Bailing out of the Mainframe Industry
By David E. Sanger
The New York Times
February 5, 1984
THE nation's computer industry should pass a remarkable milestone this year: For the first time, the value of desktop, personal computers sold in the United States - computers that were almost unheard of only eight years ago - will overtake sales of the large ''mainframe'' machines that first cast America as the leader in computer technology.
To many in the computer industry, that change is further evidence that an era of computer technology is being swept into history. ''Mainframes are dying; I am not even quite sure what the term ''mainframe'' means anymore,'' said Michael L. Dertouzos, director of the laboratory for computer science at the Massachusetts Institute of Technology, which helped to give birth to the mainframe four decades ago.
Professor Dertouzos may be overstating the case, but not by very much, many believe. Granted, mainframes are still used by thousands of big corporations, and they continue to be a very healthy source of profit for International Business Machines, the granddaddy of the industry. But the biggest segment of the mainframe business - mid-size machines that represent about four out of every five mainframe sales - has stopped growing and might soon begin to contract.
The prospect that mainframes are becoming an endangered species has forced five major manufacturers, collectively known as ''the Bunch'' (an acronym for Burroughs, Sperry's old Univac division, the NCR Corporation, the Control Data Corporation and Honeywell) to scramble for new niches in the computer industry. Their search has led them to Japan for some new products. They have been hiring away I.B.M. employees to gain new ideas. And they are trying to move into the booming market for small desktop computers linked through complex ''office automation'' networks that permit the computers to zip information from one to another.
But the Bunch corporations are latecomers. Office systems and desktop, personal computers - more precisely, microcomputers costing $4,000 and up - are dominated by such new powerhouses as Wang Laboratories and Apple Computer. And in the last two years, the desktop market has also been skillfully exploited by the ubiquitous I.B.M. As a result, the Bunch group has shared in only a tiny fraction of the revenue from desktop computer sales, which are mostly to corporations rather than to home users. Those sales are expected to jump to $11.6 billion this year from $7.5 billion last year. Meantime, mainframe sales will ease to $11.4 billion from $11.7 in 1983, with less than $2 billion going to the Bunch companies in 1984 and most of the rest to I.B.M., according to the International Data Corporation, a research firm.
THE Bunch drive into desktop systems ''looks like too little, too late,'' said Jack Hart, an analyst at International Data. ''There is a real question in my mind whether there is enough room for all of them over the next four or five years.'' Several analysts think there isn't and therefore two or three of the Bunch companies are soon likely to merge their computer operations. Honeywell and Sperry are often named as the most likely merger candidates.
The disarray among the Bunch companies has the industry wondering whether anyone other than I.B.M. will ever again be an innovative force in the mainframe arena. ''The reality today is that it is just too expensive for a small participant to do it himself against an onrushing I.B.M.,'' said Ulric Weil, Morgan Stanley's computer industry analyst.
Already, Honeywell, the biggest of the Bunch group with nearly $6 billion in annual revenue, has halted most research for its advanced mainframe and has laid off thousands of computer division employees. It has also announced plans to buy some of its mainframes from the Nippon Electric Company in Japan, for sale under the Honeywell label. Sperry is on a similar track. It introduced a new mainframe last year - made by Mitsubishi.
''When I looked at the cost of continuing to develop these large machines, even our installed base of $12 billion was not big enough to justify the R.&D.,'' said James J. Renier, who heads Honeywell's computer division. ''I decided to count on Nippon for the hardware, and we will continue on the software.''
Mainframes are the large, powerful general-purpose computers that first emerged in the 1940's from the computer laboratories of Harvard and M.I.T. In fact, those early mainframes were the computer industry and only the richest corporations and government agencies could afford them. With their endless rows of vacuum tubes, some were so large they filled a gymnasium-size room. But by the 1960's, the new technology of miniaturized circuitry made possible the advent of the minicomputer, a refigerator-size machine powerful enough to take over some of the mainframe's tasks.
Of course, a General Motors or a United States Steel or a government still needs a complex, powerful mainframe computer with enough electronic memory and circuitry to store and rapidly process the huge amounts of information that must be marshaled to print out the nation's monthly Social Security checks, or a major corporation's payroll, or to keep track of a multinational's worldwide inventories. I.B.M. has always made most of these machines - which start at $2 million apiece - and their sales continue to rise, along with those of the Amdahl Corporation and Cray Research, a supercomputer maker. But the midsize models, the sort of mainframes that a smaller corporation would buy for $200,000 and up to perform the same tasks, are increasingly being displaced by sophisticated minicomputers, and also a new generation of desktop microcomputers powerful enough, despite their size, to do some mainframe work. That sort of technology has blurred once-clear distinctions between mainframes and other types of computers.
''Fundamentally, it is hard to picture the mainframe business ever again becoming the dynamic enterprise it was in the 60's and 70's,'' said Charles E. Exley Jr., president of the NCR Corporation, which many experts consider the most successful of the Bunch companies in making the transition from mainframes to smaller systems. ''That does not mean it isn't a good business, but for growth we simply have to go elsewhere.''
One reason for that is that I.B.M. has cut off their growth at about 13.2 percent of the market because its mainframes have become the overwhelming industry standard. In the past few years, a crucial I.B.M. design for communications between mainframes has become so widely accepted that the Bunch companies - which once reveled in their independence from I.B.M. standards - are forced to make sure their computers can ''interface'' or talk to Big Blue's. That I.B.M. victory appears to kill off motivation for innovative technology, although Bunch executives dispute this.
The Bunch group, in fact, continues to introduce new mainframes, which it contends represent technological advances, although critics say the improvements are too minor to be significant. For example, two weeks ago Burroughs introduced what it described as a new generation of mainframes, and NCR has concentrated on using microcomputer technology to reduce its general purpose mainframes to about the size of a suitcase. ''We have increased our R.& D. 50 per cent and brought out a whole new line of mainframes in the past year that are the most innovative in the industry,'' said Robert F. Holmes, a senior vice president of Burroughs.
That's a contention that some Burroughs customers challenge. ''We see a real slowdown in innovation, in the concentration on being different,'' said Richard A. Lewis, data processing manager of Brulin & Company, an Indianapolis chemical manufacturer, and president of an independent group of Burroughs equipment users. ''They are no longer willing to take a chance.''
In fact, Honeywell's deal with Nippon Electric for sophisticated mainframes was an outright acknowledgement that innovation is too expensive. Mr. Renier said the decision to sell these computers with a Honeywell label required a major ''change of culture'' at the company. With Nippon doing the mainframe manufacturing, Honeywell is free to concentrate its strengths on such other fields as environmental control systems for buildings, production control systems for factories and the design of specialized computer programs.
SPERRY, whose latest midsize mainframe is built primarily by Mitsubishi, makes the same argument. ''More important than who makes our parts is making sure they all work together,'' says James B. Aldrich, Sperry's vice president of products, strategy and marketing support. ''This is all part of being a systems supplier who sells not just computers to his customers, but solutions.''
But that argument skirts the question of whether in marketing new systems the Bunch companies should risk losing their own identity by simply selling I.B.M.-compatible equipment. It's a ''horrible dilemma,'' Mr. Weil said. ''If they become too compatible with I.B.M., they face the desertion of their longtime customers,'' who could choose I.B.M.'s own equipment, he said. ''If they stay unique, they lock in their own customers,'' who would find a sudden transition to I.B.M. equipment too expensive and disruptive, ''but they attract no new customers.''
Indeed, even long-time Bunch customers are getting nervous about their mainframes. Many are apparently turning to I.B.M. to update their systems. ''Just in the past few months, we have seen a number of our biggest users defect,'' said Mr. Lewis of the Burroughs user group, which acts as a sort of consumer advocate for those who operate Burroughs computers. Mr. Hart of International Data concurs: ''Every month you hear about another Control Data site that I.B.M. has picked off.''
Even if I.B.M. weren't the huge problem it is, the cost of developing new and faster mainframe computers has soared far beyond what it was a few years ago - and often beyond the reach of companies seeking to enter the mainframe field with a new product. That was the experience last month of Trilogy Ltd., which is seeking to develop a new mainframe in a venture led by Gene Amdahl, the founder of Amdahl, with investments from Sperry and Digital Equipment, among others. The new technology for this mainframe centers on an attempt to place an extraordinary number of circuits on the silicon ''wafers'' from which computer chips are cut. In January, however, the Trilogy said that the project had been delayed at least until late 1985 by costly technical problems. And last week the Storage Technology Corporation, which made its name in ''plug compatible'' disk drives that fit I.B.M.'s machines, last week scrapped its two-year effort to build a small, powerful mainframe, saying it was simply too expensive.
Most executives of the Bunch companies acknowledge that with the evidence mounting two or three years ago that the mainframe market would shut down on them, they should have moved more quickly to develop new products, including word processors, retail sales terminals, automatic teller machines, desktop personal computers and ''distributed systems,'' the industry's ungraceful term for the networks that link hundreds, sometimes thousands of office computers.
Some argue that the Bunch companies didn't act because they simply didn't recognize the fundamental change taking place in the economics of computer power. For years, it was a basic tenet that the bigger the computer, the more economical it was to use because so much more calculating could be done at so little incremental cost. But the advent of microprocessors in the mid-1970's - based on thumbnail-sized chips packed with thousands of circuits - upset the conventional economics. ''The cost curves were completely reversed,'' said Mr. Exley of NCR. ''Today, the lowest-priced machines and the smaller ones can execute an instruction at a fraction of the cost of the largest.''
But Mr. Weil argues that the Bunch companies saw the changes on the way; they simply failed to react. ''In fact, they read the tea leaves correctly,'' he said. ''But that is the easy part. These are big, phlegmatic, multibillion-dollar companies. Implementing new strategies against a tough I.B.M. and a world of Apple's that were successful much faster than anyone would have guessed - that has been their Achilles' heel.''
When they did finally act, some did so in desperation. Sperry, for example, arranged to market a Mitsubishi- made personal computer that it concedes is a ''clone'' of the I.B.M. XT, an advanced version of I.B.M.'s personal computer, but 10 percent less expensive and also compatible with larger Sperry computers. However, analysts question whether anyone other than a Sperry user would buy the machine. Several weeks after it was announced, they point out, the Tandy Corporation, which has thousands of Radio Shack outlets coast-to-coast, introduced a similar I.B.M. clone that is faster and even cheaper than Sperry's.
Haste has also created headaches for Burroughs's OFIS 1 equipment, an office automation system brought out to compete with I.B.M. and Wang products. Mr. Holmes, the Burroughs vice president, concedes that sales of the system so far ''have only been fair.'' The system, he said, ''is an example of an outstanding product that we brought to market prematurely,'' apparently a reference to user complaints that it did not perform as advertised. In the next few months, he said, the company will announce improvements that should make OFIS 1 far more useful.
All of these new products require marketing strategies sharply different from those for mainframes. But the Bunch companies have found themselves unprepared to sell through retail outlets, a technique I.B.M. adopted quickly for its personal computers. Even NCR, which has moved faster than most in bringing word processors and personal computers to market, acknowledges major problems. Says Mr. Exley, NCR's president: ''We still have a long way to go in building a significant dealer network.''
FOR THE BUNCH, NEW SURVIVAL STRATEGIES
Until the mid-70's, NCR was still turning out thousands of electromechanical cash registers. It took a new chairman, William S. Anderson, to move the company into electronic terminals. Later NCR's president, Charles E. Exley Jr., honed in on full computer systems linked to NCR's supermarket checkout terminals and automatic bank teller machines - all capable of talking with I.B.M. machines.
Now industry experts say NCR's strength in small systems has made it among the most profitable of the Bunch mainframe makers.
The company has invested heavily in personal computers and microelectronics, and has even entered the semiconductor business. Most of these ventures are not yet profitable, but for many of them, Mr. Exley predicts, ''this is the year of the turnaround.''
For a while, it looked like Control Data might escape the problems of the other Bunch companies, as it became a premier manufacturer of peripheral equipment and a provider of ''information services.''
''Now everything seems to be coming apart,'' said Ulric Weil of Morgan Stanley. Peripheral equipment sales are down, and Control Data has had production problems with an I.B.M.- compatible disk drive while its services business has slowed.
Plato, the company's computer- based program to train workers to be more productive, turned its first profit late last year, but the company is having trouble adapting it to desktop machines. Moreover, it has decided not to build its own personal computers. ''We are casting our lot with the things we do well: delivering information,'' said Thomas W. Miller, vice president of corporate marketing.
Computers are only a third of Honeywell's business. Other segments - aerospace, defense, and factory and environmental control products - are quite healthy.
But the Information Systems division, which handles computer manufacturing, was hit hard during the recession and the comeback is slow because Honeywell is still in the embryonic stage of developing new computer markets. However, James J. Renier, division president, says he won't be caught ''sitting in the corner waiting for the Holy Ghost to descend on us.''
Honeywell has moved into personal computers, which are sold primarily as part of larger systems and not as individual retail items. The company is weak in the office automation market. But Mr. Renier, a year in the job, is optimistic: ''There is an infinite number of niches for us to exploit.''
Burroughs has sought to enter new fields through acquisition: Under the leadership of W. Michael Blumenthal, the former Treasury Secretary, it acquired a software development house and Memorex, a manufacturer of disk drives. Still, there are major gaps in its product lines, and in the next few months the company is expected to acquire a company that is strong in CAD/ CAM, or computer aided design and manufacturer.
The company lacks strength in large mainframes and small office computers, but it has a solid position in computer sales to banks, a special niche, and it has a well-established, worldwide distribution network.
Robert F. Holmes, a vice president, says the company's new attitude of ''co-existence'' with I.B.M. is paying off. ''We have found that we are replacing many I.B.M. and Honeywell installations with our equipment.''
Sperry's computer division once made some of the industry's hottest products. But those days are largely past and the Univac name was cast aside last year. Its brightest spot these days, users say, is its Mapper software, a flexible program that helps write an unlimited number of other programs to perform specific tasks.
The company has turned to other manufacturers to build small computers, and it envisions itself partly as a ''systems supplier'' that can make the machinery of several manufacturers work together. The company also sells its large 1100 mainframes to the Air Force as well as to banks and government agencies. For new technology in mainframes, the company is depending heavily on the success of Trilogy Ltd., a development project in which it has invested. Analysts call it a 50-50 bet.
GRAPHIC: graph of U.S. sales of mainframes and personal computers; graph of marketshare of major U.S. mainframe businesses; drawing
Copyright 1984 The New York Times Company