The computing world had many reasons to celebrate with the announcement of a 7nm chip made by IBM, but we closed that same article by saying that if no unforeseen events arose, the technology would reach the market in 2017. Well... an unforeseen event happened. Intel CEO Brian Krzanich recently confirmed that the shift to 10nm will be delayed until the second half of 2017. In other words, there will be a "double tock" at 14nm.
A Half-Century of Projection
Its demise has been announced so many times that the public began to see the specialized media as doom-mongers. Moore's Law has officially been with us for just over fifty years, and to be honest, it hasn't been free of modifications. The original version spoke of doubling the components in an integrated circuit annually, but in 1975, the value was adjusted to the then-current state of semiconductors, extending the time to two years. Even the term "law" stuck over time. Technically it's a projection, but with half a century of effectiveness, it's hard not to call it a law. Still, Gordon Moore himself said last March that his observation would be over "within the next decade." The industry has implemented all sorts of techniques to keep reducing component size, and with IBM's recent announcement, there was hope for an extension.
Intel's Confession
However, Intel decided to throw a cloak of sincerity. Through a financial report, Intel CEO Brian Krzanich confirmed that there will be a delay of about two years in the implementation of the so-called 10nm node, which would symbolize the debut of the Cannonlake architecture, a new tick in the structure. Today we are in what is Broadwell, also a tick, and compression of Haswell to 14nm. Compared to previous generations, the number of available Broadwell processors is relatively low, a detail that is more noticeable among desktop products. Krzanich himself admitted that the migration to 14nm has been much more complex than calculated, and that the first signs already appeared with the previous transition to 22nm.
What Does the Future Hold?
What's the outlook? For the coming months, everything will stay the same: Broadwell will be replaced by Skylake, a new 14nm design that establishes the tock, but instead of moving to Cannonlake at 10nm, there will be a kind of "double tock" with Kaby Lake, a third 14nm architecture that in theory will offer mobile and desktop chips like its predecessors. Krzanich doesn't believe this is the end of the tick-tock system, but it's a good time to think about the future. We've felt very comfortable with the stability of Moore's Law. Maybe it's time to rethink our processors? Will it be possible to increase surface area or explore 3D designs? Comments are open.
