I believe Moore's Law has always been misunderstood, even by Gordon Moore, who coined it.
Moore, who eventually became CEO of Intel (NASDAQ:INTC) and is now retired, with a multi-billion dollar foundation, wrote in a famous 1965 article that the number of transistors on a given area of chip could double every year or two as far out as he could see. He made no claims regarding other technologies, and has insisted since he was only talking about silicon.
But exponential improvements have happened in related areas regardless, partly by using the computing power Moore's Law enabled. Optical fibers now handle exponentially more data because of diodes that can distinguish colors. Hard disks can handle exponentially more data because of advances in math describing how data is inscribed on them. The Von Neumann architecture of the past has been replaced by massively parallel processing.
The basic Moore's Law equation has now allowed the complete replacement of hard disks, and optical drives, on tablets. And IBM's (NYSE:IBM) point, by revealing a new chip architecture based on carbon nanotubes, is to insist it will continue moving ahead, even beyond the silicon age.
IBM's technique is compatible with current silicon-based technology. The idea is to lay down a substrate of hafnium oxide and silicon oxide, then put the tubes into a solution with water and immerse the substrate in it. The tubes then adhere to the substrate in predictable ways.
This comes just a few months after IBM demonstrated how carbon nanotubes could create functioning transistor junctions 10 nanometers apart - less than half the distance of today's best silicon.
This is not the only basic change being contemplated for chip technology. Intel says it can produce trillions of connections per hour on its 22 nm technology - the term for that is a teraflop. Harvard and MIT are working with chips that mimic the functions of a human brain, that are in effect analog. British scientists are working on "quantum chips," working with light, that promise to increase processing many times over.
The fear, as we got circuit lines closer together on slices of silicon, was that Moore's Law would peter out somehow, that we would reach "the end" of it. What these breakthroughs show is that is not happening. We are using the abundance created by IBM to create entirely new chip-making technologies that will make today's computers seem obsolete by the end of the decade.