Intel’s European Manager of Components Marketing, Hans Geyer, addressed a press conference in London recently to reassure the world, in case it was wondering, that Intel and the iAPX-86 microprocessor architecture is going to squeeze all other architectures into the margins of the market by the year 2000. The company is keeping up with its […]
Intel’s European Manager of Components Marketing, Hans Geyer, addressed a press conference in London recently to reassure the world, in case it was wondering, that Intel and the iAPX-86 microprocessor architecture is going to squeeze all other architectures into the margins of the market by the year 2000. The company is keeping up with its prediction that it will double the number of transistors on one die every 18 months the 80586 will be two to three times as fast as the 80486 and will have 4m transistors on the die. Three years after that chip is launched the 80686 is planned to appear which will be two or three times as fast as the 80586 and will have 20m transistors on the die. Following on from the announcement of the 386SL (CI No 1,533), Intel reckons that it will have an AT motherboard on one chip by 1993. Geyer believes that the differences between various personal computers in the future will come not from the microprocessor implemented, but from things like communications software, speech interfaces, video and so on. However, Intel no longer dreams simply of dominating the personal computer market – its iAPX-86 architecture will stretch from the notebook to the mainframe courtesy of companies like NCR Corp – it will be a top-to-bottom architecture and Intel says it will account for 85% of the world’s entire installed hardware base by the year 2000. So what makes Intel think that it alone has the capability to survive when it believes others are doomed to fail? Basically, it thinks that its financial situation puts it in a winning position. Geyer says Intel has $2,300m cash in the bank which insulates it from the outside economic climate. He continued, saying its success is driven and will continue to be driven by the 32-bit microcomputer market. There is one enormous problem with the 32-bit market, however, and that is that although so far the hype of having the newest and fastest processor has ensured the growth of the 80386 personal computer market – so that since April, 80386 volume shipments have overtaken 80286 volume shipments in the US, a crossover that happened at the end of last year in the UK and that is expected to happen at the turn of this year on the continent – there are few practical reasons for buying into 32-bit architecture as yet for the simple reason that there is a dearth of 32-bit software. Geyer said that there is an enor-mous business opportunity for software companies when one considers the installed user base of 80386 users that are already out there in the market and for whom software does not exist. As an example, he cited the phenomenal market success of Windows 3.0. He admitted that more software was needed to drive the continued evolution of the 80386 market.
To this end Intel has a team writing back ends for software compilers and also provided AT&T Co with a multi-processing kernel for Unix. Geier continued by saying that the 80386 is already outshipping the RISC market and added swiftly that Intel will catch up with market demand for the 80386 and the 80486 by the end of this year. Intel now has the capacity to ship more than 10m 80386 microprocessors per year. As for second sources for the chip, well, the industry doesn’t need them, does it? Geyer explained that with a traditional second source agreement, the manufacturing process would be different, therefore the product would be different. Second sourcing would require a company with tremendous financial resources, and frankly such an agreement would be cumbersome. Cumbersome, no doubt to Intel not quite so cumbersome, perhaps, to system vendors able to get less expensive chips. Moving swiftly on, Geyer pointed out Intel’s leader-ship in research and development where the spend hovers between 12% and 18% of revenues depending on how good or bad the year was. Two-thirds of that spend goes on the development of high end microprocessors. For example, the 50MHz 80486, which is due to ship in high volume next year, is manufactured using the 0.98 micron triple CMOS process, enabling Intel to shrink the die size
by 60%. The 80586 chip will be fabricated using a 0.67 micron BiCMOS process. Intriguingly, Geyer said that the iAPX-86 architecture will not be 100%-owned by Intel when it has 85% of the market, but, he said Intel had no plans to license the architecture.