NVIDIA and Cisco Plan 400 Gb/s-per-Lane USFP Module, Targeting 12.8 Tb/s in an OSFP Footprint

NVIDIA and Cisco have announced plans for the Ultra-Dense Small Form Factor Pluggable (USFP), a high-density optical module designed to support 400 Gb/s per lane in AI factory networks. The OSFP-sized design targets up to 12.8 Tb/s of aggregate capacity with 16 full-duplex lanes.

The announcement addresses a scaling problem that network engineers know well: bandwidth per connection and connections per switch both have to rise as AI infrastructure grows.

"As AI models and inference workloads expand, AI factories need to connect more computing systems while managing network complexity, power consumption and costs."

Doubling bandwidth per lane

"USFP is designed to support the transition from 200 Gb/s-per-lane to 400 Gb/s-per-lane Ethernet AI fabrics."

The first planned USFP transceivers will include 16 full-duplex lanes in an OSFP-sized footprint for use with switches and SuperNICs. By increasing both lane count and bandwidth per lane, USFP is designed to connect more 1.6 Tb/s endpoints without adding network tiers. The design also supports future versions with 32 or more lanes.

Relative to 200 Gb/s-per-lane pluggable modules, USFP is specified to provide four times the bandwidth and front-panel bandwidth density and approximately 1.3 times the power efficiency — figures that matter to infrastructure designers trying to grow compute capacity within an existing network footprint.

Critically, USFP retains the pluggable transceiver model deployed across multiple generations of data center networks, allowing customers to adopt higher-bandwidth optical technology without replacing their entire data center infrastructure. The design is intended to maintain the deployment, serviceability, and manufacturing processes used for existing pluggable transceivers.

Connector and thermal design

USFP is designed for electrical signals of 400 Gb/s per lane and uses a two-dimensional array connector with a two-step insertion process. High-speed contacts engage after the module is inserted — an approach intended to reduce contact wear and support signal integrity while making insertion force more consistent. The connector supports future versions with 32 full-duplex lanes or more, and because it is designed for 400 Gb/s operation, it can also support improved performance at 200 Gb/s per lane.

On thermals, USFP supports power dissipation of up to 100 watts without a separate embedded cooling circuit inside the transceiver, reducing switch design requirements. USFP configurations also support optical designs that can reduce latency and power consumption by simplifying signal processing.

Building the ecosystem

NVIDIA and Cisco are working with TE Connectivity and Amphenol to develop USFP cages, connectors, and modules, with additional companies potentially participating. The collaboration builds on work to support interoperability between NVIDIA Spectrum-X Ethernet networking and Cisco Silicon One-based Ethernet networks, with intended deployments by enterprises, neoclouds, and sovereign clouds.

"NVIDIA and Cisco plan to develop an open industry specification for USFP through a community-led, multisource project."

If that open specification materializes, the shared form factor would give suppliers a common basis for developing pluggable infrastructure for 400 Gb/s-per-lane networks — the kind of multisource standardization that has historically determined whether a transceiver form factor becomes an industry workhorse or a footnote.