Edge computing is gaining ground in the vision industry. However, it is struggling with bandwidth and latency limitations when it comes to high-resolution, high-speeda and/or complex multi-camera systems. This presentation explores the bandwidth bottlenecks and technical challenges and investigates technical solutions to implement high-bandwidth, low-latency imaging with AI processing on the edge.