Why latency is the first problem
In XR teleoperation, everything an operator sees and does has to travel from their headset and controllers to the robot, be acted on, and stream back as new camera video, all fast enough to feel immediate. When that round trip is too slow, the delay between moving your hand and seeing the robot's hand respond becomes disorienting, similar to the nausea some people feel in laggy VR games, except now the lag can also translate into a robot mistiming a grasp or colliding with something it should have avoided.
Keeping this loop fast enough to feel responsive is a genuine engineering challenge involving the network connection between operator and robot, the processing time on both ends, and the video compression used to stream camera feeds. It is one of the reasons many current teleoperation setups keep the operator physically close to the robot or on a very reliable, low-latency network rather than operating over the open internet from arbitrary distances.