AI Infrastructure · Global AI networking market
Cisco Extends Open Ethernet Systems for Distributed AI Factories
Cisco’s OCP 2026 release adds 1.6-terabit silicon, production systems and optical work for AI scale-out and scale-across networks.
Cisco used the October 12 Open Compute Project summit to introduce silicon, systems and optical work aimed at open Ethernet networks for rack-scale and distributed AI infrastructure.
The company said its first G300-powered systems are shipping, introduced 1.6-terabit G302 silicon and reported P200-powered systems in production hyperscale environments. Cisco and NVIDIA also plan an open specification for 400-gigabit-per-lane pluggable optics.
The release frames these products as foundations for scale-out and scale-across AI fabrics. Performance, interoperability and deployment benefits are vendor claims; operators still need to validate topology, power, cooling, software and serviceability against their own clusters.
Official source · Source reviewed October 21, 2026.
Key details
Why it matters
The product roadmap expands open Ethernet, optics and management options for distributed AI infrastructure, while operators still need workload-specific validation of topology, cooling, power and interoperability.
Status and next steps
Cisco said G302 systems are planned to begin shipping before the end of 2026; Live Protect support for Cisco SONiC is planned for mid-2027.
Review the official next step →Project contacts and next steps
A direct project contact has not been verified for this article. Consult the linked source organization for the appropriate office.
Primary sources
- Cisco OCP 2026 announcement · 2026-10-12