Cisco has unveiled an AI factory architecture with Nvidia, building on the expanded partnership announced last month to provide validated reference architectures today.
Together, the companies are developing the Cisco Secure AI Factory with Nvidia to simplify how enterprises deploy, manage, and secure AI infrastructure at any scale.
“AI can unlock groundbreaking opportunities for the enterprise,” says Chuck Robbins, chair and CEO of Cisco. “To achieve this, the integration of networking and security is essential. Cisco and Nvidia’s trusted, innovative solutions empower our customers to harness AI’s full potential simply and securely.”
Jensen Huang, founder and CEO of Nvidia, says: “AI factories are transforming every industry, and security must be built into every layer to protect data, applications and infrastructure.
“Together, Nvidia and Cisco are creating the blueprint for secure AI – giving enterprises the foundation they need to confidently scale AI while safeguarding their most valuable assets.”
Developing and delivering AI applications require high performing, scalable infrastructure and AI software tool chain. Securing this infrastructure and AI software requires a new architecture – one that embeds security at all layers of the AI stack and automatically expands and adapts as the underlying infrastructure changes.
The new partnership on the Nvidia Spectrum-XTM Ethernet networking platform provides the foundation for the Cisco Secure AI Factory with Nvidia.
Cisco is integrating security solutions like Cisco Hypershield, to help protect AI workloads, and Cisco AI Defense, to help protect the development, deployment, and use of AI models and applications.
Building a Secure AI Factory
AI factories – data centres purpose-built to power AI workloads – are designed to be more modular, scalable and agile, but organisations must also look beyond raw compute power.
AI Factories must address new and complex security challenges. The recently published Cisco State of AI Security report analyses dozens of AI-specific threat vectors and over 700 pieces of AI-related legislation to highlight key developments from a rapidly evolving AI security landscape.
Organisations that strategically address both their AI infrastructure and security challenges simultaneously will be more agile, scale faster, and derive business value quicker.
Cisco Secure AI Factory with Nvidia is expected to build on the companies’ unique ability to offer flexible AI networking and full-stack technology options that leverage the planned joint architecture. The partnership will bring together technologies from Cisco, Nvidia and ecosystem partners into a secure AI factory architecture for enterprise customers, including:
- Compute: Cisco UCS AI servers based on Nvidia HGX and Nvidia MGX for accelerated computing.
- Networking: Cisco Nexus Hyperfabric AI and Nexus networking solutions, powered by Silicon One and Nvidia Spectrum-X Ethernet networking.
- Storage: High-performance storage from certified partners Pure Storage, Hitachi Vantara, NetApp, and VAST Data.
- Software: Nvidia AI Enterprise software platform to streamline the development and deployment of production-grade agentic AI workloads.
The Cisco Secure AI Factory with Nvidia includes security at all layers:
- Securing the infrastructure: Cisco Hybrid Mesh Firewall provides unified security management and consistent policy across multiple enforcement points, including network switches, traditional firewalls, and workload agents. This integrated approach ensures pervasive and consistent security, ranging from deep packet inspection to wide infrastructure coverage, detecting, blocking and containing adversaries. Cisco Hypershield (part of Hybrid Mesh Firewall) will, in the future, extend pervasive, zero-trust security enforcement to every AI node by integrating with Nvidia BlueField-3 DPUs.
- Securing the Workload: Cisco Hypershield prevents adversary lateral movement and proactive vulnerability mitigation without the need for patching, all from a single management interface. By monitoring and controlling process executions, file access, and network activities, Hypershield delivers deep visibility and surgical runtime enforcement within AI workloads. Future enhancements will further strengthen workload protection through integration with Nvidia BlueField-3’s DOCA AppShield for real-time workload threat detection in AI-focused virtual machines and containers.
- Securing the AI application: Cisco AI Defense empowers security and AI teams with comprehensive tools to protect AI applications from safety (e.g. off-policy, toxic behavior) and security (such as prompt injection, data privacy) risks across the development lifecycle. AI Defense integrates into existing CI/CD workflows to provide automated vulnerability testing and a common layer of runtime security across any number of models and applications. Additionally, AI Defense helps companies align to AI security standards with a single integration, including NIST, MITRE ATLAS, and OWASP LLM Top 10. Future enhancements include integration with Nvidia AI Enterprise to streamline AI security workflows.
Cisco Secure AI Factory with Nvidia will have flexible deployment options, including:
- Ready-to-deploy: Utilising Cisco Nexus Hyperfabric AI along with Cisco’s security portfolio and Nvidia technology, customers can deploy a vertically integrated AI solution that automates and simplifies the secure AI factory lifecycle from design to deployment and ongoing monitoring.
- Build-your-own: Featuring customizable modular components from Cisco, Nvidia and the companies’ storage ecosystem partners, customers can incorporate their current infrastructure and build solutions that are designed precisely for their unique environments.