NVIDIA and Cisco outline USFP optical module for AI factories
3 companies named across 6 categories, one of 613 articles referencing Nvidia. Previous coverage: PEGATRON Corporation to show AI infrastructure at OCP Global Summit 2026 (Oct 2026).
Companies mentioned
Best suited for
- Seniority
- Architect
- Job function
- Network / Network Architect
- Persona
- Network Architect
- Buyer role
- Architect / Technical Evaluator
- Decision stage
- Which to buy
- Adoption curve
- Innovators
- Technology maturity
- Emerging Exploration
- Industry
- Information Technology / AI Infrastructure / AI Networking & Interconnect / Optical Transceivers, AECs & Interconnect Components
Our classification, not the publisher's statement. Best suited for, not only for.
NVIDIA and Cisco have outlined plans for Ultra-Dense Small Form Factor Pluggable (USFP), a high-density optical module intended to support the next generation of AI factories built on pluggable transceivers.
Growing AI models and rising inference demand require AI factories to connect more computing systems while managing network complexity, power, and costs. Operators gain more from their infrastructure investments when they move additional data per watt and per switch. When a switch reaches capacity, further expansion can require additional switch layers and optical links, which increase latency, power consumption, and costs.
USFP is designed for AI factory builders to scale from 200Gb/s per lane to 400Gb/s-per-lane Ethernet AI fabrics, extending a deployment model used across generations of data center networks. The first planned transceivers provide 16 full-duplex lanes in an OSFP-sized footprint supporting switches and SuperNICs, with expansion to 32 lanes or more later. Doubling lanes and bandwidth per lane allows fabrics to connect more 1.6Tb/s endpoints, reaching 12.8Tb/s of capacity, without adding network tiers.
Compared with 200Gb/s-per-lane pluggable modules, USFP provides four times the bandwidth and front-panel bandwidth density at about 1.3 times the power efficiency. Its two-dimensional array connector uses two-step insertion, engaging high-speed contacts after insertion to reduce wear, improve signal integrity, and make insertion force more predictable. USFP supports up to 100 watts of power dissipation without a separate embedded cooling circuit within the transceiver.
All USFP configurations use optical designs that reduce latency and power consumption by simplifying signal processing, and they support a broad range of optical technologies and connection distances. NVIDIA and Cisco are working with TE Connectivity and Amphenol to build USFP cages, connectors, and modules, and additional partners are expected to join.
“AI builders need an interconnect architecture that can scale performance without locking them into a more complex operating model,” said Bill Gartner, senior vice president and general manager of optical systems and optics at Cisco. The companies plan to create an open industry specification for USFP through an upcoming community-led multi-source project.
Press release, provided by Globe Newswire on behalf of Nvidia. Read the original.