RW: Dave was at Intel for twenty-two years, thirteen of which as General Manager of the microprocessor division. In this 2005 interview, he describes the steps taken to maintain and strengthen Intel's position of dominance in the microprocessor market. From the 286, 386, 486 and the Pentium's, Dave kept Intel the dominant player in the field. Well Dave, tell us about your growing up and what was your family like?
DH: I grew up in Michigan . My dad worked in the automotive industry in a factory. I mean, he was a machine operator. Later became involved in labor relations and I was the oldest of four kids. Very, you know, we were not at all well to do. We made ends meet but to get by, always had to work and earn my own money and put myself through college.
RW: So sort of middle class?
DH: I would say we were at the bottom of the middle class in terms of income. - I came up with some ways of making money. I'd sign, first I'd sign with my dad and then I went out and enlisted other people in the neighborhood and I offered a service for 52 weeks of the year, all year long. I would mow your grass, rake your leaves and shovel your driveway and walks but it was a fixed price per week. And turned out to be a good marketing approach because, you know, there's a lot of weeks when it doesn't snow and you don't - in the spring, you don't have to mow the grass yet and the snow is gone and you get, you know, a month or two off sort of there. And you're getting paid all the time. So I'd go out and sign up my neighbors to this annual contract.
RW: Well what got you interested in engineering?
DH: I guess, you know, genetically I'm an engineer. If you're to take a tissue sample, you'd find my DNA has engineer in it. I was always fixing things, always putting things together and so I just knew I wanted to be an engineer. I thought about being a mechanical engineer and I got a job while I was in high school - my senior year of high school - as a draftsman. I took drafting in high school and at a camshaft manufacturer. And I – the Chief Engineer says well what are you going to do? Well I'm going to go into engineering, go to college, go into engineering. And he says what kind of engineering? I said well mechanical. I mean, it's Michigan . It's automotive and he says you ought to look at electrical. So I started thinking about electrical and talked to one of my professors in junior college and decided electrical was the right way to go. That was the future.
RW: So where'd you go to school?
DH: I went after two years of community college, I went out to Michigan Tech, up at Houghton , Michigan and it was - I remember going up there after two years of community college. I needed 222 quarter credits to graduate and I transferred in and I had 88. And I go to orientation and I had a two year scholarship because I had no money. And I go to the initiation and they say well, look at the person on your left and right and one of you will graduate - one of the three of you are going to graduate. And I said oh wow, that's - that's kind of tough. And then he said look at your row, there's about ten people in your row. One of you will make it in four years. And I thought oh my God, I'm going to run out of money in two years. And I don't know what I'm going to do. And so I went twelve months a year. I went summers, took the summer quarters and did my sophomore classes the first year and my junior EE classes all in the summer, then my senior classes the last year and then my electives the last summer. And my objective was to be in the top ten percent and I was number seven out of sixty-seven. So I made it on a rounder!
RW: So what was your first job?
DH : I went to Raytheon in Massachusetts , interesting - because of Sputnik and all the interest in electronics, there was lot of people interviewing engineers that year. And so I decided that since I'd never been more than five hundred miles north, south, east or west from my home, that I was going to interview the United States in my senior year because I didn't think Muskegon, Michigan was the right place to - the economic opportunity wasn't there. So I took fourteen job interview trips in my senior year, went all over the United States, decided that I was going to go to graduate school but I was going to have to go part-time because I didn't have any money. So I selected California as the place I wanted to live. And so I went to Boston . I went to Northeastern because that was a place where I could get my Master's Degree. So I was working with Raytheon Communication and Data Processing Operation on data acquisition systems. It's 1965. First - one of the first integrated circuit machines there, TO5 cans - I remember burning out a flip-flop one day that's in a TO5 can and finding out it was $56 for that flip-flop that I just shorted with a scope probe and burned up. And we were gathering actually data from a bank of filters and a receiver, collecting information on missiles that were launched as they went through the ionosphere. They emit certain signatures and collect that data and then detect, you know, what's happening relative to - we were obviously interested in continental ballistic missiles and detecting their launch. So that - that was the data system.
RW: So it's interesting, you've never designed logic with individual transistors.
DH: Not since college. First machine I worked on was integrated circuit. One of the very first integrated circuit machines started out with Sylvania SUEL logic and - and then shortly after that to TTL.
RW: So then how did you end up at Honeywell?
DH: Well I was at Raytheon for almost four years. And I was very interested in designing in the emerging minicomputer business. And I was designing mainly special purpose systems. So I was one of the system engineers on designing the FAA and router traffic control system of that era and - and then after I'd - we'd designed the architecture of the system, I went on to design the display processor. And when that was done, I thought well, you know what, I really wanted to do was design the CPU. And so Computer Control Corporation which was acquired by Honeywell became Computer Control Division and made embedded processors. That is industrial control, building controls. And they wanted to start a new development and they were looking for someone to lead the development. They had a 16 bit line of machines, the 116 being the original, 516, 416, 316, 716 later. And that was the processor, of course, in the IMP in the first node of the internet. And they - it was a 16 bit machine but Digital Equipment Corporation had a 12 bit machine, a PDP-8. And when we were against them on performance, we'd win but when we were against them in cost, we'd lose. There was a bunch of 8 bit machines that came out in the '60s. And the idea was well we'll come out with a 8 bit machine and we'll have a 16 bit machine and an 8 bit machine. We'll bracket Digital. And so I was hired to design that 8 bit machine. Of course, DEC was busy doing their own 16 bit machine and the PDP-11 at the same time.
RW: That - that's interesting because that's the level of the first microprocessors, 8 bits.
DH: Yeah. That's right. In fact, I'm designing this machine. It was inside of a 3 C's there were code names and there was always X something, X - and my machine was the X12, this 8 bit machine because that was the next number in the sequence. And I'm designing this machine and we had a memory development group. Bill Jordan ran the memory development group. And so I would be in charge of the processor design but I would go to the memory development group and contract with them to have the memory designed. So I'd go into Bill in need of core memory for this machine. And Bill comes into my office one day and he says you know, he says Dave, I'd like to put a semiconductor memory in this machine. And I said what are you talking about? And he says, you know, there's this new DRAM by Intel and I says what - what's a DRAM? And he says well it's a Dynamic Random Access Memory. And I said so what's dynamic about it? And he says well it stores the charge on a floating gate. And, of course, you have to refresh it. And I said what do you mean you have to refresh it? And he says well you have to go back and read it and rewrite it periodically. And I said how often? And he said every millisecond. I says the machine's not going to do anything but refresh itself. No, no, no we'll have special control circuitry to do that. I said this, I don't know, this do; and so what happens when you turn the power off, Bill? Well you lose all the contents. I said well core memory, you can turn the power off and turn it back on again and reboot it. He says well no, we can't do that with this. I says no way Bill. We're not going to do a semiconductor memory. So we go off and we - and Hank Bodio is designing my core memory. And Bill Jordan later goes to Intel and runs the system - memory systems business there - comes to me and he says well, you know, I have my own little research budget. And I would like to design a 1102 - at that time it was called - 1K DRAM memory, that's plug compatible with the core memory. And I'll do it on my own funds. I said well, help yourself. I don't care. I'm not going to count on it. And so Bill Rigitz was in charge of that design. Of course, Hank Bodio and Bill Rigitz both wound up at Intel later. And so, sure enough, I had a core memory and I had a semiconductor memory. And one day I was following the development of both but mainly the core memory which was working fine. And I knew the semiconductor memory wasn't working and so one day, Bill Jordan comes over he says you got to come see the memory. It's working. So he's got the memory over. And I go over and here's a printed circuit card and he has taken - they've taken and wired - soldered a wire to the lid - the gold lid of these white 1102 packages all the way around it - wire's going up this power supply. And I said well, what's that? And he said well we - we discovered that you have to back bias the substrate and you see the lid is connected to the substrate. And so by connecting a voltage three volts more negative than the most negative voltage, we could back bias the substrate and then the thing works. I said well that's great but, you know, you don't have another pin. You've used all sixteen pins. Of course, sixteen pin DIPs were the most common. I said that's not going to work. And the next day I get this memo and the subject line is "Industry Standard 18 Pin DIP". And I go back to Bill. I said, Bill there is no 18 pin DIP. How can it be? Oh there will be because we're going to make one. And I said oh that's never going to happen. Of course, it did happen. And so I - that's my introduction to Intel. I met Bob Noyce at that time when he was out on a visit and became aware of dynamic DRAMs, Dynamic Random Access Memories. And the 1102 in its 18 pin version was called 1103.
RW: And Honeywell was a big driver in that.
DH: Yep, that's right.
RW: So what then led you to go to Intel to be the memory guy?
DH: So - Computer Design magazine had come to me and asked me to write an article for Computer Design magazine on architecture. And so I'm trying to think of what to write about and I kind of watched, you know, the first integrated circuits that I'd worked on. This was like five years earlier and then how we get more integration and more integration and now we had 4 bit ALU's that we could cascade to make 8 and 16 bit ALU's. And we had 4 bit registers in a single chip. And so I wrote an article about the impact of semiconductor technology on minicomputer architecture. And kind of like the fourth guy that discovered America , I was the fourth or fifth person to discover Moore 's law. I mean, Gordon had already published and I didn't know it. So I did these charts. Russ Henzel and I, one of the guys that I worked with wrote - co-wrote this article and we'd get together on the kitchen table at his house and we're plotting out, you know, well what is the trend here and what parts, you know, what data can we plot in terms of showing the progress in semiconductors. And then we're postulating what that was going to do to minicomputer architecture. And, of course, it was going to validate the design of the X12. I mean, that was a given. And so I'd written this article and that got me kind of interested in semiconductor technology trends. And one day I'm walking by - Russ and I worked for a guy by the name of Ike Templeton and he had a cork board, a bulletin board outside of his office. And he used to cut articles out and tack them up on this board. And I walked by one day and I'll never forget, here's this article about the Intel 8080. And I thought, I don't know, there's Intel 8008's - sorry 8008. And I thought, holy cow! And I went right over to the chip purchasing group, the purchasing organization that would have the data sheets, etc. And, of course, because of our relationship with Intel on the DRAM memory, they had a data sheet. So I get the data sheet for, I mean, immediately read it and I - wow, this is really rudimentary. This is a very, I mean, this is a very simple computer. So I'm not in danger now but this is big. I went home and told my wife at the time, I said I learned two very important things today. She says what's that? And I says well, first is my job is going to change industries. Said what do you mean? I said well high volume, low cost computers that I design will be done by semiconductor companies, not computer companies. Said oh okay, that's interesting. She said what's the other one - other thing you learned? I says my job's going to change coasts because those guys are out west and I was like whoa, that was important to her. And so I knew I - she said well when are we going to move? I says not right away. I mean, the - I'm looking at this and I'm looking at the complexity of that processor and how simple it was. And - and it wasn't going to displace the things I was doing at the time right away. But I says this is going to happen in a few years. I mean, it's not going to be that long before this happens.
So I was at, you know, busy designing minicomputers and, at that time, Honeywell bought the GE computer business. Big change. I'd made myself a name. I'd won the Harold W. Sweat Engineer Scientist Award which is the big engineering achievement award that, you know, puts you kind of in a Honeywell Fellow sort of a status. And those guys tend to stay around for a long time and so everybody assumed I was going to stay at Honeywell. And when Honeywell and GE merged, they created an architecture task force. And the task force was to come up with architecture for the combined product lines. And they broke it into entry level, midrange and then high end. And they had a group focusing on each. And I got put on the entry level architecture committee. So we went to Paris and we met for a couple months, worked at Honeywell Bull in Paris and then we went to Phoenix for a month and we went to Oklahoma City for a month and we went back to Massachusetts for a month. And there were a lot of different proposals put out on what we should do. And I said it's simple. Intel then - by then I knew - the 8080 wasn't out yet but I knew of the 8080. And I said Intel's coming out with this new product, the 8080. We should just take and put that on a print circuit board with memory and IO, their large mother board, put switches and lights on the front like we always do on our minicomputers and small business machines and put it in a nineteen inch chassis and sell it as our product. Well the guys in Oklahoma City that did peripheral controllers had designed a controller out of a bunch of logic - custom logic chips. And there were like eight chips or something and I said - it was called Basic Logic Unit or BLU. And they had proposed New BLU which was going to be like three chips instead of eight chips. And so they advocated that we not use 8080 but we use New BLU. And I lost the argument. They went with the Basic Logic Unit. They missed the opportunity to be the MITS or Altair or Cromemco or whatever - whatever to create one of the earliest 8080 machines. And I said it was a geopolitical decision. The government of France was involved because they didn't want to lose any French engineers. The government - Italy was involved because Honeywell Italy , they didn't want to lose any engineers. The guys in Oklahoma City were politicking up the organization because they wanted their technology - proprietary technology. And I says ah, this is, you know, this isn't going to work. Well so it's about time for me to leave Honeywell and I happen to be on a airplane going to IEEE Architecture Workshop to give a paper at Lake Arrowhead in California . And the guy sitting next to me is doing - had done the stock offering at Microdata in southern California . He gets - starts talking to me, hooks me up with the VP of Engineering. Couple weeks later, I'm in southern California as Director of Computer Development for Microdata, working on a 16 bit machine called the 3200. And stayed there actually until I went to Intel.
RW: But you were the memory guy at Intel.
DH: So that was my first job. So when I was down to Microdata - when I went to Microdata, the guys - there were like fifteen or more people from Honeywell that had gone to Intel. The whole memory design team had gone. Bill Jordan was running memory systems. Bill Rigitz was designing the 4K - the 2104. Hank Bodio was developing memory systems. You know, there were a whole gang of people that had gone. And because I was a Sweat Award winner, they figure well he's probably not going to leave. And when I went to Microdata, they all started trying to hire me. And so I wound up only staying there for a year and four months before I went to Intel.
RW: But you were the DRAM guy.
DH: So the problem at Intel at the time was that the 1103 basically didn't work. I mean, it was a very flaky design. And when it didn't work, customers would call up and say “stop shipments because we don't need anymore of these because we can't make our computer work”! It really fell into a good spot and a lot of people had planned and used plated wire magnetic memory. So you put a oxide, ferrous oxide coating, magnetic coating on the wires and at the intersection, you magnetized that coating. And that becomes the storage mechanism. And they were using those because they were able to make them very small and therefore very fast and they were typically used for control store for their microcode. And basically plated wire just wasn't production - couldn't make it work in production. 1103 happened to be the right speed in the right place. And so it got stuck into a lot of control memories. Hank Smith had been in charge of applications. Hank left and went back to New York to do whatever he did next. And they needed somebody basically to replace Hank. And Ed Gelbach had the job of hiring that person. And he obviously needed to know about computers and he needed to know about memories. But he needed to know them like how to make them work. And when I had been at Microdata, we had a 6002 AMS - AMI - AMI or is it AMS - Advanced Memory System - AMS made a 6002, 1K DRAM, competitor to 1103. It had been already designed into a machine that I had to make work and I had to fix that memory system. And so got that working and got it in production and delivered. Well Gelbach heard about me from the - the ex Honeywell guys and - and took me up for an interview and made an offer to me. And I just was going through a divorce and had filed for divorce and my kids were living in Southern California and I turned the job down. I said I want to stay with my kids. And Gelbach later calls me up and says, you know, I'd like to talk to you again. And I said well, you know, we already went through this. I, you know, I got kids down here. I don't re - he said well come on up and just talk to me. Well I can't. Why not? Well I got a date this weekend. He said bring your date. I says bring my date? He says yeah, I'll pay for it. I'll pay for your weekend. The only thing is you got to meet me at the San Jose airport and talk to me. So we sat there and talked for an hour and a half while my girlfriend shopped. Then I went away for the weekend. I thought that was pretty good.
And made me another offer. And I thought seriously about it and turned it down. About two months later, month later, I get a call from Gelbach. I said oh no, not you again Ed. He wants to talk to me. No I don't want to talk to you. Well he says I'll come down there this time. He didn't want me to come up. I said I don't want to come up. He said I'll come down there. We'll have dinner. I said no, I don't want to have dinner. I mean, we've been through this Ed. Okay, how about have a drink with me? Okay, I'll meet you for a drink. Disneyland Hotel. We meet like 6:00 . 11:00 we leave. We haven't had anything to eat. We've just been drinking. And he tells me a story. He says, you know, the deal is that I have to get Les Vadez approval on the person I hire because this person's kind of working between engineering and marketing. And so the deal is I hire them but Vadez has veto power. We have interviewed forty-three candidates. You're the only person that he and I can agree on. I have to hire you. So here's the deal. I'll fly you back and forth. You fly up Monday morning. You fly back Friday night. I'll pay for all your airline tickets. I'll pay for your apartment up here for you to live in during the week. You know, you got to do this. You don't see your kids during the week anyway do you? I said no actually I'm working very late at night and the kids are in bed by the time I, you know, so I don't see them. You're right.
And so I wind up taking the job at Intel. So I showed up before the 8080 was introduced and I said, by the way, you really want me for the microprocessor because that - I mean, I'm really a computer guy. I can make memories work. And one of my characteristic is I basically can always make, if it's broken, I can fix it. And so the initial job was fixing, you know, when customers called and said stop shipments, my job was to get shipments started by getting the computer working. And I remember I hadn't been there very many weeks and Gelbach comes to my office and he says, go to the airport. I need you to fly to Burroughs. And the flight's at so and so time. We've booked your ticket. You need to go now. I said well I need to get some clothes. He says no. Buy the clothes when you get there. Go to the air - you're just going to make it. Go now. So I go to the airport, I get on the plane, fly out there. Salesman picks me up, of course. He's panicked. His commission's in danger. Burroughs machine doesn't work. Takes me over to the plant and I start working with a Burroughs engineer on the design of this 1103 memory that, you know, why doesn't this work. Well everybody was making the same kind of mistakes. We got it working and had to reroute the power and change the decoupling capacitors and make sure we got the clock timings perfect. And got it shipping again. And this wind up repeating a few times. And so I went to work - I hired a couple engineers and a technician and started doing some experiments with 1103 in the lab. And it became very clear that there were only three things people did wrong and they always did those wrong. They didn't distribute power and ground well enough. You really needed to have a four layer board. They're trying to go on the cheap and do two layer boards. They needed a power and ground plane because you could demonstrate to them they didn't have power stable at the corners of their board. They didn't have adequate decoupling capacitors because when that thing switched, when that clock went down, it pulled a whole bunch of current and you'd get a big spike on the power line. And you needed a good high frequency decoupling capacitor close to the chips. And then the last thing is the famous TOV. There was a pre-charged clock and a clock. And they had to cross just so. And you could get that at the generator of the clock but delivering it to all corners of the board got to be a problem if you weren't careful about the layout. So I wrote an ap note about how to design with 1103 and we hired some Field Application Engineers and then, you know, they were in business. It largely took care of those three problems. Everything was good. So I started getting involved in product definition, working with the engineers on what we, you know, what we should do in the design. And so soon I had product marketing and application engineering.
By the way, when I showed up, one of my conditions in accepting the job was - was that I wouldn't come until we had customer acceptance of the 3200 from our first customer which happened a week ahead of schedule. And the second one was that they change the job title from Manager of Applications to Manager of Application Engineering because reporting to the VP of Marketing and Sales was a big insult to me anyway. And I had to maintain my engineering integrity and I had to have engineering in the title. That was one of the conditions.
RW: Well Intel really got behind on the 18 pin versus the Mostek 16 pin.
DH : Well I was a 22 pin. You know, it was - was it 4/10 ths or 6/10 - it was a wide package. They had a 16 pin dual inline. They multiplexed the addresses and we had a straight 4K address so it's at 12 -
RW: I think it was a 4K DRAM.
DH: Yeah, 4K DRAM. Right. Was a 2104. It's - 4K is a 10 and - and 4 - 10 and two 12 - 12 address pins.
RW: At - at any rate, that was a - a big catch up to-
DH: Intel never recovered in the DRAM business, never recovered. We eventually designed a dual-inline multiplex address part and then we fought it out at the 16K level – the, 2116 and there was an issue with the output drive on our part versus Mostek's part. And Intel never really recovered a leadership position after the 1K in the DRAM business. But by this time, I'd moved over to the microprocessor business, so in '76. So I'd joined in February of '74 - in May of '74, we introduced the 8080. And - and then in '75, I had product marketing applications and then in '76, I left the DRAM business to become Marketing Manager for the microprocessor business. And I did that for two years initially under Bill Davidow and under Les Vadez. And then in '78 became the General Manager for the microprocessor business.
RW: So what was the success of Intel was really so much based on the adoption of the 8088 by IBM? Why did they do that? It was not the best part.
DH: Yeah. The - the early days of the microprocessor were quite interesting. I - when I took over, we had had a few challenges already. The 8080 was quite successful and then the Z80 came out and basically took away the business. Our 8085 never was as popular as the Z80. We came out with the 8086 and, of course, after that, Zylog had the Z8000 but the 68000 came on the scene. And it was a big competitor. Well IBM; we had won some designs in IBM in Boca Raton . The System One and the Displaywriter which was IBM's answer to the Wang word processor. And that turned out - and it's an 8085 design win - turned out to be a major design win. Jack Carston had joined Intel and I got to give Jack credit because out of TI - his father had worked for IBM and at TI he'd made a big effort getting business at IBM. And the belief at Intel at the time was IBM is totally vertically integrated. They make their own IC's. They're not going to buy from us. And Jack says we can win business at IBM and put salesmen on it and created an organization and probably as a result of that as much as anything, we had won this design win at the Displaywriter in Boca Raton .
And about this time, IBM decides that they need have a PC, that the PC - Apple is making some inroads and that they needed to develop their own PC and was it Carey was the CEO at the time. I've heard from talking with the people that were there at the IBM site later about Carey's involvement with Estridge and sponsoring Estridge to do the skunk works project down in Boca Raton . And he was given some money and told to go develop a machine. And he rented this space in a empty supermarket, a shopping center off the IBM campus and went out and bought - it took some these engineers that had been working largely on the display writer and bought them all an Apple II's and sent them home for a month and says I want you to learn this machine inside and out and I want you to show it to your kids and your wife and friends and because when you come back, we're going to design a machine to beat the Apple II. Well they came back and they said well we'll need a processor here and so they were going to use the Z80. And we had a salesman by the name of Paul Indaco who was calling or maybe it was a Field Application Engineer - he was a Field Application Engineer at the time - and he was calling on them. And Dave Gellatly was my Marketing Manager. I was running microprocessor business and Earl Whetstone was the Regional Manager of the salesmen on the account. He was the sales guy. Indaco was more the Field Application Engineer. And so we'd get reports back about this project but it was super secret. IBM was always very secretive in those days but they were especially secretive about this. So I would get these reports back from the field that the project was going on and they were going to use the Z80.
So we went down and made the pitch and said, you know, how can you beat an 8 bit machine with an 8 bit machine? You need to get - have a 16 bit machine. Of course, we had the 8086 coming along. And - and our 8086 was out. And so they said well okay, we'll start working on it. And so they decided that yeah, they really did need a 16 bit machine but they should look at which 16 bit machine they should use. Since they were using the Z80, they thought that maybe the Z8000 would be a good machine. Trying - so I'm trying to sell the 8086 versus the Z8000 and Motorola comes in and they fall in love with the 68000. And they want to use the 68000. And I'm trying to get them to use the 8086. And they start running into cost problems. They had very tight cost goals because they were going to go up against Apple. And obviously selling through the IBM organization, they had to have low product costs. And Motorola just couldn't get the 68K peripherals to work. Our peripherals were in okay shape but not tremendous shape. Zylog was having trouble with peripherals so by peripherals, the clock driver, the DMA controller, the timer counter and a parallel IO, serial IO chips. And we had this design. When we designed the 8086, there someone, I think it was in Israel, or someone had come up with the idea of because we had so much trouble with the 8080 peripherals and getting them to work, because we had all kinds of bugs and technical problems, that as a safety measure, we should design the 8086 so that you could double clutch the IO - that is to multiplex the IO as an 8 bit IO and send it - you know, do two accesses, two bus cycles to get the 16 bit quantity instruction or data. And so when we designed the chip, we put those circuits in there so that we could do that. But as we were moving along, we were feeling more comfortable about our IO chips and we had never actually bonded it out. And IBM's busy trying to hit their - their number and I'm not sure who - somebody says well, what if we took the 8088 and the 8085-8080 peripherals. We could blow them away on cost. Because the peripheral chips were sucking up all the money and that's where all the risk was. And so we went down and we pitched the 8088. And we said it's going to get to five bucks. Won't be that right away but it'll get to five bucks. And Bill Davidow was actually the one who said we got to promise them five dollars eventually. And, you know, it's a lot more in the beginning but it's going to get to five bucks. And, of course, the IO chips were second sourced by AMD and probably others. And so they were somewhat commodity priced by now because they were in the last generation. And that was - that was what it took.
They wound up making the decision go with the 8088 and the 8080 peripherals, the 8085 peripheral chips. And that was a big day when they made that decision. So IBM was so secretive in their development that they wouldn't let us see the computer. And so the original IBM PC, so they set up this lab and they had a curtain and there was a table - it was just a regular folding, you know, leg table and it was set up on one side. We could put our equipment over there. So we had a development system, our in-circuit emulator, we had some oscilloscopes and things like that and logic analyzers. And they would hook it up to the machine on the other side of the curtain. We'd say okay, we want to see the clock signal at the clock generator. Okay lets see the clock signal at the 8088 input lead. And we could look at the displays. They'd put the in-circuit emulator in. We could capture information. But we couldn't see the machine.
We never were allowed to see the machine that was behind the curtain. They were that secretive about it. And, of course, the fact that we were even there - they weren't supposed to - there were very few people that could be told. David Gellatly my Marketing Manager, I could be informed of it. I could tell Grove about it. But we were supposed to keep it very tightly controlled.
RW: But that was such a key ingredient to Intel's success.
DH: Yeah. Yeah. From the beginning, we always were trying to get into a PC. And we didn't get in the Apple and kept trying to call on Apple and get them to use the Intel architecture and - and continued after this. There's some interesting stories there. And we were in things like, you know, Cromemco and IMSAI and some of these machines but these were like hobby machines. They didn't have the software to be a personal computer. And so we had - I felt like we had one shot and that was IBM. And we had to get that one.
RW: And from that came so many successes.
DH: Oh yeah.
RW: - Intel successes.
DH: Yeah. It's - history would have been different if we hadn't won that.
RW: Well how - what was the relationship with the people that were making IBM clones?
DH: Well, you know, in the early days, there weren't any clones and Compaq was really the first company to come out and they came to us and we - we worked with them. And then the clones started spreading and in later years, you know, the Taiwanese and around the world and Asia, Europe , everybody had a IBM compatible machine. And so there was very interesting relationship always between IBM and then IBM and Compaq and the rest of the companies and Intel. IBM certainly won an advantage with the PS2 they - they clearly had an effort to take the PC back. OS2, the operating system, special deal with Microsoft that gave them advantage. And with PS2, they - they had negotiated with Intel an agreement to second source the microprocessor, the 8088 in 70, at that time I worked for Jack Carston who worked for Andy Grove and Jack had negotiated a second source agreement that they could be their own second source. And so IBM was busily designing a CMOS 286 at one point in time. And that was going to be their direction and they were going to leave the Intel processors and have their own processors and in Manassas , Virginia , they had a design group designing CMOS 286. And I kept going to IBM and saying the fastest way to a CMOS 286 is a 386 because a 386 is CMOS. And it - you don't have to use a 32 bit instruction. Just running it like a 286 if you want a CMOS 286. But no, they had to develop their own. Of course, it never got finished and wasn't - you know, they - they wound up missing the 386 generation. It was September, I think, of '85 that Compaq came out with the Deskpro 386 and it wasn't until like August the following year that IBM had a 386 machine. And, of course, during this time, Compaq just increased their market share very dramatically and, of course, they were the first real legitimate clone.
DH: It was September, I think, of '85 that Compaq came out with the Deskpro 386 and it wasn't until like August the following year that IBM had a 386 machine. And, of course, during this time, Compaq just increased their market share very dramatically and, of course, they were the first real legitimate clone.
RW: Yeah, that was - that was fantastic. Well then - then came the 486. And tell about that introduction and those wins.
DH: Well before I go into the 486, I think the 386 is the most interesting story. The programmers all preferred Motorola or Zylog because they had 32 bit addresses. And we had a 16 bit address, segmented architecture. So 64K segments and you wanted to change segments, you got to do a jump shift of the new segment and then you could only address within that 64K segment. Big pain for the programmers. They wanted this nice, flat linear 32 bit address. And we had come up with a plan for a 32 bit machine called eventually the 3200.
RW: The 432.
DH: The 432. Well in its eventual name. That's by the way, this is an interesting story in itself. When I was first early at Intel, there was a proposal made for a product called the 8816. And the 8816 was this object oriented architecture that was proposed by some of the software guys and really promoted by Terry Opdendyk who was head of the software development. And the proposal we would - after the 8080, that we would develop this 8816 is our 16 bit machine. It was only later that the 8086 was created as an interim product to fill in till we got the 8816. And so this is object oriented machine was proposed. It was very radical in its design. And so Gordon Moore asked for a task force made up of Ted Hoff, me and Terry Opdendyk. And were to go evaluate the 8816 and make a decision on whether we were going to implement this. Well we all read the specs and we got together and we were totally in disagreement. Terry was just religious about it. I said there's no way this machine's going to ever work. It goes to memory 68,000 times before it ever does anything. Everything's then directed through so many tables, by the time you ever do anything, you know, it's going to be just way too slow. And Terry says but it's going to be programmed in this high level object oriented language and that's going to make up for all of that stuff. And you're just a hardware guy and you don't understand. And I said no I understand because I know what it's going to do with it execute on instruction and how many clocks it's going to take. So we couldn't agree. So the executive staff asked for a review of - a presentation from the committee. We said we can't agree. They said well then we want to hear all three points of view. And Gordon invited Carver Mead to this meeting. And we go in and - and the three of us present. Terry, this is the greatest thing since sliced bread. It's going to revolutionize the world. I stood up and said this thing is such a dog, it's so slow, it's so complex, it's never going to do anything. It's going to go to memory forever before it even does an add. And Ted stood up and said well on the one hand but then on the other hand. And - and so we got like, you know, two people very opposed and one guy in the middle. So Gordon turns to Carver and Carver says I think we ought to do it. I think we should give it a try. And the 8816 got born.
Well I'm running the microprocessor business later and I said I'm doing marketing, you know, work and we got to have an interim machine. We got to have something because this thing is just slipping out in time. It's going to take a long time to get the software ready. So created the 8086 which then became very successful. And then the 8816 became the 8800 because it became a 32 bit machine because its object orientation really gave it - you could look at this thing and - as a variable address linked machine. And it kept on going and kept evolving and then it, you know, wasn't - it missed the 16 bit generation so it became a 32 bit generation. It was our hope there. And I said well I got to have an interim machine until that's done. And I got to figure out a way to get a 32 bit machine but I can't start over again. I hired this guy by the name of Glen Myers out of IBM research. I had met him somewhere and I was very impressed with the guy. And I hired him to be on my staff. And I says Glen, we got to figure out a way we can run the 8086 code and still have a 32 bit address so that it's upward compatible. It's got to run the binaries.
And he comes back about a week later and he says you know this is really pretty easy. I said what do you mean? He said well, you know, the 8086 because it really was related to the 8080 architecturally, has a bit in it in the op code, the 8 bit op code that says whether the address is a 8 bit or a 16 bit displacement. And he says we can make that a mode switchable bit so that we could enter a 32 bit mode with a 8 bit op code and here's the op code by the way. We've got some extras. And we'll make a rule that every 32 bit routine needs to start out with declaring itself to be 32 bits. And from that point on, that bit will mean it's either 8 bits or 32 bits on 8 bits or 16 bits. And so you'll be able to run all the 16 bit - and then when you enter a 16 bit routine, the first instruction should be “we're 16 bits”. So it sets the mode so every subroutine will always - every subroutine has got this capability. And, by the way, until then, we'll run the old stuff by just whenever you exit a 32 bit routine; you just put yourself back into 16 bit mode in case you're going to some old software. And he also had an op code that would change it for one instruction only instead of changing modes. And he says - and I said, God, that is brilliant. That is so elegant, so simple.
And so that gave birth to the 386 architecture. And the thing that allowed us to move all the 286 code and 8086 code up to the 386 and run it directly and almost all the 386 processors ran nothing but 16 bit code. Because that's what was around. It wasn't really till the 432 came along or the 486 came along that we got the - the real 32 bit address code. And, of course, meanwhile the - the 8088 became the 432 and that wound up being introduced and we spent a lot of money and never made any money on that because it - well a number of reasons. One is it never met its performance targets. As I said, it had to go to memory too many times before it could do anything. And it was hard to beat the momentum of all that software. I mean, it - once the software started building on the 286 - 8086, 286, 386, it was - it was just too hard to - for anybody to go back. Became, you know, software drove the whole thing.
RW: Intel has tried to obsolete the X86 many times unsuccessfully.
DH: Yeah, that's right. With the Itanium being the most recent one. One of my last jobs at Intel was running the server group. And I, during '90 - so that was '94 to '96 and '92 to '94, I was in charge of corporate strategy. And I said we could go into the server business. We've captured the desktop business. We got a notebook business. We need to capture the server business. And - and so Grove said okay and you organize around notebooks, desktops and servers and put me in charge of servers. And so I was trying to win over these RISC - various RISC architectures - I'm kind of skipping over the RISC story, but HP was a large, you know, potential customer. And they had a design that they really liked and showed to our guys and our architect says yeah, this works. This architecture is a great next generation architecture. I said well the only way we can do it is if it runs the X86 software. And if we do this, we have to also do the 64 bit extension to the Pentium architecture. Having done it once from 16 to 32, I knew how to do it from 32 to 64. We just go back to that bit and we just take a new op code and we redefine the displacement as being 64 bits instead of 32 bits. And - and then you could do it in a compatible way so that it runs the old routines in 32 bit mode and the new routines in 64 bit modes. And so it was clear we had a elegant architectural approach. And I said whatever we do, we going to do this machine, we've got to do the 64 bit extension. And when I left in '96, that was still the plan is do the 64 bit extension.
Somehow the company convinced itself that what became Itanium would be the 64 bit solution and that putting 64 bits in the Pentium would somehow hold back the Itanium and they held back on the 64 bits. That opened the window for AMD. Probably one of the more serious mistakes Intel's made in microprocessors was to hold back the Pentium processor to make room for Itanium because you can't do that in the marketplace. You can't hold back technology. You've got to deliver what technology will allow or somebody else is going to do it.
RW: Yeah, it's fascinating. Well the Itanium is - doesn't look like it's going anywhere. Is that right?
DH: I think I've said that you know, its days are limited clearly. I think it'll go to the microprocessor graveyard.
RW: Yeah, the X86 goes on. Well let's go back to the RISC-CISC wars. I mean, everybody thought that RISC was going to take over because in the history of computers, there's always been the new architecture versus the old. And the established companies always stayed with the old architecture because it was a cash cow. And then the new architecture would come in and take over. And that happened in mainframes. It happened in minicomputers but it didn't happen in microprocessors. We all thought that RISC was going to be the answer.
DH : Minicomputers were a different economic model, not just a different architecture than mainframes. It was a totally different economic model, a different approach. It wasn't about the instruction set. Yes, they did have more scientific oriented instructions as opposed to more character oriented instructions in the IBM line. This was about implementation and there was a raging debate inside Intel. From the time John Cocke at IBM started talking about RISC, whether we should do a RISC implementation and what's the difference between our implementation and their implementation. We had - when I had started designing computers, excuse me, when I started in computers, they were hardwired and microcode came along and my first microcoded machine was at Microdata. And I always thought that well we're adding a level of indirection here. I mean, we're going to memory during instruction execution. If we could just do this in hardware, we'd do it a lot faster. And when I came to Intel and, you know, look at the 8086, there's quite a bit of microcode and to save silicon in those early days, we were in a position of always putting stuff in microcode. But as we got more and more transistors, we started taking the more frequently operated instructions. And I remember looking at these charts that showed - we'd run different programs and we'd profile what instructions were used how many times during executing software. And could easily gain that data with-in circuit emulators and other software tools. And we'd go back and we'd reduce the number of clock cycles. I mean, that's what each generation did. So how did you do that? You took it out of microcode.
So when the RISC thing started becoming less academic and more commercial, when Berkley was doing their work, when Hennessey was doing his work at Stanford, we were already in this “reduce the clock cycles per instruction” and “profiling the distribution of instruction usage”. Still there was a big architectural argument amongst the architects. And I had these discussions about well how much faster can you make it? And, of course, with a RISC machine, you've got to do the complicated instructions in a series of instructions. So it's not just counting instructions. It's counting function and then that's a little harder to do because how well was it programmed and how good was the coding and how good was the compiler. And so the - the whole analysis, there was a big debate going on. And there were a set of people inside Intel that says we need to do RISC. We've got to break with the past. We got to do a RISC based machine. And then the main line of thinking was no, let's just continue to reduce the number of clock cycles per instruction. And John Crawford, I mean, I had lots of discussions with these guys - John Crawford would say, you know, “if we execute every RISC instruction in one clock, and we take those same instructions and we execute those in one clock and we just add some microcode to do the complex instructions, we'll have more transistors because we'll have the microcode storage and the control to make that operate. But it's not that many more transistors. And by the way, Moore 's law's given us a doubling every two years anyway. And a doubling in performance every eighteen months. And so the advantage is not that big and to go incompatible would mean all new software”.
So we did a lot of analysis inside and the conclusion I kept asking and, I would ask this question every six months - if you didn't have to be compatible, you had the same silicon process, you have the same CAD tools, you have the same design team, and the team has already designed like three or four generations of this architecture so they're experienced on it, so everything's the same, how much faster can you make a pure RISC machine given the same silicon process number of transistors. And I would ask different groups because I had a design group in Israel, I had a design group in Santa Clara, we had one up in Oregon and we were designing X86 processors later on in Folsom. And the answer always came back, thirty to fifty percent. I'd say could it be a hundred percent? Only a few fringe people that didn't seem to have the credibility of the great minds would ever say it would be that much. And the consensus was thirty to fifty percent. So I said let's assume fifty percent. Let's suppose you could be fifty percent faster. If you're doubling in performance every eighteen months, that's not a linear scale but, you know, maybe that's a nine month advantage. So if these guys came out with a processor and everything else was equal and I think we got better silicon, I think I got better design too. I got more staff and we probably have more experience with this architecture but everything was equal, they'd have a nine month advantage. But these microprocessors take - Intel was averaging three years and nine months to develop them. A RISC machine would be a little faster but two to three years to develop. So if you got one design team - this comes out every two and a half years and you got nine months and then you spend a year and nine months being behind and then - and as our business was growing, I said well, what if we had parallel developments. So instead of just pipelining the machine, let's pipeline the developments. And we would start a new development, initially it was every two years and then it was every year. And by coming out with the new implementation - so we had several implementations of the 486 that came out, I mean, you would see several die sizes over a time with new processes and several implementations of the original Pentium.
And so we'd keep coming out and we might be six or nine months behind what theoretically we could do but by always, you know, it's a stair step, and by making those steps short and close together, we actually never had a performance disadvantage. Because we're vertically integrated, because, you know, the silicon technology guys were developing specifically for our processor, it - we had very tight coupling and they would optimize and the engine - the architects and the circuit designers, you know, the process guys, were working extremely close together and the CAD tools were working very closely together, I got a high level of efficiency in terms of getting Moore's law out to the customer in time. And by pipelining engineering and having new processors, new versions of the chip, one right after the other, I was always right on kind of the edge of what the silicon process would allow us to do. And these guys just couldn't afford to have four parallel developments going in at a time. And it turns out that we just kept adopting more and more of their RISC techniques. Whenever they came out with something we'd put it in. And so maybe we would be four months later getting it but we would continually add those capabilities. And they you know, theoretically, they may have had an advantage but it's an academic argument and, in fact, they did not have a performance advantage. And it they had it, it never lasted more than six months.
RW: I was really wrong on that one. I sold my Intel stock. I said Intel is a one product company and sold my stock. Well you mentioned AMD which has been a very interesting Intel-AMD thing through the years. So tell us a little about the relationship.
DH: When I came to Intel, I was like one of the only non-semiconductor guys. I'm a computer designer. I'm not a silicon process guy. And I thought differently. Jack Carston came in and he's out of TI in the chip business and it was a given to him that we needed the second source product. And he was running microprocessors and IBM insisted on a second source and to him that seemed like something he'd have to do and so he worked out a deal with AMD to second source the chips. And worked out an exchange program where they would develop chips and the exchange was based on a complexity factor formula so that, you know, you - all chips aren't created equal. And so the more complex ones got to be worth a lot more bartering power. Well AMD figured that out very quickly and although you'll never be able to prove this that the data would show that they were exceedingly good in making things like graphics controllers - very complex so they were rich in complexity factors and so that the stuff that they were offering up to get back our processors in exchange had a high exchange value. The only thing is the contract said we had to agree. So we would agree on the definition and then they'd go off and develop this thing and it would come back, you know, hugely expensive and just economically not viable. And I think they really believed that we had to bow down to IBM and provide a second source so we would have to accept them.
And I took over from Carston and I had said all along, I just don't get this second sourcing thing. I mean, I've been in the computer business for lots of years and we design a computer and the last thing we would ever do would be take it to a competitor and say here you have it. You take all of this engineering we've done over all these years and you just have it. And then - so you can come and compete with us at our customer - this doesn't make any sense to me. I don't know why companies would do that. Well customers insist on it. I said well, you know, I'm in a different business. And I'm in a software business. I'm not in the chip business. And I'm developing and delivering microcode on silicon. And I don't think that we need to do that. And so they came in with these huge - they had a hard disc controller and a graphics chip and they were unbelievably expensive and large. And there were other solutions out there and these parts weren't going to win. I said no, I'm not going to take them. And then, of course, you know, AMD got very upset that they'd been snookered in and set up and all this sort of thing. And I said well, read the contract. I don't have to take them. And I didn't accept them. And then - then we had the big breakup and big fight between Intel and AMD and our agreement called for arbitration and there was this huge - I spent fourteen days testifying in front of an arbitrator who, of course, was being paid by the day, a retired judge. And went through this huge, long process before finally got all dissolved. And, of course, then they started developing their own. They went out and bought a couple - what was it NextGen, I think, they bought to get some expertise and started developing their own versions of the Intel architecture.
RW: Well you and Albert Yu - what was that - that relationship? What was “two in a box”?
DH: Well, Albert was, you know, you know, Mr. Engineer. He was a silicon background. He had been in technology development and quality. And really knew the silicon process side and we clearly needed a tight integration. I was the architect, you know, instruction set, you know, software interface, you know, kind of computer guy. And Albert really understood the silicon process and the silicon technology and was from that group and could talk with the technology development. So he came in as my head of development basically at some point in time. And headed up all the development activities which allowed me to really focus more on the sales and marketing and - and - and business and customer kind of relationship. And we became “two in a box” which was popular at Intel at that time. And I was the boss but we worked very well together. He was the other guy at that position but he really focused on silicon manufacturing, technology development, CAD tools. I said folks want an instruction set, peripherals, IO, marketing, positioning and customer presentations and sales training.
RW: Let's talk about Intel inside. Who came up with - with that ?
DH: So it starts back in August of 1984, with the introduction of IBM's PC AT, it's second generation machine. And IBM had their launch event and they had; I believe I was the only one non IBM employee speak. There might have been two but I think I was the only one that was asked to speak. So we go to this big launch and Grove was already pointing out that, you know, our microprocessors really the thing that's making the difference inside. And this has got a 286 in it and, you know, I get to go and talk at this introduction. I came back from this introduction - I'm pretty proud of myself that we, you know, we got to point out - I mean, they asked the microprocessor company to say something. And Grove - typical Grove, he'd find the problem in the situation. He says, you know Dave, the chassis is the same, the memory's the same, the monitor's the same, the keyboard's the same, the mouse is the same, what's different is the processor. Who knows that? You know, people got to know that it's the processor that makes the difference. And I said, God, how in the world are we going to do that? Little while later, this was a few quarters later, I'm talking to my marketing guys and - and the advertising guys and I said how are we going to make people know that it's the processor that makes the difference? And I said, you know, there was in those days an ad of a Pepsi blind taste test. And they'd take a Pepsi and a Coke and put them in a plain glass and then people would taste them and they liked the Pepsi and I said we got to, you know, people clearly - our processor passes the Pepsi blind taste test. That is, if you sat down in front of an 8086 machine and a 286 machine and everything's generic and there's no labels on it and, you know, you got the same monitor, same keyboard, same - everything's the same, you're going to move right to the 286 machine. We got to let people know that it - it's not the PC AT, it's the 286 that counts here.
And a guy by the name of Chip Shafer had left Regis McKenna and created his own ad agency and he came back with this ad idea that was just bizarre in how aggressive it was. And I looked at the thing and I said oh man, we can't do this. We can't do this. And it was the Red X campaign. And finally I says, you know, we can do this. So I went to Grove said Andy, we're tight and we're very tight - Intel was very tight budget times. They were cutting budgets everywhere. And I said, Andy if I could spend five million dollars next quarter and make fifteen million, would you let me? He said it's a trick question. I said no it's not. I'll show you why. And I took the conversion from the 8088 to the 286, date of introduction of IBM's machine, percent processor shipments and showed how the 8086 to 286 conversion was an S shaped curve. And I took the conversion from the 286 to the 386 and I plotted it on this curve. And guess what? It fell right exactly on the curve. And I said, you know, if we could get five percent more next quarter because we charge so much more for the 386 than the 286, we could make fifteen million bucks more margin. And I believe if we were on this campaign, we can do this. Grove let me do it, but he insisted that it be done scientifically. So we picked twenty-four cities with twelve matches, a control group and the group that was going to get the ad. We bought billboards and we bought pages in magazines for that specific city because you could do that for, I don't know, three or four business publications just for those cities and over a six week period of time, we had the same page in those publications every day for six weeks. And the ads, the billboard was white, solid white, and had 286 on it. Had Intel in the lower right hand corner and had red spray paint X across the 286. And that ran in print and it ran in billboards. You know, people saw it on billboards first typically and they thought that it was graffiti. Somebody had put a big X across the 286. But then they'd pick up the print and it's printed. So who's doing this? Is Intel - I mean, Intel's the only name on here. Is it somebody else? And it got a big stir. And then for the next four weeks, the same exact billboard and the same exact page in the paper, we had a white background, the letters 386 and then in spray paint SX was our low cost version of the 386 and Intel in the lower corner. And the whole message was don't buy the 286. Buy the 386.
Well at this time, most of our revenue's coming from the 286. But I got 386's to deliver and they're much more profitable. And the - we had like a fifteen percent shift in marketplace. You know, it's - instead of fifteen million, it was like forty-five million. It was a huge step in conversion because everybody buying a PC says well Intel is saying don't buy a 286, buy a 386. And so we ran a series of ads called the graffiti ads which used spray paint for a while. And that idea was starting to get a little bit tired because, you know, the impact was gone. And in Japan , we had an ad agency Nippon Denso, who had run a campaign in print and on TV in like ten or fifteen second spots on TV called "Intel In It". "Intel In It." And it had a circle around it. And there were these little cartoon characters and they used to speak in Japanese. And that campaign was having a nice impression in Japan . And so, you know, that's kind of the context that's going on.
Meanwhile, when I first came to Intel, Intel's byline was “Intel delivers” because semiconductor companies when there was a shortage of supply which shipped to the highest priced customers and not shop to the others - they had overbooked and then shipped based on price. And the salesmen would say well if you give me a little more money, I think I can actually get the parts. Well Intel would ship based on orders. And first order in would be the first order out. And that was part of Gelbach's policy of building confidence. Well by the time we were into the '80s, everybody was doing that and “Intel delivers” didn't make much sense. And I'm trying to get the word across that the microprocessor is what counts. And I got the Marcom people to change our byline to "the computer inside". So it was "Intel, the computer inside".
Well, Dennis Carter was working for me at the time running Marcom and he comes into my office one day and he says you know Dave, I think we should bring "Intel In It" to the U.S. And I says, "Intel In It" to the U.S. and he says yeah, I think that program's having good effect in Japan . We should bring it to the U.S. And I said, "Intel In It", "Intel In It", that's Japolish. How about Intel and I turned around and took a piece of stationery because we used to have memos and paper in those days - out and on the byline at the bottom I put it down and I says, the computer inside. How about "Intel Inside". He says I'll test that one too. Dennis was always great in terms of running market tests, doing everything scientifically. And, of course, that was important to Grove's support; and it was the right way to do it. And so he went out and, of course, "Intel In It" bombed and Intel Inside was popular. And so we decided we're going to start advertising “Intel Inside”.
And we started trying to figure out well how are we going to make this happen? I mean, how are we going to get people to do this? And what we decided is we had always - at the end of every fiscal year, we would have to go to our customers who were planning their next fiscal year and give them projections on the prices of our semiconductors. And our basic philosophy was to reduce the cost of any speed of the microprocessor thirty percent a year but always introduce higher speeds at higher prices. And how much higher was kind of like what we thought we could get. And so we would typically go and say okay, the 16 megahertz was going to, you know, start at this price and by the end of the fourth quarter, it'll be thirty percent lower. So I said to Grove, I said what I think we should do is we should drop it twenty percent next year. Happened to be a very good year, we didn't have much competition. There was a lot of demand. AMD wasn't doing too well. But let's offer a five percent rebate.
And we will offer a rebate if they put “Intel Inside” on the box that they ship the computer in and on the front of the unit and if they put it in their advertising. And we had a split where there was a certain percent for putting it on the box and the computer and eventually moved to the advertising. But, you know, it had to be certain size, so much white space around it, etc. And the deal was we will give you back five percent of everything you spend and it'll be in the form of paying half the price of your ads that have Intel Inside on it. And of course, IBM and Compaq weren't very happy about doing that. They weren't going to ruin their brand, but, you know, our idea was, you know, we'll put money into a fund for you - market development fund - five percent of all the processor purchases and then you can draw on that to pay fifty percent of the price of the ads.
And the Taiwanese had been asking if they could put Intel or some Intel indicator on their box to show that it had Intel inside. We didn't want to get associated just with that but Intel Inside, you know, it's not an Intel box but there's Intel inside. That was okay. And so we went out and of course, the Taiwanese and the smaller brands all, you know, loved that right away.
IBM and Compaq says no, we're not going to do that. This is our brand. We're not going to do this. And we'd go back and say well, you know, you've spent so much money. You know, it's available to you. I finally went to IBM and Compaq and I - this was a little bit arrogant. I probably shouldn't have carried it this far. But I was a signatory for checks. We had to have two signatures - above a certain amount; you had to have somebody from finance and another officer. And I could be the officer. So I had a check made out for the amount of money that five percent; it's available to you.
I finally went to IBM and Compaq and I - this was a little bit arrogant. I probably shouldn't have carried it this far. But I was a signatory for checks. And I went in about two weeks before the end of the quarter to Compaq and I went into IBM and, this was like tens of millions of dollars - these checks. And the business is highly competitive and the margins from the PC are very small. And I went in and I held up the check and I says at the end of this quarter, this will be on our earning statement or yours. That is, you will make this much more money as a company or we,depending on whether you sign, you execute on the Intel Inside program. They no, no, no, we can't do that. I tore the check up and left it on the table. That was pretty arrogant. Come back the next quarter, I got the check again - do the same thing. That was the last time I did it because they both signed up for the program. It was so clear, you know, they see this big checks and this was all profit, it didn't make sense for them not to join the program. So, I mean, there was just too much money that was at stake.
RW: Okay, but eventually you left Intel? How come?
DH: Here's a story that I haven't told because, probably wasn't that important to tell it. But back in 1982, approximately, plus or minus a year, I'd taken my first sabbatical at Intel. In the early '80s, I was working for Jack Carston so it had to be between '81 and '83. And Jack came to me and says we're going to start this commercial microprocessor systems operation. Jim Lally, who's been running development systems is going to go to Arizona to do this and - or wherever he went - maybe it's Oregon -we're going to take the development system and make it into a business computer. And we want you to take over the development systems business. And it was a Tuesday and I said to Jack, I said well, hmmm, I have to think about that. And I said I'll get back to you on Monday. He says what do you mean you got to think about it? This is a good deal. You know, what do you mean, you got to think about it? And I said well, as long as you're pressing me, I've been offered the job of president of Fairchild. And I told who was it that at that time, the guy that was running it? After Slumberger had bought it. Anyway, I told them I'd give them an answer by Monday. And if I tell you yes, you're going to announce it tomorrow. And if I leave Monday, we're both going to look stupid. So why don't we just make sense and wait until Monday, you know, and I'll give you an answer.
Well ten minutes later, my phone rings, it's Grove. Get over here. You know, he's in Santa Clara , get in my car drive over there. Walk into his office. I know what he wants to talk with me about. He said let's go for a walk. Okay. I think we're going to a conference room because he's got an open cubicle. Right. No, past the conference room, down the hall, down the stairs, out the front door, across the parking lot, down to the corner, down the street. We're a block down the street, you know, like Andy where we going? Oh I got to talk to you. Okay but, you know, couldn't we just go to a conference room? I'm - where are we going? No, I wanted to get away. Okay, what - talk to me then. He says well, I've told Gordon and the board that I'm going to retire when I'm 55. I says no. I mean, you must - you're in your forties now. He says yeah but the board asked me for three candidates for my replacement and you're one of the three names I gave them. And I said come on Andy, I don't even report to you. I'm a level below you. He said well think about it. You're the right age group. You know, a lot of the guys that you would think be candidates will be too old by then. And so, you know, it's you and Jim Lally and Terry Optendyk. Those are the guys I, you know, and I says me? Me be the next CEO after you? You know, I can't - well anyway, we had a long discussion about this and - and I turned down the job at Fairchild.
But now I'm thinking different, I mean, that day my mind went to a different place. And so Grove keeps on going and every year he'd have to submit his list to the board and he'd tell me that I was still on the list. He wouldn't tell me who else was on the list because Optendyk had left and Lally had left. And - and - but, you know, you're still a candidate. And then Grove goes away. He's like 53 and there'd been some famous CEO who had announced he was going to retire and the day he retired, he appointed one of his staff to be the CEO. And everybody else on the staff quit because they didn't get the job. And so he decided he didn't want that to happen. So there were about five of us that he put a pretty heavy stock option on those would overlap this 55 th birthday. It'd be about two years into it's vesting. So every year goes by and I'm still a candidate and I'm thinking about this and Business Week runs an article about Grove's going to retire and they show House's picture. This is the guy that everybody's expecting to replace him. Grove goes away on sabbatical, he comes back and we have dinner one night soon as he came back. And we'd periodically would have dinners together.
And he says, you know, Dave I went away and I spent a bunch of time and I've been thinking about this. I decided I'm not going to retire. And I says oh, I think that's great. I think that's great for Intel. That's great for the stockholders and that's great for the employees. And he said well what do you mean? I said well, I don't think anybody - any of us could do as good a job as you could do. I said I have tremendous respect for your intelligence, for your business acumen, for your - your you're so articulate. You're such a great leader. I said I don't think, you know, the toughest job in the world is going to be the guy who follows Grove. And he says well, how do you feel though personally? He says you told me how you feel for the stockholders. I says, well I'll tell you how I feel. I feel like I have been training for a marathon and - a long time and - and - and - and I enter this big race and I'm running along and I'm starting to move ahead and I come up over this hill and I see the finish line and there's a crowd and the photographer's are all there and I look behind me and I'm like fifteen strides ahead of the guy behind me. And when I turn back, everybody's going to their cars and there's a sign that says "race cancelled". That it just got cancelled. And he says yeah I clearly understand what you mean. He says, you know, but, I decided this is what I need to do. And I said well, you know, I understand.
But know that some day, you know, I'll probably leave. I'm not going to leave now because the options are worth so much. If I just stay out this four year option, I'll never have to work. I'll probably go be CEO of something. He says well, I understand that. And so as his 55 th birthday comes around and he makes it public, I had been advocating we really had to get in the server business. And Grove says because I was figuring I was going to leave then and Grove says well, I think we should reorganize and do notebooks, desktops and servers and you should have components and systems for servers. And so Albert will stay off and kind of shepherd the microprocessor design business but the business unit will be moved up to the platform level. And so I was enthusiastic about that and I stayed for two more years and my options were vested and everything was in good shape and I stayed a little bit past the time that - they were vested. And then I went away on a sabbatical and I started thinking, you know, if I'm going to do this, I need to do it now. And I had been approached by the people at Bay Networks and - and here's a company with a broad product line, great engineering staff. Customers really wanted an alternative to Cisco but it was lacking leadership. It was a merger that had failed. Never really merged. All it needed was adult supervision. I says, you know, that's the easy part. And this is, you know, now's the time. I came back from sabbatical and went and talked to those guys a couple times and got an offer and took the job.
RW: And were you at Bay and then they were bought by Nortel?
DH: Yeah. So I went to Bay. Bay's an interesting story. I don't know that it relates to semiconductors so maybe not part of this history but, you know, here's a merger that didn't work. Two years earlier it merged and it just, I mean, it was lacking adult supervision and I showed up and people said well, what are you going to do? And I said I don't know. They said well you don't know? I said no. I can tell you I don't know. But I'll talk to the customers and employees. Then I'll know. I'll be back in sixty days. I came back and then put together a plan and our problem there was the East Coast and West Coast were fighting with each other. And any time a development got started on one coast, the other coast would start a competing development. And there was no birth control. I mean, projects were always under-staffed. And so, in fact, I did a review of all the projects and nothing had more than fifty percent of its required staffing. So products were late and they had bugs in them and they were lacking features.
And so I just prioritized all the tasks, prioritize all the developments, cancelled half of them, moved all that staff up to the other half so that everything was fully staffed. And that was in the beginning of '97 and in 1998, fifty-eight percent of our revenue came from products that were less than twelve months old. So we totally refurbished the product line and got the company back on track. However, at that time, the internet is going crazy, the bubble was being created and the telecom equipment companies all wanted routing and packet capability. They were all circuit switched. They wanted IP. They want routing. And Cisco had just declared themselves to be the competitor of Lucent and Nortel and Alcatel, that they were going into that business and declared them to be competitors. And, in fact, there had been private discussions between Cisco and Nortel about a merger and they had been broken off. And Nortel really felt that it needed to have routing and Lucent - we had a close relationship with Lucent. We had had a technology agreement with them that I had broken off much like I broke off the AMD second source agreement. I broke off the second sourcing of our routing code to Lucent and got those rights back. And but we had a close sales relationship with Lucent. And clearly Alcatel and a few others were candidates to do business with. And I went to them all and I said, you know, we got routing capability. Your other suppliers declared themselves your competitor. We're not. We're going to work with you. We want to sell through you. Let's have a close relationship and proposed how we might work closely. And all of them thought that was great but all of them came back a little bit later and said if we do this, we're going to be dependent on you for our routing. What if their biggest competitor buys you? We'll be dead. We can't afford it. So we can't - either we have to develop it ourselves or we have to acquire you. Now it's very clear to me that our routing code in technology is of more value to their shareholders than my shareholders. They're going to pay more money for it. So I went there, the market cap was three billion. Two years later, we sold it for nine billion. Actually twenty-one percent of Nortel's stock which shortly went to over forty billion, before it went back to probably three billion or whatever today. But that's history. But clearly did the right thing for the shareholders.
RW: Well at Intel, officers of Intel teach classes and you actually learn how to run a meeting or how to -
DH: How to make decisions, how to manage for results, how to run a meeting, how to handle issue resolution.
RW: I thought that was fantastic.
DH: That's what I did when I went to Bay. I went to Bay and I said, you know, there's no consistent culture here. I got to establish a culture and so I put together four training classes and just the tyranny of numbers. I said if I teach thirty people at a time and I teach five classes, that's a hundred and fifty people, they will be able to teach all eight thousand people in the company. So everybody will learn from somebody that I trained. And I put together my own slides on managing for results which is about objectives and goals and long term goals and yearly and annual goals and tying that to performance and tying it to compensation and making it all measurable - all Intel basic stuff that's, I mean, generally taught in terms of managing by objectives. Decision making. All about pushing decisions down to the people that are responsible, the use of consultants, people that need to be advised. They don't get a chance to make the decision but you have a responsibility to talk to them before you make the decision. And then the process of making decisions - little decisions, big decisions, how to scale the process. Then straight talk was about issue resolution. At Intel, it's called "constructive confrontation". I have a different version of that which is less contentious and I think more effective. And effective meetings which is about fewer, smaller, shorter, more effective meetings. And how ineffective we can be in meetings, how we all hate meetings, how much it costs to run all these meetings and how to do it a lot better. And I taught those four classes, one a month. And then the people that I trained had to train all their people before they came back the next - I was back the next month with the next class. And so I took that Intel, you know, training comes from top management, not from consultants and brought it into Bay and it was extremely effective in terms of creating a new culture within Bay. Very well received by the employees.
RW: I also found it's very rewarding to do.
DH: Yeah, absolutely.
RW: It's fun to teach when your people are really smart and they have to learn this because you're the boss.
DH: That's right and when you teach it, you have to act it. If you go up there and teach it and then you don't live by it, you know, there's no better way to train yourself than to train others.
RW: Yeah. I did a lot of that at LSI Logic and I really enjoyed it. I thought it was effective. So anyway, back to Nortel. Then you became President of Nortel?
DH: Yeah, I was the number two guy. We had the Office of the CEO. Clearly John Ross was no - Roth was the number one guy. I was the number two guy. My job was to lead the transformation into IP technologies, packet switch technologies. John announced that he wants the new Nortel to be more like Bay Networks than the old Nortel in terms of its culture, the way that it operates, etc. Those were good words. When it came down to doing the really difficult things, the company wasn't willing to do all the stuff that it needed to do. And that became clear to me after I'd been there for about six months.
RW: So then - then Intelecom Allegro Networks.
DH: Yeah, Allegro, I'm on the Board of Azaire Networks which sells equipment to cellular operators that allows the cellular operators to access Wi-Fi or any IP access actually - integrate into their network. So they've got a lot of data network for their email and text messaging and instant messaging and all this sort of thing. They've got a data network and why not access the Wi-Fi networks when you're in those areas in those hot spots. That's good for notebooks clearly but also for PDA's. The big movement is voiceover IP. In fact, Mercury News on Monday of this week had an article about dual mode phones that have Wi-Fi and regular cellular technology and use voiceover IP when they've got Wi-Fi access because the places people complain about phone access the most are in their home and in their business. And those are areas that you can set up Wi-Fi and you can improve your cellular coverage. And it's much more cost effective for the cellular companies to do it with Wi-Fi than it is to try to put more cellular technology into these steel structures of business and remote locations of homes.
RW: You've always had an interest in cars. And does that come from your father's influence and where you grew up?
DH: Yeah. When I was at home we, you know, we'd pull engines out of cars and rebuild them and I my job in the beginning was to hand the tools to my father and to hold the light, you know, because it was typically dark. And - and I remember my dad saying, couldn't you point that in my eyes a little more? You know, because I'm a little kid here - I'm holding it, you know. And then, I would be able to do some of the other work on the car and you know, the town was automotive - Muskegon , Michigan was very automotive oriented. And so when I was in eleventh grade, I bought my first car, a Pontiac two-door hardtop. I put spinners on it and I put little - it was a two-tone kind of pea green color and I put lake pipes out the side and I put these little green lights in the wheel wells and made the interior lights green and, you know, and trying to have a hot car. And so I've always been interested in cars.
RW: And you took some classes in performance driving, right?
DH: Yeah, when I left Nortel, one of the things I wanted to do is start racing. So I went to Skip Barber and took their race driving course and took the post graduate stuff and the car control clinics in the lapping days and got my license and started racing with Skip Barber. And since then, I probably am on the track, you probably thirty days a year.
RW: And so you bought Skip Barber?
DH: Well there were a group of us - five of us at Mid Ohio and this guy Mark Patterson says you know, we should buy this company. I says huh? He said well you know it's going bankrupt, don't you? I says no. He said well just look at the financials. I mean, they're out of cash. These guys are going to go bankrupt. And we could - he said - we should buy them out of bankruptcy, buy the assets. And I said I wouldn't have the first idea how to do that. He said well that's what I do for a living. He's now the Patterson of the Patterson the buyout firm and he was with one of the big financial institutions at the time and heading there buyout business buying bankrupt companies basically and turning them around. And I says yeah but, you're not going to run this and I don't want to run it and he said well I got this friend, Michael Culver, and, you he races and they've got a company and he could do it. And I said well sure I'd be in. So then Mark starts working on putting this deal together and then couple of the other drivers. And it turns out by the end, there's five of us because more guys are finding out and they want in. And so there's five of us and we together with Michael Culver's company, wind up buying Skip Barber Racing in December 2001.
RW: I'd heard that you'd written off so many of the Skip Barber cars that it was cheaper.
DH: I had a bad year, I think it was 2001. You know, knock on wood, I've considerably improved and probably have one of the best safety records today. I went from October of '03 until July of '05 without any damage to any car. You do that thirty days a year on the racetrack, that's a pretty good record, racing a couple races a month. But I had one year, I think it was 2001 where in Skip Barber, they have this thing where you start out, you get full insurance and then if you have too many accidents, you get put on probation. And then your deductible goes up. And then if you have too many accidents on probation, you got what they call on the bubble. What on the bubble means is if you continue to have accidents, you will be disallowed to race with anyone. You will be kicked out. And you got to then have so many races without any accidents before you can get back on probation and so many there before you get back into full, regular status. I made it all the way to the bubble one year. I was close to getting kicked out of Skip Barber racing. That's about the time we bought the company. I think that was the year - December we bought the company. Expensive insurance policy.
But now I get to race for half price.
RW: And weren't you also into winemaking?
DH: I'm more into grape growing. We have our own wine - House wine. And, I mean, House wine, how can you miss that brand? What an opportunity. It's sold around the world. It's known everywhere. So - so our wine is, you know, labeled "House".
DH: House wine. The label on the back says this fine wine from a superb - the superb wine from a small winery in the Santa Cruz Mountains gives a whole new meaning to the words "House Wine". So yeah, we have a vineyard. I started out just planting kind of like landscaping and through the entry when you drive up the driveway and planted on either side. Then I planted on this side of the house. Then I planted over here and then I planted over here and then two years ago, We've got 73 acres up here so on one of the parcels, we developed a vineyard. I planted 12,000 vines two years ago there and then up at the Mountain Winery, I'm one of the four owners of the Old Paul Masson property. I planted another 12,000 vines the same time up there. So we will soon be in the realm of about 2500 cases a year - more than I can drink.
RW: Well, you know, that - that just sort of brings up another aspect of Silicon Valley that so many of the entrepreneurs have just started out as engineers, as just regular guys and made a lot of money. And what is it about Silicon Valley that breeds this culture?
DH: Well, you know, it - it's an interesting place. I told you that I interviewed the United States back in 1965 when I was graduating from Michigan Tech. And what I found in California that attracted me is just a less structured environment, less respect for authority, less hierarchy, faster movement, faster pace, more willingness to entertain new ideas and new concepts, just less structure in place. And that just fit my personality but I think it also fits the personality of innovation. I mean, the fact that we don't have a hierarchy in our companies like East Coast companies typically do, that there's not a formality about communication and who can talk to who, the fact that people tend to be valued more by what their contribution is than what their family lineage is or who they know, and that - I think that's all made for a very fertile melting pot of ideas and technology. I mean, luck has paid some role. I mean, Shockley might have gone somewhere else but, you know, he came here and started the semiconductor industry. But, you know, the other two inventors stayed back in New Jersey and it didn't really take off there. You know, venture capital really started here. So maybe there's some luck or maybe it came here because of the environment. I came here because of the environment that I saw and I just knew that this was the place where I'd fit in.
RW: We were talking about Intel and its strengths. One of the strengths, I think, is the ability to kill off projects. If it's not going to work, they kill it off. And so many traditional companies get invested in an idea, in a product.
DH: I would put it a little differently because Intel certainly has the ability to invest in things like the 8816, 8800, 432 or the Tahoe, or Itanium for long periods of time without getting a result and then, It took them a long time to get out of the DRAM business. I argued, that book Only the Paranoid Survive , I lived, you know, I saw the movie. I lived that. I used to argue that we had to invest more in microprocessors. We just won the IBM PC design line. I saw what we needed to do and I saw the returns that we were going to wind up making. I gave a talk to the executive staff once where I came in and I had a picture of the 8086 and I said you see - and I highlighted the microcode. I says I'm not in the chip business. I'm in the software business. Bill Gates has a floppy disc, and delivers his software that way. My bits are on this piece of silicon. The silicon is just a delivery vehicle. The price is not a function of the cost of the silicon, no more than the software is priced based on the cost of the floppy disc. It's based on the value to the user. And that you've got to make more investment in this business. You know, Andy wrote the book about Gordon, first I got Andy to agree that we needed to invest and then he went to Gordon. Gordon said you can't have the microprocessor business without a high volume memory product to debug your silicon technology. And I said well you can't harvest a microprocessor business if you don't invest and if we can't invest in both, we got to invest in this one. And my cost is not the critical item, unlike all previous semiconductors wherein the end cost winds up being the success factor or failure. This is going to be based on the value. And that's the Pepsi blind taste test, you know. If it works better for the user, we leverage all of the rest of the system and we'll be able to make it successful. The thing that that I really learned at Intel and I think made a difference is not that it killed off the projects that were not successful but it prioritized and fully staffed. When we developed a microprocessor - I've developed a philosophy about this from my Intel experience and I'll state it from my philosophy standpoint because it's really an Intel philosophy. If the engineering manager comes to you and says I need this much resource, he underestimated. He doesn't know the problems he's going to run into because he hasn't designed it yet. And as he goes along, he's going to find that he needs more. And the classic business manager says well can't you do it for this? And they try to squeeze the engineer. They keep trying to squeeze and you get products that are late to market, full of bugs and they're missing key features. When the engineer says I need this much resources, you have to say, are you sure that's all? You know, how much pad do you have in that number? Last time you developed, what did you think it cost and what did it really cost. And then he says well no, I can do it for this. And he marches along and he runs into a problem. You say well are you going to need more resources? Most business managers don't say that. They say can you cut your budget? I'm trying to get this budget in and what they do is they take all these development programs and they starve them to death. And they get products that are late to market, missing features and full of bugs. Intel didn't do that. They would kill projects within a particular product line. So if we were developing multiple chip ideas, we would not do the graphics processor because we couldn't fully staff them all. Instead of taking resources and trying to spread it around and yeah, we're going to do the graphics as well, or give me the other area. The problem we had was when the organization came together at the top and the guy in authority, you know, we at Intel talk about the separation of authority and knowledge. In old structured industries like automotive and steel the knowledge and authorities are at the top, but in technology, the knowledge is down here and the authority is up here. And when - when you've got that situation - so you've got the DRAM's reporting into the CEO and you've got the microprocessor reporting in and the memory systems reporting in and the 432 reporting in or Itanium reporting in separately, there's not enough knowledge up there to make those decisions. And Intel made the mistakes that it made and they clearly made some of investing way too long in some technologies, tended to be when those decisions weren't pushed down to the level where the knowledge was. And I think in the microprocessor business, we were able to manage that business and be successful because we could make those decisions. And God knows I wanted to own the graphics business. And God knows Intel was never successful in that business because it was more important, first of all, I got to have the CPU. If there's a separate cache, I need to have the cache memory to make that work. I've got to have the north bridged from the processor to the bus. I've got to control the bus architecture. I need to connect to the DRAM first. I mean, think about the - the hierarchy. Then I need to connect to the hard drive maybe. The graphics, you know, if I get enough to get there, you got to do a hierarchy of what's the most important and then you got to fully staff those projects. And I think that's the thing that Intel did very well within microprocessors, we could do that and I'm going to assume within flash memory or EPROMS or within DRAM's, they could do that. The thing that they weren't able to do is make the decisions at the top. And so they spent a tremendous amount of money on some projects and got zero revenue for long periods of time.
RW: Well I was also thinking of Microma, of bubble memory -
DH: Well Microma - that was an interesting one because, you know, Gordon is the guy that pulled the plug on that. I'll never forget Dick Boucher to run that. We had an executive staff meeting and we had Christmas business one year and finally we had a watch that sold and we had sold out our inventory. And Boucher and, you know, he's just as happy as can be and the now we can really get into the watch business. And Gordon says, how much inventory we got? None. What would this cost us to get out of the business then? Well not very much. Well let's do it. So the fact that we were successful with one watch in season and Gordon says this is a chance to get out cheap.
RW: That was a wonderful decision. Okay, well Dave, we're in your office here. Why don't you tell us about some of these products?
DH: Well one of the things we always did at Intel is the design team would create a plaque when we completed a chip development. So this is kind of a hall of fame of microprocessors. Here's a 386 with a chip and a chip photo. Here's a 486. Here's the Pentium processor. You see the chip and the chip photo. Here's a Pentium Pro. This is an interesting collection here. Here's the - the 486. There's the processor itself. There's a photo of the processor. And here's a computer plot of the 486 processor. And this is signed by the architect, John Crawford and Pat Gelsinger who headed up the development of the 486. So you get the whole enchilada here. And then over here is a - the history of microprocessors. This was put out for the microprocessor forum and it's interesting to me because it starts back about the time I joined Intel. I joined in '74 just before the 8080 was introduced. And so the 4004 and 8008 were out at that time. But this shows Zylog splitting off from the 8080 and then the 8085 and then the 8086 and 8088. First processor used by Intel, the Z8000 and the 286 and 186 at Intel and then the 386 and 486 and 386SL, 486SL, the Cyrix 486 and we go onto the Intel Pentium processor and the Pentium Pro that you saw over there on the wall. And the whole line of Intel processors. I left in '96 and this chart happens to end in '96 so it's kind of works out nice for my history. But then we've got all the competitive processors and lot of these have been forgotten about over the years but they were certainly serious competition and at times, it looked like the Z8000 was going to win and the Z80 and the Z8000 and the 68000 were going to win. But in fact, this top line of Intel products probably represent ninety percent of the revenue of this whole chart and probably ninety-nine percent of the profit it generated over the period of time. So this kind of my memory wall here. The "Intel Inside" charter member of the "Intel Inside" advertising campaign plaque. And then these are shots out of the first television ad ever run by the networking industry. And at Bay Networks we had an ad featuring the New York Stock Exchange. And so after that, Cisco and everybody else started advertising on television and then the announcement of the Bay-Nortel merger. Went there it was a three billion dollar company. Two years later, we sold it for nine billion dollars. And here's part of the story you haven't seen probably. This is little known part of the story. This is a letter dated April 2000 and it says “we offer you the position of President and Chief Executive Officer of Google and you will get a option to purchase ten percent of the stock of the company as part of your offer and you also have the right to buy another five percent at a valuation of five hundred million for the company”. So fifteen percent of the company will be yours and I didn't take the job.
RW: Eric Schmidt got it.
DH: Eric Schmidt is very happy that I turned down this offer. So I thought in case I started feeling too smart, when I look at this wall, that I really should put this up here to remind myself that I'm more lucky than smart. Not that smart. Over here's an interesting piece. This art was actually given to me by Bill Gates. I did something for them. I spoke at a conference or something. And this is supposed to commemorate Intel and Microsoft and at the center in there is the symbol of the Windows flag, of course, the four squares but it also represents a chip with bonding wires and a package. So it was sort of like Windows inside an Intel processor. He commissioned this by some artist that he knew and gave it to me back before people knew Bill very well. Here's a picture of Bill and me back from Reseller News, the top twenty-five with Bill Gates, Dave House and Ray Norda. Another piece of memory lane. I, you know, if I read that too much, then I - I can just go read the Google letter to get my mind back and realize I'm not that smart.
RW: Well why don't we - why don't you take us on a house tour. Now you had this place built right? This is to your specs?
DH: Yeah, they the history is in 1978, I bought a spec house down at the bottom of the hill and it was under development and I wanted to see what was on top of it. And I took my dog and my kids and I hiked up here and I said oh man, I want to build a house up here. And the kids said “dad you're building a house down there”. And I said I know but this is not available and when it is, I want to build a house here. It became available and I bought it and Goody Steinberg of the Steinberg Group was the architect and built a house 1998.
RW: It's fabulous so take us around.
DH: Okay. Well, some of the prized possessions are in the garage. My Dodge Viper, my SL55 AMG, the Ferrari 360. There's a 430 coming next month to replace that. And, of course, I had to carry the artwork into the garage and - and well this mural of a gas station in the garage and my gas pump over here with 38 cent per gallon gasoline available. Everybody wants to know if they can fill up on my pump. The design of the house - I asked for a house that really brought the outdoors in and the indoors out, that really flowed back and forth. So the architect designed the house so that it's got the same materials inside and out. So it's got his siding and the stone and the stone on the floor appear in both places. And then a critical item was capitalizing on the views. And so we put in this vanishing edge pool to capture the view of the mountainside. And one of my favorite sides and the vineyards that exist around here. Mount Eden Vineyard up here. We had Cinnabar Vineyards on the top of that hill. The Mountain Winery Vineyard that I planted is up here. You can just see part of it over there. Cooper Garrett is next door and, of course, our vineyards around - all around the house. And obviously bar for entertaining and a nice big living room so when all the family's here, we can entertain. And more of the views. You can really see from here the, you know, one side we've got the mountain view on Eden Valley . The other side we got Silicon Valley . Little bit of smog today. You can see the hangars down at Moffitt Field and you can see San Francisco Bay runs right down behind this tree, downtown San Jose airport and probably the biggest center of creation of wealth in the world right here. Venture capital land. One third of the venture capital in the United States and a major part of wealth creation in - in innovation and in technology right down here. And the sound of construction is where I can always modify on something to make it a little better. A dining room that gives you a pretty spectacular view on its own. This probably isn't worth videotaping but I've got to be a technologist so five inches of sound absorbing material on the back wall and two inches on the side walls and a separate air conditioning unit that pumps air through the equipment so that the fans which are all heat sensitive - you know, heat activated stay turned off. And so we're always blowing cold air through the equipment. And all the lumens you can get, actually have to run this projector at half intensity because it's so bright, it'll light up the room and wash out the blacks. So we run that at half intensity. Okay. So when we built the house, we landscaped the entry and around the house and had this driveway and I said to the landscape architect well, what are you going to do with the area along the driveway? And he said well I thought I'd plant California wild flowers. And I said how about vines? He said well how about a vineyard? I mean, we're right in the middle of wine country here. If you look up at the top of the hill, that's Mount Eden Vineyard and you can see the sparkly bird mylars to scare birds away on that vineyard up there and the vineyard to the right, that's all - actually the one on the left belongs to the Ray family, Martin Ray's son and the one on the right is Mount Eden. And then up here we've got Cinnabar at the top of this hill and then just to the right of that above the quarry is the vineyard that I planted at the Mountain Winery, the old Paul Masson property. And down below, we can't see it from where we're standing here but from the other side; we'll have the Katherine Kennedy Vineyards. And so I started planting Cabernet up here. I've got Merlot along the driveway and then Chardonnay below. More Merlot below the swimming pool, more Chardonnay on the other side of the house. And just kept planting every year and so now we've got - we've - we've - we've harvested our Chardonnay. On Sunday, we harvested the Merlot. It's a late season this year and it's a low producing season. Hopefully great quality. And then above - three rows above the driveway, we've got the Cabernet Sauvignon. And that's yet to be harvested. The world renowned House wine.
There you go. That's actually our Merlot. We don't bottle much Merlot. We've got lot of Cab and a lot of Chardonnay but as it says on the back here, "This superb wine from a small winery in the Santa Cruz Mountains gives a whole new meaning to the words 'House Wine'". So with a name like House, how could you not take advantage of it in branding. I mean, it's a little tongue and cheek!