Why Do We Need Roadmaps? How Moore's Law Has Changed the Semiconductor Industry

Why Do We Need Roadmaps? How Moore's Law Has Changed the Semiconductor Industry

Roadmaps are widely used in business, investment planning, and industrial policy and are often a figure of fun as being considered a useless formality. Nevertheless, according to the research community, they may be a key decision making instrument in the innovation industries. In the early 1990s roadmaps helped the US semiconductor companies agree upon their mutual investment expectations and integrate the empirical regularities known as Moore's Law in their decision-making process.

 

Gordon Moore and His Law

 

Gordon Moore. Photo from the collections of The Henry Ford, Dearborn, Michigan, USA.

Photographer, Michelle Andonian. OnInnovation link.

 

Gordon Moore formulated his law in 1965, when he was working as director for research and development at Fairchild Semiconductor. The integrated circuit was the emerging technology then, its idea was put forward only in 1952, and implemented in the late 1950s. Basing on the data derived from just three observations (data points) Moore predicted that the number of transistors per square inch of an integrated circuit would double every year and would consequently result in the exponential growth of computer performance every two years. In 1975 he reformulated his initial prediction, and the present-day version says that the number of transistors on the integrated circuit chip would double approximately every 24 months. The crucial component of this formula was the assumption that complication of the integrated circuits would not be followed by their cost increase.

 

Moore's Law was formulated in the early days of computer industry and turned out to be built into its very economy and became the landmark for both semiconductor manufacturing companies and companies from the relating industries using integrated circuits. It became the law not of physics, but of business and economy, and that is why it turned out to be disputed as soon as ten years after it had been reformulated.

 

Eastern Competitors

 

Mount Fuji, Japan. Hghask Ekorb / Unsplash

 

As for its subject matter, Moore's Law showed clearly that continuous investment into technology updating are followed by practically unrestricted decrease in value of electronic components. Despite the fact that Moore's forecasts had come true, in the late 1980s most players of the US electronic industry found them too optimistic. Watching the Japanese economic rise, the businessmen and politicians started to fear of losing the competition for the leadership in the semiconductor industry. In 1990 the National Advisory Committee on Semiconductors (NACS) notified the US Congress that the industry was facing serious risks: 80% of the market of memory devices, which were one of the top sectors of the electronic industry then, was captured by the Japanese companies as far back as in 1988. Moreover, the Japanese competitors were ahead of the Americans in the volume of investments — from 1984 through 1989 the Japanese semiconductor manufacturers invested $12 billion more than the US ones. As More's Law extrapolated the US experience, many people had an impression that the exponential growth of the Japanese semiconductor industry would disprove this regularity, and the Land of the Rising Sun would dominate high-tech industry.

 

According to NACS, the main problem of the US was uncoordinated investment decisions. This was especially true for such high-tech industries as electronics and semiconductors because they were dependent from R&D results. The experts from NACS considered that key factor, which had provided the competitive edge to Japan, was insufficient investment in R&Ds on pre-market stage, when innovative products and services are being created and tested. Although coordination of efforts in this regard met interests of all the industries using integrated circuits, the investment strategy of some companies required individual research projects. Besides, exchange of the information related to pre-market technology development was not encouraged, although this information flow could in no way play into the hands of competitors as it was too general. Thus the Japanese companies had two benefits: first, in the US R&D investments were restricted to the budget of each particular company; second, there was no guarantee that any of these companies would reinvent the wheel — i.e. it would duplicate a research project already implemented (and paid for) by someone. Under these circumstances, Moore's Law could quickly become a thing of the past.

 

Saving Moore's Law

 

Intel 8008. Jonas Svidras / Unsplash

 

Ironically, Gordon Moore himself had to save Moore's Law. In November 1992 he presided the conference on the issues of the semiconductor industry future in Irving, Texas, in which more than 200 scientists, engineers and technologists participated representing major US tech businesses, universities, government agencies, and research laboratories. Moore conceived the idea of "shared vision" of the semiconductor industry development for the next 15 years, and the instrument to unite this vision with the investment decision making processes in every single company. In other words, he attempted to remedy Moore Law as the law of business and make them not just the landmark, but the instrument for coordination of the main players of the electronics industry. In order to address this challenge, it was required to combine the abstract Moore's Law and specific decisions of every individual company, which had neither legal nor geographical relations. US National Technology Roadmap for Semiconductors became such an instrument. It comprised target performance and cost indicators for semiconductor components calculated following Moore's Law. The conference participants got these calculations as the target, which enabled to coordinate capital investments at the level of industry on the whole and to avoid direct interference with the confidential budgeting process or choice of specific research programs of every single company.

 

From Past to Future

 

geralt / Pixabay

 

Six versions of this document have been published since 1992. In the late 1990s, it was decided that this coordination practice shall become global so that the major US manufacturers - Intell, H&P and AMD could influence development of semiconductor devices in other countries. The US Semiconductor Industry Association has officially recognized the global semiconductor industry, and in 1998 it teamed up with such associations from Europe, Japan, Taiwan, and South Korea to create a global cooperation platform for the leading semiconductor manufacturers, as well as suppliers of equipment, materials, and software, including research institutes, consortiums and Government laboratories. It was time to make Moore's Law really global.

 

These efforts resulted in a new version of a roadmap, which is being revised annually. Its core element was a table of five columns — "technological nodes" — correlating with particular dates (years), by which a new semiconductor product was expected to enter the market. Each "node" was matched with its parameters of the components production cost and performance estimated on the basis of Moore's Law. It was supposed that every new device would double the number of electronic components every two years with the constant value; thus, aggregated production cost of components would decline by 29% annually. In the event of the economic downfall or delay in the new technologies commercialization, the industry players agreed to maintain the speed of decline in value on the historically medium level. This, in its turn, would enable the industry to maintain the rate of increase in demand for its products, which made 15-17% annually since the early 1970s. In other words, the roadmap has formalized the industry growth indicators in its early development stages and made them the targets — thus it ensured compliance with Moore's Law not only in the past, but in future too.

 

Why Do We Need Roadmaps?

 

A series of roadmaps for semiconductor industry allowed consolidating success of this industry reflected in Moore's Law — making it global and long-lasting. In other words, roadmaps are the instruments to solve the challenges of collective action, which arise in the situations when people or organizations are not willing to incur risks or expenses required for the action, which, in the end, will be profitable for all. The problem is especially burning for the innovative industry because after technological renewal every company may face the loss of its market niche and lost competition — the money invested into R&Ds, the demand for which has not yet established, could be spent for solving more urgent business problems. But these investments are beneficial to the industry on the whole as they result in cost saving and optimization of research efforts preventing from reinventing the wheel. Roadmaps offer a solution to this problem through coordination of the investment decisions made by legally autonomous and geographically distant companies from the relating industries while respecting their autonomy. 

 

Dmitry Zhikharevich