Apparently fiber isn't enough for Google, as they are again putting resource to work up in the sky. They are now said to be backing WorldVu and it's plan to put 360 satellites up into low earth orbit starting sometime in 2019. The target would be high speed wireless broadband access into the hard to reach spots on the planet.
When the AT&T and DirecTV deal was announced, one of the reasons I couldn't get excited about it was that there's only so much you can do with satellites. Of course, Google was busy getting ready to prove my thoughts insufficiently imaginative. Of course, satellites aren't all like, nor are orbits.
DirecTV's satellites are in geostationary orbit out at about 22,000 miles with all the latency limitations that implies, but it's great stuff for broadcasting. WorldVu's would be at just 800-950km, which is an entirely different kettle of fish. In fact, it's much more like Google's Project Loon and its cloud of internet access balloons/blimps. In fact, it's so much like that in terms of intent, it's probably part of an overall shotgun approach to the same problem.
Just where a cloud of LEO satellites would really fit in the world's internet infrastructure in 5-10 years is an open question. After all, that's a couple lifetimes in tech, and who knows what the internet will look like by then. But the various last mile technologies seem increasingly likely to be interconnected and co-dependent -- towers, small cells, FTTH, cable, LEO satellite, balloons, and who knows what else, all hooked into the same underlying global fiber backbones.
The more pieces Google gets a grip on now, the more influence they'll have later on. The LEO satellite cloud dream isn't a new one, (remember the original Iridium dreams). But it shares a key commonality with their Kansas City FTTH efforts. It's a dream that others have tried at and failed, became somewhat tainted as just too hard and a bad business idea, and is now perhaps going to be rehabilitated.