Cisco and NVIDIA Unite to Propel AI into the Enterprise

Today Cisco and NVIDIA announced a mutual intent to expand a partnership that brings together two of the industry’s most innovative data center networking and compute leaders. Our collaboration will help enterprises accelerate AI projects and make it easier for more customers than ever to standardize on Cisco as they modernize their data center networks.

A Unified Architecture for AI Workloads

To date, most of the industry dialog on AI has been focused on chips, computing power, and Large Language Models. As enterprises begin to take on AI workloads, though, the challenge of connecting computing resources and data together — both inside of and in between data centers and clouds — is going to be the next frontier of AI innovation. In this sense, the network is going to be the key to successfully scaling out AI in the enterprise. Under the proposed partnership, here’s how Cisco and NVIDIA intend to lead the charge:

1. NVIDIA will enable Cisco Silicon One coupled with NVIDIA SuperNICs to become part of the NVIDIA Spectrum-X Ethernet networking platform. Cisco would be the only partner silicon included in NVIDIA Spectrum-X. This is not only massive validation for Silicon One, but it also highlights the centrality of Cisco technologies — including silicon, optics, and software — in the AI technology landscape.
2. Cisco will be building systems that combine NVIDIA Spectrum silicon with Cisco OS software, allowing customers to simultaneously standardize on both Cisco networking and NVIDIA technology in the data center. Every enterprise will have a huge variety of use cases and applications for AI in their business. By enabling customers to leverage Nvidia Spectrum Silicon that’s specialized for back-end connectivity or Cisco Silicon One, both under a common Nexus software stack (NX-OS and Nexus Dashboard), we will enable an exciting new level of interoperability in the industry.

In addition, Cisco and NVIDIA also intend to invest in cross-portfolio innovations to tackle significant common challenges like congestion management and load balancing, ensuring that enterprises can accelerate their AI deployments, achieve faster time-to-value, and derive greater ROI. This is a multi-generation collaboration that will adapt as the technologies and customer needs evolve.

Overcoming Challenges in AI Adoption

Taking a step back, this anticipated Cisco-NVIDIA expanded relationship couldn’t come at a better time. There are new, game changing innovations in AI coming to market every week, and while that is truly exciting, for enterprises it’s also stressful. A recent Cisco survey found that while 97% of CEOs are advocating for AI projects, only 1.7% feel fully prepared. In that same survey, leaders pointed to three big concerns holding them back: a daunting need to modernize infrastructure, AI security and safety issues, and doubts about whether their people have the skills needed to implement and get the most out of AI.

At Cisco, we are addressing all of these issues with game-changing innovations of our own. In fact, this isn’t Cisco and NVIDIA’s first joint effort. For over a year, Cisco and NVIDIA have collaborated together to greatly simplify how enterprises deploy and manage AI infrastructure. Cisco Nexus Hyperfabric, our data center network fabric-as-a-service offering makes it easy for customers to build data center fabrics. Cisco Nexus Hyperfabric AI takes this concept even further, and with the help of NVIDIA, we designed an enterprise-ready AI platform and control plane tailor-made to support AI workloads.

On the infrastructure front, Cisco’s AI PODs have allowed organizations to quickly get up-and-running with AI inferencing. The pre-sized, pre-configured and centrally managed infrastructure stacks are tailored for specific AI use cases and industries, taking all of the guesswork out of deploying AI infrastructure. And as speed is a premium today, AI PODs provide quicker time to value, more consistency and reduced risk so customers have the confidence to move fast. To date, Cisco has validated four different AI POD configurations, with more use cases coming.

We’re also leading the industry to reimagine safety and security with AI Defense, which we launched in January. AI applications are fundamentally different and present an entirely new set of challenges for security and safety teams. Most important, enterprises need new tools for securing the AI models themselves inside their enterprise. This means not only knowing where an AI model might be in use by an employee, but also where a model is being integrated into an application by a developer. Further, they need the ability to create and enforce guardrails to ensure those AI models aren’t being manipulated to cause problems like data leakage, toxic behavior or any number of potentially massive problems.

Finally, at Cisco Live EMEA earlier this month we debuted a whole host of new solutions to enable enterprises and service providers to power AI workloads, including:

• Cisco Smart Switches, which redefine adaptive and scalable AI data center architectures by embedding services directly into the network;
• The new Cisco Agile Services Networking architecture, which provides a blueprint for service providers as they look to enable AI connectivity and monetize assured services.
• The expansion of our AI server family with the new Cisco UCS C845A M8 server, based on the NVIDIA MGX reference architecture, which brings flexibility to AI workloads. This adds to Cisco’s AI server portfolio, which includes the Cisco UCS C885A M8, Cisco’s first entry to the portfolio and first eight-way accelerated computing system built on the NVIDIA HGX platform.

We’re Just Getting Started

In the future, there are going to be two types of companies in the world: those who innovate and win with AI, and those who are left fighting for relevance. We believe that advanced, secure networking is the key to fueling AI innovation and adoption. That’s our core at Cisco. It’s deep in our DNA, and we are committed to helping all of our customers get off the sidelines and into the game. Stay tuned for more news and updates.

Share