Cisco Unveils Secure AI Infrastructure in Collaboration with NVIDIA
Cisco and NVIDIA Launch a Secure AI Factory
Overview of the Announcement
Cisco and NVIDIA recently introduced an innovative AI infrastructure architecture called the Cisco Secure AI Factory. This collaboration is aimed at simplifying the deployment and management of AI workloads while ensuring robust security at every level of the infrastructure. The initiative was officially announced during the GTC event held in San Jose on March 18, 2025.
Strategic Goals of the Collaboration
This partnership aims to empower enterprises to unlock the full potential of AI. According to Chuck Robbins, Cisco’s CEO, integrating networking and security is vital for harnessing AI’s capabilities efficiently. Both companies are leveraging their expertise to create a system that allows clients to smoothly adopt and scale AI without compromising security.
Jensen Huang, NVIDIA’s founder, emphasized the importance of embedding security into all layers of AI infrastructure. The partnership seeks to create a secure foundation for businesses to confidently scale their AI initiatives.
The Infrastructure of AI Factories
AI factories, designed to support AI workloads, are becoming more prevalent. They must not only offer considerable computing power but also address intricate security challenges. The Cisco State of AI Security Report highlights various AI-specific threats that organizations must manage. Addressing these concerns effectively enables companies to enhance agility and speed in their operations.
Key Components of the Cisco Secure AI Factory
The Cisco Secure AI Factory architecture stands out because of its integration of several critical technologies. This collaboration aims to create a secure AI infrastructure that includes:
1. Compute
- Cisco UCS AI Servers with NVIDIA HGX and NVIDIA MGX enable accelerated computing for AI tasks.
2. Networking
- Cisco Nexus Hyperfabric AI and NVIDIA Spectrum-X Ethernet provide advanced networking solutions to support AI workloads.
3. Storage
- High-performance storage solutions are facilitated by partners such as Pure Storage, Hitachi Vantara, and NetApp.
4. Software
- The NVIDIA AI Enterprise software platform simplifies the development of production-grade AI workloads.
Security Layers in the AI Factory
Security is integrated at every level within the Cisco Secure AI Factory. The security measures include:
Securing the Infrastructure
- The Cisco Hybrid Mesh Firewall ensures consistent security across various points in the network, helping to detect and block threats efficiently.
Securing the Workload
- Cisco Hypershield provides real-time monitoring and threat detection. It helps prevent unauthorized movement within the workloads and offers streamlined management from a single interface.
Securing the AI Application
- Cisco AI Defense equips security teams with tools to identify risks in AI applications during their development phase, ensuring compliance with security standards such as NIST and OWASP LLM Top 10.
Flexible Deployment Options
The Cisco Secure AI Factory offers two main deployment options:
Ready-to-deploy Solution:
- Utilizing Cisco Nexus Hyperfabric AI and associated security tools, this pre-configured solution simplifies the secure AI lifecycle, from design to deployment and monitoring.
- Build-your-own Approach:
- Customers can select modular components from Cisco, NVIDIA, and their storage partners to construct a customized infrastructure that fits their specific needs.
Future Availability
Solutions based on the Cisco Secure AI Factory architecture are expected to launch by the end of 2025, although several individual components are already available for purchase. The rapid development aims to meet the increasing demand for AI infrastructure in today’s fast-paced market.
Summary
Through their collaboration, Cisco and NVIDIA are setting the groundwork for a secure and efficient AI infrastructure. By focusing on security at all levels and providing flexible deployment options, they aim to empower organizations in their AI journeys.
