Cisco Systems and Nvidia are partnering to transition AI infrastructure from GPU acquisition to rack-scale production, integrating networking technology to create more efficient, enterprise-ready AI factory systems.

Key Points

• Cisco and Nvidia are collaborating on a "Secure AI Factory" architecture that integrates Nvidia’s Spectrum-X Ethernet networking with Cisco’s enterprise management and operating systems.
• The partnership includes a new rack-scale compute solution developed with Supermicro, featuring liquid cooling and support for Nvidia’s Blackwell and Vera Rubin platforms.
• The initiative aims to shift AI infrastructure from bespoke supercomputing projects to repeatable, industrial-grade systems that are orderable starting in September.
• The architecture focuses on "continuous token production" by optimizing the entire stack, including physical data centers, chips, networking, and software models.
• Cisco’s role serves as a bridge, connecting Nvidia’s high-performance AI clusters to existing enterprise data, storage, and application environments.

Why it Matters

This partnership signals a shift in the AI market from simply acquiring hardware to building reliable, scalable production environments that integrate with existing enterprise infrastructure. By treating the network as a core component of the computer, these companies aim to maximize the return on investment for massive AI deployments through better lifecycle management and operational efficiency.