Brand: NVIDIA | Category: Network Switches
SKU: QM9590 | Part #: QM9590 | MPN: QM9590
Contact for Pricing — Request a Quote
The NVIDIA Quantum-2 InfiniBand QM9590 delivers 64 ports of NDR 400Gb/s QSFP112 connectivity in a compact 1U form factor, offering an aggregate switching capacity of 25.6Tb/s for next-generation data center fabrics. This high-density switch architecture enables network engineers and AI infrastructure teams to build scalable, low-latency interconnects without sacrificing port count or performance density.
Built on NVIDIA's Quantum-2 platform, the QM9590 combines sub-microsecond port-to-port latency with advanced in-network computing via NVIDIA SHARP v3, intelligent adaptive routing, and congestion control to optimize workload performance across distributed systems. The switch supports bidirectional 400Gb/s per port while maintaining backward compatibility with HDR 200Gb/s and EDR 100Gb/s deployments via splitter cables, protecting existing infrastructure investments. Full GPUDirect RDMA support, up to 16 virtual lanes per port, and redundant hot-swap power supplies with flexible airflow options ensure reliability and operational flexibility. Management is streamlined through NVIDIA UFM (Unified Fabric Manager), OpenSM, and REST API integration, enabling seamless fabric orchestration across enterprise environments.
For detailed specifications, pricing, and integration planning for the NVIDIA Quantum-2 InfiniBand QM9590, contact the Omnixon Global team to submit your RFQ.
| Brand | NVIDIA |
| Category | Network Switches |
| SKU | QM9590 |
| Part Number | QM9590 |
| Condition | New |
| Capacity | 200GB |
| Speed | 200Gb/s |
| Manufacturer Part Number | QM9590 |
| Product Family | Quantum-2 |
| InfiniBand Standard | NDR (Next Data Rate) |
| Number of Ports | 64 x NDR 400Gb/s QSFP112 ports |
| Per-Port Bandwidth | 400Gb/s bidirectional |
| Aggregate Switching Capacity | 25.6Tb/s |
| Form Factor | 1U |
| Port Type | QSFP112 |
| Backward Compatibility | HDR 200Gb/s, EDR 100Gb/s (via splitter cables) |
| In-Network Computing | NVIDIA SHARP v3 (Scalable Hierarchical Aggregation and Reduction Protocol) |
| Latency | Sub-microsecond port-to-port |
| Routing | Adaptive routing with advanced congestion control |
| Management Interface | NVIDIA UFM (Unified Fabric Manager), OpenSM, REST API |
| Power Supply | Redundant hot-swap power supplies |
| Fan Configuration | Hot-swap redundant fans with front-to-back and back-to-front airflow options |
| GPUDirect RDMA Support | Yes |
| Virtual Lanes | Up to 16 virtual lanes per port |
| Operating Temperature | 0°C to 45°C |
| Storage Temperature | -40°C to 70°C |
| Certifications | CE, FCC, RoHS |
The NVIDIA Quantum-2 InfiniBand - 200Gb/s is enterprise networking hardware for data-center, campus, and branch deployments. Common roles include core/distribution switching, server uplinks, top-of-rack fabric, and edge connectivity in mixed Cisco/Juniper/Arista environments.
Key specifications for the NVIDIA Quantum-2 InfiniBand - 200Gb/s: new condition; network type InfiniBand Switch; network port config 40x 200Gb/s HDR; network speed 200Gb/s; network switching capacity 12.8 Tbps; network forwarding rate 9.52 Bpps; network power consumption 350W. Manufacturer part number QM9590. For the full datasheet with electrical, environmental, and compliance details, contact our pre-sales engineering team.
The NVIDIA Quantum-2 InfiniBand - 200Gb/s interoperates with standard switching protocols (LACP, MLAG/VPC, 802.1Q, OSPF/BGP). For specific cross-vendor scenarios — Cisco-to-Arista MLAG peering, fabric integration with Cumulus/SONiC, or migration from legacy Catalyst — our network engineers map a compatibility plan as part of the RFQ.