed for demanding GPU computing environments where performance, high memory capacity and fast interconnects are important.
Each server combines eight NVIDIA A100 80GB SXM4 GPUs with AMD EPYC processors, 2TB system memory and high-speed InfiniBand networking.
These GPU servers are suitable for organizations building or expanding:
Enterprise AI infrastructure
Large language model training
AI inference environments
Deep learning infrastructure
Machine learning platforms
GPU cloud hosting services
High-performance computing clusters
Scientific computing and simulation
AI research laboratories
Data center GPU clusters
Distributed GPU computing environments
For companies looking for alternatives to new-generation GPU servers, refurbished NVIDIA A100 infrastructure can provide access to established enterprise GPU technology.
The suitability of an A100-based system depends on the intended workload, software requirements and performance targets.
Refurbished NVIDIA A100 GPU Servers – Availability in Europe
For companies looking to purchase NVIDIA A100 GPU servers in Europe, refurbished enterprise GPU hardware can offer an alternative to ordering new systems.
Microsoft / ZT Systems C2660 servers are hyperscaler-class platforms that may become available through specialist enterprise hardware suppliers.
server-parts.eu works with an international network of hardware suppliers to source enterprise GPU servers, networking equipment and data center infrastructure.
The referenced Microsoft / ZT Systems C2660 configuration has previously been offered from UK stock.
server-parts.eu can assist with enquiries for:
Microsoft / ZT Systems C2660 GPU servers
NVIDIA HGX A100 8-GPU servers
NVIDIA A100 80GB SXM4 servers
Refurbished NVIDIA A100 GPU systems
AMD EPYC GPU servers
200Gb/s InfiniBand networking
Enterprise AI and HPC hardware
Shipping arrangements, warranty terms, available quantities and delivery times are confirmed individually. International orders may require end-user verification and applicable export compliance checks.
NVIDIA A100 GPU Servers – Customer Eligibility and Compliance
The availability of enterprise NVIDIA GPU servers may be subject to supplier requirements, end-user verification and applicable export control regulations.
Customers interested in purchasing Microsoft / ZT Systems C2660 NVIDIA A100 servers may be asked to provide basic information about their organization and intended use.
Depending on the destination and supplier requirements, additional information may be required.
This can include:
Purchasing company details
End-user company information
Intended installation country
Final delivery destination
Planned application of the GPU servers
Data center or installation information
Relevant end-use documentation
server-parts.eu can assist customers with the necessary information before confirming an order.
All shipments remain subject to applicable compliance requirements and supplier approval.
Bulk NVIDIA A100 80GB GPU Server Orders
Organizations planning larger GPU infrastructure projects often require multiple servers with matching hardware configurations
Server-parts.eu welcomes enquiries for individual NVIDIA A100 GPU servers as well as larger GPU server deployments.
We can assist customers with:
NVIDIA A100 80GB server pricing
Available server quantities
Individual and bulk GPU server orders
Complete hardware specifications
GPU memory and processor configurations
NVMe storage requirements
InfiniBand networking
Rack and power requirements
Warranty and support options
International shipping arrangements
Customers planning multi-node GPU clusters are encouraged to provide their required server quantities, technical specifications and delivery destination.
All offers remain subject to stock availability and final supplier confirmation.
Request a Quote – Microsoft ZT Systems C2660 with 8× NVIDIA A100 80GB GPUs
Are you looking for Microsoft / ZT Systems C2660 NVIDIA A100 servers in stock, or planning to purchase refurbished NVIDIA HGX A100 GPU servers for an upcoming AI infrastructure project?
NVIDIA A100 80GB GPU Servers
Special Pricing & No Lead Time
server-parts.eu can assist customers looking for high-performance NVIDIA A100 GPU infrastructure.
Product: Microsoft / ZT Systems C2660
GPU configuration: 8× NVIDIA A100 80GB SXM4
GPU memory: 640GB HBM2e total
Processor: 2× AMD EPYC 7V12
System memory: 2TB DDR4 ECC
Storage: 3.84TB NVMe SSD
Networking: 8× Mellanox ConnectX-6 HDR 200Gb/s
Form factor: 6U
Power supplies: 6× LiteOn 3000W
Availability: Upon request, subject to stock confirmation
Condition: Configured enterprise GPU server; condition confirmed per quotation
Warranty: Available upon request
Pricing: Available upon request
For current pricing, availability and full specifications, contact:
Email: info@server-parts.eu
Website: https://www.server-parts.eu
Contact server-parts.eu today to request a quotation for Microsoft / ZT Systems C2660 NVIDIA HGX A100 80GB GPU servers.
Comments