Remember the time when IBM Corp effortlessly dominated the computer systems business? How long ago it seems. Today, IBM has lost its power to dictate standards in any part of the business, and we have a new monopoly to worry about, that of Microsoft Corp and its enthusiastic cheer leader, Intel Corp. IBM was originally able to dominate the computer industry not because of any inherent excellence of its architecture and system software – in the late 1970s, specialists would have agreed that ICL Plc, Burroughs Corp and Sperry Corp all had better architectures and operating systems than IBM, but, IBM had built itself to the point where it was 10 times the size of any of its rivals. The cost of developing applications, and the penalty for getting it wrong, were so great that data processing managers became as risk-averse as trustees investing mites for widows and orphans. They had seen Xerox Corp abandon its mainframe business to a Honeywell that ran it for cash, while conducting disconcerting hot-and-cold affairs with Parisians, seen RCA Corp hand its mainframe business over to Sperry, and watched it fade away, and if they were Burroughs or NCR users, wondered whether their supplier would be next. And with each new generation of hardware, more and more decided it would be safer to migrate in their own time to the one company that clearly was not going to go away than be forced to do it at a point of maximum inconvenience in an unpredictable future.
Today, of course, IBM’s power to dictate standards is confined to the fading mainframe business and no-one cares much any more – and even in mainframes, IBM’s power is extremely circumscribed. It failed to establish Serial Storage Architecture and has ceded dominance of mainframe storage to EMC Corp to the point where it has had to rush to one-time arch-rival Storage Technology Corp for a remedy. As for the PowerPC adventure, it has turned into an embarrassing fiasco and a spreading stain on King Louis Gerstner’s escutcheon. All that has been achieved is that the PowerPC is used in the fading Macintosh, but the hidden cost has been huge. The decision to move the AS/400 over to the PowerPC is mocked by the fact that the PowerP C AS is only partly compatible, minimizing the benefits of economies of scale – at the price of largely paralyzing the AS/400 base for three years at a time when it was absolutely vital for the health of the line to that IBM made a quantum leap in functionality just to persuade the System/36 users to migrate, let alone attract mid-range virgins to the line. The decision instead to devote vast internal resources to the migration to PowerPC has probably doomed the AS/400 to fade away early next century. IBM today looks moderately successful again because, now that it has cut an enormous swathe of costs out of the business, it is so big and is in so many sectors of the market that – unlike Digital Equipment Corp which lacks its scope – it is cushioned from the full impact of poor performance in any two or three activities – remember disks, personal computers and the AS/400 were all doing badly last year.
And Louis Gerstner and his team have battened everything down and made the company reasonably stable, where the reverse is the case with Robert Allen and his team over at AT&T Corp, which should be a lot easier to run than IBM but begins to look lik e General Motors Corp when everything started falling apart. But IBM has has no chance for the foreseeable future of again dominating the industry as it once did. It has to settle for being as respected and successful as Xerox Corp while fearing that at times it will be regarded as as much of an under-achiever as Eastman Kodak Co. IBM’s twin nemeses were the original IBM Personal Computer with its MS-DOS operating system – to which, CP/M and the Z80 played John the Baptist – and the plunging cost and soaring advance in performance of CMOS, which has made the Pentium Pro as powerful as 90% of IBM’s mainframes of five years ago. The fact that by 1985, users could buy a personal computer from any manufacturer and know that shrink-wrapped software bought off the shelf would run happily on it was the most phenomenally liberating experience to users used to being locked into one hardware and software vendor. And users and IBM’s corporate victims alike began to wonder why this model was not being extended up into the mid-range. Unix seemed the least inadequate vehicle, and all the systems builders IBM had defeated embraced Unix with varying degrees of enthusiasm. But they fatally flawed the model by deciding they couldn’t compete on service, price-performance and by writing superb vertical market applications alone, they had to differentiate their various implementations of Unix by adding proprietary features. And instead of recreating the personal computer model, it was as if, when Philips Electronics NV and Sony Corp launched the audio Compact Disk in 1980, they implemented the standard differently, so that to run Polygram NV disks on Sony players, Sony disks on Philips players, you had to throw some DIP switches inside, switch them again for EMI Group Plc disks, while all the record companies had to produce several formats of each recording. And as a result, the Unix world begins to fear that it has been fattening frogs for snakes as its users begin to regard Windows NT, owned by a dictator, as the fulfillment of Unix’s promise. Just as two decades ago, users dictated IBM mainframe compatibility to the industry, today it is the so-called independent software vendors, which sound like software retailers but are in fact software developers that are dictating conformance to the Windows-on-iAPX-86 application programming interfaces, to the detriment of the Mac and of all the less than fully compatible Unixes, the AS/400, and ultimately the mainframe – and to all processor architectures apart from iAPX-86. The likes of DEC with Alpha and IBM with PowerPC will always find it an uphill struggle to keep up with the standard even running NT in a Windows NT world, because their machines will always either be more expensive or much less profitable than ones that run NT on the Pentium Pro and its successors – as what is clearly currently the wisest company in the business, Hewlett-Packard Co, has so coolly recognized. Windows NT is the only major piece of software Microsoft is implementing for several processors – PowerPC, Alpha, R-series, Sparc perhaps – as well as Pentium Pro, and is no doubt doing it in the certainty that all versions apart from Pentium Pro will fail, so that it can turn to the regulatory authorities and say look, we tried to be open and support multiple architectures at great cost, and look what happened. It would be absurd to characterize the computer industry as a natural monopoly as are activities such as water supply, but without active regulatory intervention the industry in its present form will always tend towards a monopoly.
