This third-party active optical breakout cable (AOC) is fully compatible with NVIDIA/Mellanox MFA7U10-H003, providing a 400Gb/s twin-port OSFP connection split to two 200Gb/s QSFP56 ports over multimode fiber. Ideal for hybrid AI training clusters, HPC environments, and data centers bridging NDR 400G switches to HDR 200G adapters or nodes.
Unlike passive DACs, this AOC offers longer reach, lighter weight, and better bend radius with integrated lasers for signal regeneration. Backed by our lifetime warranty and rigorous testing—save 70-90% vs. genuine OEM while enabling flexible, high-performance scaling.
Key Features
-
Extended Reach & Flexibility: Active optics for reliable performance up to 30m (series max)—far beyond passive copper—with low weight and EMI immunity.
-
Massive Cost Savings: Affordable compatible AOC for mixed-generation AI/HPC deployments.
-
Plug-and-Play: Hot-pluggable OSFP and QSFP56 connectors with pre-programmed EEPROM for seamless NVIDIA compatibility.
-
Low Latency: Optimized for InfiniBand HDR and 400GbE with integrated signal conditioning.
-
Lifetime Warranty: Fully tested.
-
Use Cases: Fan-out in multi-node GPU clusters (e.g., connecting Quantum-2 NDR switches to ConnectX-6 HDR adapters).
Technical Specifications
-
Data Rate: 400Gbps total (HDR InfiniBand / 400GbE, split to 2x200Gbps, 8x50Gbps PAM4)
-
Connectors: 1x Twin-Port OSFP to 2x QSFP56
-
Length: 3 meters (10 ft)
-
Cable Type: Active Optical Multimode Fiber (MMF, 850nm)
-
Power Consumption: ~10-12W total (typical for AOC)
-
Operating Temperature: 0°C to 70°C
-
Standards Compliance: IBTA HDR, OSFP/QSFP56 MSA, CMIS compliant, RoHS
Compatibility
Programmed for full compatibility with NVIDIA/Mellanox ecosystems and MSA-compliant platforms, including:
- NVIDIA Quantum-2 QM9700 series switches (400G OSFP ports)
- NVIDIA ConnectX-6 adapters (200G QSFP56 ports)
- NVIDIA DGX A100/H100 systems in hybrid setups
- Other OSFP/QSFP56 platforms supporting HDR/400GbE
Need longer lengths (up to 30m), DAC alternatives, or bulk quotes? Contact us—we're your go-to for NVIDIA-compatible high-bandwidth AOCs in AI workloads.
