Rack-scale AI: Cisco, Nvidia expand secure AI factory - SiliconANGLE

Cisco and NVIDIA have extended their 'Secure AI Factory' solution into the rack-scale era, delivering enhanced full-stack operational capabilities for enterprises. This advancement supports high-perfo... Read original →

The collaboration between Cisco and NVIDIA on rack-scale AI marks a pivotal shift from modular to full-stack integration in AI infrastructure. By combining NVIDIA’s Spectrum-X silicon with Cisco’s NX-OS and SONiC platforms, the solution enables deep GPU networking and compute coordination, supporting trillion-parameter model training. With liquid cooling systems exceeding 200kW per rack, the architecture enhances energy efficiency and deployment density, signaling a move toward high-density, low-latency AI data centers. This advancement strengthens enterprise AI adoption but introduces supply chain vulnerabilities, especially in critical components like chips and cooling technologies. Competitors like AMD and Intel may accelerate their own AI-native architectures to capture market share. Over the next 12–24 months, the AI Factory model will become the core infrastructure for enterprise digital transformation, pushing data centers toward intelligent operating systems and establishing new technological moats.