Dr. Rodgers' Keynote to the Graduating Class of the Dartmouth College Thayer School of Engineering: An Entrepreneur's View of American Competitiveness, June 10, 1989 | Cypress Semiconductor
Dr. Rodgers' Keynote to the Graduating Class of the Dartmouth College Thayer School of Engineering: An Entrepreneur's View of American Competitiveness, June 10, 1989
Dr. Rodgers' Keynote to the Graduating Class of the Dartmouth College Thayer School of Engineering: An Entrepreneur's View of American Competitiveness, June 10, 1989
AN ENTREPRENEUR'S VIEW OF AMERICAN COMPETITIVENESS
An Address to the Graduating Class at the Investiture of the Dartmouth College
Thayer School of Engineering
June 10, 1989
By Thurman John Rodgers, President and CEO
Cypress Semiconductor Corporation
INTRODUCTION by Dean Charles Hutchinson
I'm particularly pleased to introduce the speaker who is affectionately known as "T.J." by many people. He is not a Thayer graduate--but he should have been. And I think he exemplifies what we try to do at Thayer School. He didn't do all that bad, though. T.J. Rodgers was born in Wisconsin where he grew up and went to high school. Then he came to Dartmouth College. He was here on a Sloane Scholarship. He earned an A.B. with double majors in Physics and Chemistry, graduated first in each one of those majors and second overall in the class. Following Dartmouth, he went to Stanford on a Hearst Fellowship where he earned his MSEE and PhDEE. While there, he invented and received multiple patents for the development of new semiconductor process technologies. In 1983, he formed Cypress Semiconductor and helped bring that company to its full potential. Cypress brought the first product to market in only eight months and Cypress now manufactures 114 products, including some of the world's fastest semiconductors. Cypress has sales of $200 million per year and 1,300 employees worldwide. It is considered to be a model company and T.J., a model CEO.
But that is not all there is to Dr. Rodgers. He has other interests. These include cooking--especially Italian, French and Chinese cuisine, which I'm sure he learned while he was a Dartmouth student--and collecting fine wines, especially French wines which he did not learn while he was at Dartmouth. He is an avid jogger, logging six miles a day. I did not ask him if he did that yet today, but I'm sure you will appreciate the fact that his body clock is off a little since he just flew in from the West Coast last night. It is with great pleasure that I turn over the podium to Dr. T.J. Rodgers, president and CEO of Cypress Semiconductor.
It may seem a bit strange that I would be invited to deliver the commencement address at the Thayer School of Engineering, since I took only one course at Thayer while I was here at Dartmouth. On the other hand, perhaps my invitation is appropriate since that single electrical engineering course changed my entire life and led me to the Silicon Valley. Today, I am going to talk about entrepreneurialism, American competitiveness and our current economic battles against Japan. I will focus on a recent attack by a segment of the intellectual community which blames--believe it or not--entrepreneurs and Silicon Valley as being root causes of our American competitiveness problem. But before I address this topic, please bear with me for a few minutes while I reminisce about the path I took from Dartmouth to Silicon Valley. That story, I hope, will give you some perspective on the Valley and explain why I feel so strongly about the anti-entrepreneur arguments.
When I was graduated from Dartmouth, I applied to many top schools in physics. I was accepted at each of them, including Princeton and Stanford. Since I did not yet know what I wanted to do, I also applied to a group of business schools. The only school that did not accept me was the Harvard Business School--which was fortunate because had I gone there, I would probably not be qualified to do the job I am doing today.
In making my graduate school decision, like many of you, I sat back and took a deep philosophical look at what I wanted to do with my life after graduation. After 21 years in Wisconsin and New Hampshire, I made a firm decision to go to school where there was no snow. With that lofty reasoning, I decided to enroll in applied physics at Stanford. Had I not taken that one integrated electronics course in my last term at the Thayer school, I would probably be working in some new area of physics like cold fusion. But, that course in digital electronics was the most interesting thing I had ever done. Why would I want to continue in physics when I could be paid for my hobby?
After that course, I changed my career objective. I called Stanford admissions and asked to change over to electrical engineering. They told me that I could not jump from discipline to discipline so easily. They also told me that my fellowship would probably be cancelled if I changed disciplines. With that firm guarantee, I took everything that I owned, shoved it into the smallest U-Haul trailer rentable, and drove three thousand miles west.
(Years later when I was talking with Cypress's founders about my experience driving over the Sierra Nevada and on to Silicon Valley, all but one related his similar "U-Haul pioneer" experiences. The exception, appropriately enough, was our VP of R&D, who was born in Hollywood, California.)
When I got to Stanford, I went directly to the Integrated Circuits laboratory to get a job to pay for tuition. The IC lab said that until I had my masters degree and had passed the PhD qualifying exam, I should forget about getting in.
Fortunately for me, the director of the IC lab, Professor Jim Meindl, had a particular problem. He had to find someone to run his new epitaxial reactor, a machine used to grow silicon crystals in an ambient of pure hydrogen in a quartz reaction chamber at red-hot temperatures. The electrical engineers at Stanford were not quite ready to work on that problem, but with my chemistry background, I was a natural for the job. Professor Meindl changed his mind and decided that I could run the epitaxial reactor to pay for my tuition. Unfortunately, his first check did not come in soon enough and I had to sell my coin collection to pay for groceries.
STANFORD AND THE SILICON VALLEY
Once at Stanford, I began to learn about Silicon Valley and how it operated. My rented house in Palo Alto was only blocks away from the famous garage on Addison Street where Hewlett and Packard started in 1939. It was there that they built their first product, audio oscillators, to sell to Walt Disney for his new movie, Dumbo. I figured that Hewlett and Packard had done pretty well since their names adorned buildings on the Stanford campus. My office was near that of Dr. Frederick Terman who invented the Silicon Valley/Stanford Industrial Park concept by moving Hewlett and Packard across the street onto leased Stanford land. I took courses from Professor William Shockley, one of the Nobel-prize-winning inventors of the transistor. He was an extraordinary teacher; every bit as good as the best I had at Dartmouth, Professors Stockmeyer and Braun in chemistry and Professor Doyle in physics. I became friends with Professor Shockley and learned about Silicon Valley over dinners at his house.
Shockley was one of Silicon Valley's pioneers. In 1955, he started Shockley Transistor, a company chartered to bring the newly invented transistor to the market. He had working for him two bright young men, Gordon Moore and Bob Noyce. Noyce was from Iowa, a midwest kid like me. Noyce and Moore left Shockley Transistor in 1957 to form Fairchild Semiconductor. Shockley Transistor went downhill from there and closed in 1968. After its start-up, Fairchild got one more major addition, one Andrew Grove, a Hungarian immigrant who fled in 1956.
Fairchild, of course, became the proverbial mother of Silicon Valley. Virtually every semiconductor company in Silicon Valley spun out of Fairchild. Every year, we in Silicon Valley celebrate our genealogy at a banquet where we pass out charts which delineate the history of all of the companies in Silicon Valley. For example, Cypress spun out of Advanced Micro Devices, which, in turn, spun out of Fairchild, which, in turn, spun out of Shockley Transistor. We actually have a very short lineage compared to some of the companies in the Valley.
Fairchild is now out of business. Ironically it was swallowed by one of its own offspring, National Semiconductor.
In the late 1960s, there was a tremendous revolution which formed what I call the mega-companies of Silicon Valley--Intel, AMD and National Semiconductor--which are all now bigger than one billion dollars a year in sales. In 1967, Charlie Sporck, Pierre Lamond and Fred Bialek left Fairchild to form National Semiconductor. Sporck still runs it; Pierre Lamond has become a venture capitalist whose firm has funded companies which include Apple, Tandem and Cypress. He is currently a director of Cypress. In 1969, Bob Noyce, Gordon Moore and Andy Grove left Fairchild to form Intel. They are currently at Intel, except for Noyce who runs Sematech--more on that later. Also in 1969, Jerry Sanders, the sales and marketing guru at Fairchild, left to form Advanced Micro Devices. More about all of these companies later. In Texas, a parallel event happened. In 1969, a new venture called Mostek spun out of Texas Instruments. The president and founder, L.J. Sevin, who hails from the rural areas of Louisiana, built Mostek to $350 million in sales before becoming a venture capitalist with Compaq, Lotus and Cypress to his credit. He is now the Chairman of Cypress. If you are reading a little bit of incest into all of this, you are absolutely accurate. I have just described in gross terms the state of Silicon Valley when my U-Haul pulled into Palo Alto in 1970.
Like most other Stanford IC lab students, I immediately began trying to invent something that would warrant starting a company. That is what you did in Silicon Valley in 1970. I invented a transistor called VMOS which used hydrazine--rocket fuel--to etch silicon preferentially along certain crystal planes to create little V-shaped notches in the silicon. I remember very clearly the term "detonation" appearing on the safety label of each bottle. The transistors formed in those V-grooves were very dense and fast relative to the competing technologies of the day. Fairchild, Intel and a company called American Microsystems, known as AMI, were all interested in VMOS and they all offered me royalty agreements in 1974 to develop the technology at their companies. I decided not to go to Intel because in a personal interview with President Andy Grove, he struck me as a very tough taskmaster who would not have given me very much rope--and he probably would have been right. Five of Cypress's six founders failed with me at AMI on the VMOS technology. Although we did get the technology to work and did sell about $10 million worth of product, it ended as a colossal financial failure. In 1979, I departed AMI. I ask you, right after a major failure like that, what would you think I tried to do? Of course, start a company.
I talked to Hambrecht and Quist and said, "Here I am, available."
They said, "No thanks. You want $31 million, you are 31 years old, and you have just failed. Go work a little while and show us you can do something to make money."
I went back into the job market and got offers from Fairchild, Intel and AMD. I chose to go to work for Jerry Sanders at AMD because his company appeared to be the most entrepreneurial, the place where I could have the most freedom and learn the most. While at AMD between 1979 and 1981, I lost all hope of starting a company because the successful Japanese attack on the flagship semiconductor product, the Dynamic Random Access Memory or DRAM, started the demise of Intel, AMD, National and Mostek in that business. The semiconductor business went through a bad recession, and I reduced my own near-term career goal to moving up through AMD's hierarchy.
THE PHONE CALL
Then my life changed again. I got a chance phone call from a New York venture capitalist who was reference checking a person applying for $2 million in start-up financing. I said, "Give the money to me; I could do a lot more with it!"
He replied that he did not know I was in the market. So I flew to New York to take my shot. That venture capitalist would not risk enough money to fund a semiconductor company, but he introduced me to Ben Rosen, a venture capitalist who is currently the Chairman of Compaq Computer. Ben liked my ideas and sent me to Dallas to talk to his partner. That was L.J. Sevin, our Chairman. That is how Cypress got started.
Dean Hutchinson has already given you a synopsis of Cypress's current status, but let me add a few more facts. Cypress received four rounds of private funding totalling $40 million. We went public three years ago and raised $110 million more in two offerings. We were listed on the New York Stock Exchange last year. Of the $150 million we raised, we still have almost $100 million in the bank. We have built a company whose market value is over $500 million. We have a philosophy of employing Americans wherever possible. We now have 1,300 American employees. Most "American" semiconductor companies employ a significant portion of their people in the Far East to take advantage of sub-fifty-cent-per-hour labor rates. We stated initially that we would not hire overseas, that we would be an American company aimed at producing state-of-the-art technologies and products. We decided that when we could not afford an American to do a job, we would automate and have a robot do it. That job probably would be a lousy one, anyway. Our formula has led to consistent 20% profitability, as well as to growth and profitability performance better than that of any Japanese semiconductor company. The point of telling you all of this is that I think what Cypress has accomplished--and not just us, but many other start-up companies--is good for the country. We create American jobs, export products to reduce the balance of trade deficit, and create tax revenue to reduce the budget deficit.
ATTACK ON ENTREPRENEURS
Currently, there is an attack on entrepreneurialism which claims that entrepreneurs in Silicon Valley companies are an important reason for the Japanese incursion into our market. That inaccurate and dangerous attack has as its voice, its "Tokyo Rose," MIT researcher Charles H. Ferguson whose position has been synopsized in the Harvard Business Review: "nimble, small companies cannot survive against stable, concentrated and protected industrial alliances [namely Japan Incorporated]." When this young researcher speaks authoritatively about what is wrong in Silicon Valley, my first impulse is to dismiss him. He has never built a company or managed a significant entity. He is only 32 years old. And in my heart, I would like to believe that he did his undergraduate work at Harvard under Dr. Timothy Leary. Unfortunately, he derives significant credibility because of the public statements of some of the mega-company founders. Gordon Moore, the Chairman of Intel, has renamed venture capitalists "vulture capitalists" because they circle overhead and pick off his good people, thereby weakening Intel in its ability to compete with the Japanese. Andy Grove has said that he does not know who is a bigger enemy, the Japanese or Wall Street. Just after I left Advanced Micro Devices, in one memorable 2:00 a.m. meeting in the parking lot outside of a Palo Alto restaurant, Jerry Sanders pointed at L.J. Sevin, our Chairman, and said, "You sold out!"
These anti-entrepreneurial arguments could be dismissed as having the false logic of the classic California environmentalist--the guy who, after he gets his house built on the ridge, wants a building moratorium declared in the remaining area. After all, if the mega-company founders could build their companies with the energies squandered by Fairchild, why should the new entrepreneurs not do the same? Unfortunately, people are hypersensitized because the Japanese have recently done so much damage to our industry. As a result, the new attack on entrepreneurs is not as easily dismissed. Therefore, since some mega-company leaders support Mr. Ferguson's argument, we are forced to look more deeply to the root causes of our competitiveness problem.
The anti-entrepreneurial argument starts with a flurry of facts. You have heard them: we have too many lawyers and not enough engineers. The Japanese graduate 4.4% of their students with BSEEs, while we graduate only 1.9%. And then there is the balance of trade problem. For example, in the period from 1972 to 1987, we went from a $5 billion yearly deficit in automobile trade to a staggering $60 billion yearly deficit. During the same period, in consumer electronics a $4 billion deficit grew to a $12 billion deficit. In steel, a $2 billion deficit grew to an $8 billion deficit, and in semiconductors, a $2 billion surplus changed into a $17 billion deficit. These data were recently published in the Scientific American by Lester Thurow's group at MIT. Furthermore, the argument continues, the semiconductor industry has been attacked successfully. In 1978, the world's two largest semiconductor companies were Texas Instruments and Motorola and six of the top ten semiconductor companies were American. A decade later, in 1988, the top two semiconductor companies had become Nippon Electronic Corporation and Toshiba, and six of the ten top semiconductor companies had become Japanese. So there is a real problem and we have to face up to it.
Ferguson's solution, as outlined recently in the Harvard Business Review, proposes we in essence mimic Japan. The Ministry of International Trade and Industry (MITI) in Japan has subsidized the largest Japanese companies and organized them into a powerful cartel. You might call them Sumo companies; they are gigantic, and virtually impossible to defeat in head-to-head competition. Ferguson's answer is have us do the same thing. Let us have Sematech, a government subsidized consortium selectively directed to help only large American semiconductor companies. Let us boil down the diversity of companies in the Silicon Valley to the equivalent of Ford, General Motors and Chrysler, and let them meet the Japanese head-on in the Sumo ring. Worst of all, Ferguson proposes changing U.S. tax laws in such a manner as to dry up the flow of venture funds into Silicon Valley. I view Ferguson's proposals as losing propositions with potentially catastrophic results.
We do have problems. The patient is ill, but "Dr." Ferguson's prescription for that patient is penicillin, and I am here today to alert you to the fact that the patient is allergic to penicillin.
THE SKY IS NOT FALLING
First, let me provide you with some additional facts relating to the doomsday scenario of American competitiveness I just recounted. We do have too many lawyers, as a percentage of new grads. On the other hand, if you look at the total headcount of U.S. engineers and scientists as tallied by the National Science Foundation, we have a strength of 3.6 million versus 1.5 million in Japan. So we have a two-to-one advantage in absolute engineering headcount, although our ratios are not as good as theirs. The large semiconductor companies supporting Ferguson should avoid the topic of too many lawyers. Cypress has been in business six years and we have been sued or threatened with suit over 20 times. Today, large semiconductor companies maintain several active or threatened lawsuits against us, even though we have never lost a suit, nor have we ever paid a penny out to any litigant. Their lawsuits were and are meritless, intended only to put stumbling blocks in our path--a standard tactic used by large semiconductor companies to stifle competition. If we have too many lawyers, perhaps the large companies that use them so frequently should make less use of them, thereby creating unemployment in the legal community and an eventual shift of our national talents towards engineering.
Relative to the nay-sayers' doomsday prediction, I have some questions for you. What country owns more U.S. assets than any other country? Who is more productive, the average Japanese or the average American? And from which country do we import more than any other country? If you answered Japan three times, you were wrong three times. Japan does not own the most U.S. assets. Japan has just passed the Dutch to move into second place behind Britain. Canada exports more to the United States than does Japan. And regarding productivity, we keep getting an inaccurate message because we read quotes like this one, again from MIT:
"Growth in [American] productivity, a crucial indicator of industrial performance, has averaged only slightly more than 1 percent per year since the early 1970s. Productivity has grown more rapidly in several Western European and Asian nations."
All right, slow productivity growth is a bona fide problem. On the other hand, the actual absolute value of U.S. productivity is given in a recent study from the British Treasury:
"It is always thought that Japan was well ahead of Britain on all measures of productivity. What emerges is that overall productivity levels in Britain are close to those in West Germany and only the United States is still comfortably ahead of all other countries. For every hour of work, the Americans produce 30% more than Britains and more than twice as much as the Japanese."
So the sky is not falling. We are ahead, but the gap is closing--and we do have to worry about addressing that problem, but in the right way.
Even the automobile industry has shown promise in the last three years. In that period, Ford has increased its North American market share from 25.8% to 30.2% and Chrysler from 16.4% to 16.7%. The car companies, except General Motors, are now on the mend and starting to hold the line against the Japanese. The semiconductor industry appears to have stabilized. The U.S. share of the integrated circuit market has dropped from 49% in 1984 to 34% in 1988, but appears to be holding. (The Japanese took 39% of the integrated circuit market in 1988.) The patient is surely ill, but not dying--unless we allow "Dr." Ferguson to give that fatal injection of penicillin.
MONGRELS VERSUS SUMOS
Now that I have added some balance to the doomsday predictions, let me discuss America's historical semiconductor problems in more detail. We Americans invented the Dynamic Random Access Memory. The dynamic RAM, as it is called, is the basic memory in almost every computer. It is the highest volume integrated circuit manufactured. When dynamic RAMs became a high volume product, the Japanese took them away from us. We invented the static RAM. It is the high speed memory device used in all high performance computers. When it became a high volume product, the Japanese took it away from us. We also invented and lost the EPROM (Electrically Programmable Read Only Memory) market in the same way. The EPROM is a device which holds data even when the power is turned off. The point: we are consistently innovation winners against the Japanese and consistently manufacturing losers. Why is that?
If you look at the Japanese society, it is homogeneous, tightly structured and highly disciplined. They save much of their earnings which, in turn, serve as low-cost capital for their industries. Their cultural traits naturally produce a society which is good at manufacturing. On the other hand, that very rigid society greatly constricts individual creativity, a cultural trait we have in abundance. We invent--they take over in manufacturing.
One of the most condescending and contemptuous statements I have ever read was made by former Premier Nakasone of Japan. I was amazed that he survived in office after having made it. The Premier named two American minorities and then stated that the reason the Americans were not doing as well in manufacturing as the Japanese was that with these minorities, we were "mongrelizing" our race. We do have a manufacturing problem, but our "mongrels" have consistently beat the Japanese at defining, inventing and bringing to the market new products and technologies.
I believe a good part of the answer to our manufacturing problem is well stated in The Third Century, written by Joel Kotkin, a prominent technology writer. Joel and his co-author, Yoriko Kishimoto, believe that the Japanese incursion into our market will slow down and that we will reach a competitive stalemate with them. Their rationale for the impending stalemate is that although the Japanese are currently better than we are at manufacturing, we are catching up. Manufacturing skills are learned, not innate. We can learn them, too. As a matter of fact, the most prestigious award given to the company with the best manufacturing quality in Japan each year is called the Deming Award, named after W. Edwards Deming, the American guru of quality control theory. He was sent to Japan after World War II along with other quality control experts to teach quality manufacturing to the Japanese. Unfortunately, the Japanese learned from Dr. Deming a lot better than we did.
The Japanese-American quality gap is starting to close. There is a noticeable improvement in the quality of American automobiles which should cause that $60 billion piece of the trade deficit to start shrinking. Ten years ago in the semiconductor industry, we lived with yields of 25%. That figure means 25% of the chips on a silicon wafer were good, while the other 75% were being thrown away due to defects caused by dust and other contamination. At that same time, the Japanese were producing wafers with 75% yield. The Japanese have now improved to the 90% yield mark on their wafers, but we have moved to the 75% range. What used to be a 200% Japanese manufacturing yield advantage has shrunk to a 20% advantage. The gap is closing because we are learning quality manufacturing skills.
On the other hand, it is difficult to learn creativity. The homogeneous Japanese society cannot learn how to be a diverse society. They cannot learn quickly how to change their economic system to allow for the existence of venture capital or to tolerate employees who are "disloyal" enough to quit and form competing ventures. The advantages we have institutionalized in American Society are almost impossible to synthesize elsewhere. The old order was that we invented it, brought it to market and then the Japanese took it away from us. The new order will be that we will invent it and then we will hold it. If we follow "Dr." Ferguson's prescription to emulate the Sumo company concept, we will do little for our manufacturing capabilities and we will strike a severe blow to Silicon Valley. After all, it was our version of the Sumo companies--the mega-companies--that lost the first battles to the Japanese. Let me go back and look at those lost chip markets, but with a level of detail which the opponents of entrepreneurialism tend to gloss over. Because in that detail lies the central truth.
WHO WILL SAVE THE VALLEY?
The dynamic RAM. Intel did invent it, but then they lost the market lead to Texas Instruments, which, in turn, lost its lead to Mostek, which, in turn, lost its lead to Toshiba and other Japanese suppliers. Intel started to lose its dynamic RAM market long before the Japanese took over. Eventually, American companies lost the dynamic RAM market because the Japanese manufactured better. And also because they cut their prices, they had better quality, and they broke the law--they dumped parts below their cost. But before our government intervened, the U.S. mega-companies had been pushed out of the dynamic RAM business. However, if you look today, the largest supplier of American manufactured dynamic RAMs is the Boise, Idaho, start-up company, Micron Technology. If it weren't for Micron, the American mega-companies would have caused the U.S. to be forced out of the dynamic RAM business.
The high performance static RAM. Intel again. They first brought them to market in 1972 and were dominant from 1972 to 1980. I remember watching Intel's performance while wishing that my group at AMI could do as well. Today, Intel is out of the static RAM business. Their literature specifies the static RAMs of several other companies, including Cypress, as companions for their 80386 microprocessor. Did the "vulture" capitalists take their key people? Not at all--their static RAM technologist, his manager and his vice president are still employed by Intel to this day. What really happened was that Intel made a mistake. In the late 1970s, I and other Silicon Valley technologists, along with the Japanese, all recognized the emergence of CMOS (Complimentary Metal Oxide Silicon) technology. CMOS produces the chips that are in watches and laptop computers--low-power chips for use with batteries. We realized that CMOS capability would determine the winners and losers in the static RAM race. We saw that CMOS could not only be low in power consumption, but very fast. In our business, high speed and low power are an unbeatable combination.
On the other hand, Intel subscribed to the IBM philosophy published in 1974 that NMOS--an older technology which is as fast as CMOS, but much higher in power consumption--would continue to be dominant. In a very disillusioning speech I attended, Bob Noyce of Intel declared assuredly that NMOS would be the dominant technology for the rest of the decade. He declared that our industry would need only to pump money into NMOS to make its transistors smaller. In our business, smaller is faster, and smaller means that more transistors can be jammed onto a chip. He showed the audience of engineers a photoresist line made by a scanning electron beam that was only 0.3 microns wide--less than the dimension of the wavelength of visible light. He said, "There is the line, now all we have to do is make a transistor that small." I remember another quote, "All that is left is the ‘dog work' to get there." I was not ready to go to Intel to do "dog work." As a matter of fact, I had some personal and very much more negative plans for Intel's static RAMs. I started Cypress and--along with other competing static RAM start-ups, such as our arch-rival Integrated Device Technology--attacked the static RAM market with CMOS. Once again, if the U.S. had opted to turn its diverse system of start-ups into a cartel of Sumo companies, like Intel, we would have been virtually shut out of the static RAM market by the Japanese CMOS attack.
The EEPROM (Electrically Erasable Programmable Read Only Memory) is a cousin of the EPROM discussed earlier. EEPROMs are the chips that remember the channel you programmed on your radio, or in a bigger game, remember the terrain that a cruise missile must follow to reach its target. The leaders: Seeq and Xicor. You may never have heard of them. They are both Intel spin-outs. Xicor is an interesting story. They took a technology that Intel had developed and publicly declared to be unreliable, and developed it into a $100 million business. The name Xicor even means eX-Intel-CORporation. A year ago, Intel licensed back its own technology from Xicor. Apparently, the technology was not as unreliable as Intel originally claimed.
I want to pause here and say that I do not mean to denigrate Intel in any way. In his new book, Microcosm--Into The Quantum Era of Economics and Technology, George Gilder called Intel "the most important company in the world," a statement with which I essentially agree. I feel free saying a few negative things about Intel because they have been so consistently excellent. The point I am trying to make with these examples is that if we allow our diverse entrepreneurial base to be compressed into a few companies, and trust the national technology to them, even companies as excellent as Intel are quite capable of making major mistakes. In a Sumo economy, those errors would not be compensated by small companies with different ideas and the audacity to challenge the logic of the seers. In America, the diversity of our companies and our ideas is a national strength we must not abandon.
Intel invented the microprocessor, too. They currently have a dominant position in that field. A year ago, we also introduced our first microprocessor, a RISC machine with 29-MIPS (Million Instructions Per Second) performance. That figure means our RISC processor performs 29 million operations per second. Intel's best processor at that time performed at 4 MIPS. We outdistanced their flagship product, the 80386, by a factor of 15. Intel states publicly, all the time, that it takes $100 to $300 million to develop and introduce a microprocessor. How could Cypress, a company devoting only a fraction of its $35-million-per-year R&D budget to microprocessors possibly make a machine faster than theirs?
The answer lies in a Silicon Valley partnership to develop another technological revolution--RISC, the Reduced Instruction Set Computer. Researchers at Berkeley, Stanford and IBM discovered in parallel that if you made computer instructions simpler, the machines that run them could be about four times more efficient than current day computers, which are called CISC machines, or Complex Instruction Set Computers. The 80386 is a CISC machine. As the RISC architectures are simpler, so are the integrated circuits needed to manufacture them. Intel used 280,000 transistors to achieve 4 MIPS of performance in their 80386. We used 100,000 transistors to achieve 29 MIPS of performance. How did we do it? We worked with yet another start-up company, Sun Microsystems. We were brought into partnership with Sun by John Doerr, our common director and venture capitalist from the well-known firm of Kleiner, Perkins, Caufield and Byers. So, another notion--that American companies do not cooperate with each other--is also wrong. We cooperate a lot. Our cooperation is informal, but very powerful. Sun gave us their RISC architecture, which we could not afford to develop. We gave them our CMOS technology and put their architecture on a chip. And, we did it not for $300 million, but for about $4 million. By the way, our biggest competitor in the RISC area is not Intel or another large company, but another start-up named "MIPS." The point of all these stories is basic: competition stimulates innovation, and innovation happens faster and better when there is competition. Sumo companies protecting their rings are not innovative. I think even Intel has to run faster and think harder because of Cypress and other start-ups like us.
Although we continue to hear that the Silicon Valley is in a state of gravitational collapse which will leave behind only a few giant companies, that conclusion has yet to be supported. In the twenty years from 1963 to 1982, the decades in which the mega-companies were formed, there were 109 semiconductor companies formed in Silicon Valley. But in only the next five years, 111 more were formed. Silicon companies have been starting up at an accelerating rate. The Valley is getting more diverse, not collapsing. And the new companies are not going out of business. In the period from 1977 through 1987, of 157 companies formed, only 18 were merged or closed. The proposed consortia of American companies would stifle creativity in Silicon Valley, one of America's most potent economic weapons.
An article in the British journal, The Economist, attacked the concept of forming defensive U.S. consortia:
"The corporate strategies that have brought Japanese companies so much success in the past will bring disadvantages in the future, and so are being changed. Yet the Americans seem to be choosing this moment to imitate some of the things that the Japanese will soon no longer be doing. … Yet there is now a school of American "thought" that is calling for government-backed production cartels--or consortia, in the more fashionable phrase. These would not promote quality; they would try to promote volume. So would government-backed research consortia designed to perfect specific technologies like chips or high-definition television. Worse, each dollar of subsidy that goes into such research consortia also stacks the odds against un-subsidized competitors--and so discourages private entrepreneurs from putting up their own research money."
Next week's Fortune Magazine will have as its cover story Cypress's plan to get big without getting fat. We decided that since the start-up path worked so well for us, we would drive the company further in that direction. We have already founded four start-ups as subsidiary companies. We attract entrepreneurs, give them stock in their own company, and guide them to success. I sit on their Boards of Directors to keep them on track. Next year over 20% of our sales will come from these vigorous new enterprises. When Cypress is a billion-dollar company, my vision is that it will be ten highly vigorous $100-million companies, rather than one slow-moving, billion-dollar company. Entrepreneurs have been attacked. We have been blamed for the American competitiveness problem. I do not believe it; I do not think the data support the claims of the attackers. Cypress is run by a Dartmouth graduate, just like you. I am also proud of my English/Irish/French/German mongrel heritage. I invite you graduates--and I mean it--to join me in the competitiveness war at Cypress, or any one of the other hundreds of Cypress's in Silicon Valley. Thank you.