The BionicSoftHand 2.0 may be an impressive robotic hand, but presenting it as a fundamental component of human-robot collaboration is a bit like saying a stone pizza oven is essential for reheating a frozen pizza.
For many industrial tasks, a basic robotic gripper is more than enough, and I don't see why we would spend on such a complex hand unless strictly necessary.
It's the kind of thing that has a high price and is justified only for very precise operations, like picking up a grain of rice, not moving boxes.
Such specificity is more of a niche than a general rule for automation.