# Nvidia and Cisco Launch USFP Optical Module for AI Data Centers

> Nvidia and Cisco introduce USFP optical modules with 400Gb/s per lane for AI data centers.
- Title: Nvidia and Cisco Launch USFP Optical Module for AI Data…
- Summary: Nvidia and Cisco introduce USFP optical modules with 400Gb/s per lane for AI data centers. Nvidia and Cisco unveiled USFP, a next-generation optical module…
- Keywords: nvidia, cisco, usfp, optical modules, ai data centers, 400gbps, bandwidth density, osfp, finance, markets, Launch, Optical
- Source: BigGo Finance — https://finance.biggo.com/news/7d61cbd3-510f-42ba-8c10-1d32dd611d39
- Author: BigGo Finance
- Published: 2026-10-12T20:55:16+00:00
- Read time: 1 min
- Topics: nvidia, cisco, usfp, optical modules, ai data centers, 400gbps, bandwidth density, osfp, finance, markets
## Nvidia and Cisco Unveil USFP
Nvidia and Cisco jointly announced USFP, a next-generation pluggable optical module specification on October 12.
## 400Gb/s Per Lane
The first USFP product supports 400Gb/s electrical signaling per lane, quadrupling bandwidth density over current 200Gb/s solutions.
## 16 Lanes in One Footprint
USFP packs 16 full-duplex lanes within the same footprint as existing OSFP modules, achieving 6.4Tb/s aggregate capacity.
## Power Efficiency Boost
USFP delivers approximately 1.3x improvement in power efficiency compared to current 200Gb/s single-lane pluggable modules.
## Two-Step Connector Design
USFP uses a two-dimensional array connector with a two-step insertion design to reduce contact wear and improve signal integrity.
## Partnerships for Ecosystem
Nvidia and Cisco partnered with TE Connectivity and Amphenol to develop USFP hardware and plan open industry standard through multi-source project.
## Market Impact Uncertain
USFP remains in design planning stage with no product launch timelines; market impact depends on open specification adoption and industry uptake.
## Key takeaway

Nvidia and Cisco unveiled USFP, a next-generation optical module specification supporting 400Gb/s per lane to quadruple bandwidth density in AI data centers.