[engi_nerd]

They are not as of this very moment a viable total replacement. A valuable supplement to manned aviation, yes, drones are already doing this. In a few decades, who is to say? I don't know enough to even hope to make a prediction.

So these are problems that exist now:

The data link between the drone operator and drone is a weak link. It can be intercepted, studied, and even broken. Again, the Iran RQ-170 incident: https://en.wikipedia.org/wiki/Iran%E2%80%93U.S._RQ-170_incid... This is not an insurmountable problem, but it means that drones are not an end-all solution at this time.

Drones are not cheap themselves. The X-47 program cost is over 800 million. https://en.wikipedia.org/wiki/Northrop_Grumman_X-47B Now most of that is R&D, and should the type be produced in significant numbers, expect costs per unit to be much much lower, but doing some rough figuring, I doubt that costs will ever be much below $20 million per copy. Simply based on the sensor packages that would be installed.

Drones as they are now require more staffing (they can fly for long periods of time, and you need regular crew swaps). And they're not the sexiest things to operate (see: http://www.rt.com/usa/us-drone-pilots-exhausted-demoralized-...).