When Cisco Systems announced its Nexus Hyperfabric solution at the Cisco Live event in June 2024, the focus was very much on how the new approach would address building and deploying generative AI clusters, in collaboration with GPU maker NVIDIA. Since AI was a major theme of the Cisco Live event, and AI hype was at a fevered peak, that announcement was an essential statement of strategy and direction, keeping pace with the rest of the market and the continuous stream of similar announcements by other networking and infrastructure vendors. A key aspect of that initial announcement was that Hyperfabric would vastly simplify and manage the full lifecycle of data center networks from the cloud.

The networking switch element that Cisco brought to the mix with this new offering is called the Cisco 6000 Series. Not to be confused with the end-of-life Cisco Nexus 6000 series of switches, the new Cisco 6000 Series would be a new design that would meet the specific and unique new needs of extreme high performance (400G and 800G), RDMA over Converged Ethernet. Also included were the automated configuration and management capabilities necessary to support the anticipated needs of deploying and operating GPU cluster architectures.

At the Cisco Live event, details on the Cisco 6000 were limited, and while well-rounded and comprehensive in scope, the Nexus Hyperfabric announcement was very much forward looking, promising delivery in 2025. Since the solution was aimed at enabling enterprise AI buildouts, as opposed to the hyperscalers that are the earliest generative AI adopters and deployers, the timing for solution delivery recognized that most organizations were still in the research and planning phases.