For decades, we’ve transformed how video is captured, compressed, and distributed, yet one principle remains unchanged: live streaming still transmits every pixel of every frame, even when most of the scene hasn’t changed. In live sport, the pitch, stadium, and crowd are largely static, while only a small fraction changes. Modern codecs reduce visual redundancy, but the pipeline remains frame-based, driving unnecessary complexity, cost, and latency. As AI and edge compute mature, a more fundamental question emerges: what if live media could be distributed by transmitting change, not pixels? This session will discuss what would happen if AI could analyze the scene in real time, identify what has changed, and enable devices and networks to reconstruct the experience dynamically — shifting live streaming from continuous data delivery to intelligent, context-aware transmission.