Difference between revisions of "Oral-History:Richard Petritz"
m (Text replace - "[[Category:Ion implantation" to "[[Category:Ion beam applications")
m (Text replace - "39 Union Street, New Brunswick, NJ 08901-8538 USA" to "IEEE History Center at Stevens Institute of Technology, Castle Point on Hudson, Hoboken, NJ 07030 USA")
|Line 17:||Line 17:|
This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.
This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.
Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, Rutgers - the State University,
Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, Rutgers - the State University, , , NJ USA. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.
It is recommended that this oral history be cited as follows:
It is recommended that this oral history be cited as follows:
Revision as of 15:46, 30 June 2014
About Richard Petritz
Richard Petritz was born in Rockford, Illinois in 1922. He attended Northwestern University where, after serving in World War II, he earned a BS and an MA in electrical engineering. He then went on to complete at Ph.D. in theoretical physics. In 1949, he worked briefly with his advisor Arnold Segert at Los Alamos. In 1950, he became a professor of engineering at Catholic University in Washington, DC. He left CUA in 1958 to become a research director at Texas Instruments in Dallas. After ten years at TI, he left and formed a venture capital group called New Business Resources that focused on businesses in the semiconductor industry. NBR founded Mostek, among others; Petritz served as first president of Mostek. After NBR disbanded, Petritz went on to form NMOS, jointly owned by the British government and the private sector. After Thorn-EMI bought out Britain's share, he helped form Simtek, which specializes in high-end memory chips.
The interview begins with Petritz's education and early career as a professor at CUA. He then turns to his career managing research labs at TI. The bulk of the interview centers around his work in a venture capital company involved with early silicon valley type companies. He describes his work with Mostek, and suggests that the smaller, newer companies, like Mostek or Intel, were much more successful in semiconductors than larger, older companies. He also suggests reasons for Intel's continued success, including the fact that they were never bought. He then describes his work with NMOS and Simtek, explaining the reasons for the varied successes those companies have had. He also presents his view of the history of the microprocessor.
Petritz died on 28 April 2003
About the Interview
RICHARD PETRITZ: An Interview Conducted by David Morton, Center for the History of Electrical Engineering, June 21, 1996
Interview #272 for the Center for the History of Electrical Engineering, IEEE History Center, The Institute of Electrical and Electronics Engineers, Inc.
This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.
Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, Rutgers - the State University, IEEE History Center at Stevens Institute of Technology, Castle Point on Hudson, Hoboken, NJ 07030 USA. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.
It is recommended that this oral history be cited as follows:
Richard Petritz, an oral history conducted in 1996 by David Morton, IEEE History Center, New Brunswick, NJ, USA.
INTERVIEW: Richard Petritz
INTERVIEWED BY: David Morton
DATE: June 21, 1996
Background and Education
Well, start with telling me when and where you were born. And where you grew up, your education, and why you decided to go into physics.
I was born on October 24, 1922, in Rockford, Illinois. I am a midwesterner, and lived there until about high school. My dad's job changed, and I ended up going to Chicago. I decided to go to a technical high school, in Chicago, it was a central one called Lain Technical High School. It was comparable to large city technical schools. At that time there were over eight thousand students, all of whom were boys — it was strictly a boys school. I did not go with vocational; you could become an auto mechanic, printer, or an apprentice, or you could go through a very good college prep-type course. It is interesting how they separated you at that stage. You were grouped with people who chose the same career path, but in particular the college career path, which I chose for reasons that I am not sure right now. I chose Latin as my language, and that put me with a very good group of people. It was a smart group of young fellows, because nobody in their right mind would choose Latin as their major language. One of the things we got was two years of Latin.
Looking back on it, comparing it to today's education, it was really pretty solid. I had four years of science, starting it out with general science, and then biology, then physics and chemistry, a year of each. I had four years of mathematics up through trigonometry and beginning of calculus; I think today they go a little further than calculus. I had to take the shop courses. I had one aviation, then woodworking. It was a long school day because of those shop courses; it would go to two or three hours in the afternoon. And then the normal courses in English, history and civics and so forth. So I had, really, in retrospect, a very solid high school education. During that time frame, I did well in mathematics and physics and chemistry, and I really liked the scientific end. My dad was a graduate of Georgetown, both undergraduate and law school. I think my career path was either to become a lawyer or an engineer or scientist, and I didn't particularly like the idea of being a lawyer, so I chose the engineering path.
At that time, this is about 1940, Northwestern had received a very large endowment from Walter P. Murphy. He was quite an interesting guy in his own right. I don't know if you have ever heard of him, but he gave the school, I think, ten million dollars — which in 1940 was an enormous amount of money — to build a first-class engineering school in the middle west. There are other schools that competed for that, like IIT. Northwestern got the gift, but he wouldn't allow his name to be used. Well, it is not called the Walter Murphy Technological Institute; it was called the Tech School, the Northwestern Technological Institute, and they built a very fine school there. I was not in the first class, but was there the second semester. I was one of the earliest students; not the very first, but one of the earliest at that school. I felt fortunate to get into it, because practically everybody there was a valedictorian or something. I think I stood about fifth or sixth of my graduating class of two thousand fellows. I enjoyed sports in those days. I was on the swimming or diving team, played baseball, but when I got to college I really gave up that kind of thing and concentrated pretty much on my studies.
I started out, not knowing any better, as a civil engineer. I think that was based on my dad's saying, "Herbert Hoover was a civil engineer, so you can't go too far wrong." It was a co-op program, and it took me only a year to realize that civil engineering, at least in those days, was done largely by handbook. I did cooperative work in the summers as a roadman out with the railroad, and that was fun: driving stakes and living out the open road, as they say. But it also taught me that there wasn't really an awful a lot of room for creativity. And the guys that seemed to be having the most fun were in the electrical engineering department or the chemical, but not civil or mechanical. So in the second year, I shifted to EE. I also joined the Naval ROTC. This was just frankly before the war broke out on December 7, 1941, I think. I got started in school in 1940, but that kind of put me in line for a career in the Navy during the war. But we were fortunate that the Navy didn't need a lot of officers right then, so they left me and basically my class in school up through four years. It was February 1944 when they finally decided that they were getting ready for the invasion from England in the summer. So I essentially completed a bachelor's of science, not the full EE degree.
When I went off to active duty I got into the radar end of the business. So I had good training in electronics in the Navy; I was radar officer, if you like, and an officer in what they call the CIC: "Combat Information Center." Kind of a fun career in the Navy through 1946. We were in battle at both Iwo Jima and Okinawa, and we got a presidential citation. We had a couple of near misses from Japanese kamikaze planes, and in many ways I am lucky to be alive, because had some of the bombs gone off, the ship would have sunk. In a squadron of twelve ships, I think only five really survived the kamikaze attacks that Iwo Jima and Okinawa had. The Navy in those days wanted people to stay in, but I really didn't have the right temperament to be a naval officer, and I realized that. That is why at the first possible opportunity I got out, and went back to Northwestern. The GI Bill helped at the beginning, but also I completed my bachelor's degree and then went on for a master's in EE.
And as an assistant I had help. The dean of the engineering school was Dean Ashbach, who is one of my idols. He's a marvelous guy. His son and I were roommates, but he gave me a Walter Murphy Fellowship. That basically added quite a bit to the GI Bill and enabled me to shift to physics. He was a pretty broad-minded guy, and he's the dean of the engineering school. I went on to get my Ph.D. in theoretical physics. It turned out that the physics department was housed in the same building as the engineering school, but it was different; it was part of the liberal arts school, not part of the engineering. So it was pretty decent. You may know of Sam Goldschmidt, who was on the faculty, of the Uhlenbeck-Goldschmidt spin theory; you wouldn't probably know that name. They didn't have a Princeton-type faculty, but they had a very good physics faculty as well as a good engineering faculty.
I can still remember Dr. Goldschmidt telling me, "Well, you have to take electricity and magnetism." And I said, "I have a masters degree in electrical engineering. How can you teach me anything about electricity and magnetism." And he said, "Well, take the course and we'll see." It was totally different. It was from a much more fundamental theoretical-physics point of view, and it really was different, and I always remembered that. The great thing about physics is you learn the fundamentals really; it's less applied than engineering. And I took it in, if you like. The sequence — doing engineering first, always had sort of a practical outlook on things — but then it took me from 1947 through when I graduated in 1950 with my Ph.D. in physics to get those fundamentals down in quantum mechanics and statistical mechanics and all of the courses that you think you have learned a little bit about in the engineering school, but not really very much.
None of the experiences that I had were quite as exciting as in the summer of 1949. My thesis advisor, Dr. Arnold Segert, was quite a world-renowned physicist. We got to Los Alamos in the summer, and what happened at Los Alamos was that after the war was over, they had this tremendous collection of Nobel Prize-type physicists: Fermi, Teller, Oppenheimer, and really big names who had left after the war was over, felt it would probably take the Russians ten years to get to where we were in the atomic energy business. And what they hadn't really figured on was the secrets would be stolen by people like Klaus Fuchs and others. And the Russians set a bomb off in 1949 I think it was, because I went to Los Alamos in the summer of 1949 with Dr. Segert.
It was a tremendous experience, because all of these physicists came back, they all had a little bit of a guilt complex that maybe they left a little bit too early. That was the summer that Teller had invented, with Stanislaus Ulum, the hydrogen bomb. And John Von Neumann was there, lecturing on the MANIAC computer. George Gamoff was there lecturing on the origin of the universe, "The Gamoff Theory," and I was rubbing elbows with all of the big names.
I worked under Hans Bethe, who is a Nobel Prize physicist. In fact the work I did — I learned later on when I visited Livermore when I was doing some consulting for the Navy — I introduced myself there. The head of Livermore said, "Petritz, is that Petritz of the Petritz ray theory of nuclear bomb efficiencies?" And I said, "Yes." Well, that never got into the public literature. Apparently that work I did under Bethe was recognized and used for awhile, which was always kind of a nice ego builder if you like.
When I graduated, I decided that I would like to get into university work, although I did interview at Bell Labs. I had a job offer there, and it is hard for people to believe it today, but I thought I was really too pure a scientist, and that I wanted to do research. I didn't want to go to work for Bell Labs. I was offered a position at Catholic University in Washington D.C., as an assistant professor. Maybe I was just an instructor — just the beginning of a career, and I think I got up to assistant professor. I don't think ever made associate. But I had a fun time in Washington. That was quite a hot bed: you know, the National Bureau of Standards was there, the Naval Research Lab. Catholic University had a good physics department. I was teaching there from 1950 up until 1954, and actually I was there from 1950 to 1958. During that period I wrote about 25 papers. My general area of interest then was in semiconductors, transistors, and particularly photo conductivity, and I wrote about, as I said, twenty-five papers and some of them were pretty well recognized, enough that Texas Instruments got interested in me.
Dr. Gordon Teal was head of T.I. and had come up sort of from nowhere. They were one of the companies that put up $25,000 and got the licenses on the Bell patents regarding transistors. I thought Bell Labs and AT&T's treatment of that invention was absolutely right on, and that they did not keep it a secret. They made it available both through university people like myself. They had a summer course that I went to, and they also made it available to industries to license it. But the whole solid state electronics got off to a very fast start in the 1950s, largely because Shockley, Bardeen, Brattain, and Teal had invented the transistor and they didn't try to keep it a secret. I don't know whether there was pressure, but I suspect that there was quite a bit of pressure to keep it a secret. Particularly pressure could have come from the military, but that is just a guess.
Whether there was pressure or not, they made it available to the whole industry, and the whole electronics revolution moved enormously faster because of that. So I think that management of Bell Labs there really deserves a lot of credit for making that public and giving it the push that it needed to get it going. What was interesting, actually, was that the outcome of that was that the big companies — the General Electric's, the RCA's — they didn't do terribly well with it. It was the little guys like Texas Instruments that grabbed a hold of it, and T.I. actually hired Gordon Teal as a director of research. They really were the first.
Why do you think that is?
I think the well-established companies, first of all they had a lot invested in vacuum tubes and in other technologies. It has been very hard for a big company to change course. I am not saying they don't want to, but it's just difficult. And of course small companies can start up faster and get going and do things much quicker and more efficiently. And that has been really the story of the semiconductor industry; a lot of the major innovation has come from new companies
I was in the teaching and research role there from 1950 to 1958 and realized that that was probably long enough. And I had a nice offer from Motorola as well as T.I.. I decided to take the T.I. offer and join Gordon Teal as a director of one of his labs in Dallas in 1958. And I was with Texas Instruments for ten years through 1968 and had a really exciting career when the integrated circuit was invented by Jack Kilby at T.I. I got involved early with the planar process, helping making it a real manufacturable item through the work that I had in my lab. And there were really a lot of exciting developments of silicon technology for optical electronics. We had really quite an interesting lab that I headed up, with something like 300 people at one time and forty or fifty Ph.D's. So I had moved from doing personal research at the university into managing research at a company. It was at that stage, when the company was growing or — They might have been a one hundred million dollar a year company when I joined them. By the time I left, we were well on the way to becoming a billion dollar a year company.
Beyond the obvious differences, what's the difference between conducting your own research and managing research?
It's a big difference, and that's the big decision that somebody has to make. In the university environment, or even at a national lab like the Naval Research Lab or the Naval Ordinance Lab, where I did a lot of my personal research with a small group of people, you are basically spending most of your time on your research subject. You are writing papers, and you are doing the work yourself. The rewards there really come from publishing and becoming known in the scientific or the engineering community, and that is how you develop your reputation. At the same time you do some teaching as you do this. As for managing research, I think one of the things I did well was that I decided to make that break. I didn't kid myself, thinking I could go at T.I. as director of research in this area of semiconductors and still write a lot of papers. I think I carried over maybe for a year or so, kind of finished some things that I had written. But I didn't try to continue that. One of my colleagues, I won't mention the name, did try to do it and, really, I don't think you can do both well. Enough said on that. But I think that there are people who are research-oriented who want to be managers, but they don't want to give up the research side. Let's say that I happen to know one of those.
I made a fairly clear decision that once I got into management, it was important to make sure that the work of thirty to forty people — which eventually became more like three hundred — was well directed and choice of pride. The major responsibility of being a head of an R&D group is first of all really to be pretty confident that your people are working on useful things. At Texas Instruments, as opposed to let's say Bell Labs, we were not as interested in extending the knowledge as determining how to help build cost-effective transistors, and later integrated circuits. The pure research scientist is just working on learning things. And that's what I did when I was in the university environment. There wasn't anything terribly practical about it. A good research manager at a reasonably-sized company has to be interested in products that are going to eventually become production, I recognized. I started in the Gallium arsenide equipment called "compound semiconductors" at T.I. The opportunity came to head up the semiconductor lab, which was basically silicon, and I realized that the silicon was a material of the future. So I chose that opportunity to take over the silicon lab, which was the most important lab that T.I. had: that's where we were developing integrated circuits and the processes that go with it.
That was during the early 1960s. A lot was happening in the semiconductor business, but it wasn't happening out of places like General Electric or RCA. It was happening with little companies on the west coast, particularly Fairchild. Shockley had started a company and it attracted some very good people, very bright people. Shockley was really the inventor of the junction transistor. He was a very bright guy. He had his goals set and, because of his problem back at Bell Labs, the telephone switch, he wanted to build a diode — we call it P-N — which would be good for telephone switching. Some of the guys were under him — Bob Noyce in particular, and Gordon Moore, the people that are very famous — saw the opportunity to design silicon transistors. Up until then, the type of silicon transistors that Texas Instruments was building for the most part were grown junction. They were not a very cost-effective. They were good for the military, but sold for twenty-five dollars, that type of thing.
So what happened apparently — I wasn't part of Shockley's company — I think it was seven or eight of these fellows got fed up trying to convince Bill Shockley that they were working on the wrong thing. And they secured financing from Sherman Fairchild and started a company called Fairchild Semiconductors. They brought out a silicon planar transistor that was just head and shoulders above anything that anybody else had. And I realized quite early that these were the competitors; it wasn't the big companies. And then from Fairchild other companies spun off, so that there was a whole Silicon Valley, as we know it now. It was really built around small innovative companies. Shockley's company was sort of the origin, if you like: the fountain. Four or five years later there were literally hundreds of companies there at Silicon Valley; it is an incredible story in itself.
It also showed me personally that it didn't really take a lot of people or a lot of money to start a company in the semiconductor business. And I realized that my job at Texas Instruments was primarily one of starting new businesses for T.I., and we were quite successful at that. There is what they called the series 7400 line of integrated circuits that came out of my lab, and even today it is the most successful line of bipolar integrated circuits. It is very successful and it is still, although it is being displaced now by ASICS and other technologies. But we did that in a lab with a half of a dozen engineers, and we had salesman assigned to go out and find some customers.
New Business Resources and Mostek
Also the Fairchild story shows that you didn't really need this huge structure like Texas Instruments. In effect what they were doing was supplying the money, but this was also in the era when venture capital was just beginning to become sort of more than just a family type, wealthy people. My partner there, Richard Hanschen — he was vice president of sales and marketing at T.I. — and I decided that we would go into the venture capital business, that we would leave T.I. We both had been there about ten years, and raised some money, and started some new companies. So he and I left, I think it was the 1967 time frame or 1968; it was 1968 for me because I know I was there ten years. And I think I had been a very productive lab manager and left on good terms and just decided to do my own thing and become an entrepreneur. T.I. did have a very good entrepreneurial flavor to it, because they awarded stock options and I had a very nice stock option program with T.I., which enabled me to generate a reasonable amount of capital so that I could be on my own for a while.
I put together a business plan for a venture capital firm; we called it New Business Resources (NBR). The goal was to raise seven million dollars and start a half dozen companies in the high tech area. Dick Hanschen and I decided very early on that we were just going to do brand new companies. I had seen enough of trying to get in and straighten somebody else's problem. So we concentrated on new companies. Of course, having been head of R&D and really responsible for looking forward five or ten years, I had a pretty good feeling of what the exciting technologies were. And one of which I had identified was MOS as a way to go beyond bipolar technology.
Talking about mistakes, one of the mistakes that Texas Instruments made was that they decided that the MOS program was going to be moved from Dallas — by the way, this is public information, this is not anything that is private information — that they would move the MOS group from Dallas to Houston, and just sort of told these guys, "Look, if you want to work in MOS you have to move to Houston." Well they did not want to move to Houston, they knew that I had set up — well, was in the process of setting up — this venture capital company and they came down to see me. L.J. Sevin and Lou A. Sharif and I were very much interested in MOS and felt, well, that there would be other people in the industry that you could found a company around. But these were guys that I knew, that actually worked for me at T.I.
We put together a business plan about the same time I was financing New Business Resources, and this all sort of came together in June of 1969. The partnership was funded by Werthheim and Company plus some people that I knew, like Northwestern University, for about seven million dollars. Dick Hanschen and I were the general managers and over the next two or three years invested in six companies, all brand new companies. It was six or seven; actually seven, six of which were successful. The big success was Mostek. That was the first investment; that was the same MOS group that I mentioned. Mark Shepherd, who was president of Texas Instruments at the time, was interviewed by Electronic News when he retired some years ago. One of the questions that they asked him was, "What were some of the mistakes you've made, Mark, along the way?" I was pleased to see that he said, "Probably the biggest mistake I made was the way I handled the MOS group so that we lost a lot of good people there." It was sort of cavalier the way he handled it. It certainly set T.I. back some and enabled Mostek, which became a three hundred and fifty million dollar company, to be a very successful company.
I understand that you were the first president of Mostek?
So you had an active role in the new company?
Right. In the very early days, to get money, we made a joint venture with the Sprague Electric Company. We actually got Mostek running with a relatively small amount of capital because Sprague had put a nice new plant up when they were really not doing well on the bipolar business. They were chasing after T.I. in trying to do what was T.I. was doing, so they were behind the learning curve of the semiconductor. Once a company gets down that learning curve as T.I. did, you never catch up to them. Mr. Sprague was quite a fine gentlemen up in North Hampton, in Massachusetts. They knew enough to stop spending their money on bipolar and invest in Mostek. My venture capital partnership put the first half million up, and Sprague followed with about two million, plus they made their facilities available. So Mostek ran a production operation and did development in Sprague, in these facilities, paid rent and so forth.
What were Mostek's early products?
They were memory chips. In the early days of MOS, there was no very good way of adjusting certain properties. The surface potential governs whether it is an "N" channel or "P" type and there was some research in ion implantation. And one of the things that I did when I became partners with Sprague on the Mostek thing was that I went up and visited the research labs in North Adams, I think it is. They had a little set up there where you could do an ion implantation on a silicon wafer, one wafer at a time. The research guy doing the work — I forgot his name now, I think it is something like Richardson — said, "You know, I can set the surface potential of this MOS just by how many ions I implant." I knew that what was interesting was really looking for a way to set the surface potential. I said, "Boy, that sounds interesting," and I communicated this information to my colleagues at Mostek.
Bob Palmer, who now is president of Digital Equipment, was my process engineer. He had a physics degree background and said, "That sounds good." So he went up there with me and one of the innovations that Mostek proclaimed being first to do was to put ion implantation to work. And this gave us an edge over Intel and the other companies, because they weren't using it. I think from a processing point of view I would give Intel credit for recognizing the importance of the silicon gate technology as opposed to metal gates. They were on their own path. Mostek sort of had a unique position for two or three years, and they knew more about how to use ion implantation. And Intel and some of the other companies just didn't pick it up very fast. Well, that gave Mostek, I feel, a technological lead on the industry; it took quite a while before the two companies both began to use the two technologies. Intel picked up the ion implant eventually and Mostek eventually picked up the silicon gate. But there was about a period of two or three years there where they were going their separate ways.
From your point of view of innovation, here is innovation in a small research lab up at Sprague. My claim to fame on that one is that I spotted it and recognized that this was important. It had been sitting there, and the Sprague people knew about it but they weren't doing anything with it. We literally ran the Mostek production for probably six months by carrying wafers from the facility — where is that suburb of Boston now? Mostek rented this space from Sprague, but it was about a hundred-mile drive. We would take the wafers that were processed up to a certain point, drive them up to North Adams, irradiate them, sort of one wafer at a time, then take them up. And we did that for probably at least six months before we got our own ion implantation equipment.
But it was a nice story of how there can be sort of hidden treasures in research labs. My coming up to the research end [of Sprague] was probably worthwhile to see what they had. My Mostek colleagues really didn't think there was anything of interest up there. None of them would have gone up and looked at it. I think probably that story could be verified by a number of people. And Mostek did not think that there would be anything interesting in the Sprague Research Lab, and I didn't know that there was. But I said, "Well, let's go take a look." It really gave Mostek gigantic leap ahead of the competition to get into the MOS business in a big way. So that is really how Mostek got started, and got started in the memory business. It never deviated from its commitment to MOS.
Mostek Compared to Intel
At that same time Bob Noyce and Gordon Moore left Fairchild, which is kind of an interesting story in itself. Noyce was really in line to become the president of Fairchild, and for reasons that I don't understand they decided to hire Les Hogan from Motorola to be president. And Hogan and seven or eight of his top executives left Motorola to run Fairchild, and this did not sit very well with Bob. Bob was quite a good friend of mine in those days and he confided in me that he was going to start another company. And so he and Gordon Moore started Intel. I think it was a year before Mostek. I think it was in the summer of 1968. You can check that, but largely because, here again, big companies makes the mistake of going out and hiring somebody else, who knows when and if Intel would have gotten started. But this whole Hogan army moved in on Fairchild like the motivators, so all at once Intel gets started with two founders, Gordon Moore and Bob Noyce.
They got venture capital from Arthur Rock who made two or three phone calls to get it started. They put some of their own money into it. They made some money off of their Fairchild options. The story I have heard, they each put about a quarter of a million in so they had nice position, and of course become worth hundreds of millions. Andy Grove was one of the first people they hired as operations manager, and Andy is a very hard-nosed, a very good operating type. Noyce and Moore were more strategic. I don't think either one of them would ever claim — I don't claim either, by the way — to be a good day-to-day operating type manager. But Andy was and still is. So they had a very strong team of three people, with two scientists defining the general direction of the company and Andy running it. That group has stayed in. Noyce passed away, unfortunately, a few years ago, but it stayed together all through that, which was pretty unusual for a semiconductor industry, because it was a very successful company.
In the case of Mostek, L.J. Sevin took over as president. He was one of the three founders, and as sometimes happened when you have something like that, there was a sort of falling out between L.J. and Lou A. Sharif. And so Sharif didn't stay with the company very long. L.J. was the stronger of the two, and he was the right guy to become president and run the company. He built the company up to three hundred and fifty million dollar a year company. Mostek got into the DRAM business, which turned out to be the most cost-effective way to make a very large memory chip. It has single transistor cell, one transistor and one capacitor. It is as about as dense a memory as you can get. It became the most popular memory very quickly, and basically Mostek had some very good designers, memory designers — Bob Probstein, Paul Schroder — very innovative designers of memory chips, which Intel really didn't have. They did some of this, but while Intel's initial charter was to build bipolar and MOS memories, microprocessors were not in their game plan at all. Mostek started out pretty much as a memory company, and as the microprocessor business started to develop, they were sort of overwhelmed by the memory business. If you get into that mainstream memory business, it is very difficult to do anything else. It is just so big, and then the Japanese came into it.
So Intel started out as a memory company, a bipolar and MOS manufacturer, and their emphasis really was on bipolar early on — a lot of people forget that. In the general period of about in the early 1970s, a fellow by the name of Ted Hoff came up with the idea of the microprocessor. But Mostek stayed pretty much in the memory business and built a very large business out of it, very successful. It was so successful that it was sold in 1979 to United Technologies for something like $350 million. But the business did not have the breadth to really sustain Japanese competition. In that same time frame, Intel had shifted this business away from memories into the microprocessor area to become very successful there.
Additional Thoughts on Early Career
You were going to tell a story from a little bit earlier period.
There was a little bit that I left out in the early period that I would like to mention. And that is that I had the good fortune to be in graduate school in 1947. I think that was the year that the transistor was just invented by Bardeen, Brattain and Shockley at Bell Labs. That influenced the direction of my Ph.D. thesis. I did recognize that this was a new era where we were going to do in solids what we were doing in vacuum tubes, and my Ph.D. topic was the study of essentially fluctuations or noise in semiconductors. My advisor was a statistical mechanics professor of physics, and so I got into the research side of semiconductors very early. Once I joined Catholic University I continued that. Most of my research in that period from 1947 through 1948 was on semiconductors, transistors generally, and photoconductors. It was in the area of how they worked and what were the sources of noise, so what limited the level of signal that could be detected. But more important than that is simply that I was just fortunate to be in school at the time, and still at a point in my career where I could steer it.
The other direction that I was thinking about was nuclear, having spent that summer at Los Alamos. But I realized pretty early that particularly the weapon side of nuclear work was very confining, with no publications, and that didn't appeal to me. That is why I made what was probably one of the best decisions of my life: to go on the electronics/solid state side. And that's the research area that I pursued while I was still in the university. So my point here simply is that I was lucky to be in the right spot when the whole industry really started with the invention of the transistor at Bell Labs. It was very crude in those days: point contact transistors, and a lot of people booed this, saying, "This could never be very important." I think what I recognized was that the transistor was just the beginning. Certainly nobody envisioned putting hundreds of millions of transistors on a chip in those days, but it seemed like it had a lot of potential for growth. So that's a little bit back in the early stages of my career.
History of Microprocessor
In the early 1970s, when Mostek was doing quite well, particularly in the memory business and calculator chips using the MOS technology, their main innovation there was the use of ion implantation. Intel at the same time was developing bipolar memories as well as MOS memories; they had a very fine invention in what we call the EPROM, which enabled you to program a chip and then change it by shining a light on it. That made them a lot of money in the early days. I think my version of the history of the microprocessor and how it was actually developed is probably worth going through, don't you think?
My feeling is that one reason why it happened the way it happened was that the larger companies, the T.I.'s and the IBM's, everybody was thinking big computers. But you had to somehow put a big computer on a chip, so we were partitioning the computer into ways that could be done out of some reasonable number of chips. The real breakthrough came when, frankly a Japanese company, Busicom, was dealing with both Mostek and Intel. In the case of Mostek, they essentially contracted Mostek to build a calculator chip for the run-of-the-mill high volume calculators that today they give away. But Intel was given the task of developing a family of chips for more scientific applications and things that were not necessarily such high volume but were different, like the financial calculator. I think Ted Hoff at Intel recognized that, rather than design say eight or ten different calculator chips for eight or ten different kinds of calculators, let's put a microprocessor in there, a microcontroller that could be programmed to do whatever you wanted here. That was the four bit microprocessor called the 4004, and then later came the 4040.
I think it was frankly not at all recognized by Intel themselves how important that was. There was never any big paper given at the Fall Joint Computer Conference, or even the ISSEC. As I remember, it was written up in the trade journal Electronics. People look back and say, "Well, I realized how important that was." We think it is somewhat of an embarrassment to Intel that they didn't really realize what they had then. It didn't take too long, but there was not big fanfare or anything about it. I think the mentality was that the people were thinking of big computers and asking what they could do with a mere four bit computer, which allows essentially sixteen instructions or something. So if I had a sort of a single explanation, as often happens when you have a major breakthrough, people are looking for the wrong thing: "How do we build a big computer out of semiconductors," rather than, "Let's put a small computer on the single chip."
This is somewhat slightly different issue. You don't hear much about this simultaneous project at Mostek to build the calculator chip. Were you involved with that, too? Could you say a little bit more about what was going on there?
No, I was really not involved. I was involved on a sort of day-to-day basis with Mostek only for the 1969-1970 period. This came a little bit later than that, so I don't think that I would be the right one to talk about what happened in Mostek. I know that is what happened, but I can't really describe who did it. I know that Mostek's more important long-term contribution was really designing dynamic memories around the single transistor cell. That turned out to be a very sophisticated design challenge. They had two very brilliant guys, Bob Probstein and then later Paul Schroeder. And this really is what made Mostek the great company that it was, because it was born in the area of memory design. The calculator chip business, I think, very quickly ran out of gas at Mostek. They had maybe a couple of years of good profitability. I don't think there wasn't anything terribly unique about it; it wasn't important, but a relatively short-lived success. They then went on to try to break into the microprocessor business, but the memory business was so overwhelming. But they really weren't able to put the resources into becoming a big factor in microcontrollers or microprocessors.
What I don't know — also that you might find out talking to somebody at Intel — is when the importance of microprocessors became more apparent. I don't think that they felt it was very important at the four-byte level but it probably paved the way to where Faggin and some of these people designed eight-byte processors, and they were on the beginning of the major role. It really changed the company from a memory company to a microprocessor company. And I think Andy Grove has particularly done an outstanding job as the manager who has kept expanding the 086, this whole family of — now the latest being the Pentium. And being smart enough to recognize that the innovation, in a way, is somewhat of management to basically kill your own children, to obsolete your own products by the next generation. And you do it more timely so that the other people really never catch up to you, and this comes back to this learning curve.
And I think T.I. practiced it on the series 7400 bipolar: we got so far ahead of the industry that nobody could compete with us on a price basis. Mostek and the Japanese companies did this in dynamic RAMs, but Intel really was the one who did it on microprocessors. Motorola I don't think ever really was the big factor in the overall microprocessor. They have done well on automotive and more special things. I think Intel did a marvelous job of managing to obsolete their own products at a rate that made extremely difficult for anybody to compete. Even today they can produce a Pentium probably for half of what anybody else could produce it for. They have the volume. It's a very strong technical and business position. They have generated enough cash that they can invest in these huge factories that are needed, and it is a very strong company.
Patents and Innovation
That is an interesting perspective on how they maintain their dominance. We haven't talked much about patents. What do you think the role of patents has been in this, still talking about the microprocessor and some of the companies who were involved with that. Has that patent been important in that industry or have they been sort of secondary to the learning curve?
That's a very good question, a good topic. I was at T.I. when we negotiated our first cross-licensing patent with Phillips. That was the first major company that recognized T.I.'s integrated circuit patents. In those days, and this is back in the mid to late 1960s, the general feeling on patents was that we weren't trying to make any money off the patent. We were trying to make it so that, if for example, Fairchild had this patent and we had that, we would cross-license each other and we would make our money on the chips themselves. So it was much more of a free and open exchange of the patents. Many of the agreements involved no cash at all, and one of the tricks in starting a new company was to build up patent portfolios as quickly as possible so that one could make deals like we are talking about. The bigger companies in the early days did tend to get some licensing revenue from smaller companies that didn't have nice portfolio patents, but again the fact that it was smaller meant there probably wasn't much money involved either. That really changed. Intel had not so much patents but copyrights on their ROM's and their general structure of their microcontroller, and they fought a lot of very heavy patent battles within copyright battles, particularly with AMD.
I don't pretend to speak for either company now, but I think one of the smarter things Intel did was, they got into this mode of, "Let's just obsolete our own products and we don't care that somebody copies us. They copy us two years too late anyway." I don't think from the Intel point of view that they are paying a lot of attention to that; I think the AMD stuff was pretty well water under the dam. It doesn't matter, because you beat the guy on the learning curve. Texas Instruments, on the other hand, has dug up quite a portfolio of patents that they have enforced particularly on the Japanese, and for quite a while there, when their business was down, their patent revenue was and still is very important to them. So T.I., I would say, is the main practitioner today of making money off of patents.
Intel was using it, copyrights in particular, to try to keep other people out of the business, not to develop royalties but just to keep like AMD from selling the same thing. But at some point along the way, I don't know whether it was Andy Grove or — there was another really smart guy, I am trying to think of his name — but recognized it, "Look, we are wasting all this time and money from trying to keep people from building it. Let's just try to outbuild them." I think that is the real progress that Intel has made in the last four or five years — is very much this obsolete of its own products, don't worry about the competition. "We are just so far head of them that they aren't ever going to catch up. And in fact they are going to lose money by trying to catch up." It is very difficult to make any money once your competition is well down the learning curve: they are here and you are up here, and you never catch up. And I think that was a brilliant management decision by Intel and whether it was Andy Grove or who was just responsible for that. And at the same time just stop wasting a lot of energy. You waste a lot of energy on fighting battles in court.
What about on the memory end of the business? Or in Mostek's experience, what patents were important in any way?
Yes, I don't know that Mostek has made a lot of money. I do know that in the case of INMOS — I am jumping a little ahead of my story — but when Mostek was sold, it turned out that the patent portfolio was eventually worth more than the company was. We had some very good things, and they were things like specific circuits that made memory products faster. I have forgotten the name of the particular technique. The patents that tend to hold up best are circuit patents, circuit level which are intended if they happen to be important. You can't build this high speed static RAM without violating this patent — those have been pretty enforceable.
One more question about patents. One of the original functions of patents, historically, was supposed to be to transmit enough information for someone to duplicate the product, so did patents ever act as this vehicle of technology transfer between companies? Or were they really just a tool for licensing royalties and that kind of thing? But did anyone actually learn anything from reading patents and not trying to copy them?
That's a good question. I'd rather doubt it only because I think by the time the patent was issued the learning part was probably not terribly important. I think most of the patents in the chip industry had really more to do with being able to produce a product yourself in recognizing that the other guy could use it as well. Today it's in a phase where specific patents are bringing in some pretty lively royalties. Fairchild, I know, collected some pretty heavy royalties, particularly from the Japanese in the early phases of microprocessor. But there has been a pretty dramatic shift in Fairchild, and at least the big battles of copyrights and all that have been pretty well, just not bother with it and let's just do better.
IBM's Investment in Intel
We should get back to your career, if we can remember where we were?
Right. I think we are up to where I was telling about how the microprocessor was invented, and I tried to get into the argument of who owns the basic patent on that. There are really big squabbles over that, and I have been asked to be an expert witness by more than one side of that, and I have always stayed clear because it is just too complicated. Even though I know a lot about the early days of it, from my point of view, the first practical microcomputer microprocessor was this thing that Hoff did at Intel. I certainly wouldn't say that this was the earliest invention. This was followed I think by the work of Faggin and that group, to do an eight-byte processor. That's when people started to get interested from a broader point of view than calculators. And it has just grown from there. As I said earlier, unfortunately, the company I started, Mostek, never became a big factor in that field. They were a big factor in the dynamic and memory business, but they were sold.
I think I should add just a side note on innovation. I think one reason that Intel has been successful over a long period is that they got strong enough that nobody was really able to buy them. In the case of Mostek, when it reached the three hundred fifty million dollar a year business, a big company, United Technologies, made them a sort of offer that people couldn't resist. The stock, I think, was trading around eighteen to twenty dollars a share, and it got bid up to sixty-two dollars a share and I was very frankly personally pleased to see it be sold. That was a very good price for it; a year or two later the dynamic RAM market had collapsed and United Technologies was never very happy about it. So Intel is unique in the sense that they became strong enough early enough.
By the way, I think IBM deserves some credit here too. IBM made a very substantial investment at Intel in probably 1984 to 1985, when Intel needed some money to expand. IBM invested something like two hundred million dollars — you may want to track that down — for about a twenty percent share, rather than buying it. But I think it would have been a very bad decision if IBM had bought Intel. A lot of this innovation would have gone out the door. They were smart enough to make a minority investment. They wanted to adopt basically the 8086 family as their basic microprocessor for their personal computer business. And so they made the investment which in turn enable Intel to invest further and kept Intel as an independent company. More recently IBM sold their interest at a huge profit, and I don't think they have any ownership in Intel at all right now.
But that was a fairly important turning point where, in the case of United Technologies, they bought Mostek and the innovation and entrepreneurial spirit quickly died, and it just didn't work out very well for anybody. But those of us who were fortunate got to sell the shares at sixty-two bucks a share. That's going further into my career and I can relate back to that again.
NBR and Linolex
So I am sort of at that point that we had to set up this venture capital company to start new companies. This New Business Resources really occupied my career pretty much from 1969 through 1976. It was a seven year partnership. During that seven years, we started seven companies, all but one of which had some degree of success, Mostek being the most successful. One that I personally had a lot of fun with, which is not recognized by at least some part of the computer industry, was a company called Linolex up in Boston, or up outside of Boston, which made a computer. It was a real personal computer — it had floppy disk, a screen, a printer — but in those days it had to sell for about fifteen thousand dollars. So it was not something that everybody could go down to the store and buy.
But it was a self-contained computer, set on a desk, and really came out at a time when a lot of the competition were what we called "hard-wired machines." They were fixed for a given application: data entry, word-processing and things like this. And we recognized that the personal computer, the programmable type that could be used for many different applications, was the way of the future. I am personally proud that we helped finance a company that is recognized by the Boston Computer Society. I ended up as president for about three years. It was really recognized by the Boston Computer Society as the first real personal computer. We had a partnership somewhere along the line with the 3M company, and they eventually bought Linolex and absorbed into 3M, and it's pretty much disappeared. It's another one of those stories of big companies buying small companies.
When was this?
They bought it in 1975, I believe.
That is interesting. 3M is a diversified company, but what did they think they were going to do with it?
They were in office automation. They were in the copier business, and they recognized the importance of having a computer in the office. They made a smart move to buy that, but they basically told everybody they had to move from Boston to Minneapolis. They just lost the entrepreneurial spirit. And nothing much came of it.
Where were we now? I was kind of leading into my venture capital career. My venture capital career was basically from 1969 through 1976, a seven year partnership. As I say, we started about seven companies, six of which had at one degree or another success, the biggest one being Mostek; the second one was Linolex, the personal computer company. I won't bore you with the other ones. But they were fun companies, and they were all start-ups, which was just quite unique and even today would be unique for any venture capital company just to deal with start-ups. Seed capital, as it is sometimes called. I liked to be more on the management side of these companies, so that when the seven year period was up in 1976, we had basically two years to liquidate the partnership and distribute the stock to our partners.
World Bank and Korea
And during that two-year period of liquidating, one of the things that I did was the World Bank. Jack Kilby recognized that I was not as busy as I had been, and the World Bank was looking for somebody to go over to Korea and work with them in making their five-year plan. And they invited me to do that. I have never done any real consulting other than that, but I thought it would be an opportunity to get to know the Asian industry. And back in the 1970s — 1976, 1977, 1978 — Korea was basically an assembler, with low cost labor but no high tech. They were smart enough to realize that they needed to get into the wafer side of the business. I helped them. I don't think I have it with me, but there is a thick book that I wrote it for them. I had them send, I think, six young MBAs who could speak English, and we worked up their five-year industry plan, which they followed really quite well. I pointed out the various areas that they ought to work in. The book is planned well. I have had Korean visitors come through who know the book and said that it was right.
South Korea today is a major factor in the chip business, particularly companies like Samsung that I worked with. We used to call it Goldstar; now they have another name for it now. There were three or four of these gigantic companies that had made large investments, and I am sure the government has helped them. Out of that study I also assisted them in getting a loan for a research center. In fact I turned that project over to Glen Madland. I helped them get about a thirty million dollar grant from the World Bank to set up a research lab on semiconductors. But I haven't really done anything with the Koreans since then. Glen picked up some of that and I am not sure of whatever happened with that. From an industry point of view they have certainly done very well in setting up the business.
And when they were setting up labs, did they get their technology from specific identifiable places like ICE? Who helped them do that? I guess you weren't a part of that.
That is a good question, how much of that they just hired. They had a lot of good engineers sprinkled through the industry. I don't know. Ask Glen, because he took over the role I had. But what I recommended to them was that they do some joint ventures with smaller U.S. companies. The joint ventures with the big U.S. companies hadn't been very successful in those days, but they picked companies like Mostek, the growing companies who were more anxious. But I don't think that they did much of that either. I don't know. In fact, I am not sure how they got their technology, now that you ask me about it. They are smart people and hard workers.
Starting and Financing INMOS
But at any rate, during that period I would say that I had an opportunity because our venture capital partnership had been quite successful. The Mostek investment itself — half of a million of dollars — was worth about thirty million when the company was sold. And we had some others that were not that big of a success, but I could have continued in venture capital had I really liked. You see, in venture capital you are generally more on the board overseeing the thing rather than the management side, although in some cases, like the early days of Mostek, I got involved as a manager or president. In the case of Linolex, I ended up being president for a couple of years, but in general your real position should be more on the board investor side. So I didn't try to put another partnership together. I decided that I would write a business plan for another company.
Although it was too late to start a new semiconductor company, and we were just entering what was called the VLSI Era (Very Large Scale Integration), I wrote this plan. There is a book called The INMOS Saga that you might enjoy reading sometime. It's a fun book. How that came about is that Ian Baron had read my keynote speech at the Solid-State Conference that I mentioned. I think that was in 1975 or thereabouts. And he invited me to be on a panel of the computer meeting, AFIPS. The big worldwide computer meeting was being help up in Toronto, which was probably around 1977 or 1978. I almost did not get to that meeting because there was an airline strike in Canada, so it was quite a chore to get there, but I had made the commitment, so I did try to get there. So in the 1978 timeframe, I thought since the partnership was being wound up, it's not too late to start another semiconductor company despite what people tell me.
I was writing this business plan when Ian Baron, an Englishman who is a real genius, invited me to be on a keynote panel to represent the semiconductor industry. Basically it was a computer meeting. And there was some really bright, famous computer programming types and computer architects. It was kind of an interesting meeting because, let's say in 1978, the real impact of the microprocessor had not yet been felt by the computer industry. And particularly the programmers who were interested in these high level, very complex programs, there was a very much of "don't invade my territory" attitude. It was sort of an elitist type society, and you could sense that in the meeting. My topic was really that this personal computer — the single chip computer had been invented by now. That's what my talk at the Solid State Circuits Meeting was about, and I was not a terribly popular speaker because I was basically telling these guys that they are all obsolete. I didn't say it quite that bluntly but I said that you needed small programs that were going to be cheap. You couldn't pay fifty thousand dollars for a program to run on a computer selling for a few thousand dollars. You needed a whole new class of software and a whole new class of peripherals. You are not going to put a ten thousand dollar magnetic disk on, again, a few thousand dollar computer. I did recognize a change, and that was my message to the group there.
Ian Baron liked it very much. During the meeting, we were having a drink together and I told him that I was working on a plan to start another semiconductor company, and he showed a real interest in that. Ian is probably about 5'6", a real typical Cambridge intellectual type guy. He is very smart. I think he is probably a genius, really smart. And he says to me, "Well, I think I can help you finance that." And I really didn't pay much attention to him — he's an Englishman and you think of England before Mrs. Thatcher as being pretty obsolete — but we chatted a little bit about it. I was living in Dallas and we talked a little bit more about it. He got back to England and kept bugging me about, "Well, send me the business plan." And I really didn't do it right away. I think it was probably October before I decided that I would send him the business plan.
I had completed the Linolex word processor, but I didn't have a spreadsheet program; they had not been invented yet. At least I didn't have one. Every time I changed something in the spreadsheet, I had to go back to it by hand. It is those kind of things, by the way, those spreadsheets — Lotus-type programs helped move the whole PC industry, because they are cheap and they are really valuable. But I got the plan together, and in the initial company I think I put the name KMOS. I wanted MOS to the name. I couldn't come up with what I thought was a good name. This is just something to call it. I sent it over to Ian, but his contact was the National Enterprise Board, which is part of the British government, and it was in a little bit too far in reconstruction finance; it was a government entity whose mission was to salvage British companies that were in trouble, but they also had a mission to start new things. And they really had nothing on their plate but a lot of old things like Rolls Royce, not the car but the engines. And in this business plan.
I knew they had money. I think they had supposedly a trillion pounds or something. I was fairly venturesome saying that it was a fifty million dollar investment up front, even though venture capital companies had never been started for that kind of money. And the plan was to do basically a full line of memory: static memory, dynamic memory, nonvolatile memory and microprocessors. And Ian himself had been the inventor of what is known as the "transputer." Have you ever heard that name? It's a microprocessor that is very much communication oriented. It has ability to link up with its nearest neighbors. I can give you a copy of his book, a very far-reaching book about how future big computers would be built out of these little small computers hooked together parallel. Rather than these big serial machines that people like Cray and others were doing. And he was right-on there. So we had a good match.
On the memory side I had with me Dr. Paul Shroeder, who had been the chief memory designer at Mostek. So really the two founding partners were both geniuses, one in the memory area and one in the microprocessor area. They never got along very well together but it was fun working with them. So I sent that plan over, and Ian took it over to the National Enterprise Board. They invited me to come over. I went over at my own expense, and I stayed at Ian's house just outside of London and met with Sir Lesley Murphy for lunch one day, and basically we put the deal together over lunch. It just went real fast. They were excited about having a sort of David and Goliath story, because the Japanese picture was fairly evident by that stage, and T.I.'s and others. But they had the money and they felt I had the background talent to bring the right people together.
An interesting side note on that is that I proposed fifty million dollars on the entrepreneurial side and owned about thirty percent of the company. They came back and before not too long, "Well, you really haven't taken placement into account. So let's make it fifty million pounds," and the pound was about two dollars then. There was no start up that's ever had that kind of financing. The employee group, myself included, still owned about thirty percent of the company, so it was probably the last effort to start a full-fledged semiconductor company. We did all four major areas, and we got the thing together. There was unfortunately a lawsuit early on. My co-founding partner at Mostek, L.J. Seven, got irritated because we hired a couple of people from him, and he really tried to kill the company early on by bringing a lawsuit which we soundly won in a Dallas court.
Locating INMOS and Recruitment
So we got INMOS' finances actually closed in 1978, and then we had the decision of where to locate the company. The British side really didn't care where we located in the States, but having lived in Dallas, my partner Paul Schroeder didn't particularly like Dallas. Dallas was still smarting a little bit over the Kennedy assassination. There was a period there, it was hard to recruit good engineers, particularly from the west coast through the east coast. So we looked around and we ended coming here to Colorado for the U.S. site of INMOS, and there is a really nice building at the south end of town that we built.
On the British side it was much more of a political decision that they did not want to spend the money for a new plant. They wanted it to be an area that needed economic development and they finally decided on South Wales, rather close to England, and things worked out very well — actually Newport. So we have a very interesting design there. It was designed by a world renowned architect, Richard Rogers. I don't know if you have ever been to Paris, the Pompidou Center. Some of these world renowned architects did really a nice job. We had a very fine architect, a California architect here in the Springs that did the INMOS building. INMOS got off to a good start.
Probably the only thing to mention there is in particular a couple of things. One is, though — the word went sort of around the industry — while Petritz got the money together from the British government, where is he going to get the designer? Colorado Springs was not a big center for semiconductors, and I chuckled a little bit at that, because it turns out that more people want to live in Colorado than there are jobs for them. And I had the good fortune, a friend of mine was running the solar energy research lab up in Boulder at that time. He put one of my ads in a physics journal or something, and he had more applicants — his name was Paul Rappaport, and he is from RCA. He said if you wanted to get the very best people locate in Colorado. I also had been out to California, because it was a sort of search that we were doing as to where to locate. I stayed a weekend in Santa Clara at the Marriot. I still remember picking up the Sunday paper, the San Jose Mercury, I think it was sixty seven pages of ads for engineers — this was local, they just moved from one company to another. And I decided that California doesn't need a semiconductor company and Colorado did.
The recruiting problem was never a problem, we just got very absolutely top notch people from all over: from Europe, from east coast, California, Texas. Because Colorado is quite an attraction. A lot of engineers, scientists that like the outdoors, like to ski and so forth. We got it off to a good fast start and built up a very successful company. The major change that occurred at INMOS: this was all put together under the Callahan government. Jim Callahan ran a socialist government. The biggest stumbling block there was that this had to be entrepreneurial. I just told them, "Look, I could retire. I have made enough money off of Mostek and others. I am not looking for a job. I want to build a successful company, and the only way to do that in the semiconductor business is to make it entrepreneurial so that the employees share in the success because everybody owns stock." And that was a little hard for them to swallow, but they finally did, because I just told them that I wouldn't do it otherwise. So one of the interesting things here is under socialistic government we were able to put together an entrepreneurial company.
Thatcher Gov't. & Sale to Thorn-EMI
What happened a year later, Mrs. Thatcher is elected prime minister and we had a love-hate relationship with her for about four or five years. She loved the success of INMOS but she hated the fact that it was British government money. It was sort of like Ronald Reagan-type thinking of, so long as the government wasn't involved she would have been happy. What she didn't realize was that there was no chance of that ever being started without the government. So the INMOS thing worked out very well, the government owned, as I say, seventy percent of the company. When the chance came to sell it in 1984, Thorn-EMI made a very nice offer and she actually made a nice profit on her investment which she couldn't quite believe. Because a number of American entrepreneurs have really ripped off the British government, probably in that time frame the most famous being the car maker DeLorean. He just really raped and pillaged that Northern Ireland investment. INMOS was just a straight arrow, we didn't take any — I think DeLorean had a lot of inside deals going. Plus the car was never a success either. Much to Mrs. Thatcher's surprise, when INMOS was sold — when the fifty million pound investment sold — they got probably ninety million pounds out of it. And the employees got a nice chunk of cash, so it was very successful.
As is often times the case, when somebody buys a company, something changes. At some point, Thorn-EMI decided that they could run it better than the Americans, and I agreed to stay about a year, and I stayed a year. And at that time, the industry had turned down and they sort of panicked, and they gave up on a lot of things they were thinking. They put in their own management, John Heightly. All of my key guys, we were all basically fired at the same time — I think it was July 1, 1985. I was happy to get out. I am too old to be a good employee for two companies. If I can't be doing my own thing, I am not going to do it. I did stick around for a year to help them to make the transition. Once the authority took over and put British management in, it just lost all of the entrepreneurial flavor.
And when Thorn finally sold to S.G.F. Thompson a few years later, they didn't really get very much for it. Let's say the patent portfolio probably turned out to be more valuable than much anything else, because they just didn't follow up on the products. I guess one of the sad stories of my entrepreneurial career has been, having built or helped build — starting and helping build two very successful companies, Mostek and INMOS — neither one is really in existence now. I think S.G.F. Thompson still uses the name INMOS for the transputer. But I compare that to Intel starting out in the same time frame and still being a very successful company. From a very personal point of view, I have no real complaints because both of these companies were sold at a very good price and made some money. But it is disappointing, too, that it changed that fast, and really deteriorated that fast.
Frankly it is far more typical in the venture capital industry than the Intel story. Most companies, at some point when they are successful, someone comes along and buys them. And to digress just for a moment on that subject, for instance in the case of 3M buying Linolex, I think almost all of the companies NBR started were bought by larger companies: the Medical Information Company (Medicus) was bought by Whittiker. These big companies have a great deal of trouble getting something from zero up to ten to twenty million dollars of revenue or to a hundred million. So there is a real market for these young energetic companies that have a business; there are products that are interesting, so they get by. And it's sort of a minor miracle that Intel didn't get bought somewhere along the line because it was very successful. But I think probably one reason is that the founders owned enough stock, and they just didn't want to sell.
I have never heard the story of whether anybody seriously tried to buy Intel. One of the smarter things that IBM did was to make that investment for minority position rather than going in and buying it. When they bought Intel, I was sure it would be another story of a big company buying something that they don't know what to do with. And this whole thing could be a lot different today. That was a very smart decision of IBM, and they have profited by it not only from the investment, but how they built their own business. Most big companies aren't that smart.
Even when INMOS was sold, I put up quite a battle with Mrs. Thatcher, which is not really public knowledge. But I had RCA lined up where the two of them could take it over. I had RCA and Thorn, the company that bought it, in a position that each could take say forty percent, leave twenty percent for the employee group, and take the British government out, and leave INMOS as an independent company. But Thorn, they wanted the whole thing and they wouldn't do it. A year later, they were in serious trouble. I am not saying that INMOS wouldn't have been in trouble — I don't blame that on the Thorn management — but the typical big company wants to control everything. They are not really happy with a minority position. This is a whole subject in of itself, I won't go into it any further, but if you talk to Glen Madland about it, there are an awful lot of very young nice companies get bought, where the buyer would have been a lot of smarter to take a minority interest in it. Be a partner rather than an owner.
This is tangential, but you talked about the geographical aspects of this, for instance you mentioned Colorado. What about Wales? What were the problems in locating a high tech company in Wales?
Well my British partner Ian Baron wanted it to be at Bristol because Bristol is on the west side of England. Are you familiar with England? That's where we located the headquarters, but the manufacturing plant we put in Wales. They wanted us to go to Newcastle or Liverpool, places that would have been much more difficult. I got quite friendly with Nicholas Edwards, the M.P. for Wales. He started pleading with me saying, "You know, Wales is just like England." And in that part of Wales it was still Wales, and although it was a very nice part of the country, there wasn't really any infrastructure there. But this was basically a manufacturing plant. And we brought over a few key people from the U.S. to run it, to get it started. We provided some interesting jobs, and there weren't that many good jobs in the semiconductor industry, so we had no troubles staffing that. And it worked out fine. I think it would have been probably more difficult if we had been up in Newcastle or Liverpool. Scottish investments, that the U.S. companies have made in Scotland, have all worked out pretty well.
Nonvolatile Memory and Simtek
Where does that bring us? We are in the mid 1980s?
Yes. INMOS was sold in 1984-1985 time frame. I am pretty near to the end of this story. As I say the downturn of the industry came in that time frame when Thorn decided, as a lot of U.S. companies have done, that they wanted to get out of the dynamic RAM business. That this turned out to be a big mistake, but that's what they decided. And they also decided to give up on the nonvolatile, not the MDS-RAM but the EPROM, which is a special form of nonvolatile memory, and just stay with the static RAM's and the transputer. It was a sensible decision. I had felt that for some time, like I said this in my keynote speech at ISSEC, that the next big move in memory would be some form of nonvolatile memories that could remember after the power was gone. And the program we had going at INMOS was a pretty interesting program that was a different technology than the west coast people were using. And I had people that had developed a lot of this technology, and the understanding of it, at Bell Labs and the National Labs. Thorn decided to cut back on the INMOS, so Dr. [Gary] Derbenwick, my co-finding partner of Simtek, basically came to me and said, "Just pick up the EPROM with a nonvolatile memory program at INMOS and start a company around that." And we had assembled a lot of good engineers. I was a pretty old guy by this time — this is 1985, and I was sixty-three years old. I always enjoyed working with bright young fellows, so I said, "Well, why not."
I started out trying to raise money for it, and it's got an interesting story itself. By this time the venture capital industry had become very much more of an institutionalized industry. We did get a venture capital company up here in Colorado interested and leading an investment. During that same period though, this Dr. Watanabe that I mentioned to you earlier, a leader in semiconductor science in Japan who had become a good friend of mine. The Japanese are really interesting people; if you make a friend over there, you have a friend for life. He just wanted to help me get this thing going and he didn't want anything out of it, in the sense of a fee or anything. He introduced me to Nippon Steel Corporation. And they were at a point in time where the steel industry in Japan was in some trouble. The steel business was down and all the steel companies were diversifying. And they saw electronics as a big field to get into. I think Kawasaki made a deal with LSI Logic; there were a number of deals to put together between American companies and Japanese steel companies. Dr. Watanabe introduced me to Nippon Steel, and they were not ready to go into the big scale semiconductor business, but they liked the idea of our proposal, which was to pick out a narrower product area. And from their point of view, they could sort of tip-toe into the industry without risking a lot of money.
They proved to be very good partners. We put this thing together, closed in May of 1987. It was quite not two years after we closed INMOS. And Dr. Derbenwick, who was my co-founding partner, and Nippon Steel was going to put up a fab to be our foundry. They were good partners. They paid something like over five or six million dollars for a license to help us develop the product. And they made a couple of million dollars equity investment. Around two or three years later, they got the appetite to get into the big business. So they basically abandoned our program. They still were our shareholder, and in that time frame we secured Plessey as a partner to do the fab work. That is the British company. And we went public in 1991 and raised about twelve million dollars in public. So Simtek was already financed with some of my own money, plus Nippon Steel, and we had one venture capitalist in for a million dollars. But financing mainly on licensing revenues came close to ten million: about half from Nippon Steel and about half from Plessey. And I got it started for a lot of reasons that I won't bore you with.
The company has not developed at the rate that we hoped it would. We had financial difficulties. I think probably that there is a lesson to be learned that you might want to write down somewhere. Most of my experience had been with products that, if you developed a product, there was a ready market for it — dynamic RAM's, static RAM's, but not the transputer. The transputer really had to be designed into a system. But with the nvSRAM, I think we felt that the market would take off. "Here is a nice nvSRAM, start ordering parts." With these proprietary type products, the customer has to design it into the system, and it creates about twelve to fifteen month delay. So you don't start getting revenue anywhere near as quickly as we thought we would. So we made the sort of mistake that happens quite often in venture companies: we over-anticipated how fast the market would take off. It proceeded to spend the money too fast, and made a lot of classic mistakes that entrepreneurs make.
So Simtek did about two and half of a million in revenue last year. We are going to hopefully about double that this year, but we have been through some hard times. Compared to either Mostek or INMOS, it has been a much more difficult company to get large. But it is a different kind of company, it's a handful of people. In both cases of Mostek and INMOS, we had our own manufacturing facility. At INMOS we had over two thousand employees, about half here and the other half in England. Mostek, I think, got up to five thousand employees. Today you can build a company with twenty or fifty employees, which is what we are doing, but it takes longer. It's had its share of frustrations but it is moving along pretty well. Who knows? We keep our fingers crossed. Maybe we will make this successful.
You mentioned a couple of applications for the Simtek products. You mentioned facsimile machines, what are some other things this kind of memory is used in?
First of all, we are too expensive to be used as a main memory in a PC. Those are dynamic RAMS, and sixteen megabytes would sell for about the same price we would sell a sixty-four kilobyte — in the five or ten dollar price range — so you can't put that as a mainframe memory. Where it is used is in systems where one or two of these per system are needed, where it is important that you have rapid access to the memory, but when the power goes off — whether you shut off yourself or power failure — it retains that information. Facsimile machines or copiers, where you are trying to keep track of something. You don't want to lose it. One of the big applications coming up is to electrify the metering system for power to your house. You don't want to lose that information if the power goes off.
One of the applications we are looking at right now is airbags in cars. If it has a collision you want to know what happened, sort of like the tape in an airplane, for lawsuits more than any other reasons. So its application can be in high volume things like cars, but at least at this stage it tends to be more in moderate volume applications. It's not cheap enough frankly right now to go into your little telephone, but you can think a lot of applications where you don't want to lose the state of the information when the power goes off. And it is those kinds of things that we were looking at as our market.
I can think a lot of things that we would be nice to be able to shut the power off to conserve portable things.
Innovation by Small Companies
I think the industry, particularly the electronics semiconductor industry, has had an enormous amount of innovation and really quite a bit of it has happened at small companies. Maybe not the very germ of an idea, but the execution of it. For example, silicon gate technology was actually developed in Fairchild by Noyce and these people, but Intel was really the first to exploit it. So that the invention itself may have not have been at a small company, but it pretty well may have been a small company that recognized it and took it to market. The movers and shakers in the industry, the proactive ones, have been the smaller companies. Then they have gone and gotten big. There are a lot of successful small companies that have gotten quite big just in the last half dozen years. I think the new company formation is still going on, not to poke fun at Mr. Sevin. He made some comment not too many years ago that he didn't see that there would be any more semiconductor companies started. I think he invested in one, that company called Cypress, and that was sort of to be the last one.
Well there have been probably two hundred since then, so the industry keeps sprouting and creating new companies, not necessarily companies that can do the huge tasks that the big companies do now. I don't see anybody competing very heavily with Intel, for example, but there are so many niche areas that are like nonvolatile memory, that are opportunities. So the opportunity for innovation in the semiconductor industry, I think is still there. You need half of a dozen creative people and little bit of market knowledge and it could be done. The thing that is making that possible is — aside from the fab itself — design could be done on a PC for maybe ten thousand dollars investment, not hundreds of thousands. While design cost to do that has come down, the creative side is still there with people.
The wafer fabrication companies are now devoted purely to being a fab for you, like TSMC, Taiwan, or Chartered in Singapore. My friend Dr. Weimer is on my board. He is starting another one in Thailand or Malaysia. There are these companies that just process wafers, like that one we have got here that was processed by Chartered in Singapore, and they don't sell anything but processed wafers. So you don't need to have your own manufacturing facility. There was a network of sales reps and distributors that could sell the product, again at very low initial expense to you. So the opportunity for innovation, I think, still exists very much in the semiconductor business, and particularly in electronics in general. Software as well, so I strongly recommend it. Frankly if you are speaking at all to the IEEE, it's a good career path for young people to get involved in the electronics industry either on the software side or the chip side.
Looking back over several decades, what do you think drives innovation? We have mentioned that it has got a lot to with small companies and their structure, their freedom and people's freedom to work within them. What do you think the innovations are? There are different models, like the demand driven, or technology push, there are all these different things. Other people have put emphasis on military demand, for miniaturized circuits early on, so that has been noted as a sort of spur to encourage people to work in that area.
Yes, that is a highly debatable issue: is it market or demand-driven, or is it more technology-engineering driven? The third element of that is: What is the motivation of the people? I could probably address that better. I think my feeling about the motivation of people — why do people work harder, and do better, and be more inventive in a small company — I think it's a couple of things. One is that I think that you have more freedom of flexibility than having eight levels of management looking down your shoulder; that's one aspect. The second aspect is that people tend to fall in love with the project, without the financial side involved at all. A lot of the reasons that Simtek has survived as long as it has is because that the people really feel that this nvSRAM is really quite a unique product and don't want to give up on it.
But third, really the overwhelming thing, is the idea of ownership, the ability to create a stake for yourself and make some money in raw terms. Not that you cannot make money on a big company, big salaries and so forth, but the idea of really owning a significant ownership position in a company. Then if the company is successful, you are successful. In fact I preached to people the idea that I don't consider any startup anything successful unless the people doing it make some money out of it. This idea that it was a big success, but it didn't make any money — we had to give up on it or something — I don't consider that a success. The measure to me of success of a new company is, did the shareholders, the entrepreneurs, and the financiers — everybody needs to make money on this, not just a few people. As I said, even Mrs. Thatcher made money off of INMOS. I consider the financial side of it extremely important. It's a very strong motivator, and particularly if it is coupled with doing something that you like, you feel like you are David and Goliath situation. There is a lot of that as well, but I would not minimize the economic drive.
Now on this question: is it demand-driven or technology-driven? Probably because I have come up through the technical route, I think the major inventions are not demand-driven. It is like we were talking about: we were looking at how do you do a big computer, and here Ted Hoff designs a little four byte thing. That wasn't demand. I guess you could argue it with somebody who uses it as a cheap calculator, but that's sort of what created the market. The calculator market could have gotten along without that. So I tend to favor the technical side and have to be prejudiced to some extent. A lot of the success of a small company, a new company, or a failure, depends on assessing the market. It is difficult to do that because of what I call the bean-counter approach to assessing the market. If there is no product like that out there, it's not like Proctor and Gamble going out and counting how many women watch what on Tuesday and not something else. The classical market research I don't think is a very big factor in a truly successful, innovative company. Because the market probably isn't there already. You are creating the market by having brought something out. That's my view on it; you can argue the other side of that too. I tend to think that it is much more driven by an invention than it is by a demand.
Importance of Gordon Teal
A good example of that, by the way — I would like to get this on the record, now that I think about it — is the role of Gordon Teal in the semiconductor industry, which I really think has been not properly recognized. Here is a good example of somebody who was a quiet, very wonderful, ethical gentleman, Ph.D. from Brown, Texan at Bell Labs. He might have been in his 40s when the transistor was invented, or at least late 30s, and there was a group at Bell Labs that they called the metallurgy group. They felt that this should be done on poly-crystalline material. Gordon used to tell me the analogy to a perfect vacuum: if it is going to be a solid, it should be as perfect solid as you can have; vacuum tubes work better as perfect vacuums. And so he set out to develop single crystals of first germanium and then silicon. And he literally got no real support from the Bell Lab management. It was almost a bootleg project.
And I cite this as an example of how large research organizations can miss things like that. This was an individual researcher who almost had to put his equipment together and take it down at night type of thing; it wasn't a fully-sponsored program at all. And he, by the way, would be an interesting guy for you to interview. And as I say, I really feel that he should have a Nobel Prize. He should have been in on that. It should have been four people: clearly Shockley, Bardeen, Brattain and Teal. But the thing that really made the industry grow was single crystals. I don't know whether it was politics or what, his contribution there was enormous. That's my example of the demand side. They knew that they needed a new material but a very powerful group at Bell thought, "Well, we will do this with poly-crystalline," which was cast silicon.
Innovation in Semiconductor Industry
Another example by the way, while I am talking war stories, and I don't mean to pick fun at the big companies, but I think of two of the leading companies — IBM and Bell Labs, Bell Labs in particular — really should have been a leader in the integrated circuit business. But their management structure, the way they were organized, prevented it. They had some very powerful people in key positions who coined the term "right-scale integration." They felt the way to go was to put, for example, little four-element gates and then connect them together; it's sort of a tinker-toy approach. IBM did the same thing of what they call SLT — Solid Logic Technology — where they had relatively small chips and they flipped them over. The initial 360 series was not built around monolithic integrated circuits.
The industry was led by people like T.I. and Fairchild, and we believed in a monolithic chip where you don't put a lot of chips together. The economics of putting chips together is a packing thing. It has never been able to compete; if you can do it on single chip, it's going to be cheaper. So the two of the largest companies were relatively slow in recognizing that. As much as they, in particularly in the case of Bell Labs, innovated by inventing the transistor, they were not a leader in the integrated circuits business, and even today they are not. IBM has pretty well caught up. So there are a couple of examples. You notice the powerful research labs were not leaders at all in the early days of integrated circuits.
I mentioned the Gordon Teal story to you. I will certainly not be embarrassed to put my name on it. As I mentioned last night, for several years I have nominated him for the Nobel Prize, and I have had some strong backing. Bardeen backed it for several years, but we haven't been able to get it. And I think not only the invention itself but the way he did it is quite a story. He lives in Dallas by the way, he is still alive. He is probably in his mid 80s now. He is not in the greatest of health; he would be, I think, a very interesting guy for you to interview about innovation. He has had some very good ideas on it. He really is the guy got me out of the government research, from the university thing into a really nice important job at T.I. I have a lot of respect for him and it is well deserved..
Who else is in that category that would be an unsung hero? The Intel people have all been very well recognized, and properly so. As I said to you, I think the basic patent on integration, I don't think that Kilby and Noyce have gotten any serious consideration as Nobel Prize winners, but have certainly gotten very good recognition of their early work on patents. I guess that is about it.
I think there have been, in my personal view, relatively two important forecasts of the industry. One is, and is very well-recognized, Moore's Law — Gordon Moore. And this was recognizing that the complexity was doubling about every two years. And he made this observation quite a long time ago. I think it is probably the most important sort of general forecasts of the industry, which is just recognizing that, and saying that it was going to continue, and it has happened. The other one that I considered mine is that I recognized that the industry was going to really divide. That the solid state technology was going to sweep all the way from material through components, memories through processors, that the whole system would be put on a chip with silicon. The software would eventually control it, and that's what the industry would be. And that was done very early back in 1965, 1966 when that was not at all obvious. Today, I think people would say well that is obvious. Sort of like they say Moore's law is obvious. Moore's law is well-recognized, but I don't know that my papers had any real impact. I mean it has just happened. Moore's law would have just happened whether he had stated it or not. I think the fundamental truth would have happened. The only glory is saying, "Well, I said that back thirty years ago or something." But I can't think of anything quite that fundamental in the industry. I think that is probably enough of that.
- 1 About Richard Petritz
- 2 About the Interview
- 3 Copyright Statement
- 4 Interview
- 4.1 Background and Education
- 4.2 Navy, GI Bill and Advanced Degrees
- 4.3 Catholic University
- 4.4 Texas Instruments
- 4.5 New Business Resources and Mostek
- 4.6 Ion Implantation
- 4.7 Mostek Compared to Intel
- 4.8 Additional Thoughts on Early Career
- 4.9 History of Microprocessor
- 4.10 Patents and Innovation
- 4.11 IBM's Investment in Intel
- 4.12 NBR and Linolex
- 4.13 World Bank and Korea
- 4.14 Starting and Financing INMOS
- 4.15 Locating INMOS and Recruitment
- 4.16 Thatcher Gov't. & Sale to Thorn-EMI
- 4.17 Nonvolatile Memory and Simtek
- 4.18 Innovation by Small Companies
- 4.19 Importance of Gordon Teal
- 4.20 Innovation in Semiconductor Industry