While I can understand the reluctance to upgrade computers in terms of cpu/motherboard because it does take longer to take out, clean up and put in new parts. A gpu change can take literally a couple of minutes including unplugging everything and putting it all back.
There is no reason to buy now for games up to 5 years from now.
What is the point of spending even if you went with 980ti's, over 2k for one year of having flaky but epic when it works performance at the expense of having 3-4 years of worse performance than you could otherwise have.
Lets say 5 years
1st year, unmatched performance
2nd year, 2 of the next gen cards will surpass performance due to scaling issues beyond 2 cards. 2 cards also works better in a hell of a lot more games than 4 games do.
3rd-4th year 1 next next gen card will match your current setup easily but in every game, with new features, smoother and with no sli issues at all.
4-5th year, a single next next next gen card will be beating your now 4-5 year old quad card setup handily.
Lets say 980ti's and £2200 all for great(when it works) performance for a year but meh performance the second 16nm chips come along. Your other option is
1st year 2x 980ti's £1100
2nd year 2x 1080ti's £1100 - £600 for the old cards = £500
3rd-4th year 2x 1180ti's £1100 - 600 for the old cards, £500.
By the third year you spend the same amount, have much more reliable performance less noise and faster performance. For any games where sli is plain broken you'll have double the performance the second year, 4 times the performance after that, etc.
This is also the single worst time to upgrade. End of 28nm, drop to 16nm next, HBM and 2-4 times the bandwidth, DX12 and architectures tuned to next gen engines with dx12 thought about upfront. At best if you absolutely(for a reason I can't fathom) have to have 4 cards and keep them for 5 years, wait till the first 16nm parts. 16/14nm + HBM2 is going to cause a massive shift in performance AND what devs can do with that bandwidth, new features to take advantage of it.