Vision-Language-Action Models on the Tech Tree