Cover Story

IBM: Trying to out all the pieces together

By Marilyn A. Harris in Armonk, N.Y., with bureau reports
Business Week

April 21, 1986

Imagine if RCA Corp., way back when, had developed one vitrola for classical records, a different one a few years later for jazz, and a third for rock. It would have created havoc for the record-buying public -- and for RCA. Over the years, International Business Machines Corp., in its drive to reach into every corner of the computer market, has created just this kind of situation. Since 1970, IBM has designed more than 15 different computer "architectures," none of which use the same software. It was inevitable, says a former IBM software specialist, that "all those designs would bite IBM in the behind."

Ouch. It's happening. Even IBM's tight-lipped executives are starting to admit that the differing designs of IBM's mainframes, minis, and micros have become a sore point. Customers are squawking, and some are turning to rivals with products more compatible with one another. "The emphasis is shifting to intelligent networks," says Francis R. Gens, an analyst with International Data Corp. "IBM's got the basics, but the pieces just aren't in place yet."


That isn't the only headache facing the world's No. 1 computer company as the industry slump rounds into its second year. Margins on the IBM Personal Computer are eroding as cheaper imports clones force price cuts, and so is its market share. In the midrange, customers say, IBM's machines haven't performed as well as those of competitors. Most irksome of all, the track record of its latest mainframe series, the top-of-the-line 3090, better known as Sierra, has been disappointing. PCs and midrange systems give IBM a well-rounded product line, but mainframes, with their 60%-to-80% gross margins, contribute about half of the company's gross profits.

The pressure is on all over IBM, from product development and marketing to management. The company is shearing the losers from its business product line and putting its considerable resources behind its three strongest architectures. In five years, it vows, it will be a much tougher competitor. "We're investing for the long term," says President and Chief Executive John F. Akers. "We don't think anyone [will be able to] touch us."

But all the turmoil isn't helping IBM's bottom line. On almost a weekly basis, Wall Street analysts are lowering their estimates for IBM's 1986 earnings, though even the most bearish of them still expect an improvement over last year's $10.67 per share. The latest consensus: per-share earnings of $11.50 to $12, down 50 cents or more since February. Daniel Mandresh, an analyst for Merrill Lynch Capital Markets, believes that the computer giant's first-quarter earnings per share will be up only 6%, to $1.71.

Allen J. Krowe, IBM's senior vice-president for finance and planning, wasn't disputing the estimates as IBM's first-quarter results neared release on Apr. 11. The reduced expectations have kept IBM shares about 15 points below a recent high of 161 in February. Although IBM once predicted that it would be a $100 billion company by 1990, that won't happen unless the industry and IBM can result their 15% annual growth rate.

Of course, everything's relative. Even when IBM has problems, the scale of its success makes rivals' mouths water. With $50 billion in 1985 sales, $6.5 billion in earnings, and a market capitalization of $92 billion, it is still the world's most profitable and valuable company -- the only one in the information processing industry with a AAA rating. No competitor can match its investment. Last year, IBM spent $4.7 billion on research, development, and engineering, plus $3 billion on automating its factories. Added together, that's $1 billion more than the total 1985 sales of Digital Equipment Corp., IBM's closest rival. With such strength, Akers says, "I don't care who you are in this industry. No one can compete with the IBM company."

Competitors, however, believe they can -- by preying on IBM's weaknesses. Led by DEC, several are doing well by playing to what customers describe as their No. 1 need: the ability to have their machines, of whatever make, "talk" to each other effortlessly. The Agriculture Dept., which recently bought an IBM mini for each of its 2,800 branch offices, chose DEC to supply its Washington headquarters because DEC's software for moving around computerized documents was better than IBM's.

There is a reason IBM has so many different computer designs (table): Each was developed to solve a specific problem. The 370 mainframe architecture, dating from 1970, has its roots in the era when computers were exotic number-crunchers protected in air-conditioned "glass houses." Those machines were powerful. But programming them was -- and still is -- too complex for customers lacking a big support staff. In 1977, IBM belatedly entered the burgeoning small-business market, which had been pioneered by such minicomputer makers as DEC and Data General Corp. To do so as quickly as possible, it sacrificed compatibility with its larger systems when it designed its System/34 minicomputer, ancestor of the current System/36. The same thing happened with the PC: that's why it needs an extra level of software -- introduced only this month -- to talk to larger machines that hold much of the corporate data customers now find they want to tap.


Other lines grew to meet more specific needs: The 8100 was built as an office automation system. The Displaywriter was aimed at the word processing market. As each new design and product emerged, competitive fiefdoms grew up around them within IBM. "No one really cared about IBM -- just about their own projects," recalls one veteran.

IBM cannot simply drop all the systems and start over again. The customers who own them have made big investments in software and training and expect Big Blue's traditional service and support. So IBM will need every bit of ingenuity it can muster to straighten out the mess.

Clearly, it has to do something: Even a committed customer like Prudential Insurance Co., which spent the bulk of its $100 million data processing budget last year on IBM products, is getting fed up with the jumble. "My biggest gripe with IBM is that they don't provide enough of the connectability we need," says Joseph N. Carroll, Prudential's information systems vice-president.

In the face of this mounting pressure, IBM has finally developed a more focused strategy for its general-purpose computers. Jack D. Kuehler, senior vice-president of worldwide development and manufacturing, says the company will emphasize three architectures "to the exclusion of everything else." Those are: the 370 line, comprising all IBM's mainframes; the System/36 minicomputer line; and the PC family, which includes the PC-XT, the PC-AT, and the PC Convertible portable. The heat is on the tens of thousands of IBM engineers and programmers to homogenize these families, top to bottom. The goal, Kuehler says, is to "make it effortless to move applications among different computers."

Some imminent improvements:

* The PC family. To stimulate flagging demand and combat cheaper copycats, IBM is adding more powerful models that will be able to perform several tasks simultaneously and communicate more easily with other PCs in a network.

* The System/36. This year the three-year-old machine will be given more "horsepower" to serve larger departments of companies. Also in the works: software that will let customers pass data to and from System 370 computers -- and make it easier for IBM to compete against DEC and Data General.

* The System 370. Over the next several years, its architecture will be extended from the mainframe down to the desktop. That, analysts say, may mean the gradual phase out of the System/36.

The concentration on three systems will force a "head-on collision" with DEC and Data General in the office market, says IDC's Gens, because it will give IBM customers a clearer growth path and an easier way to share software and data. But it will cause plenty of pain and expense to customers left with the systems that are passed over, he says. Products based on other architectures, like the 5520 administrative system and the 8100 office system, will get new software so that they can exchange information and coexist with the three major lines. But they won't be enhanced in the future.

IBM executives blame their financial performance on other factors, which are beyond their control. Foremost among them : U.S. capital spending patterns, which Akers and his lieutenants say have been disrupted by uncertainty over tax reform. Loss of the investment tax credit alone, Krowe says, would effectively increase the $2 million to $8.5 million price of a Sierra mainframe by 18%.

Indeed, outside the U.S., IBM's earnings last year grew a healthy 19%, to $3 billion. In the hotly competitive Japanese market, the rise was 34%, to $406 million. But North American results -- dragged down by slower capital spending -- pulled overall earnings to 0.4% below those of 1984. Says Akers: "We didn't realize, nor did economists or other companies, that American business was going to stop investing in 1985."

To bolster the bottom line, IBM has instituted rigorous cost controls. Expenses on travel, conferences, and other items deemed nonessential were cut by $700 million below planned levels in 1985. This year's anticipated cut is closer to $1 billion. Inventory levels, up more than 20% in each of the past two years, to $8.6 billion in 1985, will be held to 5% growth this year.

IBM is trying to slim down its 406,000-person work force, too. It has nearly frozen U.S. hiring, and for the first time since 1975 its domestic force will be reduced this year, even though the company will continue a longstanding no-lay-offs practice. IBM will rely on attrition and wide-scale retraining and redeployment of people from staff and manufacturing positions. While some company-watchers speculate that IBM may also initiate a special early-retirement program, Akers says no such move is planned this year.

A gradual companywide decentralization, already under way, should also help make IBM more efficient. A January reorganization of its 22,000-person marketing force is moving about 2,000 staff people into frontline sales and service jobs. In early April, the Americas/Far East international subsidiary was split in two. A level of management was eliminated there and staffers were dispersed into the two groups. More such moves are planned. "We can make this company a little simpler," says Akers.

No one is certain when this retrenchment will end. Akers says he sees no letup in the stubborn industry slump but vows to keep investing, just as subbornly, to improve IBM's technology, manufacturing, and volume. "We're going to stick to our knitting," he says. When business does pick up, competitors could find IBM dropped a lot fewer stitches. 

GRAPHIC: Picture, JOHN F. AKERS, President and Chief Executive, IBM, ROB KINMONTH

Copyright 1986 McGraw-Hill, Inc.