In his highly influential 1965 article, “Cramming more components onto integrated circuits” (Electronics, Volume 38, Number 8, April 19, 1965), Gordon Moore made his now-iconic prediction: For the then coming decade, the industry could expect to see a doubling of the economically optimal number of devices in integrated functions roughly every 18 months, starting with the current number of about 50. He included a plot that illustrated his conclusion:
Though Moore could not have guessed to what extent the semiconductor industry would latch on to this graph — the visual expression of what they came to call Moore's Law — it's attractiveness to a broad audience is unmistakable. Plotted against a logarithmic vertical axis, the sparse data suggests a reasonably straight line — sufficiently straight that the 10-year prediction looks more than rational — it looks like an obvious and inescapable conclusion (despite that it had been, to that moment, neither).
And thus it was. After all, who in the industry — be they in the technical, management, or investment communities — doesn't love a straight line especially one backed by data (sparse or otherwise)?
What I've found endlessly more interesting, however, is the less-straight plot the article included illustrating Moore's insight into the evolution of semiconductor technology:
About this plot, Moore wrote:
For simple circuits, the cost per component is nearly inversely proportional to the number of components, the result of the equivalent piece of semiconductor in the equivalent package containing more components. But, as components are added, decreased yields more than compensate for the increased complexity, tending to raise the cost per component. Thus there is a minimum cost at any given time in the evolution of the technology.
So, at the heart of Moore's insight was an understanding of evolving fabrication techniques and their critical economic implications with respect to growing circuit complexity. Moore was careful not to attach his discussion to a particular semiconductor material or fabrication technology, though he was clear that silicon would dominate as the basic material because of its abundance, low cost, and readily formed oxide. But even in 1965, it was clear to Moore that, for example, GaAs would be an important material for microwave applications.
With regard to linear circuits, Moore correctly observed that the benefits of evolving fabrication technologies would not bring benefits in proportion to those for digital functions. Among the key limiting issues he recognized was the difficulty in fabricating reactive elements of sufficient value and Q. Already apparent, however, was the distinct advantage integrated circuits would have in linear circuits due to the ability to fabricate well matched devices that thermally track one another – traits that remain key to analog integrated devices to this day.