Goodbye, 'tick-tock': Intel to delay 10nm processor switch until 2017

14nm to enter its third generation as Moore's Law collapses

Intel has today confirmed what many had feared - that it will enter into a third generation of building 14 nanometer resolution processors - bucking Moore's Law and effectively ending the company's so-called "tick-tock" manufacturing strategy.

Intel now intends to make the transition to 10nm manufacturing in the second half of 2017.

In accordance with Moore's Law - that says the number of transistors per square inch on integrated circuits should double every year - Intel had been alternating a shrinking chip size ("tick") with enhanced microarchitectural improvements ("tock").

The company had been carrying out this method since 2006, with NetBurst the first "tick" and Core the "tock" on 65nm, finishing in the present day with Haswell and - later in 2015 - Skylake, which will be the architectural "tock" on 14nm.

Cannonlake - built on 10nm - was supposed to be next, but as Intel already accidentally revealed via a leak last month, there will be a new 14nm processor first, and it will be called Kaby Lake. This, effectively, will be a second "tock" on 14nm.

Intel CEO Brian Krzanich, on a conference call after the company's Q2 earnings (which were slightly better than expected), elaborated on the lack of chip development, saying that while Intel has "disproved the death of Moore's Law many times over" the last two technology transitions have, for Intel, "signalled that our cadence today is closer to two and a half years than two".

In other words, the straight line on which Moore's Law depends is starting to plateau. Krzanich added that Intel will "always strive to get back to two years" in tick/tock scheduling, but that he was "not sure" whether such a thing was possible, and that Moore's Law could actually come to an end permanently.

Moore's Law was already "updated" in 1975 from its original claim that processing power would double every year - changing to Intel's current two-year cycle.

It is thus increasingly difficult to refer to a scientific "law" that has changed twice as any kind of law at all.

There could be more bad news for an uncertain Intel, as IBM last week revealed a new 7nm "super chip" composed of a silicon-germanium alloy, rather than just the standard silicon. Germanium aids electron mobility, allowing a level of current flow around the tiny chips that silicon alone can't handle.

Of course, Intel is presumably investigating enhancements beyond silicon as well, but IBM's unexpected pull ahead at a time when Intel is faltering could present an interesting new spin on the time-honoured chip race.