The gap between the narrow quarter-by-quarter focus on much of Wall Street and the kind of trends-based medium-term analysis that more prudent investors require could not have been put into sharper focus than it was by the Wall Street Journal’s analysis of just where IBM Corp is today. The piece, which appeared in US editions on Friday, made it clear that IBM’s storming 1994 fourth and 1995 first quarter performances were generated on the back of an unusual confluence of happy circumstances that can hardly be relied on to keep the pot boiling in the medium term. And the piece highlighted an aspect that is worrying fans of chairman Louis Gerstner: for the first time since he arrived, and in stark contrast to what he was saying even two quarters ago, he seems to have caught the infection that so tainted all his predecessors: be is beginning to believe IBM’s own publicity. True, he cautioned that the comparisons for the most recent quarter were not at all challenging, but the euphoria with which he greeted the annual meeting is disconcerting. The biggest difficulty faced by analysts of IBM is that the company is so big that by its very nature, execpt during the very worst times, it generates cash in such enormous quantities that the mighty tide can be manipulated to mask almost any underlying weaknesses quarter by quarter. The problems faced by anyone analysing IBM without rose-coloured spectacles is thrown into sharp relief by Daniel Mandresh of Merrill Lynch & Co, who is a bull of IBM. When he looks at IBM piece by piece, Mandresh admitted to the Journal I reach bad conclusions. But you have to consider the bigger picture, he says: IBM is so huge that profits can swell
Errors in forecasting
on just so-so gains in revenue. He is going for record earnings per share next year, and IBM might manage it – but if it does it will be because its errors in forecasting have come to its aid rather than creating problems. And that is what is happening now: demand for monolithic mainframe MIPS is much higher this year than IBM forecast. That means that it is sold out of really big mainframes for the rest of the year. That hardens prices in the market, and the premium compared with the levels to which it had to discount two years ago feeds straight through to the bottom line. But the Journal pins down what we have sensed but for which we have had little hard evidence: the demand is all for the big old monolithic machines, not the new CMOS multiprocessors. That is particularly good news in the short term because the former carry higher margins. But IBM plans to phase out manufacture of the monolithic machines, and specifically does not have any successors on the drawing board. That might not have been too much of a problem if the competition had bought the IBM ticket. But Hitachi Ltd and its acolytes have just announced a new generation of top-end monolithic machines, and the Amdahl Corp camp is still in the monolithic game too: those companies, which at the best of times serve only a fraction of the market, have long known how to make money on much smaller volumes, and calculated that they can still do good business even in a declining market: by announcing termination of the top-end machines, IBM has potentially handed them a much bigger share of a shrinking pie on a plate. There is no question that the one-to-six or -eight way CMOS machines are extremely interesting to much of the low-end and mid-range mainframe market: what we find much harder to swallow is that the biggest users, traditionally IBM’s most profitable source of business, will happily switch to clustered 48-way configurations with all the attendant risks of major transitional problems.
Software time bomb
Moreover with IBM devoting its resources to making existing software run acceptably on these very different configurations, its development dollars are being diverted from adding all the new things that major users have always demanded of MVS year by year. Much of the existing software is effectively on a care-and-maintenance
schedule until the transition is complete. The Journal records that the CMOS machines account for only about a third of total MIPS being shipped by IBM, and, worse, will only account for about 16% of mainframe revenues because it is priced much more keenly. Yet IBM had planned to end monolithic mainframe production this summer. And that two thirds-one third split is exactly what one would expect if the smaller users were happily taking the CMOS machines while the larger users were sticking firmly to big iron. IBM is forecast to ship 200,000 MIPS this year, up 29% on last year, yet Mandresh concedes that they will generate 5% to 10% less turnover. And there is the software time bomb: prices have to fall if IBM’s mainframes are to compete with client-server systems from the likes of Hewlett-Packard Co and Sun Microsystems Inc. Even the most ardent adherents concede that mainframe revenues will decline 25% next year – yet this is the business that is still generating more than half the profits. Everything below the mainframe has to subsist on much thinner margins, and even if IBM finally gets the personal computer business right by the end of the year, a Personal Computer Co firing on all cylinders and taking back ground lost to the likes of Compaq Computer Corp cannot be expected to generate margins better than a third of those IBM has traditionally enjoyed on mainframes, margins it is beginning to approach again, but for all the wrong reasons. Some analysts do worry. If anything, IBM is almost as dependent on the mainframe as it was when it started losing money four years ago, Gary Helmig, a former IBMer who follows the company for SoundView Financial Group told the Journal. But this year, 80% of the users SoundView regularly surveys plan to migrate applications off the mainframe and onto smaller, cheaper systems. I’m very excited for IBM to think the mainframe is still their key business, because every day t hey’re doing that is another day they aren’t getting on with life, says Sun chief Scott McNealy. The AS/400 does not even get a mention in the Journal piece, but IBM again has two choices – either a very profitable AS/400 business in terminal decline, or a much less profitable business that grows. The company has never been able to reconcile the fact that the AS/400 and the RS/6000 – and increasingly servers from the Personal Computer Co – address exactly the same markets, with the result that neither does as well as it could if the other were not there in the same company. When IBM was enormous and the competition was starved, those kinds of conflicts didn’t matter: IBM could just about get away with its boast that far from confusing the customer – which it did, endlessly – it was simply offering the customer more choice. But those were the days when the next nearest competitor to IBM was one tenth the size, and in each individual business – software, services and so forth, IBM’s business was 10 times the size of the nearest free-standing competitor.
Legendary IBM hubris
But today, when Hewlett-Packard and Fujitsu are each almost half the size of IBM – and growing much faster – it matters very much and is something IBM can no longer afford. It desperately needs the kind of harmony that reigns between Hewlett-Packard Co’s twin product camps, the HP 9000 and the HP 3000, where a well-thought-out strategy laid down more than a decade ago enables Hewlett to get the maximum out of each line. Most worrying for IBM’s acolytes is the return of the legendary IBM hubris: In some ways, IBM has changed Mr Gerstner – who declined to be interviewed for this story, writes the Journal – as much as he has changed IBM. Like old-time IBMers, it says, he dismisses companies in the personal computer industry as a ‘tiny set of the business’ – and that at a time when Compaq alone will grow to be a quarter the size of IBM this year. He depicts, the Journal goes on, Microsoft as a muscle-bound marketer with unimpressive technology. He writes off the information highway as ‘90% hype.’ Yet he is fiercely commi
tted to challenging Microsoft with personal computer software. Despite the glaring weaknesses, IBM’s admirers are unconcerned, concludes the Journal. Each quarter they’re coming through with better earnings, says Charles Mayer, manager of the Investco Industrial Income Fund, with about 600,000 IBM shares, who does not worry that much of the profit comes from mainframes. They’re sold out through the end of the year – that’s not all bad. That’s the kind of thinking that got IBM into such deep trouble in the first place, and the kind of thinking that will fulfil the prediction that rather than resume real growth and retake the mantle of industry leadership, could turn into another Unisys Corp. And sad to say, a Un isys 10 times the size would still be no match for a Hewlett, an Intel Corp, a Microsoft Corp.
