The use-cases discussed in TFA all seem pretty low-power, mostly recon-type stuff. That said, it'd be interesting to contemplate what the right proportions would be to actually run a hungrier payload.

At a certain point, "perpetual" also starts to run into problems of mechanical failure. Then again, if these things are cheap enough and/or able to limp home in the case of a snapped cable or failed prop motor, maybe it still works out.

Seems like the video talked about wanting to make it into an internet communication network, which just doesn't seam feasible to me.

Then again, I'm just a programmer. I recognize I have no grasp of the power requirements for that kind of project.

The inexpensive nature of them seems a major draw. I wonder if they can detach cables and have the wings fly home on their own while having the center parachute down?

Limping home would be the easy part. It would be really impressive if the system had the ability to attach the tether of a replacement "blade" in flight, extending/replenishing the formation on the go.

But we are hardly able to keep conventional solar glider UAVs in the air, I suspect that adding the drag of a tether moves the goalposts far out of reach.