Specifications
Product
NVIDIA/Mellanox MMS4A00-XM 1.6T OSFP224 XDR optical transceiver
Primary PN
MMS4A00-XM
NVIDIA Order PN
980-9IAH1-00XM00
Related PNs
MMS4C10-XM (980-9IAU0-00XM01), MMS4C11 (980-9IAU1-00XM01)
Form Factor
OSFP224 (OSFP1600), IHS / closed finned top
Aggregate Rate
1.6 Tb/s
Port Configuration
2 × 800G (2xDR4 / DR8)
Electrical Lanes
8 × 200 Gb/s PAM4
Wavelength
1311 nm
Reach
500 m over single-mode fibre
Optical Connector
Dual MPO-12 (2 × 800G breakout) or MPO-16 point-to-point
DSP
Broadcom Sian3 BCM83628, 3 nm
FEC
No additional optical-side FEC required
Application
NVIDIA Quantum-X800 air-cooled switches (Q3200-RA / Q3400-RA)
Protocols
InfiniBand XDR, Ethernet 800GbE
Thermal
Integrated heat sink, finned top — matches the switch cage
Compliance
OSFP MSA, CMIS
Cross Reference
Third-party OSFP-1.6T-2xDR4H modules interoperate
Availability
Volume — quote for allocation
Overview
The MMS4A00-XM is NVIDIA's first-party 1.6T XDR transceiver for the Quantum-X800 generation. It is a twin-port module: one OSFP224 package presenting two independent 800G links, so a single cage feeds two endpoints or two switch ports. Inside, eight 200G PAM4 lanes drive 1311 nm optics over 500 m of single-mode fibre — the DR8 reach class that covers the vast majority of intra-row and intra-pod AI fabric cabling.
Two things matter when specifying it. First, the port semantics: an 'OSFP 1.6T for Quantum-X800' can be either 2 × 800G (twin-port, dual MPO-12) or 1 × 1.6T (P2P, MPO-16), and the two are not interchangeable in a breakout design. Second, the housing: the IHS/closed finned-top variant M M S 4 A 0 0 - X M fits the air-cooled Quantum-X800 cages, while liquid-cooled platforms take a different heat-sink variant. Getting either wrong produces a link that will not train.
The module family also includes MMS4C10-XM and MMS4C11 (980-9IAU0-00XM01 / 980-9IAU1-00XM01) and the 2xFR4 reach variant MMS4A50-XM (980-9IAS0-00XM00) for dual-LC duplex fibre at 2 km. QS Compute supplies the NVIDIA PNs and the compatible OSFP224 optics, DACs and ACCs for the same cages.
Key Benefits
Twin-port 1.6T halves the cage count per 1.6T of fabric bandwidth. 500 m DR8 reach covers row and pod-scale cabling without regeneration. Broadcom Sian3 3 nm DSP gives stable 200G/lane BER margins. IHS finned top matches the Quantum-X800 air-cooled cage so no adapter or airflow re-work is needed.
Applications
1.6T InfiniBand XDR leaf-spine links between Quantum-X800 switches; NIC-to-switch runs for ConnectX-9 SuperNICs; 2 × 800G breakout to legacy 800G endpoints; and AI factory pod cabling where per-port power and thermal budgets are fixed by the switch platform.
Request a Quote — NVIDIA MMS4A00-XM (980-9IAH1-00XM00) — 1.6T OSFP224 XDR 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 →