Can Moore's Law Continue, and Does it Matter as Much?

There has been a debate about Moore's Law for some years, with some observers arguing the pace of development cannot continue, while others are hopeful the process can be extended.

Moore’s law says that computer power doubles, for the same cost, about every 18 months. That sort of advance obviously underlies huge advances in the cost of doing complex things at lower prices, over time. Of course, there are many cost elements for products, computing power and storage being relatively minor, in many cases. But if chip costs were the only constraint, a $600 iPhone would cost $18.75 in 2020, $0.59 in 2030 and overall power or cost improving 1,000,000 times by 2050. 

That won't happen because the other cost components (manufacturing, marketing, operations, other physical components and research do not follow Moore's Law. As outside plant engineers always point out, the cost of digging trenches to lay optical fiber never follows Moore's Law, for example. 

But even its inventor has said "Moore's Law is dead." Gordon Moore, its inventor, says "It can't continue forever." The reason is physical. 

"In terms of size [of transistor] you can see that we're approaching the size of atoms which is a fundamental barrier, but it'll be two or three generations before we get that far -- but that's as far out as we've ever been able to see. We have another 10 to 20 years before we reach a fundamental limit. 

But Intel has been able to keep up with the pace predicted by Moore the past forty years. Can it continue for another 40? Some think it might be possible, but fabrication techniques will have to change, perhaps based on different elements than silicon. 

While Moore’s law is often expressed and discussed in terms of power, it has also worked on the cost side of the equation. Every two years we get the same computing power for half the cost. Not only has this seemed to hold for Intel’s chips but other electronics as well.

In 1972, Hewlett-Packard introduced its first pocket calculator, the HP-35, retailing for $495, about $2,600 in today’s dollars. You can buy a scientific calculator today on Amazon for $0.69 plus shipping. Basic calculators cost even less. Extrapolating Moore’s law from 1972, cutting the price by 50 percent ever two years, yields a projected price $0.005 today. Accounting for non-electronic components, assembly and shipping, brings it close enough to make my point.

As another example, consider the Kindle. It was introduced in November 2007 for $399. A better model now retails for $189 -- a drop of more than 50 percent at retail in three years. Of course, there are experience curves at work, that typically do lead to lower retail prices for manufactured goods of the same model, over time. 

A countervailing idea is that our fundamental constraints are not those imposed by chip technologies, but by human ingenuity. PC makers years ago realized that the value of a device was no longer a constraint on value. Rather, applications have assumed that role.