Cisco announces $5 trillion data center plan... leading "ultra-low latency" AI innovation

Cisco Chief Product Officer Gitu Patel recently discussed the rapid growth of AI infrastructure, emphasizing that “ultra-low latency” is becoming a key factor. As modern enterprises increasingly adopt “always-on” intelligent systems, networks must deliver nanosecond-level response times alongside large-scale performance. To meet this demand, Cisco is building a chip-based full-stack AI platform, with future data center construction costs potentially reaching $5 trillion (approximately 7.2 quadrillion Korean won).

Patel stated, “In the short term, we underestimated the capacity needed to meet AI demands,” adding, “Current supply cannot keep up with demand.” This shift requires a reevaluation of infrastructure from data centers to edge devices. Cisco plans to ensure AI autonomy through network designs that feature ultra-low latency, high security, and high energy efficiency.

This is not only about improving network performance but also about designing systems at the network layer to operate within trusted boundaries. Patel proposed effective network strategies that balance data integrity and the stable operation of AI systems, which will be further strengthened through collaborations with companies like NVIDIA and Vast Data.