Well, at microwave frequencies (think anything at 10Ghz and up, really) you're already talking about directional antennas. The RF engineers will call anything above 1Ghz microwave, but that's really splitting hairs.
Fiber is a waveguide and not much different in that respect from coaxial cable (or for that matter twister pair). What matters is what kinds of frequencies the waveguide will accomodate and what kind of attenuation those frequencies will experience. This matters for lots of important reasons.
As an example, take RG-6 coaxial cable (the kind you probably have in your home):
At 100Mhz (VHF) the attenuation is about 2.0 dB/100ft.
At 700Mhz (UHF) the attenuation is about 6.0 dB/100ft.
Now, take a look at Corning SMF-28 singlemode fiber:
At 1310nm (229.644 THz) the attenuation is 0.35dB/kilometer.
At 1550nm (192.4 THz) the attenuation is 0.22dB/kilometer.
So, fiber is an exceptional waveguide at very high frequencies, which makes it uniquely suitable for high-bandwidth communications. Compared to all of the other waveguides, it's the most durable, most compact, and most future-proof solution.
Except when you can't use it.
SO, what's left? Wireless probably. Well, there is free-space optical which uses the same 1550nm frequency but has the problem of aligning the transmitter and receiver to have a completely free line-of-sight. Then there is the fact that attenuation through free space isn't the same as within a fiber. Attenuation at 1550nm ranges from 0.2dB/km (clear) to 100dB/KM (foggy). So, the atmosphere is a shitty waveguide for optical frequencies.
But there is still a chance with wireless, since we aren't restricted to the propagation characteristics of optical frequencies or the directionality. With advanced modulation techniques (CDMA/OFDM), it's actually pretty easy to fit many more bits/hertz than you could with fiber transmission systems. It's just that that sort of heroics is not needed for fiber systems since it's easy to get extra bandwidth with another fiber.
Now, when you are talking about "last-mile" types of solutions, it really depends on what the alternatives are and what the "load-factor" is for that method. There is a big difference between phone calls and streaming audio for example when it comes to wireless systems. All of those things need to be taken into account.
IMHO, the end-game is probably going to be coax to a smaller set of homes using whatever flavor of DOCSIS is available at that moment. Most CableCo networks are mostly fiber anyhow. Once you get to a certain "cluster size" the choice of fiber to the home versus fiber to the local node and then coax to the house becomes an engineering question. For lower bandwidth or other services, you'll see wireless broadband displace some take-up of coax-base services. The only reason I think this will happen is because the coax is already deployed (and so it will be used). If we didn't have almost universal penetration of coax into the average home, there would be no question that most broadband would be wireless to the home.
The FTTH systems now being deployed have lots more in common with cable systems than is commonly understood. Your fiber doesn't actually go ALL the way to the central office. Your fiber is split at some point in the neighborhood (from 4-64 homes) and then is carried to the central office.
Fiber is a waveguide and not much different in that respect from coaxial cable (or for that matter twister pair). What matters is what kinds of frequencies the waveguide will accomodate and what kind of attenuation those frequencies will experience. This matters for lots of important reasons.
As an example, take RG-6 coaxial cable (the kind you probably have in your home):
At 100Mhz (VHF) the attenuation is about 2.0 dB/100ft.
At 700Mhz (UHF) the attenuation is about 6.0 dB/100ft.
Now, take a look at Corning SMF-28 singlemode fiber:
At 1310nm (229.644 THz) the attenuation is 0.35dB/kilometer.
At 1550nm (192.4 THz) the attenuation is 0.22dB/kilometer.
So, fiber is an exceptional waveguide at very high frequencies, which makes it uniquely suitable for high-bandwidth communications. Compared to all of the other waveguides, it's the most durable, most compact, and most future-proof solution.
Except when you can't use it.
SO, what's left? Wireless probably. Well, there is free-space optical which uses the same 1550nm frequency but has the problem of aligning the transmitter and receiver to have a completely free line-of-sight. Then there is the fact that attenuation through free space isn't the same as within a fiber. Attenuation at 1550nm ranges from 0.2dB/km (clear) to 100dB/KM (foggy). So, the atmosphere is a shitty waveguide for optical frequencies.
But there is still a chance with wireless, since we aren't restricted to the propagation characteristics of optical frequencies or the directionality. With advanced modulation techniques (CDMA/OFDM), it's actually pretty easy to fit many more bits/hertz than you could with fiber transmission systems. It's just that that sort of heroics is not needed for fiber systems since it's easy to get extra bandwidth with another fiber.
Now, when you are talking about "last-mile" types of solutions, it really depends on what the alternatives are and what the "load-factor" is for that method. There is a big difference between phone calls and streaming audio for example when it comes to wireless systems. All of those things need to be taken into account.
IMHO, the end-game is probably going to be coax to a smaller set of homes using whatever flavor of DOCSIS is available at that moment. Most CableCo networks are mostly fiber anyhow. Once you get to a certain "cluster size" the choice of fiber to the home versus fiber to the local node and then coax to the house becomes an engineering question. For lower bandwidth or other services, you'll see wireless broadband displace some take-up of coax-base services. The only reason I think this will happen is because the coax is already deployed (and so it will be used). If we didn't have almost universal penetration of coax into the average home, there would be no question that most broadband would be wireless to the home.
The FTTH systems now being deployed have lots more in common with cable systems than is commonly understood. Your fiber doesn't actually go ALL the way to the central office. Your fiber is split at some point in the neighborhood (from 4-64 homes) and then is carried to the central office.