Intel / Technology

Factory of broken dreams: Will the Googleplex be next


Intel Closes Down the Hudson Manufacturing Facility

IHudsonIntel just announced that it would close the Hudson Manufacturing Facility it acquired from Digital Equipment Corp in 1997.  The press release from Intel  said “the Hudson plant has been used to make a variety of low-end chips found in many electronic devices.”  The closure of this plant will not mean much to people other than the thousand or so people directly affected.  The exception to this are the few pioneers who vision and drive created it in the first place. They dared to dreamed big and lost.  While they understood the potential of semi-conductor technology and large scale integration, like all of us at that time, they did not understand that the computer industry was about to be turned sideway and that companies that were vertically integrated would fail like Digital would fail.

Imperfect Vision

It must have been around 1980 when Steve Techier and Roy Moffa proposed to Ken Olsen, the CEO of Digital Equipment Corp and eventually to its board that the company should establish a state of the art semi conductor capability including a manufacturing facility.  At that time, Roy was my boss at Digital. I had joined less than a year earlier.

Digital was using bipolar chips to constructs it midi computers.  These were not microprocessors – just parts of a computer.

Both Steve and Roy reported to Dick Clayton who in turned reported to Gordon Bell, Vice President of Engineering and Chief Technologist.  By the way, it was Dick that was responsible for bringing me to Digital from Israel.   I would attend Dick’s staff meeting from time to time and can remember Steve and Roy talking about their proposal for the Hudson facilities.  Frankly, I did not understand much of what they were saying since I had no background in semi-conductor technology. But I remember how inspired and enthusiastic they seemed.  It is strange to note that later I would become a Corporate Officer of the most successful semi-conductor company in the world, Intel Corp, less than ten years later.

Roy died a few years ago.  I wrote about him here.  Steve is living in Florida. I hear from him on Facebook from time to time.  You can listen to an interesting interview with him here.

I would like to apologize in advance if any of my “facts” here are wrong and if this post is a bit too technical for most of my readers.

Avram the Engineer

PDP11-23GAt this time, 19080, I was a senior manager in the Central Engineering organization at Digital, and responsible for small computer hardware.  That included all PDP-11’s using the Q-Bus such as the PDP 11/23 as well as support for PDP-8s including Decmate, Digital’s proprietary word processor.

Frankly, I was in heaven.  I loved computer technology and design and had used computers extensively for over ten years.  I could design computer hardware and write software even thought  I had no formal engineering education.   Now, I was trilled to be leading a very large group of engineers and making important design decisions for the second largest computer company in the world.  I was 34 years old.

The PDP 11/23 used four a chip set manufactured for us by Western Digital.  The larger PDP-11‘s managed by other groups at Digital were based on the Unibus architecture.  It is interesting to note that the Central Engineering organization responsibilities were separated by the bus type.    The Unibus could only address 18 bit. One of my first accomplishments at Digital was figuring out how to extend the Q-Bus so it could have a 22 bit physical address making potentially more powerful than it’s big brother, the Unibus. An address space of 16 bits only allows for a memory of 64 kilobytes to be addressed directly.  A number of years later,  Bill Gates is allegedly quoted as saying “.. 64KB ought to be enough for anybody.” This may not be true story but I like it never-the-less.  By extending the address space of the  Qbus to 22 bits, 4 mega byes could be addressed.  Today, the microprocessor use in most desktop computers have an address memory that are millions of times larger than that.

Digital also had a single chip version of the 11 architecture called the T-11 offered a single board computer.    But the  T-11 (T for Tiny)  was not the first microprocessor the company used.   The Decmate used a CMOS processor manufactured by Harris. It was actually a 6100 12 bit processor which emulated the PDP-8 instruction set. Later, we developed a more advanced chip set called the J-11.

The T-11 was fabricated in NMOS.  It only contained 13,000 transistors. Today’s desktop microprocessor will have several billions of transistors.  Back then 13,000 transistors seemed like a lot to me.  Just ten years earlier I had been designing medical research equipment using single transistors.  I doubt that many of today’s engineers have ever held a single transistor.

I was very familiar with the T-11 chip since in my previous position as Director of Medical Computer Systems for Mennen Greatbatch, I used it extensively in embedded patient monitoring systems I designed in Israel.     I loved the T-11 and tried to get Digital to sell it on the open market and make it a standard. I went to the management committee with an  opportunity to license it to Atari, the leading electronic game company but they were not interested.  They did not want to sell the chip outright and probably looked down on the gaming industry.  Later, we develop a more advanced chip set called the J-11.

Intel introduced the 8088/8086

A few years earlier, Intel had introduced the 8086 and 8088.  These microprocessors had about 30,000 transistors.  The 8088 which was a 16 bit microprocessor but used only an eight bit memory bus. It was the processor that IBM selected to use in the first IBM PC.  The power of the T-11 was not in transistor count but in the software and tools that were available for it and could have easily become an industry standard.

My Own lack of Vision

Pro 350A year later, I was tapped by Ken Olsen to lead Digital’s efforts to create what would later be known as a personal computer.  Our product line was called the Professional Series.  It was based on the F-11 chip set and later utilized the J-11 chip set. We had a very different design objective than the IBM PC which was under development at the same time.  We were trying to build an elegant, easy to use computer system for a professional person with a rich amount of software and the ability to network (remember this was 1981). Ken wanted it to be so easy to use that both his minister and secretary would used it.  I described this as the system for “Clerks and Clerics”.

We failed at that for a number of reasons that I have discussed before.  Once of those reasons, i have to say, was we over engineered our product. It made it expensive and late to the market.  We did not ship our product until early 1993 more than 18 months after IBM.  We had a proprietary architecture and continued on the vertical integration model including our own sales force and computer stores.  But frankly, I still think the Pro was a beautiful and  amazing product. I still have one.

The Computer Industry Turns Sideways

Pro 350During this time, the leading computer companies such as IBM, Digital, HP, Data General and Prime were all vertically integrated.  That is to say they designed all the parts of their computer products from CPUs, Disks, terminals, printers and software. It was at lot of work and was very expensive which resulted in the products and services being expensive.

The IBM PC was the catalyst that changed everything.  And it was pretty much an accident.  But it was the spark that created the explosion that blew up the computer industry including IBM, the very company that created it.

The IBM  PC was the first major product of the new era of horizontal structure.  Different companies would provided different parts of the product from disks, processors, operating systems, applications and distribution.  The IBM folks did not realize that they were laying the foundation for the greatest change the computer industry had ever experienced. The IBM team did it because of time to market, limited budget and lower product cost.   But from that moment on, the clock was counting down on the  major computer companies including IBM.

It was Aug. 1980, when William Lowe the IBM Lab Director of IBM’s entry Systems in Bocca Raton, came meet with IBM’s  corporate management team including the CEO, Frank Cary, to discuss how IBM could participate in the personal computer space which was just beginning to develop with products from Apple, Tandy, Commodore etc.  IBM had learned a lesson about disruptive technology when it failed to comprehend the importance of Mini Computers.   Lowe, propose that he could create a personal computer in 12 months (it usually took this long to get a proposal  agreed to at IBM) if we could go outside the company for critical technologies.

So project Chess was formed. Lowe took assigned twelve people to the project including Bill Sydnes and Lew Eggebrecht both of who ended up working for me in 1983 when I was President of Franklin Computer Corp. Lowe was promoted and given a different assignment and the late Don Estridge took over the project.   I meet Don at a dinner in 1981 just after the IBM PC was announce. Bill Gates and Steve Jobs were there as well.  I was 35 and they were 25 years old.

The IBM Chess team had decided to us the Intel 8088 processor.  They thought they would get the Operating System from Digital Research Inc (DRI)  and were looking to Microsoft to get the BASIC programing language.  At that point, Microsoft had 40 employees. The legend has it that Gary Kendall the CEO of DRI wanted to fly his airplane on the day that the IBM PC group wanted to meet and/or Gary’s wife would not sign an NDA.  So when they were meeting with Bill Gates up in Seattle about BASIC, they asked if Microsoft might have an OS.  Bill lied and said he did and told the IBM folks he would send a proposal.  He knew of an OS for  the 8086/8088 called QDOS, written by Tim Patterson at a company called Seattle Computer Products. Bill bought the OS from Patterson for $50,000.  He licensed it to IBM and called PC-DOS.

Intel did not develop the 8086/8088 for the PC market.  That market did not really exist.  It was developed to be used mostly in embedded applications.   IBM was familiar  with Intel products.  The IBM team considered  Motorola  68000 but it was still in development so they went with Intel.  The decision to use Microsoft’s OS and Intel’s processor where taken without any thought to the long term impact but resulted in creating the most powerful hegemony, ever.  It would be Microsoft and Intel which was sometimes referred to as Wintel that would rule the Personal Computer Industry until Google and Apple changed everything once again.

Bring in the Clones

Compaq and a number of other companies were formed to create compatible IBM computers.  At first, they copied PC-DOS.  IBM would send undercover security people to industry shows and ask if the computers being demonstrated were compatible with the IBM PC.  If the company said yes, they would ask if they could test thier program.  The mostly got a yes.  The program they ran, copied checked to see if they were infringing on IBM’s copyrights.  If they were, the company got a letter from IBM. But IBM was reluctant to go after any of these companies because they had spent years dealing with government charges of monopolist behavior.

Microsoft wisely had secured the ability to modify and license PC-DOS.  They released a version of the OS called MS-DOS and licensed it to everyone that wanted to build a PC clone.   There were a number of compatibility issues in the early years of the PC but they were over come  fairly easily.

VAX/VMS

By this time, Digital was going through a major technological transformation from the PDP series of 16 bit computer to the VAX which was its first 32 bit computer.  Gordon Bell, the technologist/Architect of Digital, understood the power of the VAX and wanted to unify the company around it and its operating system, VMS.   That’s when Steve and Roy proposed that Digital develop its own semi conductor capability.  Their first target was a single chip implementation of the VAX architecture called the Micro VAX.  I think it shipped in 1984 but I left Digital in 1983.  Interestingly enough, it used the bus architecture I had developed in my first year at Digital.   The VAX was eventually replaced by Alpha a 64 bit RISC  architecture.

Palmer,  Angel of Death

In 1985, Digital hired Robert Palmer to head up its semi-conductor activities.  Since I was long gone, I have no insight into why he was brought in or what really happened.  By 1990, Ken Olsen was slowly loosing control of the business.  He was replaced by Palmer in 1992, probably one of the worst choices for CEO a board could have made. Palmer did not understand Digital’s business and most importantly, he did not understand it values and culture. While, Ken was really no longer able to lead the company, almost anyone would have done a better job than Palmer.   As the company continued to decline, Palmer sold off one part of the company after the other.  Pretty soon there was nothing much left.  Eventually,  Digital was sold to Compaq Computer in 1998, a company who got its start in 1982 with a PC Clone.  It would have been impossible, al least for me, to imagine in 1982 that Compaq would buy Digital just 16 years later.

Intel Acquires Hudson and a CEO is born

In 1997, Intel bought the Hudson Facility from Digital as part of a settlement having to do with alleged patent infringements.  Palmer was basically liquidating Digital by selling off all its parts.   The person that Intel sent to manage Hudson then is the current CEO of the company, Brian Krzanich, who got his start as a process engineer.  I have to wonder if the curse of Hudson is over yet?

What if Digital thought of itself as a Software and Networking Company

I also wonder what would have happened if Digital had partnered or acquired a company like Intel back in 1980 and eventually turned itself into a software company.  At that time, VMS, the VAX Operating System and DecNET (the most powerful networking software available) were superior to anything else in the market.  Microsoft, was just a small company whose primary source of revenue was selling the BASIC programing language which they had ripped off of Dartmouth University.  They had just bought QDOS which became  PC DOS and eventually MS DOS. Of course, IBM would never have used Digital’s software or processor so I am not sure how Digital would have Digital would have created the standard.

I certainly would not have taken this approach back then and I can not blame anyone for not recognizing the world was about to change.  By the way, it is an interesting detail to learn that Digital with the fifth company to register an Internet Domain (www.digital.com) in 1985.

Now of course the same thing is happening to Microsoft and Intel.  The world of computer rotated again and they are falling off.

 

 

10 thoughts on “Factory of broken dreams: Will the Googleplex be next

  1. Avram – Fascinating history that bringsd much back for me (and enriches the background of my book “Doing Capitalism”). The only addition I would make is to note that QDOS allegedly stood for “Quick and Dirty Operating System”!

    Like

  2. I remember using the “.dec.com” domain, not “.digital.com.”. Those of us there during the Palmer years could see the downfall coming when corporate branding people started enforcing the “digital” name, and denying the “DEC” tradition. Everyone including Bill Gates called it “DEC”. Why give up such a powerful brand?

    Like

  3. This should cause a major depression in Hudson. Back in the 80’s I lived in Hudson (but I worked for Data General, not DEC). What a grimy, grey, depressing area it was. The list of companies that grew up and died in that region: Wang, Computervision, Apollo, Symbolics, LMI, Data General, DEC… After a while I think I figured it out. Every winter virtually every employee came down with the flu; many days each winter it was impossible to get in and out of Logan (even if you could get there on time to catch a flight). Each time I would visit silicon valley in the winter I realized all the companies there had 100% of the workforce operating at 110% capacity while people in the east were chipping ice off their windshields. All I can say is, it is hard to fault Intel for their decision.

    Like

  4. Worked for Intel for 18 years, and have been inside the Hudson plant. It was getting long in the tooth for a semiconductor factory, and needs to be retired. Besides the manufacturing center for Intel now appears to be Ocotillo in AZ., a much better climate.

    Like

  5. The only time I ever flew in a helicopter was went I went to Hudson from Logan to investigate licensing the Alpha architecture. DEC had a different approach to cost!

    One correction, 16 bits can address 1 Mbyte. Bill G is supposed to have said that no one would ever use more 640K, because at the time the rest of the 1 Mbyte was used by DOS.

    Like

  6. I enjoyed this a lot. My mom worked at Hudson in the 80’s and I graduated from a Z-80 to a PDP-11/23 in high school. I was so proud carting my RL02 discs around. Great history. Wasn’t a lot of the PA-RISC (later Alpha) done in Hudson?

    Like

  7. “What if Digital thought of itself as a Software and Networking Company”… I remember attending a DECUS event (remember DECUS?) during VMS’s ascendancy, where a DEC and employee and I were discussing the future of DEC, and we both agreed that DEC should be thinking of itself as a software company not a hardware company.

    As for the re-branding from “DEC” to “Digital”, I seem to recall some sort of lame ad campaign with the slogan “Isn’t every computer a Digital computer?” or some such. And of course there was the re-branding of VMS to OpenVMS. Evidence of directionless management rearranging the deck chairs on the Titanic.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s