The adherence by manufacturers to a Law that suggested the number of transistors on a given area of silicon would double every 18 months has been impressive. Yet silicon can only go so far; after a while, you begin to run out of atoms to play with.
The industry isn't quite at that point – foundries are still pushing hard towards 10nm production and IBM has revealed a 7nm test chip – but the 'brick wall' is now in sight.
So it's no real surprise that Intel is having to slow the cadence of its product development programme. It has always adopted a 'tick tock' approach; a 'tick' takes devices to the next process node, while a 'tock' sees an improvement to the microarchitecture.
For some years, the 'tick tock' has happened on a two year cycle; now that frequency is being extended. In its latest financial results presentation, Intel admitted the introduction of the next generation Skylake processor would be six months later than anticipated. The 14nm Broadwell processor was similarly delayed. It's now 'tick … tock'.
Why should we be surprised? Building devices with such microscopic dimensions poses huge technological challenges and the smaller the dimensions, the longer it will take to solve them.
But the world has known for some years that it would, at some point, prove impossible to follow Moore's Law. That's why, alongside, More Moore – the pursuit of Moore's Law – there are More than Moore programmes which look to improve product functionality through other approaches. While scaling will hit the wall in the next few years, innovation in the electronics industry certainly won't.
