MCX755106AS-HEAT ConnectX-7 200GbE Dual-Port QSFP112 NIC

MCX755106AS-HEAT ConnectX-7 200GbE Dual-Port QSFP112 NIC is a high-performance NVIDIA Mellanox network adapter designed for AI servers, GPU clusters, high-performance computing systems, enterprise data centers, and high-speed storage environments.
Based on NVIDIA ConnectX-7 technology, the adapter features dual QSFP112 ports and supports high-speed Ethernet and InfiniBand networking. The MCX755106AS-HEAT ConnectX-7 200GbE Dual-Port QSFP112 NIC supports 200GbE Ethernet and NDR200 InfiniBand, providing high bandwidth and low latency for demanding AI and HPC workloads.
Key Features
-
NVIDIA Mellanox ConnectX-7 architecture
-
Part Number: MCX755106AS-HEAT
-
Dual-port QSFP112 connectivity
-
Up to 200GbE Ethernet
-
NDR200 InfiniBand support
-
PCIe 5.0 x16 host interface
-
PCIe extension / Socket Direct ready
-
Secure Boot enabled
-
Crypto disabled
-
Suitable for AI and GPU server clusters
-
Designed for HPC and data center networking
-
High-bandwidth storage and server connectivity
The MCX755106AS-HEAT ConnectX-7 200GbE Dual-Port QSFP112 NIC is suitable for modern AI infrastructure where large datasets must move efficiently between GPU servers, compute nodes, and storage systems.
Product Specifications
| Item | Specification |
|---|---|
| Brand | NVIDIA Mellanox |
| Model | MCX755106AS-HEAT |
| Series | ConnectX-7 |
| Product Type | Ethernet / InfiniBand Network Adapter |
| Ethernet Speed | Up to 200GbE |
| InfiniBand Speed | Up to NDR200 |
| Number of Ports | 2 |
| Port Type | Dual QSFP112 |
| Host Interface | PCIe 5.0 x16 |
| Socket Direct | Supported / Extension Ready |
| Secure Boot | Enabled |
| Crypto | Disabled |
| Form Factor | HHHL |
| Application | AI Server, GPU Server, HPC, Data Center |
| Typical Use | AI Clusters, HPC, Storage Networking, Cloud |
AI Server and GPU Cluster Applications
The MCX755106AS-HEAT ConnectX-7 200GbE Dual-Port QSFP112 NIC is designed for high-bandwidth AI training systems, GPU computing nodes, enterprise data centers, HPC clusters, and distributed storage platforms.
Its 200GbE Ethernet capability helps provide fast data transfer between servers and storage resources, while NDR200 InfiniBand support makes it suitable for low-latency compute fabrics in AI and HPC environments.
Typical deployment scenarios include:
-
AI training clusters
-
GPU server interconnects
-
HPC computing systems
-
Distributed storage platforms
-
Enterprise data centers
-
High-speed storage networks
-
Private cloud infrastructure
-
Server network upgrades
-
Low-latency compute fabrics
200GbE and NDR200 Networking
The MCX755106AS-HEAT ConnectX-7 200GbE Dual-Port QSFP112 NIC supports 200GbE Ethernet and NDR200 InfiniBand networking.
This provides system integrators with flexibility when building AI or HPC infrastructure. Ethernet can be used for general-purpose data center and storage networking, while InfiniBand is commonly deployed for latency-sensitive compute fabrics.
NVIDIA firmware documentation identifies this model as 200GbE by default with NDR200 InfiniBand capability.
PCIe 5.0 x16 Interface
MCX755106AS-HEAT uses a PCIe 5.0 x16 host interface, providing substantially higher host-side bandwidth than PCIe 4.0 and PCIe 3.0 generation network adapters.
The MCX755106AS-HEAT ConnectX-7 200GbE Dual-Port QSFP112 NIC also supports a PCIe extension option and is Socket Direct ready, making it suitable for high-performance multi-socket and GPU server architectures.
Secure Boot and Crypto Configuration
MCX755106AS-HEAT is configured with:
-
Secure Boot: Enabled
-
Crypto: Disabled
This makes it suitable for deployments where secure firmware boot is required but hardware Crypto acceleration is not necessary. NVIDIA’s official ConnectX-7 documentation lists this exact security configuration for the AS variant.
About Us


Why Choose MCX755106AS-HEAT?
The MCX755106AS-HEAT ConnectX-7 200GbE Dual-Port QSFP112 NIC combines modern ConnectX-7 architecture, dual-port QSFP112 connectivity, 200GbE Ethernet, NDR200 InfiniBand, PCIe 5.0 x16, and Secure Boot support.
It is suitable for customers requiring:
-
200GbE AI server networking
-
NDR200 InfiniBand connectivity
-
Dual-port QSFP112 interfaces
-
PCIe 5.0 server platforms
-
GPU cluster interconnects
-
HPC computing environments
-
High-speed storage networking
-
Enterprise data center upgrades
FAQ
Q: What is MCX755106AS-HEAT?
A: MCX755106AS-HEAT is an NVIDIA Mellanox ConnectX-7 network adapter featuring dual QSFP112 ports and support for 200GbE Ethernet and NDR200 InfiniBand.
Q: Does MCX755106AS-HEAT support 200GbE?
A: Yes. NVIDIA lists this model as supporting 200GbE Ethernet.
Q: Does MCX755106AS-HEAT support InfiniBand?
A: Yes. It supports NDR200 InfiniBand connectivity.
Q: How many ports does MCX755106AS-HEAT have?
A: It features two QSFP112 network ports.
Q: What PCIe interface does MCX755106AS-HEAT use?
A: It uses a PCIe 5.0 x16 host interface.
Q: Does MCX755106AS-HEAT support Secure Boot?
A: Yes. Secure Boot is enabled.
Q: Does MCX755106AS-HEAT support Crypto?
A: No. Crypto is disabled on the AS version.
Q: Is MCX755106AS-HEAT suitable for AI servers?
A: Yes. It is suitable for AI servers, GPU clusters, HPC systems, distributed storage, cloud infrastructure, and other high-bandwidth data center environments.
Q: Can CubeCore supply MCX755106AS-HEAT in bulk?
A: Yes. CubeCore supports tested hardware supply, configuration verification, bulk procurement, project-based sourcing, and global delivery.

