The Future of Moore’s Law
Moore’s Law, first formulated in 1965 by Intel co-founder Gordon Moore, observes that the number of transistors on an integrated circuit doubles every one to two years.
Transistors are devices which allow for the amplification or attenuation of an electrical signal, and thus form the foundation of modern electronics. Transistors can be grouped together to form a unit known as a “logic gate,” the elementary unit of decision-making which form the basis for all computers. As such, we can see how advances in transistor manufacturing, specifically the ability to produce them smaller and faster, allows for drastic increases in overall computing power, as more operations can take place in less time.
As an insight, Moore’s Law has had a profound influence on the development of electronics in the five decades since it was first formulated, holding up remarkably well over time. In many ways, Moore’s Law became a self-fulfilling prophecy: microprocessor manufacturers such as Intel and AMD would set their research & development targets with Moore’s Law in mind, in order to compete with other manufacturers who they expected would also be setting similar targets.
Moore’s Law has not gone unchallenged. Many technologists, including Gordon Moore himself, have at various points predicted the slowing down of transistor shrinkage and the breakdown of Moore’s Law. So far, Moore’s Law has outlasted its naysayers, but evidence shows that the pace of improvement is slowing somewhat. This can be attributed to two factors: one economic, the other technological.
Improvements in transistor size and speed are affected by economics. Moore’s original articulation of his law was not about transistor size in absolute terms, but transistor size combined with economic feasibility. After all, if making transistors twice as small makes them ten times more expensive, then these smaller transistors are not really a viable improvement. It is the shrinking of size combined with a decrease in cost that allows computers to continually improve. We may be reaching the point where transistors are becoming so small that additional size reduction will result in significantly increased manufacturing costs, making additional shrinkage unrealistic for mass production.
In addition to the economic factors, there are technological and physical limitations to transistor size. Currently, transistors have been shrunk to about 14 nanometers in size. Intel predicts that they will be able to manufacture transistors as small at 5 nanometers within ten years, but after that continued shrinkage may not be possible. Chip manufacturers have resorted to other techniques — 3D “stacking” of chips and multi-core processors rather than direct shrinkage of transistors — as a way to continue to realize gains in speed.
Moore’s Law has defined progress in computing for the last fifty years, but is unlikely to stay the paradigm for much longer. Instead, new approaches and novel advances will be necessary to keep computing power advancing.
Comments are closed