First off, as an Australian and a nerd, I am very proud.
Good news: there is now a formally verified microkernel. 8,700 lines of C and 600 odd lines of ARM assembly. Awesome.
Bad news: it took 200,000 lines of manually-generated proof and approximately 25 person-years by PhDs to verify the aforementioned microkernel.
Conclusion: formal verification of software is not going to take off any time soon.
25 man-years is not that much, if we really valued it. If the process is modular, 100 persons do 25 man-years in 3 months. Now, I have no idea by what method this was performed, but surely it has to at least be modular over different programs, so that we could, if we really cared about it, have a formally proven whole distribution in a few years.
It is by no means out of reach.
But you are right that it would take all the required man-years away from developing time, so we can be quite confident that it will not be adopted, it would slow down developing time too much to justify the cost for most developers.