I'm not sure which exhaustion counter you've been looking at. I've been keeping an eye on a number of exhaustion predictions for the past few years and they have been reasonably consistent (i.e. +/- 6 months). The allocation policies have been changed over the years and this has extended the amount of time we have, but not by much. Obviously exhaustion predictions can't take into account policy changes until they are at least visible on the horizon, so I do expect it'll be extended a bit more, but I'm honestly not expecting that extension to be more than a few months. New policies will also probably start making it much harder for people to get IPv4 addresses, so increasing the pressure to migrate onto IPv6 before the IPv4 addresses are exhausted.
there are no even halfway accurate estimates of that date.
And _that_ is why ISPs need to act now (actually, several years ago) to prepare themselves. This *is* going to happen, the longer they leave it, the more chance they will be caught with their pants down when it actually happens.
There are certainly short-term gains to be had by sticking your head in the sand and pretending that there isn't a problem. Unfortunately the cost of having to drop everything and roll out a whole new network at crunch time is going to be very expensive, far outweighing those short term savings. Sadly, business these days seems to be all about short term gains at the expense of long term viability.