Moore’s Law in the 21st Century

This month we celebrate a remarkable 50th anniversary for the Electronics Industry.  On 19 April 1965, Electronics magazine published Gordon Moore’s 4-page paper which set out the criteria of what became known as Moore’s Law.  The industry never looked back!

Electronics asked Moore to predict what was going to happen in the semiconductor components industry over the next 10 years. He speculated that by 1975, it would be possible to fabricate as many as 65,000 components onto a single silicon chip occupying an approximate area of 40 square mm. His reasoning was a log-linear relationship between device complexity (higher circuit density at reduced cost) and time.

Gordon Moore of Fairchild Semiconductor and later co-founder of Intel:

gordon moore

There has been much written on Moore’s Law but one of the most interesting parts of the IEEE 35th Anniversary publication was the accompanying cartoon in Moore’s paper alongside the predictive technology graphs.  How true this turned out to be:

Cartoon from IEEE

What was the Core Message of Moore’s Paper?

Semiconductors and specifically integrated circuits (ICs) are arguably the most complex things ever mass produced by humans and also one of the most positively disruptive technologies ever developed.  Semiconductors and ICs now touch almost every part of our day to day lives both commercially and domestically.  There is no doubt that the extraordinary pace at which they have evolved was strongly influenced by Moore’s article published 50 years ago.

He noted that the maximum number of components that could be fabricated before the rising probability of failure made it uneconomic to add more, was doubling at a regular pace. He suggested this could be extrapolated to forecast the rate at which more complicated chips could be built at affordable costs.

A Rule of Law or an Observation?

This insight or observation, if you like, became a fundamental driver for the semiconductor industry and, on reflection, was not really a ‘natural’ law but rather an astute, predictive observation which became a target for the industry’s engineers and their managers.  In other words, computing’s pace of change has been driven by human ingenuity rather than any fixed law of physics.

In fact, the ‘Law’ itself has changed over time. While the 1965 paper talked of the number of ‘elements’ on a circuit doubling every year, Moore later revised this a couple of times, ultimately stating that the number of transistors in a chip would double approximately every 24 months. In the beginning, he felt it was just a way of chronicling progress, but it became a benchmark that the various industry participants and players recognised as something they had keep up with or become less technologically competitive.

The End of Moore’s Law?

Intel (the company that Moore went onto co-found after he left Fairchild Semiconductor) spends ever increasing amounts on research as well as in their manufacturing plants and on supporting innovation from their suppliers to stay on target. In 2013, the firm’s ex-chief architect Bob Colwell made headlines when he predicted Moore’s Law would be ‘dead’ by 2022 at the latest. Specifically, he said it would be impossible to justify the costs required to reduce the length of a transistor part, known as its gate, from the values of the time (about 27nm) to less than 5nm (1nm = one billionth of a metre): the space occupied by about 10 silicon atoms.

New ‘Laws’ for the 21st Century

It can be argued that raw processing power is not the only target now in the mobile age.  Things have moved beyond simply trying to offer more processing power at a reasonable price.  A major challenge to our industry now is to add functions to portable and mobile devices while ensuring that their batteries are not over-taxed.  The focus is not just on the processors but a whole set of components that become key enablers: thin displays, camera modules, flex rigid circuits, 2D and 3D packages, embedded SiP on both rigid and flexible substrates, Anylayer HDI and surface mount high density connectors.

The decoupling of reductions in feature size from actual improvements in processing power as circuits required for error-correction, redundancy, read- and write-boosting for failing static RAM, as well as circuits to track and adapt to performance variations and complicated memory hierarchies consume more and more chip area suggest a need for a new “Law” which will signpost the way ahead in semiconductors. The same could be said for portable devices, as user-oriented functional aspects become more important for product performance than sheer processing power.

Samsung mobile phone

The Way Ahead

The important impact of Moore’s Law to Intel and others, has not been the fact that it is a law; it has only become one because Intel and the industry have made it so by managing to adhere to the basic rule of doubling transistor densities every 18 to 24 months for most of the past 50 years. It is in fact an ‘OBSERVATION’ that has sustained, galvanised and led a company’s successful commercial growth.

BPA has over 40 years of experience in analysing and predicting technologies and their impact on the electronics value chain.  Contact us to see which ‘OBSERVATIONS’ can be relevant to your business and how they can enable you to drive growth and profitability through leveraging technology.

In our next mailshot we look at how an observation can be hugely valuable – like Moore’s –but can also be deceptive and can easily mislead the industry into damaging decisions.

Get in Touch

BPA Consulting Ltd

Tel: +44 (0)1306 875500
Skype: bpa-consulting

Company No.: 3709460

Latest News