Now that the dust is beginning to settle on IBM Corp’s devastating announcement of $6,000m in pre-tax charges against fourth quarter figures, accompanied by another 25,000 job losses, after 40,000 in 1992, it is becoming clear that if anything, the company’s problems are being under-estimated, certainly not over-estimated. For a start, the size of the […]
Now that the dust is beginning to settle on IBM Corp’s devastating announcement of $6,000m in pre-tax charges against fourth quarter figures, accompanied by another 25,000 job losses, after 40,000 in 1992, it is becoming clear that if anything, the company’s problems are being under-estimated, certainly not over-estimated. For a start, the size of the charge is out of line with the number of additional people to go, making it certain that still further job cuts are in the pipeline, so that employment, currently at about 300,000, may not start to stabilise again until it gets down to around 200,000. That means still more charges this year, when the prospects for any net profits at all look dire. The US is recovering slowly from recession, but the now significantly more important European market is plunging downwards, and the outlook for Japan this year looks horrible.
Moreover it is not at all clear that the recovering US market will translate into an uplift in mainframe sales: much more likely that large users will take the opportunity to rethink their entire data processing strategy, and in most cases, a slow move away from mainframes for all new purchases now seems much more likely than it did a year ago: if IBM is cutting investment in mainframes, shouldn’t any user not completely locked in make sure that he isn’t left holding any mainframes when the music does stop. Gartner Group has been muttering about an $18,000m hole appearing in IBM’s revenues as the mainframe starts to fade: the assumption after the latest round of cuts has to be that that chasm is going to start opening up this year. It is difficult to see how IBM can save the mainframe now: its one hope was to do something similar to what Sperry Corp did with the 1100/60 when it built a superscalar machine – it had two parallel instruction paths through the CPU – out of off-the-shelf ECL bit-slice microprocessors – dramatically reducing the cost of building the machine. The difficulty now is that IBM has hopelessly devalued the mainframe in users’ eyes by discounting it so heavily over the past year or two, which means that any new technology machine conceived specifically to give IBM a quantum leap forward in competitiveness would have to be priced dramatically more cheaply by comparison not with its current list prices but with the heavily discounted prices it has been offering, and it seems unlikely that that is possible. It has made things worse, too, by leaping into facilities management and encouraging customers to see it as a Good Thing. For the operator, a single facilities management contract is a dead loss: the way you make money out of the business is to get 10 contracts in the same region, and consolidate all the business on half the number of – albeit somewhat bigger mainframes that the customers had in place. That’s a whole lot of mainframes IBM won’t now sell, even if the whole customer base remained true to Big Iron. While IBM is not going to disappear altogether – not yet awhiles anyway, its dominance of the computer industry is over for the forseeable future, and probably for good. If it is lucky, it will slip back and become a worthy and solid, if dull company like another one-time high-flyer, Xerox Corp, which had to undergo a crash course in Japanese design and manufacturing to make anything of a comeback in copiers, dabbled fairly unsuccessfully in financial services, but has never recaptured the spark that made it a Wall Street darling three decades ago. If IBM is unlucky, it will be more like General Motors Corp, a beached whale floundering from crisis to crisis. There seems little doubt that in a couple of years’ time, it will be a very much smaller company than it is today: it is now only just over twice the size of Fujitsu Ltd, between three and four times the size of Hewlett-Packard Co, but the likelihood is that once it really begins to shrink, it will shrink so fast that it will cease to be the world’s largest computer company well before the end of the decade. Unisys Corp was the second largest US computer
company at $10,000m when at its creation; it shrunk 30% before it stabilised, and in the meantime has been passed by first Digital Equipment Corp and then Hewlett-Packard Co, and has been caught by Apple Computer Inc.
Great god mainframe
If IBM does as well as Unisys has done, it will stabilise at around $48,000m, but even that may be optimistic. The problem is that IBM’s promising businesses have been shamefully neglected in the worship of the great god mainframe, so that the AS/400 and RS/6000 businesses are not nearly as big as they needed to be to start filling in the yawning chasm. Where the AS/400 is a $14,000m a year business where nearly all new sales are upgrades, it should by now be at least a $20,000m business, and should be stealing market share from IBM’s competitors. The biggest AS/400s aren’t nearly big enough, and the machine will soon start suffering from the stigma spilling over from the top-end mainframes of being a proprietary system: to be assured of a solid future, it now needs the same kind of root-and-branch makeover that created the AS/400 from System/38 – this year, and there is no sign that anything that radical is in the plan. Equally importantly, the kinds of companies that were first drawn to the AS/400 grow so fast that the top models in the line need to be far bigger than they are now, and should always be just too big for the biggest customers, so that those customers aren’t eternally worried about running out of steam and having to divert resources to make do and mend with a less than satisfactory stop-gap. As for the RS/6000, as jokes go, it’s a very good joke. But if, on IBM’s projections, it grew 50% last year, that means that it represented just $2,250m of business, or a mere 3.9% of the total business IBM is likely to report for 1992. Data General Corp would sell its former chief Ed de Castro’s soul to be doing $2,250m in Unix now, but IBM isn’t Data General and it can’t meet the payroll on a string of Data General-sized businesses. As for the personal computer business, there were never enough reasons for IBM’s mainframe users to buy PS/2s rather than Compaq Computer Corp machines: with the mainframe base facing meltdown, even that quasi-captive market will soon vanish. It’s probably too late to come up with PS/2 models that are so much better terminals for AS/400 users that none of them would touch anything else, so that the personal computer hardware business is just another commodity business for IBM – very big, but very subject to the periodic booms and busts of the whole market and desperately thin on margins.
OS/2 is said to have cost upwards of $2,000m in research and development: if so, that’s $2,000m down the drain, which may not seem much to a company that is flushing away $10,000m in pre-tax charges with last year’s figures, but it still feels like a fat wad to the rest of us – and is now desperately needed if the AS/400 is to be saved. Whistling in the dark, IBM talks of 1m copies having gone out last year – at the same time as Microsoft was shipping 1m copies a month of Windows 3.1. OS/2 is as relevant as Unisys’ CTOS – a niche operating system that no-one would feel the need for or bother to invent if it wasn’t there already. IBM could have a big commodity chip business one day but it will be a whole lot harder to build one now that it would have been five years ago when the need for cash was so much less pressing. The same applies to the disk drive business: IBM has what are by all accounts some very good 3.5 drives for the OEM market, but so many of IBM’s OEM customers and business partners have been burned in the past that there is a natural reluctance to commit to an IBM product now, however attractive and however cheap it is. Better and safer to pay a few dollars more for a bit less to Seagate Technology Corp or Conner Peripherals Inc or Quantum Corp, who can be trusted to take care of their OEM customers. Once one realises just how dire is IBM’s plight, it is understandable that calls for John Akers’ resignation are not louder: he led the c
ompany into this mess, and deserves the task of leading it out.