For more than a decade, Intel was driven by a "tick/tock" development model. The "tick" took place one year and was a manufacturing process shrink, from 45nm to 32nm, for example. The "tock," which took place one year later, was a whole new microarchitecture, such as Penryn to Nehalem.
For about a decade, tick/tock worked well. Intel choked the life out of the insurgent AMD and dominated the x86 market for a long time—and still does. But the limitations of physics have caught up with the company, and ticks are becoming much harder to come by.
As it is, Intel delayed the move to 10nm by adding a third generation of 14nm chips called Kaby Lake. The shrink to 10nm was planned for next year's Cannonlake processor. Now word is that might be delayed another year, with 10nm coming in 2018. And the next shrink, down to 7nm, won't take place until after 2020.
Last year, Motley Fool writer Ashraf Eassa spotted a job listing for Intel's India research office looking for engineers to work on an advanced process technology specifically geared for 7nm designs and for release in 2020.
Since then, the job listing has been updated, pushing back the launch date by two years and does not mention 7nm technology:
The India Lab specifically, in collaboration with MRL-US and Intel product architecture teams worldwide, will spearhead the research and advanced development of Microprocessor Cores in the 2022 and beyond timeframe. By conceiving of and prototyping radical approaches, the Lab will aim to deliver much greater CPU power and area efficiency while still delivering industry-leading performance. The microarchitecture and design of these advanced CPUs will be aggressively co-optimized with Intel's sub-10nm technology nodes deep into the next decade.
Intel has said it plans three waves of 10nm technology: 10-nanometer, 10-nanometer+, and 10-nanometer++, to be released with Cannonlake in 2018, Icelake in 2019, and Tigerlake in 2020. Since Tigerlake is supposed to be the "tick," the next generation after Tigerlake will be the new microarchitecture, which does not get a shrink. So, by this schedule, Intel doesn't get to 7nm until 2022.
It doesn’t really matter if number of transistors double again
Does this matter to anyone beyond the Tom's Hardware/Anandtech propellerhead crowd? Not really. Sure there are power savings to be had with a process shrink, but look at what is being done with the current generation. Kaby Lake is going to unleash some powerful tablets and 2-in-1 laptops. My desktop processor is a Haswell generation, and I feel no great compulsion to upgrade. My SSD and GPU were much more important upgrades.
For years, Intel lived and died by a marketing maxim masquerading as a technology maxim: Moore's Law. Why should you care that transistors double every 18 months, except that you are now out-of-date technologically and should upgrade? That was the real genius of selling Moore's Law to the public. It codified planned obsolescence better than any marketing guru could hope for.
But now physics has caught up with processor design, and processor performance is so good that it really doesn't matter any more if the number of transistors doubles again.