NVIDIA Mellanox® MCP7H60-C002 — 200GbE Passive Copper DAC Splitter Cable, QSFP-DD to 2× QSFP28, 2m
Technical Identity
The NVIDIA Mellanox® MCP7H60-C002 is a 200GbE Ethernet passive copper Direct Attach Cable (DAC) in a hybrid splitter / breakout configuration. One end terminates in a single QSFP-DD module running 200GbE (8× 25G NRZ); the other end branches into two QSFP28 modules, each carrying 100GbE (4× 25G NRZ) downstream. The "C" suffix in the part number identifies the LinkX Ethernet copper family, and "002" denotes the 2-metre fixed length. The assembly is fully passive — no signal-conditioning electronics, no external power, sub-microsecond latency. Refurbished Grade A: checked by T.E.S IT-SOLUTIONS engineering (see Inspection and checks).
Specifications
What This Is NOT
NOT an AOC (Active Optical Cable). This is passive copper. No optics, no laser, no signal regeneration. Reach is therefore limited to short data-centre runs; for any reach beyond a few metres at 200GbE, use an MCA / MFS / MFA series active or optical cable instead.
NOT a transceiver. Fixed-length assembly with permanently attached ends — connectors cannot be swapped, re-terminated, or repaired in the field.
NOT a 1-to-1 cable. This is a splitter/breakout — the QSFP-DD end is one switch port, the QSFP28 ends are two independent downstream devices. The trunk side is logically split into two 100GbE channels.
NOT InfiniBand. The "C" code in the part number designates the Ethernet LinkX line. For InfiniBand HDR splitter cables of similar topology, see the MCP7H50-H… series.
NOT 400GbE. The QSFP-DD form factor supports up to 400GbE on 50G PAM4 SerDes, but this assembly is wired and rated for 200GbE (25G NRZ) only.
NOT bundled with adapters, transceivers, or rail-kit accessories — sold standalone.
Compatibility & Constraints
QSFP-DD (trunk) side — switch port: NVIDIA Spectrum-2 (e.g., SN3700, SN3700C), Spectrum-3 (e.g., SN4600C, SN4700) and other QSFP-DD-cage 200GbE Ethernet switches. The switch port must be configured for 1 × 200GbE or 2 × 100GbE breakout mode, depending on the firmware.
QSFP28 (branch) side — adapter / server end: NVIDIA ConnectX-5 EN, ConnectX-6 Dx, ConnectX-7 single- or dual-port 100GbE adapters with QSFP28 ports. Each branch leg is treated by the host as an independent 100GbE link.
Protocol: Ethernet only. Verify that the switch port is set to Ethernet mode (not InfiniBand) before deployment.
Reach: 2 m fixed — intended for intra-rack interconnect between a top-of-rack 200GbE switch and two adjacent 1U / 2U servers in the same rack.
Cage MSA: QSFP-DD MSA on the trunk; QSFP28 MSA on each branch leg.
Power: drawn from host port (no external supply required).
Pre-deploy checklist: confirm switch firmware supports the breakout profile, confirm the matching speed/FEC settings on both ends, and confirm that two adjacent QSFP28 host ports can be served by a single trunk run of this length.
Frequently Asked Questions (FAQ)
What is the NVIDIA Mellanox MCP7H60-C002?
The MCP7H60-C002 is a 200GbE Ethernet passive copper DAC splitter cable from the NVIDIA Mellanox LinkX family. It carries 200GbE on a single QSFP-DD trunk and breaks out into two 100GbE QSFP28 legs, in a 2-metre fixed length.
Is this an active or passive cable?
MCP7H60-C002: Fully passive. No optics, no SerDes retimers, no external power. The cable is essentially controlled-impedance twin-axial copper with EEPROM at each end for vendor identification.
Why does the trunk side use QSFP-DD if the speed is "only" 200GbE?
MCP7H60-C002: QSFP-DD provides 8 electrical lanes versus QSFP28's 4. At 25G NRZ per lane, 8 lanes deliver 200GbE on the trunk and split cleanly into two 4-lane × 25G QSFP28 ports of 100GbE each. The same cage can scale to 400GbE on PAM4 SerDes, but this specific assembly is NRZ and rated for 200GbE / 2× 100GbE.
What switches and adapters can I use this with?
MCP7H60-C002: Switch (trunk side): NVIDIA Spectrum-2 / Spectrum-3 QSFP-DD-cage Ethernet switches with breakout-capable firmware. Adapter (branch side): NVIDIA ConnectX-5, ConnectX-6 Dx, or ConnectX-7 single- or dual-port 100GbE NICs with QSFP28 ports. Standard MSA compliance also makes it physically compatible with third-party QSFP-DD / QSFP28 equipment from Cisco, Arista, Juniper, Dell, and HPE — provided the port mode supports a 200GbE-to-2× 100GbE breakout profile.
Where is the MCP7H60-C002 typically deployed?
MCP7H60-C002: The 2-metre length is intended for intra-rack server-to-ToR interconnect: one switch port serves two adjacent servers, doubling effective port count on the 200GbE leaf. Common applications include high-density cloud edge racks, HPC compute fabrics, and 100GbE-class virtualisation hosts where 200GbE ToR ports must be broken out cost-effectively.
What condition is supplied by T.E.S IT-SOLUTIONS?
MCP7H60-C002: Refurbished Grade A. Every unit is checked by T.E.S IT-SOLUTIONS engineering (see Inspection and checks). Anti-static packaging, global B2B shipping, and pre-sale compatibility consultation are included.
Why buy from T.E.S IT-SOLUTIONS
T.E.S IT-SOLUTIONS is a Haifa-based NVIDIA Mellanox specialist serving HPC, AI, and enterprise data-centre procurement teams since 2011. Every cable is identified by MPN, NVIDIA P/N, lane count, FEC profile, and supported switch family before it ships. Every customer gets free pre-sale compatibility advice for switch / adapter / cable pairing, multi-currency B2B invoicing with worldwide express options, and a direct line to our engineering team — not a call-centre script. See tes-itsolutions.com.