My Cart
NVIDIA/Mellanox C8180 ConnectX-8 InfiniBand & Ethernet SuperNIC Single-Port OSFP PCIe 6.0 x16

NVIDIA/Mellanox C8180 ConnectX-8 InfiniBand & Ethernet SuperNIC Single-Port OSFP PCIe 6.0 x16

  • Brand: NVIDIA/Mellanox
  • Product Code: C8180 (900-9X81E-00EX-ST0)
  • Availability: In Stock
  • $850

Part Number: C8180 (900-9X81E-00EX-ST0)


Brand: NVIDIA/Mellanox

Adapter Card : NVIDIA/Mellanox Original C8180 ConnectX-8 InfiniBand & Ethernet SuperNIC, Socket-Direct XDR 800Gb/s PAM4 200G/Lane, Single-Port OSFP, PCIe 6.0 x16, HHHL, Tall Bracket


The NVIDIA/Mellanox C8180 ConnectX-8 SuperNIC is engineered to supercharge hyperscale AI computing workloads, AI factories, and next-generation cloud data centers. Delivering dual-protocol support for both InfiniBand XDR and high-speed Ethernet at up to 800Gb/s (PAM4 200G/lane), this single-port OSFP network adapter optimizes bandwidth, lowers latency, and maximizes cluster interconnect efficiency.

Built with native PCIe 6.0 x16 host connectivity and an integrated PCIe Gen6 switching fabric, the ConnectX-8 SuperNIC eliminates external PCIe switch chips to simplify server architecture while unlocking massive system throughput. It features an advanced hardware Data Path Accelerator (DPA) powered by a RISC-V event processor, along with SHARP (Scalable Hierarchical Aggregation and Reduction Protocol) in-network computing to offload collective operations (such as All-Reduce and All-to-All) directly at the NIC level.

Key Features

  • 800Gb/s High-Bandwidth Architecture: Supports XDR 800Gb/s InfiniBand and 800G/400G/200G Ethernet protocols over a single OSFP interface.
  • Native PCIe 6.0 x16 & On-Board Switching: Integrates native PCIe Gen6 switching and Multi-Host/Socket-Direct architecture for direct CPU/GPU access without external PCIe switches.
  • DPA & RISC-V Hardware Acceleration: Features a dedicated Data Path Accelerator (DPA) and Packet Stream Processor (PSA) for programmable packet parsing, flow control, and SDN offloads.
  • In-Network Computing (SHARP v3): Offloads AI compute aggregations directly to the network, drastically accelerating multi-GPU and multi-node AI model training.
  • Advanced Congestion Control: Telemetry-based adaptive routing and congestion management reduce tail latency and stabilize large-scale GPU clusters.
  • Hardware Security: Integrated encryption engines and secure boot protection ensure end-to-end data safety in multi-tenant environments.

Technical Specifications

Part Number / SKU C8180 / 900-9X81E-00EX-ST0
Network Interface Single-Port OSFP (OSFP224 DR8 compliant)
Data Transmission Rates InfiniBand XDR / NDR / HDR / EDR & Ethernet 800G / 400G / 200G / 100G
Host Interface PCIe 6.0 x16 (Socket-Direct / Multi-Host extension options)
Form Factor & Bracket HHHL (2.69 in. x 6.58 in. / 68.50mm x 168.40mm) with Tall Bracket
Offload Technologies RDMA / RoCE, SHARP, DPA (RISC-V), PSA Engine, Encryption & Secure Boot
Operating Temperature Operational: 0°C to 55°C | Storage: -40°C to 70°C
Supported OS Linux, Windows
Compliance & Applications RoHS Compliant | 1.6T/800G XDR Switch to 800G/400G HCA Topologies

Write a review

Please login or register to review