Dell Advert - July 1987
From Personal Computer World
You'd better believe this... or you won't believe our prices
Michael Dell started out at the age of 13 selling mail-order stamps, and by the time he was at High School he was earning some $17,000 a year selling newspaper subscriptions with the aid of an Apple II. Whilst at college, he started buying parts to upgrade older IBM PCs that dealers were desperate to get rid of and would re-build them to order, by asking customers what they wanted, phoning up a parts supplier in Dallas, flying up there to collect them and hauling the whole lot back in a car and trailer to Austin. By the age of 19, Dell was making over $30,000 a month, at which point he dropped out of college to run his new company full time. By 1987, at the age of 22 and only three years after starting PC's Limited, his company was turning in revenues of $70 million a year. Compared to IBM, Apple or Compaq's revenue, this might have been small beer, but it proved that the PC market still had room for new ideas.
Michael Dell, © Personal Computer World, July 1987In the summer of 1987, PC's Limited launched in the UK as Dell Computer Corporation, based in Bracknell, Berkshire, where it continued with its direct-to-consumer model. Mail order had a spotty record in the UK, with Sinclair's infamous "28 days delivery" actually meaning "some time in the next few months" having tarnished the idea. However, maybe its time had finally come - perhaps Dell could actually deliver when it said it could, or perhaps it was the fact that it actually offered proper professional-level on-site support, but whatever it was the company proved successful in its direct-sales approach.
Back in 1985, PC's Limited was the focus of attention at the November Comdex Show in Las Vegas as crowds flocked to see its breakthrough sub-$2,000 IBM AT clone. Whilst some of its early lower-spec machines were built for it by American Research Corporation, by 1987 it was designing and building its own hardware, and became one of the first companies to release a "hybrid" machine which could support the as-yet-unreleased OS/2 - the much-delayed Microsoft operating system intended for IBM's new PS/2. At the time it was expected that IBM's new Micro-Channel Architecture would become the next dominant hardware platform and that it was only a matter of time before official MCA-PS/2 clones came out, so machines like this were seen as a stop gap.
However, IBM was hoist by its own petard as the ISA market - the original IBM PC architecture that it had created - proved to be too big to usurp and MCA never really caught on. In fact by the end of 1987, IBM's PS/2 only accounted for 38% of the company's sales and just 9% of the entire market, whilst Amstrad had even overtaken Big Blue in terms of actual units shifted, selling over 100,000 distinctly non-PS/2 or MCA machines during 1987. Even by the end of 1988, the year when 386-based machines were the dominant "thing", the OS/2 user-base was still considered as "small", but it was still hoped that this would improve as more software specifically for the new operating system came to market. It was also hoped that the falling prices for the older 286 machines - which could still run OS/2 and some of which were under the magic £1,000 barrier by the start of 1989 - would encourage take-up.
Wait years for a bus, then they all come at once
There was nothing particularly new about the idea of expansion buses - which were simply a way of extending the data and memory address pins of the central processor out into the world at large - as architectures like S-100 and SS-50 dating from the mid-1970s had offered sometimes 20 slots of expansion, but the idea had somewhat disappeared in the era since the Commodore PET (although Apple kept it going in the Apple II). However, the re-appearance of the general-purpose bus on IBM's 5150 - the original IBM PC - was perhaps more significant than that machine's use of new-fangled 16-bit processors, because it represented an actual standard that third-parties could build against. This became quickly entrenched, so much so that when IBM launched its improved AT bus in 1984, it essentially kept its original bus for backwards-compatibility but lashed on an extra 8-bits to support the new 16-bit architecture, and when 386's rolled around, the AT bus was still the only game in town. However, it was getting increasing outmoded due its lack of support for 32 bits, so to address this IBM created the MCA bus for its PS/2 machines, which it unleashed in April 1987.
MCA fixed a lot of the drawbacks of the older bus and supported 32-bit CPUs, but it also obsoleted hundreds of plug-in cards overnight and it was the weight of this legacy baggage that was perhaps too much, as towards the end of 1988 a group of clone manufacturers, led by Compaq and including HP, NEC, Wyse, Epson, Zenith and Olivetti, got together to release the "Extended Industry Standard Architecture" bus standard. This group, known as the Gang Of Nine, provided a spec which - much like the change from the original IBM bus to AT - simply lashed some extra connectors to the old spec to ensure backwards compatibility, but which would also promise to be as good as, if not better, than IBM's MCA. The spec was hazy, and the challenges of supporting 8-, 16- and 32-bit cards all at the same time were significant and it was considered a bit of a kludge, but the choice was now there - between MCA, EISA and Apple's outlier NuBus.
Meanwhile, IBM was expecting that its move into MCA, and the fact that this time it was backed up with the power of patent and copyright law, would convince the rest of the world to upgrade with it. So much so that IBM abandoned the PC market completely, leaving it behind to what it assumed would be a rapidly-dwindling band of cut-price cloners. This backfired somewhat as the consensus seemed to be that "the PS/2 has poor software support and is too expensive - thanks, but no thanks". By the end of 1988, IBM seemed to have realised that the world wasn't going to switch to MCA just because it said so, and so it started licencing its new architecture to clone manufacturers such as Dell, Tandy, Apricot, Olivetti and Amstrad. However, the bus standard was still controlled by IBM, as this time clone manufacturers had to pay for the privilege to use it - to the tune of 2-5% of sales - whilst there was even a claw-back for historical "un-licenced" clone activity.
Whilst EISA itself didn't dominate, apart from perhaps in servers, the ISA part of it - essentially a standardised AT bus - did, and thanks to the fact that IBM had defensively patented the MCA bus, plus the momentum of the huge user-base that had already built up around the older AT standard, it would the ISA bus that would stick around and rule the PC world until PCI kicked it into the long grass in the early 1990s.
Meanwhile, American Research, for its part, was also selling its own machines directly via mail order - in a deal which also included some limited on-site support - which for the UK and the wider European market meant importing its machines from Taiwan but assembling and quality-checking them in Croydon. It was fairly unusual in the clone market as it wasn't selling its machine on the basis of price, but on speed, reliability and the fact that it was planning (like everyone else, obviously) to stay around in the micro industry. It clearly had some chops as its ARC 286 Turbo was rated the fastest computer in a test run by Arizona State University, beating micros by IBM, Tandy and Epson.
The end of the golden age
Dell, and companies like it, represented the end of the "golden age" of microcomputers and the start of the commoditisation and standardisation era, where the industry had more-or-less completed its transition from hundreds of companies selling hundreds of different machines to a few dominant companies and hundreds of also-rans all selling the same thing - clones of the IBM PC. Commodore might still be selling its Amiga, Atari the ST, Apple the Mac and Acorn the Archimedes, but with the exception of Apple, which managed to cling on to a 5-10% and occasionally 15% market share thanks to its ownership of the designers-and-DTP market, and - indirectly - Acorn's ARM spin-off, it was game over. This transition could even be seen in the state of once-legendary computers shows like the West Coast Computer Faire - the show which had introduced the public to machines like Commodore's PET and the Osborne 1 - which by 1986 was becoming something of a disappointment and no longer, according to a letter in Personal Computer World, "an opportunity to feel the pulse of a young, truly innovative industry", with similar sentiments continuing into the end of the decade when Popular Computing Weekly commented on how boring the 1989 Which Computer? Show had become if you were a home micro owner, with only Atari and Commodore having any sort of presence.
The conclusion seemed to be that it was the end of the gold rush of random-machines-of-the-month that ended up in the loft gathering dust, but was now the era of maturity, where the computer was no longer a hacker's hobby or a home games machine, but had graduated to a general-purpose tool in the world of work - an environment that demanded consistency, reliability and - above all - conformity.
text and otherwise-uncredited photos © nosher.net 2019