Specifications
NVIDIA Part Number
MMS1X00-N5400
Form Factor
QSFP112 (quad-channel, 112G PAM4 per lane)
Data Rate
400 Gbps
Wavelength
1310 nm
Signalling
4× 112G PAM4
Channel Configuration
4 lanes in a QSFP112 footprint
Optical Connector
MPO-12 (8 of 12 fibres used for 4× 112G)
Link Type
400G DR4-class 500 m link without an in-module gearbox
Media
Single-mode fibre at 1310 nm
Cooling
Integrated external heatsink on the module body
Host Compatibility
NVIDIA ConnectX-8 (400GbE autonegotiation validated)
Switch Compatibility
NVIDIA Spectrum-X switch fabric
Platform Use
Connecting NVIDIA GB300 AI Station systems at 400GbE
Ecosystem
NVIDIA end-to-end AI networking optics
Overview
The NVIDIA MMS1X00-N5400 is a 400 Gbps QSFP112 optical transceiver operating at 1310 nm. QSFP112 means a quad-channel module with 112G PAM4 per lane, so 400G is carried over four lanes rather than the eight lanes a QSFP56-DD module requires — the module is more compact than the OSFP form factor used at 800G and 1.6T, and needs no in-module gearbox to drive a standard 500 m DR4 single-mode link.
The module uses an MPO-12 optical connector, with eight of the twelve fibres in use for the 4× 112G channels, and carries an external heatsink on the module body rather than relying solely on cage-side cooling. NVIDIA specifies these optics for connecting GB300 AI Station systems to Spectrum-X switching at 400GbE, where the link autonegotiates immediately on the platform's ConnectX-8 NICs.
For AI cluster builders, the significance is end-to-end consistency: NVIDIA-qualified optics matching the NICs and switches in an NVIDIA AI factory reference platform, removing the interoperability risk of mixing third-party 400G pluggables into a Spectrum-X fabric.
Key Benefits
Native 400GbE on four lanes: QSFP112 keeps a smaller module than QSFP56-DD and avoids an in-module gearbox for 500 m DR4 links. Validated in platform: autonegotiates on NVIDIA ConnectX-8 and interoperates with Spectrum-X switching. Single-vendor fabric optics: NVIDIA-qualified end-to-end for GB300 AI Station deployments.
Applications
NVIDIA GB300 AI Station scale-out links · Spectrum-X 400GbE AI fabrics · AI cluster and AI factory data center cabling · 400G DR4 500 m single-mode links
Request a Quote — NVIDIA MMS1X00-N5400 QSFP112 1310NM 400GBPS OPTICAL TRANSCEIVER
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →