Click for next page ( 55


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 54
Keynote Address Gordon Moore Chairman Emeritus, Intel Corporation Dr. Moore opened his address by complimenting conference participants on the high level of discussion he had heard that day in terms of both technical sophistication and the ways in which scientific advances could benefit society. In previewing his remarks Dr. Moore said that he would have a different emphasis, focusing on the practical issues facing the research community and the semicon- ductor industry. From this perspective four areas stand out as important: the changing environment for research and innovation, the evolution of the semicon- ductor industry, future challenges for the industry, and the role of international cooperation in meeting future challenges. Noting that his career had focused on the details, where he had tried to build new and strong structures on the founda- tions of others, Dr. Moore hoped that his remarks would shine light on how to build foundations for the future. CHANGES IN THE RESEARCH AND DEVELOPMENT ENVIRONMENT The contributions of large industrial research laboratories, which have been so important in the past, have been diminishing in recent years. Competitive pres- sures and corporate downsizing have prompted the reduction in the size of indus- trial research laboratories. Corporate research and development (R&D) has also become much more short term in the past several years. It is harder than ever for corporations to capture the fruit of an unexpected R&D breakthrough. Thus, they tend to focus on R&D that is closely related to their core businesses and that is "reasonably predictable." Even with the phenomenal contribution that research has made to society in 54

OCR for page 54
GORDON MOORE 55 the past several decades, it would be difficult to make a case that the companies that conducted the research captured its fruits. This makes it difficult for corpo- rate leadership to justify fundamental long-term R&D. It is more sensible to focus on the short term, on research that is directly related to the business. However, the need for fundamental research is as urgent as ever maybe more so than ever before. But because of competitive pressures on corporations, we must look elsewhere for such research. That is where universities and govern- ment-supported laboratories come into the picture. Dr. Moore said that he had a bias toward university research when the issue is framed as a choice between directing funds to universities or to government laboratories. As he explained, with university research, "even if the research fails, you still get the students." Increasingly, however, the output of university research, not just the supply of trained researchers, is important. EVOLUTION OF THE SEMICONDUCTOR INDUSTRY Turning to the semiconductor industry, Dr. Moore noted how the industry had been a direct beneficiary of the industrial research system. The invention of the transistor at Bell Laboratories was a seminal event, signaling the birth of an industry. To convey a sense of the industry's size, Dr. Moore pointed out that the transistor is the highest-volume manufactured product in the world. Each year there are more transistors manufactured than printed characters of any type- newspapers, magazines, books, and copies of documents. There are about as many transistors made each year 10~6 to 10~7 as there are ants on the entire planet. Put still another way, each year the semiconductor industry makes 10 million to 20 million transistors for every person on earth. Declining Unit Costs In addition to the size of the industry, Dr. Moore said that the other distinc- tive feature of the semiconductor industry was declining costs. Several years after the invention of the transistor, a study estimated that the cost of making a single transistor would soon fall to 60 cents. Today, a 64-megabyte dynamic random access memory (DRAM) chip with approximately 65 million transistors costs $8. That is a little less than one-eighth of a microdollar per transistor, or 120 nanobucks. The continuous decrease in costs and the corresponding improvement in performance have driven semiconductor industry revenues to $150 billion an- nually, and the industry in turn supports a $1 trillion electronics industry. Manufacturing Challenges Commenting on manufacturing challenges facing the industry, Dr. Moore remarked that the semiconductor industry stands out among high-technology in

OCR for page 54
56 KEYNOTE ADDRESS dustries for its complexity and fast-changing technologies. Some semiconductor product lines, such as DRAMs and other memory chips, have become commodi- ties. Others, such as higher-end microprocessors, have a great deal of intellectual property embedded in them and command substantial margins in the market- place. Semiconductor technology is also very flexible in that there are many market niches and specialty products. The technology enables custom products to be made for specific applications or firms. As a result of the technology's flexibility, there are hundreds of semiconductor companies and thousands of firms that sup- port them. Rapid Technological Change Rapid technological change is another defining characteristic of the semi- conductor industry, Dr. Moore continued. As an example of fast product obsoles- cence, 80 percent of Intel's revenue in 1997 came from products that had not been introduced as of January 1 of that year. New products mean new plants. The plants themselves today cost multiple billions of dollars, and the equipment within must be replaced every few years. After several generations of equipment, semi- conductor plants must be scrapped and new ones built. The only constants are extremely rapid change and falling prices; on average, prices for semiconductor devices fall 20 to 30 percent per year, with the price of some devices falling as much as 50 percent in a single year. There is no other industry like the semiconductor industry, and it leaves much room for blunder by corporate leadership. Dr. Moore noted that, for this reason, industry leadership has changed many times in the years of his involvement in semiconductors. He recalled that he helped found Fairchild Semiconductor in 1957, which was an industry leader into the mid-1960s. Today, even though the Fairchild name has been revived, the company no longer exists as a semiconduc- tor maker. The industry's dynamism creates enormous opportunity for start-up compa- nies, Dr. Moore continued. The strength of U.S. high-technology industry rela- tive to that of other nations has been its ability to generate start-up companies to exploit technological opportunities. From his own experience, Dr. Moore said that it was much easier to move swiftly and efficiently in developing new things in a start-up environment. A small focused group of committed entrepreneurs can accomplish tremendous things. As a company grows, it inevitably develops a bureaucracy that makes it difficult to develop new products or exploit new tech- nologies. Dr. Moore also emphasized that start-ups rarely develop new technolo- gies but typically find new ways to exploit them in the market. Silicon Valley contains numerous examples of this.

OCR for page 54
GORDON MOORE 57 Smaller Is Better With the declining price of the transistor, Dr. Moore said that the industry has been operating under a variant of Murphy's law: "As things get smaller, ev- erything gets better simultaneously." As the industry packs more circuits on a chip, performance improves as circuit connections grow shorter, thus increasing operating speed. Moreover, with more of the system residing on a single chip, the entire system becomes much more reliable. All this occurs, Dr. Moore reiterated, in an environment of dramatically falling costs. Silicon Real Estate Recasting the issue somewhat, Dr. Moore said that what the semiconductor producer sells is real estate the area of the 200-mm (by today's standard) silicon wafer. On average, that silicon wafer represents $1 billion per acre, maybe a half a billion for DRAMs and several billion for microprocessors. The main challenge for the industry is really a real estate development problem: how to pack more circuits on the wafer with greater functionality in such a way that, while main- taining historical rates of price decreases, the revenue stream from the wafer re- mains at $1 billion per acre. FUTURE CHALLENGES FOR THE SEMICONDUCTOR INDUSTRY Dr. Moore began his discussion of future challenges with some historical observations on the industry's marketplace evolution. In the past the semiconduc- tor industry grew by "assimilating the value added of its customers." By that Dr. Moore meant that innovations such as the integrated circuit allowed, indeed re- quired, semiconductor companies to take on the design tasks that its customers- such as computer systems makers had done. This was a tough sell, essentially telling a customer that may have had a division specializing in circuit design that its division was no longer necessary. This led to Bob Noyce's second great con- tribution to the semiconductor industry, namely, the idea that semiconductor firms should sell integrated circuits to system houses for less than it would cost them to assemble the circuits themselves. Noyce's idea led to the growth of high-volume manufacturing in the semiconductor industry and to the cycle of declining costs. This development fundamentally changed the industry's interface with its customers. Semiconductor firms began engaging in circuit design, then moved into logic, and today new computer designs are essentially done by the semicon- ductor industry. Approaching Limits As for the future, Dr. Moore said that it was his belief that the current tech

OCR for page 54
58 KEYNOTE ADDRESS nology for semiconductor manufacturing, photolithography, would be able to pro- vide advances in performance and declines in cost for another three or four gen- erations. At that point the atomic nature of matter would begin to be an impedi- ment to further advances from photolithography. The wavelength of light used in photolithography would become larger than the desired circuit features to be etched onto the silicon wafer. The Roadmap Before we approach those limits, Dr. Moore stated that a great deal of re- search is necessary to get the most out of photolithography. The main guide for this research agenda is the National Technology Roadmap for Semiconductors, developed by the Semiconductor Industry Association and managed by SEMATECH. The roadmap, by extrapolating technology needs into the future based on past performance, lays the track so that the semiconductor industry's technology locomotive can keep moving forward. The roadmap has proved to be a powerful concept in the semiconductor industry, and Dr. Moore said that it is a tool that other industries could emulate. Its main virtue is defining the research agenda for the mainstream of an industry and charting a way to meet goals. Beyond Photolithography Moving beyond photolithography will present new challenges. Dr. Moore described three techniques to replace photolithography, which have varying de- grees of promise: . . X-ray lithography the use of x-ray shadow masks has been tried; how- ever, there are still substantial problems with this approach. Electron optics this area has shown some promising early results, but there is still a long way to go from the laboratory to production. Extreme ultraviolet (EUV) using normal optical techniques in conjunc- tion with multiple reflectors holds promise, yet EUV may not be able to use light that is smaller than 13 rim in wavelength. Two further challenges affecting the industry's future are design in a few years, a chip will contain 1 billion transistors, and it will be a huge challenge to improve design techniques so that ever-more complex designs can function reli- ably, and change in wafer size Dr. Moore noted that the conference would hear more about the change to the larger (300-mm) wafer size when Bill Spencer, in his presentation, described the I300I initiative of SEMATECH. Dr. Moore explained that, in the past, equipment and plant costs were low enough that a single company could change wafer size on its own. But today equipment costs are so high that, as the industry shifts from 200-mm wafers to

OCR for page 54
GORDON MOORE 59 300-mm wafers within a few years, all must change at once. The equipment manu- facturers simply cannot afford to simultaneously manufacture two generations of equipment. ROLE OF INTERNATIONAL COLLABORATION With respect to the role of international collaboration in the research chal- lenges facing the semiconductor industry, Dr. Moore remarked that he is unsure whether government-to-government collaboration is the correct approach. How- ever, he said that international collaboration among firms and universities holds great potential. Because the research challenges in making devices with circuits with ever-smaller line widths are enormous, international collaboration might be useful, as it is in the conversion to the 300-mm wafer size. The change to 300-mm wafers is, of course, already a subject of international collaboration in the I300I project. Some radically new approaches to semiconductor and computer technol- ogy, such as quantum dots, quantum computing, and DNA computing, are pos- sible candidates for international collaboration, particularly among universities. In conclusion Dr. Moore said that the semiconductor industry has had a phe- nomenal run in exploiting new technologies. The industry has benefited from the support of the U.S. government and the industry's own initiatives, such as the Semiconductor Research Corporation, which develops new technologies at uni- versities, and SEMATECH, which focuses on manufacturing technology. How- ever, large research challenges must be met if the industry is to continue its ad- vance. Collaboration is clearly one means of meeting these challenges. Dr. Moore closed by saying that he was gratified to see an entire conference devoted to scientific and technological collaboration. He expressed the hope that expanded transatlantic collaboration would result in a breakthrough with as big an impact on society as the semiconductor.