Cisco is evolving its Secure AI Factory, offering a full‑stack, security‑fused AI infrastructure platform, combining its own silicon and networking solutions with Nvidia and new partners like Red Hat for enterprise‑ready data centers.

Unveiled last September, Cisco's Secure AI Factory gained a slew of updates at Nvidia’s recent GTC event, bolstering Cisco's focus on the data center space, chief among which are its security solutions being embedded into the offering.

Newly added to the fabric were tools like Cisco’s Hybrid Mesh Firewall, which sets consistent security policies across endpoints, including network switches; and AI Defense, which performs automated vulnerability testing across applications, agents, and models.

Cisco also added Red Hat software tools that are now available alongside ongoing support for the Vast Data Platform and Nvidia’s AI Data Platform.

Speaking to SDxCentral at the recent GTC event, Kevin Wollenweber, Cisco's SVP and GM of data center and internet infrastructure, said themes from Nvidia CEO Jensen Huang’s keynote around agentic AI and the security implications of Nvidia’s NemoClaw framework for the popular OpenClaw platform underscored the timing of Cisco’s moves.

“The seminal moment for OpenClaw was that adoption pace, and now people were scrambling to introduce security and make sure we can wrap those things up," Wollenweber said. "And that's really the selling proposition for a lot of our enterprises.”

The AI Defense part of the wider Secure AI Factory stack also sees support from OpenShell, part of Nvidia’s GTC-debuted Agent Toolkit, which adds controls and guardrails to govern agent and claw actions. Cisco suggests this will help entrepreneurs racing to adopt the ability to deploy agents to manage critical workflows without compromising security.

Enterprises, Wollenweber argued, aren’t just shopping for point products, but for “overarching architectures,” with Cisco attempting to showcase Secure AI Factory as that framework.

“When you think about a Secure AI factory as a full stack, bringing in both an independent software vendor (ISV) partner as well as other general software partners, you can buy Secure AI factory from Cisco that includes all of our components from Splunk, but also Nvidia's AI enterprise software and now Red Hat's AI software stack as well," Wollenweber said.

Beyond security tools, the updates also see available hardware widened, with Cisco adding support for its custom Silicon One G300 – a recently unveiled switch line capable of supporting up to 102.4 Tb/s.

Cisco also extended support for new server lines powered by Nvidia hardware. The chip giant’s Spectrum-6 Ethernet switch silicon forms the basis for Cisco’s N9100.

Wollenweber said the addition of Nvidia and its own offerings provides diversity of supply chain for silicon.

“If you look at a large-scale build, having some diversity of supply chain helps," Wollenweber said. "So Cisco is very different from most of the other Nvidia OEMs, and we could sell just an AI factory that includes the Spectrum-X technologies, GPUs, and everything else. Or we can actually insert our own Silicon One technologies into that same ecosystem, into those reference architectures. So we can actually supply Spectrum-based forwarding silicon and Cisco-based forwarding silicon in a common architecture to give them some diversity as well."