IT'S PCs VS. MAINFRAMES--EVEN AT IBM
Is IBM really serious about remaking itself? That has been the question on the minds of analysts, investors, competitors--even IBM employees--ever since Chairman John F. Akers last year called for a thorough overhaul of Big Blue's operations. While Akers' reorganization--a plan to chop the $65 billion giant into a series of more manageable Baby Blues--is still very much a work-in-progress, this fall IBM is showing that its reform movement is dead serious.
The sign? The IBM mainframe, the biggest cash cow in the computer industry, is no longer the sacred cow of Armonk, N.Y. For years, while everybody from Digital Equipment Corp. to ambitious startups took aim at the mighty mainframe, within IBM there was a giant "No Hunting" sign posted around the mainframe business. Sure, minicomputers, then microcomputers, began to match multimillion-dollar mainframes in processing speed. But because the mainframe business was so profitable to IBM and because all of IBM's top managers, including Akers, had risen through the mainframe ranks, nobody within IBM dared poach on the large-systems business. That has helped preserve IBM's mainframe market share. But it has also meant that, increasingly, IBM's minicomputer and microcomputer divisions were fighting with one hand tied behind their backs.
Not anymore. IBM management seems to be facing up to the inevitable--that revenues from sales of traditional mainframes will be flat, at best, for the foreseeable future and that the big machines are losing ground to the relentless advances of newer technology. And if IBM's Enterprise Systems Div. is going to lose mainframe sales to minis and micros, then they might as well be IBM minis and micros.
GREATER FREEDOM. In other words, Big Blue is freeing--and in some cases, it is encouraging--other divisions to encroach on the company's once sacrosanct mainframe turf. On Sept. 1, for example, the Application Business Systems (ABS) group came out with a new AS/400 minicomputer that competes directly with small IBM ES/9000 mainframes. And, according to ABS General Manager John M. Thompson, still larger models are in the wings.
What's more, IBM's Adstar Div., which is responsible for disk drives and tape drives, brought out a so-called disk array for the AS/400--a gang of small, relatively cheap disk drives that together provide mainframe-caliber capacity, reliability, and performance. In years past, IBM headquarters would likely have blocked such moves. But Ray S. AbuZayyad, the general manager of Adstar, says Akers has encouraged IBM's business units each to act more entrepreneurially, in their own interests--even if that means smashing old barriers. "None of us is taking each other for granted," he says.
In personal computers, as well, Big Blue has removed the upper limits. The company's IBM-PC Co. subsidiary, which was formed Sept. 3 (box), is building PC-compatible network servers that sell for as much as $150,000, designed by startup Parallan Computer Inc., of Mountain View, Calif. Working in conjunction with a network of desktop PCs, a Parallan computer can take on big commercial mainframe jobs such as processing retail transactions for as little as one-tenth of the cost. Under the old IBM rules, such a blatant overlap with the mainframe business would have been unthinkable.
"You'll see more overlapping," promises ABS General Manager Thompson. He describes the new phenomenon as "healthy" because it plugs the "gaping holes" that IBM's centralized management inadvertently left open. While trying to protect the mainframe from "self-impact," IBM created opportunities for competitors, Thompson says.
This has been particularly true at the low end of the mainframe market, where for several years customers at medium-size companies have been saving money by "downsizing" work to lower-cost minicomputers and to PC networks. To help IBM win more such business, ABS is giving the AS/400 extra hardware power and software functions, Thompson says. The ABS chief even hired 150 engineers from a mainframe-development facility in Endicott, N.Y., when the laboratory's activities were merged with those at IBM's central mainframe-development site in Poughkeepsie, N.Y.
PARALLELISM. Meanwhile, Big Blue's own software group has started to redirect a key initiative away from mainframes. Called AD/Cycle, the scheme was designed at least in part to keep mainframes at the center of computer networks. Mainframes were to be "repositories," storing the key information needed to run a corporation. Now, AD/Cycle is being reworked so the repository can be a microcomputer-based server.
Even the folks at Enterprise Systems are figuring out how to come up with more appealing alternatives to conventional mainframes. In mid-September, the division is expected to outline plans for Genesis, a line of systems based on work begun at Groupe Bull, the French computer maker in which IBM bought a 5% stake early this year. The machines will be IBM's first so-called massively parallel computer products, designed to incorporate hundreds or even thousands of powerful microprocessors, such as those used in IBM's RS/6000 engineering workstations. A handful of those machines already can do the work of traditional supercomputers, but Genesis will handle commercial jobs, too, such as recording thousands of banking transactions per second. Genesis isn't likely to ship until next year, however.
Is the mainframe dead? Not any time soon, if IBM has anything to say about it. Big Blue has a 75%-plus worldwide market share in the big-ticket, high-margin systems. Even now, that's an awesome cash machine. But it looks as if IBM has indeed read the handwriting on the wall and is now willing to kill a few mainframe sales itself. Not only will that deny sales to IBM's competitors, it may give the company a stronger hand in the markets that will be most important in the future, such as servers. And it helps create a truly new Big Blue.John W. Verity in New York, with bureau reports