I think you are right and wrong. Yes silicon pipelines take $billions and many years to go to from a new implementation to a fabricated, tested, and shipped product. Around a year for a minor revision.

However Intel has significant flexibility on hitting price/performance points by tweaking enabled features (memory channels, ecc, virtualization support, accelerated crypto, etc) and of course playing with clockspeed and power.

So the new x-series for instance has all the marks of a rushed product launch rebranding existing chips (the LGA20XX xeons). So things like overheating VRMs, poor heat spreader design, minimal clock headroom, and poor availability. In fact the top x-series isn't expected to ship until the end of the year.

Also Intel knew Epyc was coming so while quietly shipping skylake based xeons to large cloud providers while waited for AMD to show their hand with the Epyc launch. Only after Epyc did Intel announce details on the already shipping skylake xeons.

Suddenly Intel's talking about desktop chips with more than 4 cores after a decade of quad cores (Q6600 in 2007). Think that would have happened without Ryzen?

I also expect sudden changes when AMD ships the the Ryzen based APUs.

The 18 core i9-7980XE is widely thought[1] to be a direct response to the 16 core Threadripper announcement. It was announced late, with much less info. It has a expected release date months later than the rest of the i9 lineup, and is using higher core count silicon that Intel has previously only ever used in Xeons.

[1]http://www.anandtech.com/show/11464/intel-announces-skylakex...

I could imagine them stockpiling chip designs for when they experience competition or sales pressure.