Moore's Law, initially formulated by Gordon Moore, then Chairman of Intel, first appeared in a 1965 article in the 35th anniversary edition of Electronics, "Cramming more components onto integrated circuits." It asserts that the complexity of minimum-cost semiconductor components has doubled regularly each year since the first prototype microchip was introduced in 1959.
Throughout the 80s and 90s, Moore's Law began to be rephrased by others in terms of the number of transistors that fit on a chip of fixed size, or the computational power per unit cost. This remarkable law has held strong at least up to the writing of this article, in 2005. In addition, a number of Moore's-variants of exponential growth have appeared in the development of LED lights, resolution of brain scanning devices, mass use of inventions, number of genomes sequenced, availability of RAM, size of magnetic data storage, and fastest possible data transmission speed.
What makes the success of Moore's Law all the more fascinating is that Moore only had 6 years experience with microchips as a basis for his assertion, but nonetheless it has held for 40 additional years. The death of the law has been predicted multiple times, but it has kept chugging on. Industry experts expect a hiccup in it around 2015, when conventional photolithographical techniques will reach their final limits.
Photolithography uses light beams to etch features into a chip, meaning that etching smaller features requires smaller wavelengths of light. Photolithography is already edging into the ultraviolet range. Going much further than that is difficult because of the large energies required to produce smaller-frequency waves. Therefore, other alternatives, such as DNA computing, nanocomputing, 3-D chips, or something unprecedented, will have to be used to ensure the exponential growth of computing power.
Economy, in at least one area, has had to give in order to sustain the continued exponential growth of Moore's Law. That area is the initial capital required to make a modern microchip-manufacturing plant, currently on the order of $1.5 - $2 billion (US dollars). Research and development costs to push Moore's Law past photolithography are likely to be on the same order of cost, if not several times larger. But as the demand increases for computing power used for a variety of applications, it is very likely that the necessary supply will be created to meet that demand.