The Irresistible Compulsions Of Darwinism
Why hasn’t the chip market grown beyond $300 billion in the last three years?
Why don’t new PCs deliver much more performance than old PCs?
Because chips can’t deliver improvement like they used to.
In the old days transistor counts doubled, speed doubled and power consumption halved every couple of years or so.
That meant the products of a new generation of chips had to be bought by equipment makers – there was no choice for the customers – if the equipment makers didn’t buy the semiconductor industry’s latest products, their equipment products were obsolete, uncompetitive.
But today, although transistor counts still keep growing, performance doesn’t go up much, if at all, and power doesn’t decrease much, if at all.
So there is no longer the same compulsion to buy the products of the semiconductor industry.
Gordon Moore used to say his worst nightmare was to wake up one morning and find that the world no longer wanted more computing power. That moment has come. At least it’s come for the PC.
“Intel’s fastest Core i7 desktop processor went from a 2.93GHz quad core in 2009 to 3.5GHz quad core in 2012,” pointed out Linley Gwennap in a recent Microprocessor Report, “performance is increasing at just 10% a year for desktops and 16% for laptops – a far cry from the good old days of 60% annual performance increases.”
That means chip companies can’t rely on Moore’s Law for their competitiveness any more. They have to think smart. That’s what Imec does, what the European R&D programmes do – bio, sensor, MEMS, optical, new materials, new applications.
Which is why semiconductor companies which rely for their success on the old digital competitive differentiators of denser, faster and less power have a suspect future.
And why, when Intel said it was putting $13 billion to building fabs to make denser chips, its shares went down 6% after a previous 30% drop over the last six months.
It will take a lot for Intel to give up its reliance on the competitive advantages of Moore’s Law, just as it will take a lot for Intel to give up pushing x86 as a mobile processor standard but, in the end, we all have to adapt or die.