surfing the stardust

Intel’s IBM PC Design Win

Several people suggested I write up what my view was of this historic design win, in light of the fact that I was Intel’s first “IBM Corporate Account Executive” during the period that the design win decision actually happened, and the design effort got well underway (1978-1980).  I recently published this on the Intel Alumni website.

I was hired in Dec 1977 to be Intel’s first Major Account Executive for Burroughs, GE, and IBM. This was a factory based, customer oriented, marketing job focused in support of the Field Sales Team and coordinating and developing strategies for these accounts corporate wide and Internationally. Burroughs was Intel’s largest revenue account, GE was considered to have possibly the largest growth potential for microprocessors, IBM was viewed as a crap shoot.

In 2Q 78 I was relieved of duty for the other two accounts and assigned full-time to IBM.

Gordon Moore, Intel's IBM Account Sponsor

Gordon Moore, Intel’s IBM Account Sponsor

At that time Bob Noyce was CEO, Gordon Moore was President, Arthur Rock was Chairman, Andy was COO, and Ed Gelbach was the ExecVP.

I was hired by and worked for Gordie Campbell who worked first for Hank O’Hara, Sales VP and then for Jack Carsten, Sr. VP of Sales and Marketing. I held this job from Dec 1977 unit Jan 1980.

The mission of this position was twofold: (a) to be a central repository and focal point for information on how we were doing in penetrating IBM (b) using this information to develop tactics and orchestrate Intel’s resources to win designs throughout IBM. As I put it to Gordie one afternoon at lunch “I want to get so many hooks into that big fish that he’ll never get off the line”. During that period I helped organized and presented in Intel Architecture seminars at all US and European IBM development labs. In that period I was the only Intel exec to meet with the IBM staff at all of those development labs and in all their relevant business segments.

To put the “PC Design Win” in perspective it is important to note that as of December 1977 Intel had had no (zero) direct business orders with IBM.  One of my first tasks was to look through all Intel Disti Salesout reports for 1977 and find out whether we had any business with IBM. I was able to find about $500,000 of business, primarily nonvolatile memory (mostly eproms) and a few development systems, all through disti.

In Jan 1980, as I moved on to the Special Products Division (non-volatile memory), I had the pleasure to review with Gordon Moore, who was our Corporate Sponsor for IBM, the 1980 Volume Sales Agreement worth $100 Million which we had negotiated with Poughkeepsie and which was signed by Ed Gelbach.  In that Agreement we had agreed to $100,000,000 in business minimum for 1980; most of it direct. From less than $1M in business to over $100Million in 36 months.

This VSA was negotiated with the Poughkeepsie Corporate Procurement Group, representing all the IBM sites. It covered everything from single chip micro controllers in typewriters, & copiers, the Caribou Dynamic RAM, loads of EPROMS, development systems, Memory Boards, Multibus Boards,and 808x family for multiple products including the Display Writer word processor and the System 23 Datamaster (5322),which was the first Intel based, what then IBM was calling “unofficially”, “personal computer”.

IBM Boca Raton's Datamaster - precursor to the PC

IBM Boca Raton’s Datamaster – precursor to the PC

(For more on IBM’s use of the term “personal” computer see: IBM Archives: “The next “personal computer” out of the gate was the IBM 5110 Computing System, announced by GSD in January 1978”).

To step back and understand this almost preposterous growth it is important to know there were 2 big and very, very unique memory deals which propelled our commercial ties, before any hint of serious microprocessor uses. These were for memory devices with the mainframe guys represented by Data Products Division (DSD).

– In 1Q78, facing an industry wide shortage of dynamic RAM, IBM decided to “retrofit” all their internal use System 360 & 370 memory and sell it to external customers. That caused them to come to us for our IBM board knock offs to use just for their internal 360’s and 370’s of which I had determined they had about 50. Bob O. (Boe) Evans came to SC 1 main conference room and gave the Exec Staff and a couple us staff guys a lecture as to what he expected. He bragged that his next big Systems release, the 3033 family, was going to have up to 30MB of main memory. He called Jack “Hey Sonny” as he chomped on his big cigar. This was a massive deal valued north of $25M. The guys who deserve credit for closing this were Tor Lund, Jim Saxton, and Gordie Campbell. The field had little involvement and this one was set up by an exec to exec call, I suspect from Erich Bloch, (recipient of the National Technology Medal, head of the Poughkeepsie Development Lab, and VP of the Data Products Division), to Gordon – they were very close friends.

– In 2Q78 the field conveyed a message that IBM wanted to talk to us about doing a stacked 32Kb DRam



which was code named “Caribou”.  By soldering two dual inline packages (DIP) one on top of the other, this would deliver a 64Kb socket stuffer which would help them deal with the continuing shortage.

This would be in the “external customers’” system memory. This was the first time such a high volume and critical device would not be a proprietary IBM design. Ron Whittier was the GM of Intel’s Memory Division and his engineering team made the magical device a reality. Barry Cox and Scott Gibson drove the marketing. The negotiating team which flew into
Poughkeepsie and were locked up in the CR all day with IBM purchasing and finally closed the deal was Ed Gelbach, Jack Carsten, Barry Cox, George Popovich, and me, Gerry. I broke the impasse in mid afternoon by proposing a “strawman” that allowed us to get our price in years 1 and 2 and then drop price below the SLRP cost forecast in years 3 & 4. This essentially put a massive monkey on Whittier’s back for driving the learning curve. Ron performed masterfully and drove significant “way profitable” margin all through 6 years of the program.

What these wins did was give Intel a special status above Motorola and TI who’d had a pretty good business selling nuts and bolts but nothing like these deals. The field saw this opening and blasted through it.

Ed Gelbach, the soul of the Intel Sales Force

Ed Gelbach, the soul of the Intel Sales Force

In those crucial years of 1978 and 79 the key field was the Poughkeepsie sales office. The players who made it happen were:
Bill Gsand, District Sales Manager
George Popovich, FSE
Joel Cohen, FAE,
Bill Clemow, FSE
(supported by Cathy Clemow, admin)

In 1978 and well into 1979, there was essentially no coverage in most of the Development Labs. Hank had given Gsand carte blanc to go call on the labs even in other Intel regions without the local sales teams. Because any volume purchasing would be done in New York, and the general ban against using non-IBM technology, the local teams pretty much didn’t care since their business level was chicken feed. The accepted wisdom in Intel was that any future significant business with IBM would lie with DSD and be driven by their New York mainframe operations, since we had great memory products and insatiable demand and that there was little home for Microprocessors outside of some OEM rack mounted boards for specialty OEM (industrial) solutions. Most thought GE was going to be our biggest customer for processors.

The only real local sales office energy in those design win days were Larry Gast and his FAE John Leediker covering Austin and Mike Barton in Rochester. So George and Joel took off and started calling on the design people at Boca Raton,



Austin, Lexington, Boulder, San Jose, Rochester, Raleigh, upstate New York, and most importantly General Systems Division Headquarters in Atlanta. GSD was critical because they were in new business and were running faster and looser than the traditional mainframe business of IBM’s Data Systems Division.

The technique we used was, starting in 1H78 to send a covert team of Popovich and Cohen to meet quietly with a handful of renegades, for example a design team at the GSD Entry Level Systems Unit in Boca Raton, or their counterparts in the Office Products Division in Austin – introductions were by word of mouth passing interested parties to George and Joel. Then in 2H78, after we had concluded the two big memory deal negotiations, we were invited to conduct two day “open house” seminars in each of the labs. They were positioned as “educational” with Poughkeepsie and IBM refused to give us the attendee lists. Business cards were quietly passed at the breaks.

The factory sent their best presenters which were headed by

Dane Elliot, Master Architecture GURU

Dane Elliot, Master Architecture GURU

Dane Elliot, on logic and software
Bob Greene, on memory
John Beaston, on “chip set” devices

There were other players on parts of the rest of the “Davidow Whole Product” but these 5 people deserve the missionary credit for all the initial big wins:
Bill Gsand, George Popovich, Joel Cohen, Dane Elliot, Bob Greene

If an Intel Medal of Honor were to be given, it would have to go to George. I am sure there would be no argument with this conclusion.

In 3Q 1978 we had our first ever non-undercover seminar at IBM Boca Raton.

George, Dane, Bob, and I were staying on houseboats at a hotel called The Marina Bay Club

the famous Marina Bay Club, Ft. Lauderdale

the famous Marina Bay Club, Ft. Lauderdale

and after a long dinner and partying it was at least 3 am when we turned in. Never the less we were up at 7 and walked into the Development Lab at 0800. I recall Dane Elliot talking 5 aspirins at breakfast – I said “Dane are you allowed to take that many” and he said something like “I don’t know but whatever it does to me it can’t be as bad as my headache”.

There were about 100 people in the audience in the large cafeteria as we started. The agenda was my “Introduction to Intel”, then Dane with the microprocessor and software story, and in the afternoon Bob Greene on memory. I remember Dane kicking off with something like “Well what we’ve got is a basic Von Neumann architecture….” which immediately put the dozens of computer architects in the audience into a peaceful mood as the talk began to move towards registers, bus architectures, instruction sets, High level languages. Dane wow’d them of course, extemporaneously drawing architectures on the white boards and handled all the questions beautifully. Then Bob Greene brought his deep, sonorous tones to memory architecture, the pros and cons of on board or off board memory controllers, and solid state memory solutions.

Bob Greene, the orator of memory

Bob Greene, the orator of memory

It was 7 hours later when we broke up and the audience, and as far as I could tell, was still present and accounted for. The next day Joel Coen and George came back and handled development environments – “Blue Boxes, PLM, and ICE” … yummy.

The thing that most people would never realize was that general purpose, non-proprietary, microprocessors had been in the market for 7 years and it was clear to all that this technology was on an exponential “usefulness ramp”, yet IBM’s design engineers had been proscribed from using it. They were like desert plants yearning for a rain.

Four weeks later George P. asked me to meet him to GSD Headquarters in Atlanta. George (Spike) Beitzel, the President of GSD, had appointed one of his staff officers, Peter Stern, to develop the full proposal on GSD taking “external microprocessors” into their products in an across the board sense, to be presented to IBM Exec Staff. This would end up including OEM products like the Series 1, word processors products, printers and copiers. We met with Peter for several hours, Lou Eggebrecht the design manager for the Intel project in Boca, and Dennis Gibb, the software lead, were there, and the main issue was “continuity of supply”. He brought up the subject of second sources, either with IBM or other companies. He wanted to know if Bob and Gordon would be willing to discuss these issues with their top staff. I took the action item to set up a series of meeting to address the issue. which began to take place in November. Peter said there was as yet no agreement on whether to use an Intel microprocessor in any of the GSD product line but they were seriously interested. He would not discuss the application.

Things then went quiet in Boca. Meanwhile George and Larry Gast had been working on an interest in the 808x in the GSD lab in Austin which ultimately turned out to be the 8086 based Displaywriter word processor and we had another big seminar there in November. And things were hopping in Rochester where the I/O specialists for GSD lived.

The big bang happened the week before Thanksgiving 1978 with a call from Boca to George asking that, if IBM placed an order, not now, but sometime before Christmas, whether Intel could ship 40 MDS Blue Boxes

Intel's Secret to Success - The "complete product"

Intel’s Secret to Success – The “complete product”

all outfitted for 8085 development by the end of the year in order to fit the IBM financial budgeting.

The hitch was that the project’s IBM champions believed couldn’t get internal agreement and approval for a few more weeks. I worked that with our development systems guys and we committed to IBM. The order came through about December 21st which meant it had to ship over the Christmas week. But it did, and the rest is history.

These systems were used for the development of 2 products in Entry Systems in Boca – the Datamaster and the PC. By the end of 1Q79 Larry Gast was getting many more order in Texas most of which were flowing through disti’s. They were upgraded & outfitted for 8086 and 8088. Our theory was that as the Displaywriter development matured through 1979-80 that the development tools moved to Boca and Rochester to work on the PC.

The question which I have never heard authoritatively answered was, while clear that the Datamaster acted as a “Red Herring” to protect the secret of the PC, whether that was a result of happenstance and convenience or if it was intentional from day 1. Only Lou Eggebrecht, Peter Stern and Dennis Gibb can really answer that completely. Philip Estridge might have a good opinion if he were alive today but in fact as he was the Series 1 GM at the time he was not involved in the decisions in 1978 and 79 when the design win happened & work was well underway.

Hope you enjoy these recollections.


Single Post Navigation

One thought on “Intel’s IBM PC Design Win

  1. Michael Filler on said:

    Thanks for the background!

    I was fortunate to become the Intel Senior FAE, (Field Applications Engineer) in August 1980, stationed in the Poughkeepsie office.

    We made regular trips to Bldg. 708 (IBM’s procurement office), and also to Kingston NY, Burlington VT, Atlanta GA, among others.

    Memory sales (the above mentioned Caribou piggy-back package) but also 8048, were the big volume leaders at the time.

    I presented the 8051/8751 architecture to a meeting of some 200 engineers. Thankfully I had read the data sheets front to back, so I could handle tech questions like “what is the interrupt latency” (7ms, with 1MHz clock, IIRC), and other minutia. That is, after all why they hired me.

    It was exciting to be there when the Peanut (PC), the Display Writer, and all the other stuff came to be mature and released products.


Leave a Reply

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

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

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s

%d bloggers like this: