Interview with Don Brooks
February 8, 2000
Sunnyvale, California

RW: Don Brooks went from college to Texas Instruments where he learned semiconductor manufacturing. After being recruited to Fairchild, he became the last CEO of the California Company, reluctantly selling it to National Semiconductor. Then he took over the reigns at Taiwan Semiconductor, making it the first financially successful Silicon foundry.
Well, Don, tell us a little bit about the early days, background.

DB: Well I was fortunate to get to work with Kilby. I had co - opted SMU and worked in the transistor lab, trying to develop the planar process; we were chasing Fairchild at the time, and gotten behind, and when Kilby wanted to commercialize the IC, he was looking for someone who knew planar diffusions and this kind of thing. So, when I came out of school, he hired me as one of six engineers to go to work on a commercialization of that first that integrated circuit that TI which was called Series 51. And from there, various engineering management positions, even into wafer fabs, product engineering until I guess it was probably 19...I left school in '61 it was probably ten years later, you know, that I took over a department - took over the linear department division, maybe department, I forget which one it was, at TI which happened to be a very good one to get, as you know, and linear has always been profitable, stable, continuing to grow. So my background really came kind of - I think you, in this business, what you do the first five years is very important because I think that builds the foundation of what you're good at in the future. And I think it's hard to cross over if you're process to design - if you design to software or something like that. So my background really came from fabs, working on bipolar planar products back in those days.

RW: What was Kilby like to work for?

DB: Oh the gentle giant, big man, I guess he was six seven or something. The thing I always remember about Jack is he had these long gray benches that were workbenches at TI and they were standard issue, you know, like the military or something. And all of us would sit up there at night after waiting for tests to come out or something and we'd be hanging, our feet would be about two foot off the floor but Jack would sit on these things and his feet would sit flat on the floor, you know. He was a very practical man. One of the stories that I remember, in fact I may observe this and we were talking about how to build a four-function calculator with a single chip, and we were talking about what this calculator would look like. And we had one of these plastic things that kept notes, paper, you know, they were like yeah by yeah and he picked this thing up and he said, I think this is about what it should look like. And really out of that came the idea of how to design that calculator that ultimately cost nine ninety-nine I guess or something like that. And because we had just integrated all four functions on a single chip and we wanted to take that thing to market. So he was that kind of man. He had a kind of - if you look at his invention versus Noyce it was a very practical, you know, you had four etched parts, you know, discreet parts, and they were wired, were Noyce was much more sophisticated solution with the planar diffusion and Jack was a very practical man and his ideas were pretty well grounded from that standpoint.

RW: You mentioned calculators, TI's drive to vertical integration in consumer markets seemed to have such great promise. Certainly here in the valley we saw this TI giant that was crashing around and destroying everything in its wake and we were always afraid that they were going to come after us. But then it kind of petered out - what happened in that?

DB: Well from my viewpoint, there was never - there was a parallel effort to vertically integrate while broadening the product line, which meant focus was - went from, you know - it was really a lack of focus. Something like a consumer product area, which is the home computer, calculators, watches, this kind of robbed semiconductor activity was much of its resources. It diverted money, it diverted people and Sheppard pushed so hard that the - end products like home computer used semi-conductor products - that it somewhat screwed up both, from the standpoint, home computer was out using one of the TI microprocessors instead of the industry standard Intel micro processors, so it screwed that up, while at the same time probably, semi-conductors were sort of putting as much effort in that - as they did into that product line and should have been competing with Intel head-up and as a result, wasted that effort as well. So it kind of - the desire to vertically integrate, horizontally expand was a nightmare. I mean it was just - it had a lot of problems with it. Now, you know, one of the things I remember that just shook the halls of TI was when Intel left the memory business and they decided to get out of the DRAM business. At that point in time, the Motorola's, the Ti's, the Fairchild's, everybody had made this effort to be in all segments of the business, that would have been memory, that was microprocessors, I want to do peripherals, I want to do this, I want to do that, and it goes on and on. And when Intel made that decision, I mean everybody just shook their head wondering why would they get out of the DRAM business? Nobody has ever deceased the breadth of their product line unless they dropped discretes or something. And so this had a tremendous bearing on me. I know that for the first time I began to think differently about things. To give you an idea, I took over the memory effort in 1978. We had, oh I don't know, maybe two small fabs I guess and devoted to memory. But if you look at the product line it was, and it seems kind of funny now, it was nonvollital EPROMS, it was SRAMS, it was DRAMS, it was CCD images. And I think there was one other, any one of which could have filled a fab. So with five or six different segments, very limited resources to design these products, they had three times as many as they needed. And so one of my first jobs was to start to, just for myself, was to start to decrease the number of segments - try to get it down to a manageable group so I could double up the design resources part the engineering resources, process engineering resources to insure, you know, the maximum opportunity of being successful in any one segment. But, I went to Fairchild, I went over to Fairchild and found the same thing and one fab, the one over in Palo Alto. Went over to Fred Longo. I said, Longo how many - how many memories you got? Oh I got EPROMS, I got BSQUARED, I got SRAMS, I got DRAMS. I says, if any one of them's successful, how many wafers can you run at a time? Oh twenty thousand wafers, probably a year, I mean I don't think you can run two, three thousand wafers, any one of those would gobble that up. So we all had that same problem. And it had grown the business by adding product segments, be they transistors, dials, resistors, if you want to go far enough back, or be and TTL linear and ultimately memory and so forth, if you want to go forward. And nobody thought about contracting what they're doing and focusing and making it work. It was going the other way. But it was stupidity, I mean, it's what killed the big companies, what killed the Fairchilds. And the decision, I give the credit to the guys in Intel though to get out of businesses so they could make that one business work. It was a very big decision.

RW: Of course they got out of watches, out of CCD, bubble memory. They've gotten out of a lot of them.

DB: Yeah. The DRAMS could have been all consuming in itself. I mean, you know.

RW: Well it's interesting too that because TI was - they were such management gurus, so much interest in management, management was actually writing books on how you run a business and so on. And I've read those and most of it's pretty good I think and still current. What happened, was there a disconnect between the theory and the practice?

DB: Well, the biggest problem they had with management was the Sheppard-Busey regime up to the time that Jerry Jenkins took over, which was I think the turning point in which TI started to recover, was - it was a very centralized organization. It was one that - I guess it manifested itself in a lot of presentations in the boardroom. It manifested itself in the continuing questions being sent down and you spent all of your time answering and responding. So not only did you as a division manager, but you had to support the - give a lot of presentations and be and write a lot of explanation, but you also to support the boss' idea of group manager and sometimes you had to support the president, you had to present to the board. The frustrating part of TI during that timeframe was that you had no time to manage the business. In fact my attitude was, in that time, is that there is a big black cloud, it is raining like hell and I'm going to pull out this umbrella and shelter my people and let them work while I feed the upper management. In fact, when I quit I told Fred, I said, you know, Fred, if I get another promotion around here and get taken away from the businesses anymore I think I'm going to get sick. You know, I just - I can't live with this anymore. He was always laugh last. He'd laugh and say, "Yeah you're right," he says, "I'll try to explain it to Mark, da, DA, DA DA, but that's the way he wanted to run things."

RW: But why did you leave?

DB: Just really that. You know, I thought Sheppard was a bully. I thought he tried to manage the thing from the top. He never let the management of the organization have enough time much less enough resources and latitude to run the business the way they thought to - thought they should. And it became one that just wasn't fun, you know. You know, with TI it wasn't ever really profitable either, so it's not profitable, it's not fun, and it's Dallas, Texas. What else do you need?

RW: Who recruited you to Fairchild?

DB: Mickey McCray, I guess, no? Is it Mickey McCray? Yeah, Mickey McCray was the headhunter. He got me with Tom Roberts. I had - I went there in '83 and I guess it had been like '80 they had come after me before. And so Roberts, I got to know pretty good during the eighty time period. And he - I forgot who called who in '83 but anyway, I was kind of fed up and looking for something else to do. And so we got together then and from there I was introduced to Michelle Vio who was running the electronic part of Schlumberger then. And it was understood from the first that I would come in as Executive Vice President and that I'd move on to be President. I had to place my foot firmly and talk Tom Roberts back to get him out of there ultimately, but it took two years, but it - he did.

RW: He made a lot of changes - he and Schlumberger made a lot of changes that ultimately resulted in the demise of the company. What did you see when you got there?

DB: Well first, if you think of Schlumberger as being the way it grew up and it became a successful company by farming small business units, which is usually a blue truck, a dead computer and some kind of down hole measuring equipment. And a Texas Aggie with a technician running around the southwest measuring things down hole. Then this guy's responsible for getting it done, getting the results back, getting the billings done, and everything else. I mean it was the most perfect decentralized unit you can imagine. And they tried to apply that to semiconductor industry. So the first thing I thought was that, you know, thirty-two different manufacturing areas on a five hundred thousand dollar a year business. It was just incredible. Everybody had their own fab, everybody had their own packaging test, everybody had this, had that. And it was just grossly uneconomical. And in contrast, the industry's gone to foundries, big fabs, you know, separate the business units from the manufacturing units, this kind of thing. And I spent all the time I was there trying to build big fabs and have productivity areas like Japan, Nagasaki, Japan fab them and built them to Sony and shut down this inefficient stuff so that a wafer could be manufactured at reasonable cost. The other thing I found was that I guess it hit me one night as I was contemplating what I was going to do, that probably more than half the people over there wanted me to fail because they'd just like for me to get out of their hair and go on do the nonproductive little things that they'd been doing for the last five or ten years and they weren't really interested in seeing the company profitable, upgraded, and so forth and so on. They just wanted their eight to five and get out of there. So I was convinced that I had to move people out of the company, out of some jobs, and I had to do it very quickly to ever get the right kind of people, the right attitude, the right centralization, the large capital intensive business, and to be able to find a way to attract the right design and application talent. So as it was, from the time I took over president in '85 to... '87 we sold it to National, they just took too much - it took too much and too little time to really get it done. It was kind of interesting to see Fairchild come back public now. We sold it for two hundred thousand and it comes back for a billion-two.

RW: Tom Roberts struck me as being awfully arrogant.

DB: Yeah, I think - I don't think there's any question, and probably Tom would admit, to the success they had at Schlumberger and you have to remember he started - he was a West Point graduate, started out in sales at Schlumberger and at - at IBM, I'm sorry, moving to Schlumberger in the financial area and he became CFO. So, two things that are a little hard to understand is why there would be this arrogance, if you've come from an industry that's totally different than the one you're in - it would seem like you would be a little bit taken back, it would take a while to learn it. It didn't seem to bother him. And the second thing is that, you know, he's from a financial background and I guess the belief was if you could manage one industry or one segment of one industry, you could manage anything. But he surely didn't - it didn't map over. He was a contrarian, you know. My favorite story is that he and I were having a lengthy, not discussion, argument about whether we should centralize or decentralize the sales force right after I was there maybe six months. And I was for centralizing the sales force and he, like everything else, wanted each unit to have its own sales force and I thought this was very inefficient. And I was explaining to him and he finally ran out of things to say intellectually, so he came back with something like, well John Young and I discussed this and figured this out a year ago. And it's the right way and it's the way it's going to be. Well I walked out of there, I went to my office, and there's the Electronic News and it says, HP Centralizes Sales Force. So I wrote across there and I said John Young finally got it right and when are you, you know, and sent it back to him. But he was a contrarian; he wanted to do things different than the industry did things.

RW: Well, I was trying to recruit a fellow out of Fairchild who had been there quite a while and didn't want to leave. And I got the last copy of Fairchild Progress, this was before you came on board, which was the newsletter that was sent to the customers and it covered all the Fairchild products. And it said from now on, this was going away and if you want - if you're interested in our diodes, write - send your business card to this location and if it's TTL, do this, and linear. And I thought, Gee, that's pretty dumb, because typically the same engineer or buyer, buys all those things - and this isn't like my chromo watches and DRAMS. These are really very similar products and should be sold by the same sales force. It was pretty obvious. And to think that you would decentralize your manufacturing or your engineering, which you might well do, but you don't go to the customer as well we're IBM in Europe and we're IBM in New York and we're different. You go there, you're IBM, we offer a solution. So, I took this to this fellow and I recruited him as a matter of fact. So it was good news for me. So...

DB: I think probably the demise of Fairchild is probably pretty well understood. You know, a lot of people had a lot of ideas because so much of the people in the industry came from Fairchild. I think what was probably more interesting to me about the Fairchild experience was the Fujitsu proposed merger. And to refresh your memory there, it became obvious to me, first as the Chairman Rebu died and then Vier took over as chairman of Schlumberger. And it was very obvious to me at a cocktail party one night, that we weren't long for this world. Fairchild was going to be sold and maybe all the electronics business that we were in was going to be sold. And with that in mind I went to try to figure out who to sell Fairchild to and, you know, I'd try to think about with General Motors and I said if you look at what products overlap between what General Motors buys and Fairchild makes is poor. So I finally decided that the thing to do was to try to sell it to one of the Japanese because the Japanese had a need. And that need was they didn't understand the western market and on the other hand, they had good manufacturing capability and so forth and so on. So we put together the merger and basically the products would be marketed in the western world under the name Fairchild, in the Asian world under the name Fujitsu. We'd manufacture, at the, less cost for manufacture and we would design and market geographically. And so Fairchild would have an organization that would embody what was Fujitsu in California, in the United States, in Europe. And Fujitsu would do the same thing in Asia. And not to overdo this, it ended up we lost it because we could never get to the Justice Department. It was never denied, we got up to 18-wheelers worth of documents that had been sent there and they still had to make a decision and finally it just came unglued because people started to get nervous and all these things.

RW: The claim was it was a military business.

DB: Yeah, Cray computers.

RW: Was that this would give Japanese control over these two primary areas.

DB: Right, and the super computers, correct. They - the thing I find interesting about it is that the problem still exists with the Japanese companies today that we were trying to fix back then. And this was, you know, twelve years ago. I think Fujitsu, had we put that together and had we learned how to decide what R&D, which design, which products to take to market, what to do on and ASIC level in the U.S. versus the standard product, had we rationized all those resources and put that together, I think we would have had the pattern for a Japanese company to be successful in the future. And I think that was the lost opportunity out of that, only to see it go to national where it's just - I thought it was a very poor decision on their part because they only put linear with linear and TTL with TTL, it didn't bring any value added to it.

RW: How much did National pay?

DB: As I remember, they only paid a hundred million cash. I had an LBO offer in at a hundred and eighty to two hundred, I can't remember, but there was a fifty million in paperback. But Schlumberger never accepted the LBO because, in my opinion, the board never wanted to be embarrassed, they wanted to get off the books, if it would have been a success, it would have embarrassed them because they couldn't make it a success. If it had been a failure if they didn't collect the fifty million, it would have been an embarrassment because they would have to do a - it's like two out of three in football you know, two interceptions, two out of three things are bad. It was kind of what happened.

RW: Well we've been covering TI and Fairchild, but tell me about your early days.

DB: Well I grew up in Oakcliff, Dallas. I guess dad probably never made more than ten thousand dollars a year but - which was pretty good in those days. But it was always a battle economically, but enough to enjoy life. And he and I did a lot of sports together. We fished, when he came back from the war in 1948, we would fish probably half the weekends. We played baseball - played baseball, basketball up through college. So it was a pretty competitive family in terms of he always stressed, you know, be competitive, be determined, be focused on what you're doing and all this kind of thing. Also with the proper amounts of stress on academic side, to go to school. So I lived there, I went to SMU in 1958 I guess it was, and I went there because they had a co-op program and I had gotten married and, with children, and it was the only way I could go to school. So co-oped and got through school in four years even on the co-op program but largely because TI let we work and study at the same time. So I had a lot - I had loyalty to TI, it's probably why I was there from '61 to '83, so that loyalty that they provided a vehicle for me to be able to go to school. Got one sister, she's three years younger than I am. I have two children now, both approaching forty - each having a grandkid for me. So, that's about it.

RW: Do you think that the recruiting of Texans into TI made it a too parochial, not enough outside thought?

DB: No, I think that may be a little bit of a misconception. First is, the man who made, in my mind, the man who made TI was Pat Haggerty. Pat came from the East Coast somewhere, I don't know exactly where. I do think Sheppard and Busey had that myopic view of things, although there was many managers scattered around. You know, the guy at Compaq for example was German - what was his name? The one who just left Compaq. I think if you go back, and look you'll find a good mix of people across the world and across the United States - I think we just - I think Sheppard just made some very serious mistakes about his management style. You put he and Busey together, Ralph Hawbreak if you like, the three of those, they managed semiconductor there for so long. They just made some bad, bad mistakes and Mark say this and right through that management chain, nobody questioned it; nobody took issue with it.

RW: Well after Fairchild got sold, what did you do then?

DB: Oh, fooled around with venture capital, was doing some investment with Fujitsu, but really trying to decide what it was I was going to do. So then in '91, I decided to take the position as the President of TSNC over in Taiwan. But during those - between '88 to '91 was primarily doing investments for and representing Fujitsu in those investments. They kind of had me in a consulting agreement, I guess they - whether they really wanted to do those things or whether they wanted to give me something to do through the transition since it had failed, but probably not important.

RW: Well the Silicon Foundry concept had been around a long time. The ELSI was founded on that initial concept and they only added the software and design issues later when they discovered that they had to have somebody - somebody had to do these designs for them and nobody knew how to do that. But I don't think anyone ever made any money at it, did they until Taiwan?

DB: Well, I don't think so. I know Intel as well as LSI - attempted to do some Silicon Foundry work and of course probably the most foundry work was being done in Japan, in the early 90's and of course, the Japanese would give and take. You know, they'd give some foundry capacity and some technology, but they wanted some product information in return. Also a shortcoming there was they were, you know, an analogy would be my attitude about the foundry business is I'll make the cookie dough, you guys come with the cookie cutter and, you know, you can do a circle and you can do a star and you can do an apple, you can make them any way you want to, but you've got to use this dough. Their attitude was, well I want to see your cookie cutter, you know, the Japanese did. And I want to know how well it's put together and I want to know how sharp it is. And so they could, you know, pardon the expression, screw around with this deal for years - months, probably more appropriate, a design and therefore were really - had a problem getting the product to market. And then getting enough capacity was always a problem. So - so tips in technology, probably the first fab company, although they were doing business with the Japanese, was very anxious to see someone come on who was really dedicated to doing fab. And as a result, this very complementary relationship between fab this and foundry was easy to put together.

RW: By the negotiation that you speak of that the Japanese did, do you mean was that on the design per se or on the design rules and on the process?

DB: No, they wanted to know quite a bit about the design and it was always under the- and I don't know the truth here, it could be one of two things - one is to make sure it yielded right. They want to know how you designed it, what kind of tolerance, or they just feed the thing back to their own designers on how to do some of the more elaborate circuitry. I assume it was the latter and because my customers, fabless semiconductors companies, always had to fight to preserve intellectual property and also be able to get from the Japanese enough capacity and technology to make their business run. Our's was not - the Taiwanese on the other hand, in a pure foundry relationship, we were always wanting to make it yield. But we started out with the premise that we'll provide the service to people and they can do the design, stay within the design rules. I mean, if they don't, it won't yield. And they bought wafers that don't yield; I mean it's a little bit their problem. If they take the design rules and they compromise those, my position with the people in the foundry is don't worry about that. If they're stupid enough to do that, you know, it's going to create a problem. Now once the circuits played and we took the responsibility to include the yield, we would go back to people, go back to our customers and say, we'd have to understand why at the defect level we were running, this thing's yielding forty percent and it ought to be yielding eighty five. So we have to know what it is about this circuit, how it operates, what this test is that's always failing, what does that imply. So we have to go to work to include yield, but at the first cut, we just treated it like, hey, you do the tooling, we'll do the processing.

RW: Well, product engineering is what you were referring to. You know, actually the first semi-fabless company was LSI Logic and Toshiba. LSI had a metal capability only, bought wafers from Toshiba until '84. We were Toshiba's largest customer. But there was never a category for semi-fabless, it's not known.

DB: Semi-semi, huh?

RW: What was Morris Chang's contribution in all of this?

DB: Well I think that - let me just say what the history is - the government asked Morris to start a second business after UMC. I forget the guy's name now - Chinese guy in the government that he did so much interface with - but asked him to start a separate business and Morris recommended the foundry business. There's a lot of fathers of foundry, you know, Bob Towsk claims that he gave the idea to Morris. He's probably the most vocal but anyway there's more than one father of this business if you go around listening. But, to Morris' credit, he went and raised the money, put the deal together with Phillips and got it going. He hired - I was the third President and I think one of his contribution was that here you have this Chinese workforce with Chinese American educated management, but you better put somebody up on top that understands how to do business in the western world or you're going to fail. And so the second contribution I say is he kept a president in there who understood the business and the customer base. It was easy for me to understand how to market these wafers because, you know, I had set in the chairs of my customers before and so I understood what the need is and I understood what the risks were and everything else. Now, when I got there in 1991, you know, Morris was not all that positive about the foundry business. In fact, he told me we're building fab two, A and B when you come, they've already been started - A's already started and B is on the drawing board, but Don I wouldn't anticipate there being a fab three, don't get your hopes up. Well my view was just the opposite, I had looked at the numbers, I knew what the wafer costs were, I knew what the yields were in those fabs and I thought this was a tremendous opportunity but I wasn't going to argue with him. I'd just make it a financial success and things would take care of itself. So, when he speaks of being this visionary about this business, I don't think he had any idea what he was getting started. Typically from '91 to '95, '96 and when it got to be a billion plus business, he mostly stayed in Taiwan. He was Chairman of Weiss, he was chairman of ITRIC, he was Chairman of TSNC, he was Chairman of Vanguard. We'd see him once a month. I think it surprised him - he's always been a conservative man, a contrarian. I think he always told me the business was, you know, going to crap out and you're over - investing and everything, but he didn't get in my way when it came down to doing it. So I think when he looked up in '95, '96, and saw what a success it was going to be and how important it was to Taiwan, I think he was just as surprised as anyone. Now I think when I told him I wanted to come back, I think he seized that opportunity to place himself as president and get some of the rewards that he thought he was due and boy he's done that. He's probably the most powerful industrialist and industrial manager in Taiwan now.

RW: I understand after the earthquake he shook things up with the government to get power back.

DB: Yeah, yeah. Well he's very influential there now, very influential.

RW: How come - of course, Europe tried to get into the semiconductors from forever, since the 50's, mainly unsuccessfully? What's the difference with Taiwan?

DB: Well, I could point to four or five things that I think's different from Taiwan. First is, it is a homogeneous, very - it's the US a little bit of the 50's, you know. It's a team effort. People don't want to stand out, they want to be the same. The pressures that go with that on a social level are just incredible. If one of the people don't - doesn't pull their weight, it jeopardizes the bonuses and the profit sharing of the other people, I mean they get drummed out. So that's one thing - nobody wants to be different; people want to be a part of a working unit like that. Now where else can you say that in the western world? Nowhere. Second is, these people are intelligent. If you look at their high school education, their work force is intelligent. And of course, if you look at the number of engineers that we're able to put on the task, probably two to one compared to the western world because the base salaries are that much less. So you can get people around a problem, get people around a machine who are technically oriented and that works well. The next thing I would say is that the management, most of the guys who reported to me down two levels, which is probably twenty guys and gals, all had some experience in the US - they worked for Intel when they got out of Stanford or Texas, wherever they went to school, they all worked for Motorola, Fairchild, or TI or someplace. English was good, they understood the western world, they come back and they're a tremendous benefit of translating - when I said hey, here's what we've got to do to service the customer, translating that down to the people and getting it understood. So that's the other thing that's quite unique, much better than the Japanese from that standpoint. The Japanese didn't spend that much time here actually working, very few of them did. Now, when you get to that point and you - what's really important is that lets you get to some level of success. Now you've got this machine running called a fab and in the case of TSMC and UMC, maybe they have eight fabs. They're all located in one city and they are this tremendous cash generation machine. They take a billion dollars worth of revenue and let's say you've got thirty-five percent net profit because you don't have any taxes, so that's three hundred and fifty million. Say you've got two hundred and fifty million worth of - or two hundred million worth of depreciation, so it's three fifty, two hundred, and you're at five fifty. Now you got five hundred and fifty million dollars worth of cash flow from operations, so you're looking at a fifty-five, maybe sixty percent cash from operations to reinvest. So by the time I left there, I could build a billion dollar fab a year and under that scenario, because I was getting close to two billion in sales, getting close to over fifty percent of it was cash. So I just calculate how much money I had to spend and that was next year's capital bills. Now, who's going to enter the business when you've got two guys sitting there generating that kind of cash, which today is probably a billion dollars a year worth of cash from operations, who's going to go in and invest against you? I mean, who's going to take that risk? The other thing that's very powerful about the model now is the concentration of these fabs becomes a very powerful thing. You calculate how many etchers you need, you calculate, you know, five and a half. Well okay, let's just buy five. We've got surplus equipment over in the other fab. We're qualified over there, blap. So you don't buy another fab - you don't buy the sixth etcher. If something goes down, instead of losing production, you're qualified over in the other fab so you move things around. Just the opposite of Intel doing something in Hillsboro or, you know, Oregon, and Arlington or something else. This thing is just concentrated. And as a result, I think today if you did everything right in the United States, it's going to cost you twenty percent more per wafer than it cost in that environment in Taiwan. And so...

RW: What's the effective lifetime of a fab?

DB: Well, fab 2A, 2B were - at TSMC, were put together in the early 90's - '90 and '91 and both are still in production. So they've been about ten years. But that's about it. They were half-micron technology and yeah, you can still sell some half-micron technology. But they start to lose the volume. Of course, they're depreciated so the cost basis is pretty low. So they could sell wafers for - six-inch wafer for three hundred fifty, four hundred dollars and still make some money. RW: Leaving Taiwan; Semi, and when you got in, why you went back - why you went back into the business and into the present time. And then if there's anything else that you'd like to - you know, I thought the parts about, you know, why Taiwan were - well so there's a lot of money required for these fabs, how is it raising money in Taiwan? DB: Well the Taiwan investors, which in a lot of cases are individuals, there's a lot of wealthy individuals, wealthy families there, and they provide a tremendous amount of equity to industry to start new companies. And they're particularly enamored now with the high techs since so many high tech companies have done well - either fabless semiconductors or computer companies, whatever the case may be. They have their space, which I would call is a - they would like to invest DB: I think it's - I believe it's going to be a lot the same in that they are going to look at tasks that have a high manufacturing content, not particularly a labor content, but a high manufacturing content. And if it's got - if it's technically complex but proven, then they'll have the engineers to put on the task and make it run very efficiently just like Taiwan will. And I think you'll be able to find enough equity in China to be able to start these companies via Hong Kong or even via Taiwan. I never wanted to see TSMC or UMC in China. There was no customers there. It's - there's no particular advantage because the equipment is going to cost you the same, the power is going to cost you the same, the water is going to cost you the same, all these things - so you don't bring anything from the cost standpoint, there's no customer there so there's no service benefit. And of course, the Taiwan government wouldn't allow you to transfer advanced technology there anyway. So...

RW: You talked about the cost of the equipment and the power and so on. How do you rank the cost of a fab as percentage-wise - so you got depreciation as what?

DB: Yeah, well you - the depreciation is probably your biggest cost. I had - my payroll at TSMC was less than ten percent of revenue. And so it's probably third or fourth component, probably third. Probably utilities, if you throw gases, power, all this into it, it's probably number two I would think. But the labor content itself, the payroll content, which is all the management, the engineers and the - they were less than ten percent.

RW: Profit sharing. Does that pay a part in Taiwan's success?

DB: Yeah and I tell you, that's where Bob and Bob Tao and Morris Chang and I think Bob started but I'm not - yeah, I'm pretty sure he started it and then Morris picked up on it and both have really made a contribution. Let me describe how this profit sharing works. Let's take the TSMC model and you make - you've got let's just say a billion dollars in sales and you got a, probably the best year ran some four hundred million in profit with no taxes because they were on tax holiday. The way the profit sharing is calculated, you take something in the area of eight percent of profits. So let me just be rough here and say - let's take ten percent - so now you take forty million and of the four hundred million in profit, you take forty million and you convert that to shares at par value and par value is forty cents. So you've got forty million at forty cents a share. So it creates about a hundred million shares - if my calculation is right - but the important point here, you've taken forty million dollars worth of profit at par value, which is about one tenth of the market value. So you've really taken four hundred million dollars worth of market value stock and given to the employees. So you've given a hundred percent of the profits back to the employees. Now this represents - usually represents - I'm probably on the high - side with the numbers, particularly the ten percent, even the eight percent - probably looks more like five or six percent dilution every year instead of ten percent dilution. Okay, so you only gave sixty to seventy percent of the profits back to the employees. So employees will kill to make that plan and to be successful because I mean you've got engineers that are making a couple hundred thousand dollars a year in profit sharing. You've got managers that are making a million dollars a year in profit sharing because that - probably there's two thousand people sharing in that. You go back and calculate the average and you look at the span between the lowest operator to the president and there's big factors there that makes the motivation very high to get your hands on that profit sharing. Now this will work as long as the business grows and keeps being successful because the shareholder values keep going up, you know, the people keep - even with the five percent, six percent dilution, you know, as long as the stock price goes up, shareholders don't care. If the stock price goes down, profits are going to the employees instead of to the shareholders, then somewhere out there the day of reckoning has to come. But to this point, it's a very powerful incentive.

RW: Well you said that you left TSMC, you wanted to come home.

DB: Yeah.

RW: So what happened?

DB: Well I did come home. You know, for six years I had a house either in Wai Shuang Ch'i near Taipei or Dashi. And so I, you know, I lived in the Chinese community and that's what I wanted to change because I had to get - I wanted to get back to my family here - my grandkids and my son and daughter and so forth and their family, which I did. I came back in 1997. Now, the agreement when I resigned, that Morris and I pulled together, was that I would have - I would stay on the Board at TSMC and I would manage the US operations here, which was the wafer tech up in the northeast and I'd oversee the sales office, although I wouldn't have any direct executive responsibility, I would be watching those. About two weeks before I left, he decided that he didn't want me to participate in the board so I just resigned totally. So, shortly thereafter UMC, Bob Taft came around and said, hey we'd like you to be on our board. We'd like for you to live in the US, work in the US, help us with our service strategy, our process offering strategy, build up a sales force, build up an image, build a relationship with customers, and so I said fine. I can live in the US if I can help this company, travel back maybe four times a year. So it satisfied both sides. So that's when I went over the UMC and I still work with them. I'm still on the Board of Directors. I don't have any executive position now and I'm usually not around here very much, but I'm trying to slow down because most of my effort now, on a day-to-day basis, is in the venture capital business. I've got a couple of funds I've raised, I raise some fifty million now and out trying to find the next broadcom or the...

RW: It's amazing that these two hardware guys, these valuations on the dot-comers seem awfully high because they don't have any fabs, they don't have anything tangible. So it seems outrageous that they have such a valuation.

DB: There's a few winners in there, but I think there are going to be very few.

RW: Well thanks very much.

DB: Thank you I enjoyed the interview.