In a form 10-K filing from early 2014, Intel (INTC 2.11%) described its "tick-tock" product development methodology as follows:
As part of our R&D efforts, we plan to introduce a new Intel Core microarchitecture for desktops, notebooks (including Ultrabook devices and 2 in 1 systems), and Intel Xeon processors approximately every two years and ramp the next generation of silicon process technology in the intervening years.
Intel just filed its form 10-K for the most recently ended fiscal year. Notice what the company had to say about its tick-tock methodology now:
As part of our R&D efforts, we plan to introduce a new Intel Core microarchitecture for desktops, notebooks (including Ultrabook devices and 2 in 1 systems), and Intel Xeon processors approximately every two to three years and ramp... the next generation of silicon process technology in the intervening periods. [Emphasis added.]
This is a subtle difference in wording, but it appears quite meaningful. No longer does Intel seem confident that it can stay on its traditional two-year schedule for rolling out a new microarchitecture. This is reflected in a revised time frame. Moving to more advanced semiconductor manufacturing technologies while also reducing costs per generation is known to be tough and to only be getting tougher.
Intel is unlikely to intentionally slow things down, though
I think Intel's internal goal is to stay on the more traditional two-year cadence. Indeed, when the company's Broadwell family of processors was significantly delayed as a result of 14-nanometer manufacturing difficulties, Intel still kept the Broadwell successor, Skylake, on track to launch within a matter of months of its original schedule.
An Intel representative was recently quoted in the press as claiming the 10-nanometer follow-on to Skylake (known as Cannonlake) would launch in "early 2017." Intel subsequently retracted that statement, but I wouldn't be surprised if it proved to be true. In that case, it would be a little under a year and a half between the launch of Skylake and the launch of Cannonlake.
For some context, if we look at the history of past releases, Intel launched its first 22-nanometer part, Ivy Bridge, in April 2012. Haswell, the 22-nanometer "tock" came in June 2013. Broadwell didn't show up until October 2014, Skylake is looking like an August 2015 deal, and Cannonlake could be a January 2017 launch.
The time between Haswell and Skylake looks to be about two years and a quarter. If Cannonlake's successor shows up exactly a year after Cannonlake arrives (remember, delays almost always happen), then the delta between Skylake and Icelake will be a little under two-and-a-half years. Factor in the inevitable delay, and -- voila -- it winds up being about three years between major architecture introductions.
It all makes sense now
At the company's investor meeting, Doug Freedman from RBC Capital noted during a question-and-answer that all through the company's presentation, very little mention was made of this "tick-tock" methodology. Intel's management was pretty hesitant to confirm or deny whether the "tick-tock" development methodology was still in place, but the recent form 10-K filing clears things up.
Intel is still trying to adhere to the traditional "new architecture every two years" cadence, but it seems that -- after having worked so well for so long -- the technical and economic realities of moving to new manufacturing technologies mean that Intel might not always be able to achieve that. So, it seems that in the interest of transparency, Intel has officially redefined "tick-tock."
