Dan Warren, Head of 5G Research, Samsung Electronics R&D Institute UK In the synopsis of the event there are three types of ‘x-haul’ (which overlooks ‘cross’), three types of ‘x-ran’ (which overlooks ‘O’) and seven reference points acting as the interface from one part of the network to the next. How did we end up with all these options? Does no-one remember how much trouble even the suggestion of ‘A-ter’ interface caused? Today, we are in a very different place, where a network is made of a radio interface and a massively distributed compute platform supporting a (using the term advisedly) sliced-and-diced mixture of software-based functions in every possible topology you could think of. Is it realistic to believe all of this can co-exist in a single operation? And if they do (or even, if it is a subset of them), doesn’t this make the network nigh-on unmanageable? Or does this inexorably lead us to discuss AI, and a network where some of the x-haul bandwidth is needed just for perpetual network reconfiguration and optimisation?