Moore's Law

Moore's Law principle states that since the number of transistors on a silicon chip roughly doubles every two years, the performance and capabilities of computers will continue to increase while the price of computers decreases. It is a prediction made by American engineer Gordon Moore in 1965.

The law is a description of the trend in semiconductor production. It is not a natural process but one that is fueled by scientific and technological progress and necessitates constant innovation to continue. The law is based on observing the exponential growth of electronics over time, which revealed no signs of technological stagnation.

• Moore's law was an observation made by Gordon Moore in 1965. Moore's empirical observations serve as its foundation. From observed statistics, he extrapolated the number of transistors on a microchip, doubling annually.
• The majority of this growth in chip density is due to four primary factors: die size, line dimension, technical brilliance, and technological innovation.
• Over the years, the law has seen a decline. The weakening results from the growing complexity involved in creating cutting-edge technologies.

Moore's law was one of the best technological predictions of the last 50 years. Gordon E. Moore predicted that components on an integrated circuit (IC) would increase twice yearly. His postulation became known as Moore's Law and was confirmed true in 1975. The majority of this growth in chip density is due to four primary factors: die size, line dimension, technical brilliance, and technological innovation.

According to Moore's observation, one of the major attractions of integrated electronics is its low cost. This benefit grows as technology progresses; a single semiconductor substrate can produce more complex circuit functions. The cost per component in simple circuits is almost inversely proportional to the number of components. However, the cost per component tends to rise when adding more components, and lower yields make up for the rise in complexity. As a result, there is a minimal cost at any given time in the technological evolution. For example, the prediction for manufacturing cost per component in 1970 was only a tenth of what it was in 1965.

In his own words, Moore describes that at first, it was just an observation, an attempt to predict that this would be a way to make electronics cheaper. However, the industries worked out on a continuous rate of improvement, and various technology nodes come along regularly to keep up with the advancements. Therefore, all the business players recognized that if they don't move that fast, they will fall behind the technology, thus pushing the growth further and faster.

Let's take a look at two examples that can demonstrate Moore's law better:

Example #1

Let’s take the case of Intel Moore’s law. In 1971, Intel introduced the Intel 4004 with a transistor count of 2250. And in 1974, the Intel 8080 processor came with the ability of 6,000 transistors. Two years later, Intel introduced the Intel 8085 processor with 6,500 transistors in 1976. In 1978, the Intel 8086 came with a transistor count of 29,000. Then, the Intel 8051 came in 1980 with 50,000 transistors, followed by the Intel 80186 with 55,000 transistors in 1982.

Finally, in 1985, the Intel 80386 had a 275,000 transistor count. This came to be known as the Intel Moore’s Law. From the above data, it is evident that there have been increments in the transistor counts over the years with a period of two years.

Example #2

The relevance of transistors per silicon area, as opposed to transistors per chip, has changed, which is an alternative explanation for the deviation from Moore's Law. One of Moore's initial three technology drivers for Moore's Law, chip size, illustrates this in the industry trend. By the mid- 1990s, the chip area's growth had slowed to 10% per year (approximately) after growing by about 20% a year throughout the 1970s. The chip area, however, had flattened off by the end of the 1990s'. The 1997 SIA roadmap forecasted a DRAM chip size larger than 11 cm² for 2009.

In the same way as ten years ago, it is uncommon to find a chip larger than 2 cm². While there was anticipation for another shift in the Moore's Law graph, the lack of growing chip size as a driver for complexity increased. Instead, the rate of feature size reduction increased, allowing the scope of Moore's Law to remain on course.

In the second decade of the 2000s, the slowdown in the growth of the chip building arena was evident. Intel's production of 10-nanometer chips is one of the most notable examples. Intel was one of the pioneers in the chip industry and the leading torchbearer of the law. When Moore's law was at a two-year growth rate, The Company released them in 2019, five years after the previous generation of the 14-nanometer feature chip. 

The end of Moore's law and slow growth was not sudden but evident from 1999. The researchers at Intel worried about not being able to make transistors much more capable and smaller than 100 nanometers in size by 2005. There were physical hurdles, such as the quantum effects of electrons wandering off. However, the technological industries evaded them for years by inventing new methods, such as using extreme UV (ultraviolet) radiation. Visible light wavelengths became thicker, and they could carve out only silicon features of only a few tens of nanometers. These came with growing expenses.

The most advanced chip manufacturing facilities are also getting expensive. Many expect a fab's (semiconductor fabrication lab) costs to rise by at least $16 billion by 2022, with an average annual increase of 13% (2020). As a result, there are barely three companies to produce the next generation of chips. This is a decline from eight in 2010 and 25 in 2002.

According to Nvidia's CEO, Jen-Hsun Huang, the end of Moore's Law is nearing. He opined that CPU scaling has greatly increased transistor counts over the past few years. Yet there hasn't been any noticeable increase in speed. In comparison, throughout the same period, GPUs have advanced significantly in speed. Compared to expectations, CPU performance advancements over the past few years have been sluggish. As a result, Intel has devoted more attention to lowering power requirements and enhancing performance in low-power environments.

Moore's law has historically been interpreted as stating that CPU performances will double every 18 to 24 months, but this is untrue. Moore's law predicted the doubling of transistor counts rather than raw performances. Companies have increasingly prioritized power efficiency and component integration than solely concentrating on increasing transistor counts and clock speeds. CPUs don't scale as well as they once did due to the emergence of specialized processors designed to handle artificial intelligence and deep learning tasks. However, few sections of the technology community are still hopeful about the law not dying yet.

This article has been a guide to Moore's Law and its definition. We explain whether Moore's Law is dead, the end of Moore's Law, and its examples. You can also go through our recommended articles on corporate finance –

• Walras’s Law
• Law of Large Numbers
• Okun’s Law