The 2025 OCP Global Summit is just two weeks away. If you haven’t registered yet, now is the perfect time to do so. Read on to become inspired to be part of this incredible event.
Leading the Future of AI
This year’s theme for OCP is “Leading the Future of AI.” Our Cisco contingent is going to highlight the amazing possibilities of artificial intelligence. Trust me, it’s a discussion you won’t want to miss.
This theme aligns perfectly with Cisco’s strategy. Over the past 12 months, we have introduced groundbreaking innovations aimed at simplifying, securing, and future-proofing data centers as they run an increasing volume of AI workloads. These advances empower organizations to confidently scale their AI initiatives and showcase our unique role as a trusted partner. We’re supporting hyperscale builders, neocloud providers, enterprises, and service providers, enabling their entire ecosystems to adapt to the demands of AI-driven workloads.
Cisco will be strongly represented at the summit, covering key industry topics ranging from AI Networking to SONiC and P4. Let me break down some of these topics to give you a sense of what you can expect to hear from us.
AI networking: Built for speed, scale, and simplicity
Cisco AI Networking is an essential infrastructure solution for advanced data center buildouts, unifying frontend and backend connectivity through standards-based Ethernet. This approach delivers exceptional scale, intelligence, and operational simplicity.
Designed specifically for AI workloads, Cisco high-performance infrastructure is based on the purpose-built Silicon One architecture. This foundation offers lossless, high-bandwidth Ethernet connectivity that supports both AI frontend data flows and backend GPU cluster communications seamlessly. The network fabric is engineered for low latency and features 400G and 800G Ethernet capabilities, along with intelligent packet routing that is designed to respond dynamically to real-time network conditions. It helps ensure efficient and reliable traffic steering across the fabric, helping customers unlock consistent performance and reduce job completion time (JCT). These enhancements ensure predictable and efficient data movement within networks. Last but not least, scalability is another key differentiator of Cisco AI Networking with support for large-scale GPU clusters.
Cisco is redefining the operational experience for deploying and managing AI clusters. Frequently, customers invest in large GPU clusters, only to spend months getting them operational, leading to a poor return on investment. Cisco Nexus Hyperfabric AI aims to reduce the deployment time from several months to two weeks or less. By pre-planning the fabric layout, creating cabling plans, and providing online guided assistance for installers, customers save a significant amount of time on deployment, shortening time to revenue for AI applications.
In partnership with ecosystem partner NVIDIA, Cisco provides premier AI frontend and backend infrastructure, paired with advanced storage solutions from VAST. This collaboration enables customers to build AI infrastructures that support complex data fabric workloads. Enterprises can securely utilize their data for agentic AI applications on a large scale. The latest release of Cisco Secure AI Factory with NVIDIA introduces new use cases, including the acceleration of retrieval-augmented generation (RAG) pipelines. This enhancement facilitates faster data extraction and retrieval, ensuring that AI agents have immediate and secure access to the data they need, precisely when they need it.
Security is integrated from the ground up with a zero-trust design that protects AI models, data, and operations throughout the entire AI workflow—from data ingestion to inference. This heightened security is supported by unified threat response mechanisms.
Overall, this comprehensive approach ensures a secure, scalable, and high-performance AI infrastructure tailored to the needs of our hyperscaler, neocloud, and enterprise customers.
At the 2025 OCP Global Summit, don’t miss the Cisco presentations on Tuesday, October 14: “Transforming Deployment Challenges into Opportunities with Cisco AI Networking.” and “Redefine Networking. Awaken Cognitive AI with Cisco”.
P4 and Cisco Silicon One: Flexible, high-performance networking
Programming Protocol-Independent Packet Processors (P4) is a high-level data plane programming language designed to separate the intent of packet forwarding from its implementation. Cisco Silicon One uses P4 to create a flexible and efficient system for processing packets. It adopts a run-to-completion model with a programmable pipeline, enabling the device to adjust packet processing according to network demands.
With P4, Cisco can offer various features like IPv4/IPv6 routing, SRv6 forwarding, Ethernet switching, security policies, and more—all through P4 applications. The use of a common P4-based SDK and code across different Silicon One devices allows for quick feature development and ensures consistent performance across various products, whether fixed routers or modular systems.
P4 is essential for Cisco Silicon One to provide a flexible and high-performance networking platform that can readily adapt to the changing demands of network operators, protect customer investments, and foster innovation in diverse environments.
Want to learn more about Cisco Silicon One and P4? Register for the P4 workshop on October 13 at the 2025 OCP Global Summit.
SONiC: Open networking in the cloud
Cisco recognizes the strategic importance of open networking and the potential of Software for Open Networking in the Cloud (SONiC) to accelerate innovation and provide customers with greater choice. Our commitment to SONiC has never been stronger. As a SONiC Premier member, we hold a seat on both the Governing Board and the Technical Steering Committee. However, our involvement goes beyond simply holding these positions—we actively contribute to the SONiC community. In fact, Cisco is one of the top five contributors to this community, providing code, features, and expertise to enhance the platform.
Our aim with SONiC is to offer customers a transformative path that emphasizes openness, flexibility, and community-driven innovation, while utilizing our reliable hardware and ecosystem.
As of today, SONiC has gained traction in data centers, and Cisco supports three key architectures:
- IP/BGP fabric: This reliable, scalable design for 2-tier and 3-tier Clos architectures supports both IPv4 and IPv6.
- VXLAN EVPN fabric: This fabric is ideal for advanced multi-tenant networks, featuring EVPN multi-homing and host mobility for seamless workload migration.
- AI/ML fabric: Optimized for demanding AI/ML applications, this design includes a lossless, non-blocking structure, advanced traffic management features, and integrated AI observability tools for large-scale clusters.
Additional information will be provided at the SONiC workshop, which is being held alongside the OCP Summit on Thursday, October 16. Cisco will present three key topics during this workshop:
- “Driving Innovation in Network Infrastructure: Key Advances from the SONiC Routing Workgroup”
- “Device Local Diagnosis”
- “SONiC-VPP: A Platform for Data Center Network Testing Roles”
Cisco is not just adapting to the new AI era—we’re leading it—with groundbreaking innovations delivered at breakneck pace. Stay tuned as more announcements are coming before the summit.
Register for the 2025 OCP Global Summit
