Intel only just recently launched their 7th Generation Core series of processors. And those were built on top of their most recent iteration of the 14nm process, improved upon from Skylake. It appears that they’ll be using 14nm once again, with applicable improvements, for their 8th Generation Core, Cannon Lake instead of moving to 10nm. This will likely apply to Coffee Lake as well.
Cannon Lake will continue to use 14nm
Remaining on the 14nm node isn’t necessarily a completely detrimental thing, though it does call into question the actual yields that Intel is able to currently get our of their 10nm process. Keeping it at 14nm, even though it’ll be slightly improved over the previous generation, won’t bring the full amount of optimizations for performance and efficiency as would a new node plus an improved architecture. Though Cannon Lake will still presumably bring with it around 3% increase in overall performance per core. We’re hoping that any yield problems can be overcome quickly.
Something else of note is that during Intel’s Investor Day, it was revealed on one of their slides that data centers and the enterprise segment will be priority for the next process node. Traditionally Intel has had a mobile-first strategy, where new architectures were developed for lower power use-cases and released in the U and Y series first, the desktop K second and server processors a distant third. This could very well be in response to the potential performance improvements that are projected for AMD’s server iteration of Zen. They may want to continue to entice their data center customers to use their processors, even if Zen is less expensive in server guise. Specifics are sparse at the moment, though we expect Intel to clear that up in the near future.