Intel and AMD: The Early Years
Bob Noyce and Gordon Moore, formerly of Fairchild Semiconductor, founded Intel in 1968. Jerry Sanders and other former Fairchild employees founded AMD in 1969. Despite their former association, the founders of the companies had a tumultuous and even hostile relationship through 1986. That’s when IBM said it would only agree to use Intel’s iAPX family of microprocessors (X86) if Intel presented a second source. Not wanting to bite the hand that was feeding it (and feeding it quite well), Intel agreed by licensing the X86 architecture to AMD.

Judge Phelps’ Recount
“The fundamental problem was both Intel and AMD had sprung from the same womb, which was Fairchild,” said Arbitrator J. Barton Phelps, Judge of the Superior Court, retired. “Both companies shared their technology with respect to design of semiconductors. Intel gave AMD the rights to the X86, and Jerry \[naturally\] thought he would have the rights to the 80386.” This created “a lot of hard feelings between Andy Grove (of Intel), Gordon Moore, and Jerry Sanders.”

According to Judge Phelps, a “very complicated” 10-year product exchange contract between the two companies that began in 1982 basically gave each company the right to produce the other’s products by awarding points based on the “process complexity factor,” as follows, according to the actual contract (and entered as AMD Exhibit 10):

“Process Complexity Factor” shall mean a number which reflects the average relative circuit density of the process. The Process Complexity Factor for processes currently in production are specified on Exhibit 11. The Process Complexity Factor on future processes must be mutually agreeable. For similar processes which are related to each other by a liner shrink factor, the Process Complexity Factor of the related processes shall bear the same ratio to each other as the ratio of the areas of the same circuit layout executed in the two processes. Other similar processes (related in a way different from a linear shrink factor) shall be based on the following equation

Where:

SC = 6 transistor static RAM cell size in mil2
MP = metal pitch = metal width + metal spacing in µm
PP = poly silicon pitch = width + spacing (field interconnect)

 

Each time a company licensed a new product to the other, points were awarded. Points could then be redeemed to gain access to a new technology, kind of like redeeming frequent flyer miles to get a free ticket.

The hard feelings started when Sanders told the press how much money AMD was making on the exchange agreement. These comments prompted Grove to write a note in October 1984 to his senior vice president David House, later entered as AMD exhibit 711 in the case:

DH  May you ‘even out’ the balance w\[ith\] these bastards soon!” \[emphasis in original\]

Many other similar inter- and intra-office notes were exchanged within and between the two companies. With each company suspicious of the other, Intel decided to go it alone on the 386. AMD dubbed this plan monopolistic, which was funny because AMD seemed to be okay with a monopoly so long as it was part of it.

When Grove said he wasn’t going to second-source the 386, AMD sued Intel, contending the contract based on points had been breached. Essentially, Intel rejected AMD’s quad-pixel data manager and hard-disk controller products, either of which would have earned AMD enough points to access the 386.

AMD didn’t take Grove’s decision lightly and decided to sue Intel for the rights to Intel’s 386 microprocessor architecture. In addition, AMD sought $2.3 billion in damages, an amount that Phelps was later quoted as saying was “more than the total budgets of eight states.”

When the case was about to go to trail, both companies decided they could get a fairer ruling by using a fully empowered arbitrator instead of a trial by jury. Judges are bound by more rules than arbitrators, and arbitrators have considerably more leeway to run the hearing. Arbitrators also have more decision-making power than a judge, and the decisions carry more weight and are less appealable.

Phelps was trying a different case when he got a call from an attorney he knew well, Terry M. “Without providing any detail, he asked me if I had time to try a case that would take five to six weeks,” Phelps said. He then indicated that he would probably have time but would like to learn more before agreeing to try the case.

Phelps next received a call from a different attorney, Lois M., who asked if he would have time try a case that would “last a while.” He asked if it was the same case that Terry M. had asked about. Apparently surprised that anyone else had called him, she indicated that the case would last longer than four to five weeks, but still did not provide details on what the case was about. So, Phelps set aside some time for both parties to come in.

Both companies petitioned for arbitration on April 10, 1987. Phelps, a retired superior court judge of 16 years, was chosen because of his considerable experience with complex civil technology-based litigations and private arbitrator experience.  This particular case was presented to him as a simple arbitration. The California Civil Code in arbitration required that neither party had “discovery,” meaning neither party could get documents from the other, tying the hands of each party.

“Both companies wanted discovery so they could peek at the other’s files,” recalled Phelps, who then issued a subpoena duces tecum that provided 52 days for each party to examine the other’s documents. According to Phelps, the subpoenas were “terribly broad” and resulted in “file cabinets full of documents.”

Continued on page 2