Intel has been using their “Tick-Tock” processor development methodology for quite a while now. A “tick” would represent a new node and a “tock” would represent a new architecture. The latest true “tick-tock” that we saw was Ivy Bridge to Haswell, where Ivy Bridge was the “tick” down to 22nm and the “tock” was Haswell. Well it loos like Intel is moving away from this “Tick-Tock” release schedule and is moving towards a strategy that will leverage a single process node during a 3-year cadence.
Here is the official word from Intel:
“We expect to lengthen the amount of time we will utilize out 14 [nanometer] and out next-generation 10 [nanometer] process technologies, further optimizing out products and process technologies while meeting the yearly market cadence for product introductions.”
So for 14nm this would be Broadwell (2014), Skylake (2015), and Kaby Lake (2016). This new strategy will involve three phases, first the arrival of the new process (Broadwell), then followed by the architecture phase (Skylake), and then the final phase would optimization (Kaby Lake).
We are not exactly sure how much of a performance difference you will see from the architecture phase to the optimization phase, but we should see some IPC improvement.