In 1965, Gordon Moore, who was then the director of the Research and Development Laboratories of Fairchild Semiconductor and later became the co-founder of Intel Corporation, wrote a piece in the magazine Electronics about the future developments in the semiconductor industry. He had observed that the number of components on an integrated circuit (IC) has doubled every year since the production of the first planar transistor in 1959 and that this development will continue at least for the next ten years. In the next few decades the number of transistors on an IC increased dramatically, which resulted in a dynamic and rapid development of new product categories.
Moore mentioned three main reasons contributing to this development: first of all, manufacturers were able to produce bigger chips with less defects at remaining costs. This was mainly due to the use of optical projection of lithography masks on the wafer instead of contact printing. The second reason was the ability to improve the rendering of ever-finer line widths and images. The third reason was what Moore called “circuit and device cleverness”, by which he meant that manufacturers could make use of more space of the wafer area, resulting in the charge-coupled device (CCD) by using controlled light beams instead of chemical means. However, manufacturers could no longer rely on “circuit and device cleverness” after a while, which is why Moore speculated that the plot on the number of components on an IC would take a gentler slope after 1975 with the number of transistors being doubled every 18 months.
As can be seen in the graph above, Moore’s speculation turned out to be right or at least became a self-fulfilling prophecy in a sense that it was used as the industry standard to predict the future development in the production of semiconductors. The number of components per area unit increased rapidly over the next decades, which allowed manufacturers to improve the processing power of computers. The reason of this connection is that a variety of factors that improve the processing power require more transistors including parallel execution of instructions, deeper pipelines or bigger cache.
In 1995, Moore reflected on his prediction and compared the development of the IC with the development of random access memory systems and the density of bits per area unit, which has also doubled every one to two years. But not only random access memory systems showed a similar development, but also the area density of HDDs and flash memories.
Graph 2: Own illustration, data was collected by my own from several sources
For several years now experts are arguing that the Moore’s law will become obsolete at some point because of a variety of reasons such as the limits of the lithography technology or the problems of heat caused by the increasing number of transistors. It might be true that the components of an IC cannot be smaller than atoms. However, there are a variety of other technological developments which could further improve the processing power. This includes research on the design of chips that differ from CMOS chips such as 3D chips, specialized chips or chips that are based on new materials and technologies including photonics, spintronics or neuromorphing. Moreover, researchers are already working on prototypes of quantum computers which could further push the boundaries of processing power.
The developments in the semiconductor industry and the improving processing power allowed manufacturers to develop the microprocessor and together with the increasing aerial density of memory systems it paved the way for new product categories and devices such as the minicomputer, the personal computer, phones and other small devices or small computers that are implemented in physical systems.
Graph 1: Data was taken from github