Cisco and Nvidia have teamed up again. This time it’s to deliver a high-capacity pluggable optical system that promises to let customers more quickly and easily build Ethernet AI fabrics.
Introduced at the Open Compute Project (OCP) Global Summit 2026 this week, the Ultra-Dense Small Form Factor Pluggable (USFP) module supports 400 Gb/s-per-lane electrical signals, allowing up to 12.8T of capacity for building an infrastructure that spans networking, security, AI systems, and high-performance AI fabrics. The idea is to help customers address the pressure AI infrastructure is placing on the connections between GPUs, switches, security appliances, and other systems, according to Cisco.
Increasing bandwidth density can help customers build higher-capacity systems without proportionally increasing the space consumed by connectivity hardware, according to Bill Gartner, senior vice president and general manager of Cisco’s optical systems and optics business, and Will Eatherton, senior vice president, networking engineering.
“USFP will provide a path supporting both 200G and 400G SerDes based systems and ASICs, helping customers scale from today’s 1.6T fabrics,” Gartner and Eatherton wrote in a blog post. “With 16 lanes capable of operating at 400 Gb/s per lane, USFP will deliver 6.4T of aggregate capacity, four times the bandwidth density of today’s 1.6T pluggables,” they added.
“The architecture can scale to 32 lanes, providing a path to 12.8T of aggregate capacity. It also creates an opportunity for approximately 1.3x greater power efficiency through the use of 400G SerDes technology, helping operators manage growing power and cooling demands,” Gartner and Eatherton wrote.
“Our work has been guided by a shared goal: helping customers accelerate AI adoption while maintaining flexibility, openness, and choice,” Gartner and Eatherton wrote, referring to Cisco and Nvidia’s ongoing collaboration.
USFP aims to preserve the serviceability of pluggable optics; modules can be replaced in the field, and customers can continue using familiar deployment and maintenance practices, according to the authors.
In addition, “by combining a path to higher-capacity fabrics and improved efficiency with familiar operational practices, USFP helps customers scale AI infrastructure while maintaining the flexibility, interoperability, and consistency required at scale,” the authors wrote.
“Scaling AI factories requires more bandwidth per connection and more connections per switch. When a switch reaches its capacity, further network expansion can require additional switch layers and optical links, increasing latency, power consumption and costs,” according to Nvidia’s announcement of the news.
“USFP is designed to enable AI factory builders to scale from 200 Gb/s per lane to 400Gb/s-per-lane Ethernet AI fabrics, while extending a deployment model proven at scale across multiple generations of data center networks. With USFP, customers can embrace next-generation optical technology and meet the performance demand of AI workloads without overhauling their data center infrastructure,” Nvidia stated.
“With USFP, Nvidia and Cisco are building an ecosystem to reinvent pluggable optics — connecting more computing with greater efficiency to power the next generation of AI factories,” Gilad Shainer, senior vice president, Nvidia, said in a statement.
In addition, Cisco and Nvidia plan to create an open industry specification for USFP through an upcoming community-led multi-source project, the companies stated.
Interestingly, the USFP technology is not considered part of the current partnership between Cisco and Nvidia.
“The Nvidia Cisco partnership has been great over the past year or so. We’ve been doing a lot together on Spectrum X, both on the switch silicon in their switches as well as the Spectrum X technology on top of their own switch silicon devices,” Nvidia marketing director Brian Sparks said during a pre-briefing on the news. But USFP is “not part of the partnership. It’s just Cisco and us and others in the ecosystem believe that USFP is the right solution moving forward.”
Silicon One expansion
In addition to the USFP news, Cisco also announced an expansion of its Silicon One portfolio with a 51.2-terabit-per-second top-of-rack (ToR) Ethernet switching chip aimed at high-density AI and cloud data-center networks. The new Silicon One G302 delivers 102.4T capacity and features Cisco 1.6T optics and direct-to-chip liquid cooling integration.
The G302 will typically sit at the rack level and gather traffic from servers then connect the switch efficiently to the larger data center network, Cisco stated.
The G302 is an expansion of the 102.4 Tbps Silicon One G300 portfolio, which Cisco said is shipping now and features Intelligent Collective Networking, which supports shared packet buffers, path-based load balancing, and proactive network telemetry to handle large-scale, bursty data center traffic, AI training, inference, and real-time agentic workloads.
The chip is at the heart of two new fixed form 64-port switches: the 3RU air-cooled 8133 and Cisco’s first commercially available water-cooled switch, the 2RU 8132.
The G302 is planned to begin shipping to customers before the end of 2026.