AT&T Corp’s Robert Allen has rather awkwardly put the spotlight on his oppo over at IBM Corp, and in light of Allen’s bold move to break up AT&T before it is clear that the company in trouble puts into stark relief the contrast with IBM Corp, which is still in big trouble and, many believe, would be doing a whole lot better as a clutch of independent companies. Now that Allen has bitten the bullet, Gerstner is going to have to do a whole lot better than his facile remark that because most companies in the industry are forming alliances with firms with complementary skills, IBM was right not to break itself up. That presupposes that IBM is the best company in the industry at everything it does and that if it were looking for a complementary partner, each part of IBM would choose its sibling rather than an outside company. But after the fiasco of Ramac 1, do you really suppose that the mainframe side would really choose the IBM Disk Drive Co instead of EMC Corp?

Loveless marriages

In tapes, wouldn’t it be rethinking its allegiance to Magstar and be talking earnestly to Storage Technology Corp or one of a number of other tape drive suppliers? Now that the chips for the second generation CMOS mainframes turn out to be rather less powerful than promised, might the mainframe folk like to be free to turn to one of the Japanese manufacturers that is really skilled in super-fast CMOS? Can it really be true that after 18 months of a crash programme to use common parts and subsystems throughout the company wherever possible, the CMOS mainframes still don’t use the same power supplies as the SP2s and the high-end AS/400s? If they did, IBM would surely have enough of the things in stock and on order that it would not now be suffering the embarrassing delay to shipments of its flagship product for the autumn season. But if they don’t, what’s the point of those three competing product lines being in the same company? Where’s the synergy? Where’s the benefit? If IBM doesn’t have one person specifying a single power supply for each of the three or four levels of power conversion that must be all its product lines need, then it is gaining no benefit from having all those competing products in one company. Split it up! For years, IBM has been like an extended family of people all living in the same vast and rambling house, all locked in loveless marriages. Much healthier to let them all go off and get into bed with people they actually like, instead of forcing the AS/400 people and the RS/6000 people to live together in unholy matrimony and allow ill-disguised hatred to fester. Even the move to common chips between the AS/400 and the RS/6000 is not working: gossips are saying that those in charge of the latter have knocked back the PowerPC 620 for future RS/6000s because it does not offer sufficient performance improvement over the Power2 chip set. IBM boasts that the AS/400 software has been written in such a way that it is a breeze to lift it up off one processor architecture and plonk it down on another: suppose the next PowerPC earmarked to go into the AS/400 turns out to be an embarrassing disappointment, wouldn’t the boys in Rochester like to be free to do a quick rewrite of the underlying code for the forthcoming Hewlett-Packard Co-Intel Corp RISC? The inescapable fact is that the AS/400 and the RS/6000 will never achieve their true potential while they remain under one corporate roof: a quickie divorce is the answer.

Plonk it down

Deciding how to split IBM up is tricky, not because it is difficult but because there is more than one option for most parts of the company. The AS/400 is big enough to stand on its own feet, and on its own would be much better placed to work out the best strategy for containing the twin threats from Unix and Windows NT. Let it go, and instead of selling things like Mapics, send it on a buying spree for AS/400 applications developers whose products wouldn’t be converted to run under Unix. AS/400 processors would be bought in, as would all peripherals. Embarrassingly

, given IBM’s 1990 and 1991 boasts the thing would be number two or number one in the Unix world by now, the RS/6000 is too small to survive and thrive on its own against the likes of Sun Microsystems Inc, and it is not clever enough to succeed by differentiating itself in the way that the smaller Silicon Graphics Inc has done. It therefore has to remain with the mainframes, bringing the wretched Power Personal Systems business with it; the mainframes will be free to get their chips made by the best CMOS merchants in the business. Mainframe software has to stay with the mainframes, but maintenance might be better off going its own way with a five-year contract from IBM under its belt, and become part of an independent Integrated Systems Solutions Co, which would have to find itself a less ridiculous name. It is unarguable that the storage systems business has to cut out on its own: it has superb technology coming out of its ears, and would have to take with it all the development people that have fed it up to now, but its contracts with its former siblings should be far shorter than the usual five years: the sooner it is clear that the company is unviable on its own and it gets gobbled up by someone like Seagate Technology Inc that knows how to sell OEM disk drives and get major new products to market on time – and working – the better.

It would sink

The same rather seems to apply to the chip business: perhaps floating it as a free-standing company would be fairer to shareholders, but a trade sale to Motorola Inc, which has already started runing rings round IBM in PowerPC designs, would really be the best solution for all concerned. The personal computer business? Cut it loose! Stand alone. Sink or swim! All current indications are that it would sink, but given that even a dog of a business like Lexmark International Inc can do sufficiently well freed from the dead hand of Armonk that it now reckons it is ready to make an initial public offering of shares suggests that even the personal computer business might surprise us all. Despite all the unaddressed structural problems and looming disasters when the mainframe business really begins to fade, Lou Gerstner would be Wall Street’s bluest of blue eyed boys but for one thing. IBM may have a lovely balance sheet and a spiffy profit and loss account, but it ain’t growing, and it is extremely difficult to see where growth is to come from. When AT&T was separated from the Baby Bells, it was an anaemic little $33,000m a year company: today, what was then Long Lines alone is a $49,000m business, and hardware is another $20,000m – and Bob Allen reckons – and we’re sure he’s right – that each will be easier to manage and will grow faster apart than they would have done undivided. John Akers got many things wrong, but he was finally on the right track just before the boom fell: now IBM’s recovery appears stalled, Lou Gerstner needs to have the courage to admit that he was wrong: for IBM’s sake, break it up!