Cisco And Nvidia Plan 12.8-Tbps Optical Module for AI Networks
The proposed USFP design uses 16 full-duplex lanes and targets four times the bandwidth density of pluggable optics using 200-gigabit-per-second lanes.

Cisco and Nvidia are developing a pluggable optical module for AI data centers that targets up to 12.8 terabits per second, with a design intended to increase network capacity while preserving a compact form factor.
The proposed Ultra-Dense Small Form Factor Pluggable, or USFP, would support 400 gigabits per second per lane across 16 full-duplex lanes. Its footprint would be comparable to current OSFP modules.
The companies estimate USFP would provide four times the bandwidth density and about 1.3 times the power efficiency of pluggable optics built around 200-gigabit-per-second lanes. Those are projected specifications for a design still in development.
TE Connectivity and Amphenol are working with Cisco and Nvidia on cages, connectors and modules. The companies also plan to develop an open industry specification through a community-led, multi-source project. No product shipping date has been announced.