Another draft, this time create on June 22nd, 2016. The title may have been written down in a hurry, and seems a bit click-baity. I suppose it needs a bit of nuance: the point I wanted to make here is not about the early x86-CPUs that were made as a second source under license from Intel. It has to do with the CPUs from the 386 onwards, which AMD marketed under their own brand, and at least initially, without a license from Intel.
I’m not sure how many people would actually agree with the notion that AMD shouldn’t have made x86-based processors under their own brand. But it is interesting to see why and how AMD actually got to that point.
My draft contained the following links:
http://jolt.law.harvard.edu/digest/patent/intel-and-the-x86-architecture-a-legal-perspective (working archived link: http://web.archive.org/web/20170120173519/http://jolt.law.harvard.edu/digest/patent/intel-and-the-x86-architecture-a-legal-perspective)
In short, it is about the legal history between AMD and Intel. AMD and various other companies acted as a second-source for Intel chips, which meant that a license was in place between Intel and the second-source companies. Intel supplied the licensed designs to these companies, so they could build the Intel designs as-is, which meant that these second-source chips were 100% compatible and interchangeable with the Intel ones. This wasn’t necessarily Intel’s choice, but IBM insisted that if they were going to use Intel as a supplier for their PC product line, they wanted secondary sources, to rule out supply problems, if anything happens at Intel (imagine that… we have come to know Intel as ‘Chipzilla’, but back then apparently they were still a small and potentially unreliable source to the giant of that time, IBM).
But in the era of the 286, things started to change. On the one hand, IBM was losing its grip on the market, now that clones were flooding the market. Which meant that IBM could no longer put such demands on Intel, as Intel could in theory do without IBM as a client now. The market was big enough. So Intel wanted to get rid of the second-source deal. They did not provide designs for the new 386 CPU to second-source companies, and they did not renew the deal, so it would just run its course and that was that.
On the other hand, some second-source companies would not just stick to the designs provided by Intel. For example, NEC created the V20 and V30 CPUs, which were pin-compatible with the 8088 and 8086 respectively, but implemented the instructionset of the 80186, and had some performance-tweaks. Other companies would not go quite as far, but would still modify the designs to fine-tune the performance for their manufacturing, and allow higher clockspeeds. So eventually we would get 808x-like CPUs at 10-12 MHz, and the 286 was eventually pushed to 20 and even 25 MHz (by Harris, a company that has since left the mainstream CPU market).
So, if Intel didn’t provide AMD with the 386 designs, how did AMD build a 386 anyway? By reverse-engineering. And this is where AMD is starting to do dubious things, in a legal sense. AMD figured that the license deal SHOULD have given them access to the 386 design, and by that logic their reverse-engineering would not be illegal.
Clearly Intel did not agree with that, and sued AMD for their 386, which was based on Intel’s design, so technically it was an unlicensed copy. The AMD 386 was held up in court for many years. Intel released the 386 in October 1985, and AMD had their copy ready in 1990. By March 1991, the case was finally settled, and AMD could release their Am386 onto the market.
Now, people commonly think AMD won the case, meaning that Intel was in the wrong, and AMD actually *did* have the rights, according to the license. But if you actually read the outcome of the lawsuit, Intel was technically correct, and AMD technically infringed on the license. However, the judge ruled that x86 was too important for the market for Intel to have a monopoly on the instructionset. So Intel was forced to offer AMD a new cross-license agreement. As the Harvard article writes:
While the arbitrator found that Intel was not obligated to transfer every new product to AMD or accept any of AMD’s products, he also decided that Intel had breached an implied covenant of good faith to make the relationship work. Id. at 1013. But this legal strategy paid off for Intel. By refusing to give AMD a license to the 386 and leaving the issue tied up in a lengthy arbitration, Intel excluded AMD during a critical period of growth in the personal computer market, during which the PC platform’s market share grew from 55% in 1986 to 84% in 1990; leaving Apple’s Macintosh at a distant second place with just 6%. Without access to Intel’s technical specifications, AMD took over five years to reverse-engineer the 80386, which was substantially more complicated than and contained almost ten times as many transistors as the 8086.
So in essence, this is a more or less arbitrary decision, based on antitrust laws, rather than an enforcement of license terms. Technically speaking, the Am386 wasn’t legal, and Intel was right, legally. But it was *made* legal by the ruling. And that is unusual. If Intel wasn’t as big and successful as it was, it should just have been able to claim ownership, copyright and trademark on its 386, given that Intel was the sole designer and developer of the product, and had retained all rights. But as you can read, the fact that the x86 became so successful, and it was so difficult for AMD to reverse-engineer, was used against Intel, and in AMD’s favour.
Which is why I would say: AMD should never have made the Am386, from a legal standpoint. But they just bluffed their way through, and managed to crawl through the eye of the needle. I suppose sometimes crime does pay.