Cisco expands Secure AI Factory with NVIDIA support

Cisco has expanded its Secure AI Factory with NVIDIA to support rack-scale deployments, extending the companies' joint AI infrastructure work.

The update adds support for NVIDIA GB300 NVL72 and NVIDIA RTX PRO 6000 Blackwell Server Edition systems within Cisco's architecture for AI data centres. The design combines networking, security, observability and automation in a validated full-stack approach for organisations building and operating AI systems.

The companies are positioning the latest version of Secure AI Factory as a way to address the operational and security demands of larger AI installations. As businesses move from early deployments to broader infrastructure roll-outs, they are looking for pre-integrated designs rather than assembling separate components themselves.

Under the expanded design, Cisco is adding support for NVIDIA Spectrum-X Ethernet networking and integrating its Nexus, UCS, Hyperfabric and security products with NVIDIA AI Enterprise software. The architecture is intended for both greenfield and existing environments.

Broader stack

The offering includes Cisco AI Defence to protect AI applications and workloads, Hybrid Mesh Firewall for segmentation and policy enforcement, and Splunk for monitoring and analysis. It also includes automation tools such as Cisco Nexus Dashboard, UCS Manager and Intersight, intended to simplify deployment and operations.

The rack-scale support centres on systems designed for larger training and inference workloads. Cisco said adding NVIDIA GB300 NVL72 is aimed at customers building AI factories that require tightly integrated compute and networking, while support for RTX PRO 6000 Blackwell Server Edition targets enterprise AI workloads and model deployment.

NVIDIA's role in the joint design spans compute, networking and software. The latest architecture incorporates NVIDIA AI Enterprise for data science and production AI use cases, alongside NVIDIA accelerated computing platforms.

Channel model

The system is intended to be sold through Cisco's partner ecosystem, including channel partners and system integrators working with customers on infrastructure planning, deployment and management.

Jeetu Patel, Executive Vice President and Chief Product Officer at Cisco, said the expansion reflects the changing requirements of customers building larger AI environments.

"The next era of AI infrastructure can't just deliver raw compute; it must harness that compute through validated, full-stack architectures. Cisco Secure AI Factory with NVIDIA brings together the infrastructure, software, and security needed to help businesses deploy AI at scale, while managing risk, reducing complexity, and enhancing operational reliability," Patel said.

The announcement builds on a partnership the two companies have been developing around AI data centre infrastructure. Cisco has been seeking to strengthen its position in AI networking and security as customers increase spending on data centre systems needed for model training and inference.

Security is a central selling point of the expanded design. Organisations need to manage risks not only around infrastructure but also around models, applications and data flows as AI systems become more deeply embedded in business operations.

The launch also reflects growing market demand for pre-validated AI infrastructure stacks from major technology vendors. As businesses face pressure to move AI projects from pilots into production, suppliers have been presenting bundled designs that combine hardware, networking, software and management tools in a single architecture.

"As organisations race to transform with AI, many are struggling to justify the time and expense of piecing together complex technology from dozens of vendors. Cisco Secure AI Factory with NVIDIA gives enterprises a fast, trusted path to build and secure scalable AI infrastructure, using integrated solutions and validated reference architectures, while preserving flexibility and easing operations," Patel said.