Seen from the air, IBM Rochester sprawls across northwest Rochester's landscape. The complex is the development hub and the leading production site for IBM's fastselling family of business computers, the Application System/400. IBM Rochester's 7,600 employees also have pioneered many quality initiatives now being applied in IBM plants worldwide. f IBM Rochester were an independent company, its 1991 revenues of $14 billion would make it the second-largest computer company in the world, behind IBM Corp. and ahead of Hewlett-Packard Corp. (IBM officials say the $14 billion figure would be worldwide revenues if the Rochester-devised Application System/400 computer, its IBM-made software and connecting equipment were manufactured, sold and serviced by a single corporation.)

But of course, IBM Rochester is not a separate company. It is part of a larger entity that has struggled mightily in the last few years, a company that shed 29,000 employees in 1991 and is on schedule to eliminate another 20,000 in 1992. Even in Rochester, home to the enormously successful AS/400 minicomputer, employment is down to 7,600, a drop of 500 from a year ago.

In response to its own sluggish performance, IBM has embarked on an ambitious, even radical restructuring. The plan holds risks and rewards for the Rochester operation.

In IBM's new world order, a history of success does not guarantee a division's future. IBM Rochester's survival as a development lab and manufacturing facility will depend upon how nimbly it can continue to respond to its customers and its competitors.

Not all of IBM Rochester's pressure will come from outside competitors. The restructuring of IBM into what has been described as a ``federation of independent companies'' promises greater autonomy for all IBM divisions, but also greater competition between those divisions for clients and corporate resources.

Newsletter signup for email alerts

``Decisions that we make here no longer have to be made for the greater good of IBM,'' says Robert M. Unterberger, general manager at Rochester.

What does that mean? Although IBM will maintain a single sales force, each division will, in theory, be able to more aggressively promote its own products to current or prospective clients.

``Everyone is now a competitor,'' says Unterberger, who is also director of application business systems, responsible for worldwide manufacturing and planning for the AS/400, which now competes against IBM's mainframes and the rapidly growing RISC/6000 workstations. ``I don't believe we'll be out there bashing other IBM lines,'' he says, ``but we will be aggressive in outlining the attributes of our technology.'' Keeping score

ne sign of the new autonomy: IBM Rochester's willingness to release its revenue figure, which reflects sales of the AS/400 and related equipment. Previously, IBM refused to release revenue figures for separate divisions.

``We don't track it down to the exact penny, but if you're keeping score, $14 billion is the figure,'' Unterberger says.

On another level, IBM Rochester is taking advantage of its liberation by establishing new channels of distribution, such as selling AS/400s through Wang Laboratories and the low-end models by catalog or establishing stronger business partnerships with software and equipment vendors.

One software vendor that provides imaging technology for the AS/400 has noticed the change. Metafile Systems Inc. in Rochester was recruited as a business partner when the AS/400 was first developed. The company, which agreed to write software for the new machine, was included in a database of vendors provided to IBM customers. IBM's own imaging systems division complained, however, and Metafile's status as business partner was revoked. Since the announced restructuring of IBM, Metafile is once again working with IBM.

Dave Andrews, managing partner for ADM Consulting in Westport, Conn., sees significant advantages in the restructuring for IBM Rochester. ``What they're really doing is insulating the AS/400 from the decline being felt in other areas, like the mainframes,'' he says.

IBM's restructuring might not be good news for all at IBM Rochester. Employees here develop and assemble tape and disk drives. The division has been spun out into a separate line of business, Adstar, with headquarters in California.

Karl D. Shurson insists that storage products, as well as the workers he supervises as manager of storage products manufacturing at Rochester, are important to IBM's future. But with revenue from that division down 19.9 percent in 1991, some analysts speculate the business is vulnerable.

Although Unterberger denies IBM has given each plant a quota of job eliminations, he can't rule out further cuts at Rochester. ``As the needs of business change, the need for certain skills will change,'' he says. Quality control

o far, the characteristics that produced the AS/400 and earned a Malcolm Baldrige National Quality Award in 1990 are still in evidence at IBM Rochester.

Until recently, for example, IBM's AS/400 minicomputers were assembled in a regimented process that began when a worker applied interior and exterior labels to a hollow, 5-foot-2-inch case and then bid that case goodbye. Someone else would install the power supply. Yet another person would install the storage devices, and so it went on down the line.

The process was similar to the one Henry Ford first employed almost 80 years ago.

Barb Mickelson, a production specialist, complained along with employees from each of IBM's three shifts that the process imparted no pride of ownership, no satisfaction from seeing a project through from start to finish, and, when things went wrong, no sense of accountability.

So Mickelson and her colleagues decided to try using a single worker to assemble each unit.

IBM's employees are happier now, Mickelson says, and quality has improved. Under the new system, the average number of hits per production cycle is one or two, as opposed to five under the previous system. And the average time to assemble each machine has dropped from more than 20 hours to just less than 17.

In February, IBM Rochester unveiled a new generation of AS/400 computers, the second major upgrade in just 10 months, to rave reviews. Peter Burris, an analyst with International Data Corp. in Framingham, Mass., says the new lineup reinforces the AS/400's preeminence and ``extends a heritage of excellence coming out of Rochester.''

IBM Rochester has a history of defying expectation. In 1966, IBM hired a staff of engineers and programmers to establish a development lab in Rochester. Three years later, that lab produced its first computer, the IBM System/3.

The System/3 was designed for use by small businesses. Its most notable features -- a simple programming language and, eventually a plentiful supply of ready-to-use software -- made it an instant success. In the next 14 years, the plant produced ever more powerful and popular successors: the System/32 in 1975, the System/34 in 1977. The following year, however, IBM Rochester launched a project that turned out to be a misstep. Dire straits

he name System/38 seemed to indicate that the new computer was a successor to the other machines, but it was not. In ``The Silverlake Project,'' a chronical of the development of the AS/400 written by three IBM employees, the System/38 is described as a ``leapfrog jump in technology and capability.''

Unlike the System/3 line, the System/38 was somewhat difficult to use. To make matters worse, it was incompatible with the System/3 line and its software. Customers would have to rewrite all of their data.

A substantial number of IBM's more sophisticated customers made the switch. Others waited to see what IBM would come up with next. And that's when other competitors began to take market share away from IBM.

IBM Rochester tried to protect its flank in 1983, when it introduced the System/36, which was compatible with other computers in the System/3 line. The 20,000 users who had invested in the System/38, however, were stuck. For them, the new computer was a step backward. And for System/34 users, the new model was too little too late. Many had drifted to DEC or to other midrange manufacturers; others were not impressed.

IBM Rochester had lost its technology edge in microcomputers.

IBM, sensing it was in desperate straits, began to develop a mini-computer with the code name Fort Knox, a machine that would unite all of IBM's disparate computers. In other words, Fort Knox would be compatible with each of the other IBM mid-ranges, including the System/38.

The authors of ``The Silverlake Project'' describe how, beginning in 1982, IBM's corporate office took responsibility for Fort Knox away from IBM Rochester. The incredibly complicated task -- the equivalent, the authors say, of ``combining the features of a sports car, a station wagon, a compact, a luxury car, and a pickup truck in a single, all-appealing vehicle'' -- was spread out among four development labs.

The assignment proved as unwieldy as suspected. In 1985, after three years of work and hundreds of millions of dollars in research and development, IBM abandoned Fort Knox. The AS/400 is born

he decision to abandon Fort Knox left IBM Rochester in a perilous position. For the huge installed base of System/36 and 38 machines, which were becoming obsolete, IBM Rochester had little to offer in the way of upgrades and no new products in development. A typical development cycle lasted four or five years, but IBM Rochester could not afford to wait that long.

The AS/400 was born of desperation.

Pete Hansen, a programmer and engineer, and his team of researchers in the Rochester development lab concluded that not only did they have to build a machine, but they had to do it in less than three years. They estimated the cost of development at $1 billion. IBM corporate headquarters gave its approval in early 1986.

Twenty-eight months later, the AS/400 was introduced to the world, and it became the most successful computer introduction in IBM's history. While other minicomputer manufacturers have seen sales grow at a three percent annual rate, the AS/400 has averaged 10 percent to 12 percent growth, with gross margins of 56 percent.

Unterberger, a site manager at IBM's facility in Tuscon, Ariz., came to IBM Rochester two months after it received the Baldrige award. As ABS director, he reports to John M. Thompson, general manager of the ABS division at IBM headquarters in Armonk, N.Y.

Though Unterberger is the senior IBM executive in Rochester, he shares responsibility for continuing development of the AS/400 with David Schleicher, who led programming efforts for the first AS/400. Schleicher is director of the development lab, where about 2,700 of IBM's Rochester workers are assigned; he also reports to Thompson.

Unterberger says his job is to ensure that the technical development of the AS/400 remains responsive to customer needs and that manufacturing systems are in place that will minimize the time from development.

Unterberger says he wanted to make sure when he came to Rochester that his staff did not view the Baldrige award as an end, but as a step along the way. Fortunately, he says, ``Rochester is not a team that by its nature is complacent or feels entitled to success. It's a group that already had a plan for continuing improvement.'' Customer involvement

he AS/400 upgrades introduced last February were representative of such a plan. They came only 10 months after the last upgrade, and within months of recent product announcements by Hewlett-Packard and Digital. In addition, these were substantive upgrades, offering significant improvements in price and power and available for immediate purchase.

Unterberger says the compressed development schedule was achieved by establishing ``an almost seamless'' relationship between development and manufacturing. ``That means designing for manufacturability.''

In addition, customers were brought in every week ``to help validate the design,'' Unterberger says. This process of customer involvement has become standard practice throughout IBM, but it was first used extensively by the Rochester operations in the development of the AS/400.

The latest lineup of AS/400's continues to change the notion of what a minicomputer is. The bottomrung EO2, at an average cost of about $12,000, provides small businesses with a single-source alternative to local area networks, readily available software and compatability with all other AS/400 models. The top-of-the-line EO8 and EO9 models can support up to 2,400 workstations, making them a realistic alternative to some mainframes or for companies downsizing from older mainframes.

Don Kryzer, owner of Brothers Analogic Inc., a St. Paul company that specializes in information systems project management and systems services to U.S. companies in Asia, recently spent a year evaluating a local area network system vs. the AS/400 for a client in Hong Kong.

The AS/400 won out, Kryzer says, in part because of its software solutions and its ease of use. ``I don't think IBM ever intended the AS/400 to be as user-friendly as it has been,'' Kryzer says.

IBM Rochester's challenge is to continue to prevent a flight of users and software vendors. The only way to do that, Unterberger says, is to continue giving customers what they want. Right now, customers are saying they want an AS/400 that can operate in a heterogeneous environment, and the AS/400 is moving in that direction.

``Our real vulnerability is if we don't allow the AS/400 architecture to evolve to an open one,'' Unterberger says. ``The outstanding question is, `Will we evolve fast enough?' ''

Given Rochester's track record, Burris, for one, is willing to bet that they will. By 1995, the AS/400 will be seven years old and, given the already remarkable pace of development, a completely different system. ``I suspect IBM Rochester will have a pretty solid story to tell,'' he says. Eric Wieffering is a staff writer for Corporate Report Minnesota magazine, which is based in the Twin Cities. The article appeared in the May issue of the magazine.p