Indeed, now I'm in a bad spot because if I had jumped on skylake, then picking up a second gen 6-8 core in 2019 would have been tolerable. Now? lol I either pick up a refresh that might be leaped in a years time or stay stuck on this dinosaur for another year.
I don't plan on getting a new gpu until 2018 anyways, so I might just end up biting the bullet and do a total rebuild then. High end gaming is going to suck, but I think it'll probably pay off better in the long run. Then again it was thinking like this that got me here in the first place so ¯_(ツ

_/¯
I've been thinking about this a lot recently - especially since upgrading to a GTX 1070 and confirming that a 2500K is really holding things back.
I'm also regretting not having picked up a 6700K now that we know the 7700K is going to perform exactly the same at the same clockspeed.
The 7700K may clock a
little higher, and you get a newer platform (mainly Optane and 4K Netflix support) but I would have preferred to upgrade and get this performance last year instead.
It appears that Skylake-X is going to be about 9 months away and I don't want to wait that long.
I've already been delaying a new build for more than a year now.
I'm thinking that it may be best to spend the minimum amount possible on 7700K-specific hardware, and then sell it to do a full new 7900K build when Skylake-X launches.
Spend big on RAM now because everything will still be using DDR4 and that isn't going to change.
Get the fastest 4x16GB package I can, since Skylake-X will be quad-channel and 64GB is the maximum that will be supported by the 7700K.
Pick up a 7700K in January, whatever the cheapest Z270 motherboard with an Intel NIC is, and get a mounting adapter from Noctua for my NH-D15.
It will still be a current platform when the 7900K ships (Coffee Lake is 02/18) so it should hold most of its value.
Buying the cheapest Z270 board available minimizes what I could lose on it.
There's little reason to buy high-end motherboards now anyway. I don't need WiFi/Bluetooth/Thunderbolt/RGB lighting/Audio.
This spreads out the cost a bit (buying the RAM now, setting money aside over the next 9 months) and gives me the option of waiting for Coffee Lake (mainstream 6-core) if I decide that the 7900K (enthusiast 8-core) is going to be too expensive.
I'm thinking that going from 4 cores to 8 rather than 6 is the right move for me though. Coffee Lake is likely to still be using dual-channel memory rather than quad-channel, and games seem to be requiring more memory bandwidth now too.