The "power hungry motors" part is a bit misleading, these robots very much have electric motors in the air compressors needed to drive them. In addition, because they are driven by pneumatics they are much less efficient than their electric counterparts and thermodynamics says this can't improve much either.
There are a host of other problems with pneumatics too. Bandwidth, or how fast one can move reliably, is limited because air is compressible. In fact, bandwidth of pneumatic systems is often lower than that of the human walking gait. TL;DR pneumatics typically can't move fast enough to move with a person.
The other big problem, in industrial robots at least, is that pneumatics isn't reliable enough compared to rotary electric actuators. In general, systems on industrial robots need to have a Mean Time Between Failures(MTBF), or how long they can run until they break, of 100,000 hours or they are not profitable over humans. Pretty much every industrial robot today has an MBTF of 100,000 hours, this is around 10 years of continuous operation. Pneumatics can't get anywhere near this, and I suspect the gripper in the article can't either.
In fact industrial robotics has shown a trend of moving away from pneumatics, hydraulics, electric cable drives, and even electric linear actuators in favor rotary electric actuators.
That being said soft robots probably have a future, it just isn't pneumatic. The big problem with soft robotics right now is that there aren't very many soft actuators and all of them suck. Though in 2 years this could very well change...