The KV Cache memory usage of a Transformer is proportional to sequence length, number of layers, number of attention heads and head dimension, growing linearly with context, which constitutes a hard memory ceiling for robot scenarios that continuously receive sensor streams.