Seriously, if DPR put a small fraction of his funds into developing such a thing, it would be nearly unstoppable. The real challenge is the design of the physicals, do you go with quality, or just easily replaceable parts? If I can print out most of a drone, do I really care about anything else than it will last for 'n' number of flight hours?
Yes, one approach would be a meshed swarm of redundant, inexpensive units covering an area. Another would be fewer, more expensive units large enough for a phased array of WiFi transmitters like a Vivato panel,
http://www.vivato.com/ .
Even better if the flexible solar materials come to production, then you're talking about integrating the cells right into the lifting envelope, using thin films. The only real bulk would come from what your payload is, and the support infrastructure including energy storage.
One possibility is to incorporate reflective dish concentrators into gimballed spherical lifting cells which can track the sun to focus light onto a high efficiency photovoltaic or mechanical system. Catalytic hydrogen production may be an option as well.
"Cracking" hydrogen from available water is an excellent idea, and if you could get the efficiency up there - maybe even use that for the motors themselves, retaining power just for periodical hydrogen extraction and transmitter power.
Hell, what about parasitic induction using a conductor and existing RF/Microwave signals? Is it enough to run some minor circuitry while keeping the solar generated power just for 'cracking' hydrogen?
Too bad spectrum is licensed. Open competition among transmitters with high signal strength might have led to enough wireless power in the air to actually capture. In reality, though, the wireless space would have probably subdivided into smaller and smaller low powered cells and evolved into an open mesh network in an unregulated environment.
