Big Blues: Mainframe Slowdown And Stiff Competition Put Pressure on IBM

New PCs and Minicomputers Take Over Many Tasks At Lower Cost to Users

How IBM Is Fighting Back

By Paul B. Carroll and Hank Gilman, Staff Reporters
The Wall Street Journal

November 23, 1987

Mainframes, the multimillion-dollar computers that are the workhorses of big corporations, are becoming a less crucial part of the computer landscape, and that may bode ill for International Business Machines Corp.

Today, mainframes provide some 60% of IBM's profit. They have been considered a birthright at IBM since it captured the market two decades ago. But the world is changing.

For one thing, most large companies are already up to their eyeballs in mainframes. No longer is it a matter of, say, putting a company's entire accounting program on a mainframe but, rather, maintaining and improving the program. So mainframes don't run out of capacity as fast as they did, and the company doesn't have to buy a more powerful one so soon.

For another thing, increasingly powerful smaller machines -- personal computers, or PCs, and minicomputers -- are soaking up much of the demand for additional power. They can perform many of the tasks that used to require processing by a mainframe. Eventually, some think, more mainframe capacity may be needed to coordinate the smaller machines, but for now, at least, they are cutting into mainframe demand.

It all adds up to slower growth in mainframe sales. What's more, Big Blue's share of that growth is declining right now as Japanese competitors gain at IBM's expense.

"The industry has matured," says William Moore, who was a member of IBM's executive committee until he retired last April. "Looking from the sidelines, the business is so large now that to try and have, say, a 10% compound growth rate each year is maybe something you can't do."

Depending on how severe the pressures become in the years ahead, they could create cracks in IBM that would let its smaller competitors in all segments of the computer industry proliferate and prosper. And Clyde Prestowitz, formerly a top Reagan-administration negotiator on high-technology trade issues, thinks IBM's health will largely determine whether the U.S. stems Japanese inroads in technology.

"Among American high-tech companies," he says, "IBM is really the only one that can slug it out with the Japanese."

IBM officials declined to discuss its mainframe market for this story, but they have recently said they think mainframe growth will pick up again despite softness this year. They are implementing both a hardware and a software strategy aimed at ensuring continued solid growth. Chairman John Akers is the man making the hard decisions.

In any case, IBM will remain the biggest kid on the block in the computer industry. It is a power in PCs and minicomputers, too. More important, it has a stranglehold on the market for software that runs on IBM-compatible mainframes and believes it can return to 20% operating profit margins -- from 12% in this year's third quarter -- by transforming itself into a company that focuses much more on software and services.

But many outsiders say mainframe pressures bode ill for IBM's hope of returning to the go-go growth rates of the early 1980s, when it was creating the equivalent of a Digital Equipment Corp. each year. The Computer and Business Equipment Manufacturers Association, a trade group, estimates that U.S. mainframe sales will climb just 3.3% a year from 1986 through 1996.

That is less than half the growth rate of the prior decade, and is dwarfed by the 8.8% growth rate expected for minicomputers and the 14% annual increases forecast for personal computers. The trade group expects the personal-computer market to surpass the mainframe market in 1989 and to be more than twice its size in 1996.

"People can't even use the {mainframe} horsepower that's out there," says David Turner, a vice president of National Advanced Systems, the National Semiconductor Corp. subsidiary that markets IBM-compatible mainframes. "Historically, every new generation doubles the power for the same price. . . . You wind up shipping fewer systems."

IBM has long contended that growing use of minicomputers and personal computers would make companies need more mainframes -- called "big iron" in the industry -- but many corporate users dispute that. They note that processing power can cost as much as 150 times more on a mainframe than on a personal computer.

Hospital Corp. of America, for instance, has spent tens of millions of dollars buying NCR Corp. minicomputers. Besides being less expensive than mainframes, the minicomputers give the different departments in its hospitals more control over their own computing operations and typically speed up the computing. The company most recently bought an IBM mainframe two years ago and still has plenty of capacity.

"I think you'll see more processing power deployed at the local level," says Thomas Cato, vice president of Hospital Corp.'s information services. "This has very definitely reduced demand for mainframes."

Edward D. Jones & Co., a St. Louis-based brokerage firm, has bought $30 million of personal computers and minicomputers from Digital Equipment. They will let the firm put work stations in its more than 1,300 branches, thus letting its brokers more easily handle customer documents and generate reports. Richard Malone, a principal in the firm, says it's hard to say just how much the new computers reduce its mainframe needs, because some things simply wouldn't have been attempted if mainframes were the only alternative. But he says his Digital purchases did diminish his appetite for mainframes.

Beyond the purchase price, data-processing managers say smaller computers are easier and less expensive to operate. Irene Nesbit, a consultant helping NBC set up its computer operation for the summer Olympics, says it costs about $30,000 for minicomputer software that will let sportscasters call up the background of, say, the Thai athlete who stuns the world in solo synchronized swimming. Similar software for mainframes, which were used in the past, would cost more than $80,000, she says.

Besides slowed growth of the whole mainframe market, IBM must also contend with some pesky competitors that have been stealing pieces of IBM's pie. IBM still controls about 85% of the market for "big iron" that runs software written for its own mainframes, but it has lost a few percentage points of market share over the past year, in particular to Amdahl Corp. IBM officials also say they lie awake nights worrying about NEC Corp. of Japan, which they credit with superb technology even though it doesn't yet have much of a presence in the U.S. market.

Amdahl, which is 45% owned by Fujitsu Ltd. of Japan, and National Advanced Systems, which markets mainframes made by Hitachi Ltd. of Japan, have regained the credibility they lost with corporate customers when IBM embarrassed them in the early 1980s with technology they couldn't quickly match.

Many big customers are attracted by the hefty discounts Amdahl and NAS offer. Also, customers who used to play it safe by buying IBM have become more willing to take a chance with a competitor as they grow more sophisticated about computers. One reason they are more willing to wander is that they have seen Digital Equipment bloody IBM's nose in the market for minicomputers and have seen Apple Computer Inc. and Compaq Computer Corp. more than hold their own despite IBM's onslaught in the personal-computer market.

"It's been a long road" for IBM's mainframe competitors, says Thomas M. Blodgett, a senior vice president with Southeast Bank, Miami. But "today more than ever," he says, "they're looked upon as a reasonable and prudent alternative; I don't think anyone believes the stories IBM used to put out that they don't work and they're disasters." He figures he is saving $5 million over five years by buying mainframes from NAS.

James Franklin, vice president of information services at Crowley Maritime Corp., says he had reservations about leaving the IBM camp but decided to do it because he could save "big bucks."

"There is tremendous pressure on us to reduce costs," he says. "Therefore we took the risk. Management no longer just rubber-stamps our request for equipment the way they once did."

IBM officials have said in public forums that the company will counter-attack with new hardware and a new emphasis on software. The latter would be aimed both at increasing software revenue and at driving up mainframe demand, which tends to rise as new software becomes available and many prospective purchasers find they need new mainframe capacity to use it.

Early next year, IBM is expected to speed up its mainframes by 20% or so. Later on, it is expected to roll out operating-system software that would even more sharply increase the processing speed and would finally let programmers take advantage of some of the fancy new features of the current hardware.

Then, by 1990, IBM is expected to announce a whole new line of mainframes, starting the product cycle all over again and making its competitors play catch-up for a while. Amdahl and NAS say they won't stay behind for long, but they have said that before. In the past, it has taken them a year or even two to match IBM.

Product improvements will also let IBM recapture at least temporarily some of the market share it has lost to Amdahl, NAS and others in data-storage devices, a lucrative part of the mainframe arena. IBM, whose share of the market for IBM-compatible data-storage devices had slipped to 75%, recently announced that next year it will ship products that are much faster and can hold far more information than is now possible.

Even as hardware profit margins keep falling, margins have been rising on IBM's mainframe operating systems, indispensable software on which IBM holds a virtual monopoly. Even when customers switch to competitors' mainframes, they still need IBM's software if they want to run the tens of billions of dollars of applications that corporate America has written for its IBM mainframes. And because IBM generally leases the software to customers, rather than selling it, price increases in that software bring in revenue from all the IBM hardware in use, not just the new machines being sold.

Robert Djurdjevic, president of Annex Research, a computer research and consulting firm, says IBM has been raising prices of mainframe operating-system software 30% to 35% a year and raised the price on one such system 82% last May. He says profit margins are so high on the software that IBM makes more money on it than on all the minicomputers and personal computers it sells.

IBM says it raises prices only to reflect improvements in the software. "Our growth is absolutely not coming from price increases," says Earl Wheeler, vice president for programming.

Whatever the case, IBM officials say that overall operating-system revenue will continue to grow 20% to 30% a year, far faster than the rest of IBM's business, and they are devoting enormous resources to it. IBM has been increasing its programming staff about 10% a year and now has a staff of 26,000 just in the U.S. -- nearly four times as large a group as Amdahl's entire work force.

As part of a broader emphasis on software, IBM has also begun emphasizing what it calls Systems Application Architecture, or SAA, and says that this will bring enormous increases in demand for its hardware and operating-system software.

SAA represents a set of standards for programmers arranged so that corporate customers will be able to easily move software between different size machines, something that now requires hefty rewriting. For example, someone could take a program that he wrote on a personal computer to improve his productivity and have someone move it easily to a mainframe so that a whole department could use it.

In addition, the SAA standards should make it easier for IBM's machines to communicate with one another and should let users operate a program with the same commands no matter what size computer is being used.

Industry analysts applaud SAA, though they say that Digital Equipment got there first and that the full impact of SAA won't be felt for a few years. They are less certain about how long IBM's product improvements will let it stay ahead in mainframes and storage devices. And they even have some reservations about how well IBM will do in mainframe operating-system software, largely because of a recent arbitration ruling in a dispute with Fujitsu.

That settlement will require Fujitsu to pay IBM an undetermined but large amount of money, but could make Fujitsu IBM's first real competitor in mainframe operating-system software.

IBM could, of course, develop other parts of its business to take over mainframes' role as its engine for growth. But Big Blue hasn't been exactly riding high in personal computers, minicomputers and telecommunications, and the fierce competition in those three businesses makes it unlikely that they can step into mainframes' traditional role.

In fact, analysts have been scaling back their estimates for shipments in 1988 of IBM's 9370, a strategic minicomputer that IBM had internally dubbed the Vax killer because of what it was supposed to do to sales of Digital's Vax minicomputers.

IBM speaks hopefully of the ventures it has formed with software companies and says that revenue growth in application software will outpace even that of its operating-system software. But application software is a tiny part of Big Blue's business and results have been mixed to date. Says Bernard Goldstein, who is a partner in Broadview Associates, an investment-banking firm, and a former president of Adapso, a software trade group: "IBM doesn't have the power in this end of the industry that they had before."

More promising are the steps IBM has taken to cut costs and to move thousands of employees out of staff jobs and into sales positions. Those steps are expected to improve earnings this quarter and lead to a far stronger 1988.

Still, says Ben Rosen, a prominent venture capitalist and the chairman of Compaq: "Nobody's afraid of IBM anymore."