Cover Story

What's Next for Digital?

DEC Has Cooled Off -- But Ken Olsen Has Confounded The Skeptics Before

By Leslie Helm in Boston, with John W. Verity and Geoff Lewis in New York, Thane Peterson in Paris, and Jonathan B. Levine in San Francisco
Business Week

May 16, 1988

Ken Olsen saw it once before, in 1982. For several years before then, Digital Equipment Corp., the company he started in 1957, could do no wrong. Computer buyers and investors went overboard in their enthusiasm for DEC. Then, suddenly, they fell out of love. DEC became beset my management and product problems, and its earnings fell. Analysts, customers, and competitors began to find flaws in its strategy. And Olsen, hailed as a visionary during the flush times, was second-guessed and, occasionally, ridiculed.

On the surface, at least, it's starting again. After the company's longest-ever winning streak, four years in which its revenues more than doubled and it became a potent No. 2 to industry leader International Business Machines Corp., sentiment is turning away from DEC anew. With IBM's recovery finally under way, investors are betting more on Big Blue. DEC's stock, which dropped 25% in the October market crash, has since fallen about 25% more and now hovers several points below IBM's. Young upstarts and longtime competitors alike are eroding DEC's technological lead and chipping away at its markets. "The slowdown in DEC's momentum is becoming more obvious by the day," says Robert Randolph, an analyst at International Data Corp.

Part of the problem, as Olsen puts it, may be that "success can deteriorate a company." The addition of thousands of new employees in 1987, combined with a smaller-than-anticipated sales increase, has pushed overhead expenses 37% ahead of last year. So far in fiscal 1988, which ends on June 30, these swelling costs have pared DEC's pretax profit margin to 14.9% from 17.2% a year ago. Even though revenues jumped 17% in the third quarter, ended Mar. 26, earnings were about flat -- and some 10% below analysts' estimates. For the full year, things may not look as bad. Wall Street expects an 18% profit increase, to $ 1.3 billion on a 21% sales gain, to $ 11.6 billion.


But it's the long term that has DEC observers worried. The company has prospered since 1984 by expanding sales of its hugely successful VAX minicomputers from their customary scientific and technical niches into the broader commercial market. The key was a single computer design, or architecture, available in many sizes and easily linked into networks -- a combination no other company offered. Now, for the first time, serious competition may be developing to DEC's approach. Dozens of computer makers are backing research efforts on so-called open systems -- standardized software that would tie together all brands of computers and perhaps blunt DEC's selling edge.

Meanwhile, DEC's results are suffering from its negligible presence in personal computers. Its minor role didn't seem to make much difference when the company was faring better. But suddenly DEC is missing out on the fastest-growing computer market. And to make matters worse, networks of new and increasingly powerful personal computers are eroding DEC's minicomputer sales.

The most haunting question, however, is whether DEC's 11-year-old VAX design is now outmoded. Newer technology enables machines from Hewlett-Packard Co. and Sun Microsystems Inc., among others, to deliver the same performance at lower prices. And if the VAX can't be adjusted to compete, DEC faces a major technical hurdle. "It's a tougher market than it has been," says Gerald A. Paxton, a DEC marketing vice-president. "There's a little less business out there, and there are more players."

It takes more than that, however, to worry Kenneth H. Olsen. The single-minded -- some say stubborn -- 62-year-old engineer has hewed to a vision that, even years ago, critics said was outdated. "In the early '70s, when the computers-on-a-chip were developed, everybody said our business would disappear," he recalls. Having proven such forecasts premature, says Olsen, "we feel more optimistic than ever." He calls it "irrelevant" that new microcomputers and souped-up scientific machines outperform comparably priced VAXes. "The real problem is how to make a customer's whole enterprise more productive." That, he says, means companywide networks -- not getting hung up on the performance of individual machines.

So far, it's hard to argue with his logic. From Hong Kong to New York, from Toys 'R' Us to Chase Manhattan Bank, from brokerage houses to insurance companies and manufacturers, large corporations depend on DEC. Virtually all of them also use IBM machines. But they've grown impatient waiting for Big Blue to merge its incompatible product lines and hone its own networking skills. Close to 50% of DEC's sales now come from commercial accounts, up from less than 40% in 1984.


That improvement shows up graphically in DEC's share of the midrange computer market -- machines selling for $ 15,000 to $ 700,000. DEC now gets 25% of those sales, up from 19% in 1984, while IBM's share has contracted from 24% to 16%, according to Gartner Group Inc. The improvement shows up in DEC's overall numbers. In 1984, its earnings were $ 329 million on sales of $ 5.6 billion. Last year they were $ 1.1 billion on revenues of $ 9.4 billion. Its share of overall industry revenues is nearly 10%, up from 7% in 1982. To Olsen, it all makes perfect sense: "If we have what the world needs, we should get a bigger share."

The question is: Does DEC have a corner on what the world needs? So-called minisupercomputers from Convex, Alliant, and Multiflow outperform the biggest VAXes by 100% or more in scientific Jobs -- and cost about the same. And such companies as Stratus Computer Inc. and Tandem Computers Inc. produce machines that can far outperform VAXes in transaction processing -- a $ 21 billion market, dominated by mainframes, that's growing by 30% a year. DEC "is falling behind the power curve," says Jonathan Fram, a computer analyst at Bear, Stearns & Co.

An entire new class of computers may soon make this more obvious. They're based on the emerging technology called RISC, for reduced instruction-set computing. This concept increases computing speeds by simplifying a computer's basic circuitry. In various forms, RISC is being adopted by IBM, Hewlett-Packard, Motorola, and Sun. HP, which sank nearly $ 400 million into its RISC project, is rebounding and is now heralded by some analysts as "the new DEC."

Meanwhile, Sun a maker of engineering workstations and a close partner with American Telephone & Telegraph Co., has developed a RISC microchip, called Sparc, that it claims can provide twice the power of a VAX design at less than half the cost. Unisys, Xerox, and Britain's International Computers also plan to use Sparc. And now a group of 17 computer companies, including Data General Corp., has signed up to use Motorola Inc.'s answer to Sparc: its 88000 RISC microchip. Within two years these chips will produce a flood of all sizes of computers, all competing with VAXes and selling for less.

Olsen retorts, "RISC is irrelevant." No matter how speedy the technology may be, it will be several years before software exists to make RISC machines as useful to customers as DEC's VAXes are now, he argues. "The network is the system," Olsen reiterates. "It's the most exciting thing going on in computing."

He is one of the few chief executives left in the business with the technical credentials to make such judgments. After seven years at Massachusetts Institute of Technology's Digital Computer Laboratory, Olsen started DEC on the theory that engineers and scientists wanted small machines that didn't need the "priesthood" of programmers required to keep mainframes running. Most major product proposals still have to pass his scrutiny, and few escape unchallenged. "If I ever wrote a book," he laughs, "it would be a list of things I said 'No' to."

This Yankee conservatism permeates Olsen's life. Though he's worth an estimated $ 270 million, he has never strayed far from his simple New England roots. He dislikes travel and spends much of his spare time with Aulikki, his Finnish-born wife. The couple live in a modest home not far from DEC's headquarters -- the now famous onetime woolen mill that Olsen chose for its low rent 30 years ago.


The Olsens seem remarkably nonchalant about their millions. Calling home from the mill last Oct. 19, Olsen recalls, he told his wife that the 42-point drop in DEC's stock had reduced their net worth by $ 109 million. "Don't worry," she said. "We have one row of potatoes left, and we haven't touched the parsnips yet."

Six months later, and another $ 60 million or so poorer, the devoutly religious Olsen is philosophical. He once called Wall Street's short-term view "immoral." Analysts "are only interested in you today," he said last summer before the crash. "They will drop you tomorrow, anyway." A falling stock price isn't "trouble," he adds now. "Trouble is when your kids take dope."

Still, the coming challenges to DEC's networking preeminence must give Olsen pause. Next he must parry IBM's renewed attempt to hamstring DEC in midrange computers. IBM's earlier effort, a scaled-down version of its mainframe design called the 9370, has failed to match the VAX. But a new IBM minicomputer, code-named Olympia/Silverlake, is due out in June and could fare better. It's a follow-up to IBM's popular System/36 and System/38 computers, and will be able to run the thousands of software packages already written for those machines.

And IBM's network system, called SNA, for systems network architecture, is entrenched in large corporations, the great majority of which remain committed to IBM mainframes. A recent International Data Corp. (IDC) survey of 107 major IBM customers found that 93.5% plan to expand their IBM-based networks rather than bring in DEC.

The most serious challenge, however, ultimately may come from below -- from personal computers. The growth of DEC networks pales compared with the proliferation of low-cost personal computer networks: DEC sells about 350,000 terminals and computers each year. This year, says IDC, some 3 million personal computers should be bought for networks. William D. Strecker, DEC's vice-president for product strategy and architecture -- and the designer of the original VAX -- thinks these computers will turn to DEC "as soon as they find it hard to make it all work together."

But maybe not: "Local area networks of personal computers are being substituted for minicomputers," says Amy D. Wohl, an office-computing consultant. The outlook is for more of that. Sales of personal computers in the U.S. will grow by 25% annualy through 1992, predicts Gartner Group. But sales of midsize computers will grow only 11.5% annually, the company says.

Those trends pinpoint what critics believe is one of DEC's rare but serious failings: After a costly flop in personal computers in 1982, DEC has yet to make a major new effort to produce and sell them. Instead, it has developed ways to weave IBM PCs and, soon, Apple Computer Inc. Macintoshes into its networks. The company insists that is the right way to go. "We are in a better position than anyone to be the No. 1 integrator of desktop devices and end up owning the network," declares John F. Smith, DEC's senior vice-president for engineering, manufacturing, and product marketing. But some outsiders remain unconvinced: "The real growth in the computer industry comes from supplying a new crop of computer users with something they have never had before," says John C. Levinson, an analyst at Goldman, Sachs & Co. "DEC is not in touch with that trend."

Even former allies wonder if personal computers are a blind spot for Olsen, who oversaw DEC's last effort, code-named KO. "I took [Lotus Development Corp. founder] Mitch Kapor to lunch with Ken Olsen one day, and Ken started [criticizing personal computers] and pointing out how reliable the DEC VT100 terminal is," says Barry James Folsom, formerly group manager for DEC's Rainbow personal computer group and now vice-president of Sun's East Coast Div. Kapor, who made a personal fortune selling software to personal computer owners, "tried to tell him that people don't care -- they want to run spreadsheets or whatever program they fancy at their own desk," Folsom adds. "I don't think Ken Olsen fully understands what's going on."

Competitors couldn't agree more. As they see it, Olsen's Achille's heel is his devotion to the VAX, the machine he considers his perfect brainchild. VAX and VMS, the operating software that controls its basic functions, are exceedingly good at time-sharing, which essentially divides the power of a single processor among many terminals. But today, selling minis against personal computers presents a greater marketing dilemma than it did a decade ago. Now inexpensive microchips mean that every desk can have its own computer for little more than the price of a terminal.


Soon the trend to open systems may pose an equally difficult sales challenge. Spurred by the desire of customers such as the U.S. government to cut costs and simplify programming with standard basic software, this movement had been gathering steam. Already, AT&T and Sun are working together to replace the dozens of varieties of AT&T's Unix operating system with a single, standardized version. Due out next year, it's being hailed as the means to establish networks that could connect not only different sizes of computers but different brands.

"Snake oil," says Olsen. Some customers do say that, because of its design, Unix is less secure against outside tampering than DEC's VMS operating system. Moreover, Olsen contends, networking is technically so complicated that no customer in its right mind would trust company business to software that was controlled by a committee made up of mutual competitors. "The open network will be as exciting as a Russian truck," Olsen quips.

It that's true, however, DEC's labs are spending a lot of time on trucks these days. Its Unix staff now totals 250 software developers. That's up from 50 two years ago and about equal to the staff working on VMS.


DEC's seemingly split personality on such issues may be a calculated strategy. As the company's foremost spokesman, Olsen's priority is to carry the banner for products that earn the most -- large VAX minicomputers with hundreds of terminals and proprietary software. But for customers who want it, DEC is increasingly eager to sell its version of Unix, called Ultrix. One reason: Even enthusiastic DEC customers such as Citicorp are interested in snake oil. "We ought to be independent of hardware vendors," says Arthur Iger, systems manager at Citicorp. "We want commodity-priced hardware."

Even if DEC wants to avoid this trend, it probably can't. Unisys, NCR, HP, Wang Laboratories, Data General, and even IBM have all jumped on the Unix bandwagon. "Unix is giving DEC pressure all over the place," says Dale Kutnick, an industry consultant. Market researcher IDC estimates that Unix was used on 10% of all computers sold last year vs. 7% for VMS. By 1991, Unix will be on 20% of all computers sold, IDC predicts, vs. 8% for VMS. Perhaps with this in mind, Olsen recently delivered a closed-circuit TV message urging DEC salespeople to sell more Ultrix. A low-priced, Ultrix-only VAX may even show up next year to compete with products from Sun, Apollo Computer Inc., and others.


But DEC's focus remains on its proprietary systems. Over the past five years, it has beefed up its marketing and service organizations to match IBM's. "We planned to be equal to IBM in service and quality," says Olsen. "We were surprisingly successful."

The engineer of DEC's big push into IBM territory is John J. Shields, senior vice-president for sales. Unlike his predecessors, Shields has overcome the boss's aversion to flashy marketing programs. He has built an enormous sales, service, and support organization -- 16,000 staff members were added last year, bringing the total to an estimated 65,000. He has invested millions in new sales centers to help penetrate key industries worldwide. One, occupying an entire floor in a Manhattan office building, is rigged with mock trading rooms and back offices to demonstrate the benefits of VAXes for brokerages and other financial services companies. And for the first time in years, DEC is advertising on network television.

For the most part, however, Shields is focusing on 600 U.S. companies, and 500 more overseas, that are leaders in their industries. He believes that once these companies buy, their competitors will have little choice but to do the same. And it may be starting to work.

DEC already has made such high-profile conquests as Chemical Bank, Perdue Farms, and Bankers Trust. "Nowadays, as an MIS manager, you aren't doing your job if you haven't explored the Digital alternative," says W. David Power, a consultant at Temple, Barker & Sloane Inc., a Boston-based consulting firm. VAXes are being entrusted with critical jobs -- money transfers at Chase Manhattan, for instance -- that used to run solely on IBM mainframes. One big reason: "It's so easy to implement DEC solutions," says Frederic Melul, a senior vice-president at Credit Lyonnais. The French bank recently chose VAXes over IBM gear to automate its stock- and bond-trading operations.

That's why Olsen says DEC's future is more of the same: more VAXes, larger and smaller, and continuing improvements to its networking software -- "to raise the bar for IBM." And no plans to switch to RISC or any other fad. Says Olsen: "We have more VAXes on the drawing boards than ever before." One major advancement is a new series of multiprocessor VAXes that promise to improve vastly DEC's cost-competitiveness.

This VAX-forever strategy has its defenders. Multi-processing, says Omri Serlin, a Los Altos (Calif.) market consultant, should do more for DEC than any possible switch to RISC. Moreover, he thinks that DEC can slash engineering and manufacturing costs by designing a limited number of basic processors and combining them as building blocks to create systems across a wide performance range. And although DEC must still overcome some major software challenges to make that work, there should be no need to rewrite the entire catalog of VAX software, as a change to RISC would require.

These bigger VAXes could even give mainframes a run for the money. Although DEC insists that it isn't really very interested in the mainframe market, it continues to inch in that direction. DEC's largest multi-processor VAXes already compete with mainframes for jobs such as managing large data bases. With forthcoming software they could do rapid processing of financial transactions. Next year DEC is expected to unveil Argonaut, the most powerful VAX ever. It could compute with small IBM mainframes for jobs such as running a division of a major corporation.

At the other end of the spectrum, DEC will try later this year to move its engineering workstations into the market for office workstations -- and complete with personal computers. "Workstations? I'm not giving that to anybody," says Olsen. The new strategy will include a "personal VAX," at less than $ 4,000 the lowest-priced one yet. But that machine would run only VAX software. Still another forthcoming product would use Intel Corp.'s 80386 microprocessor, adopted by IBM and Compaq Computer Corp. for their most powerful personal computers. It would run common software. But DEC has no distribution through computer retail chains -- the main source of personal computers for business.

With its broadening product line, DEC has begun to resemble IBM in significant ways. Like the bigger company, it must weigh its current customers' investments in software when it ponders major changes in technology. For example, DEC can't afford to make sweeping revisions that might make obsolete the programs in which customers have invested millions. And, like IBM, DEC has grown accustomed to the high profit margins that only proprietary hardware and software deliver. DEC is starting to suffer some withdrawal symptoms, just as IBM once did. A shift in demand to low-priced VAXes accounted for much of the third-quarter earnings disappointment. While they are both proprietary and profitable, those machines don't earn as much as larger ones do.

"DEC's problem is that its business model is based on charging what in the microprocessor world is an outrageous amount of money for minimal performance," contends William N. Joy, vice-president for product development at Sun. If he's right, that's also reminiscent of what happened to IBM. Its bloated cost structure, rppmised on high-margin mainframe gear, fell victim to more cost-effective minicomputers -- mainly VAXes. Now, DEC could be learning the same lesson. "They have to start competing on price and performance," says Rosemary M. Mahoney, manager of digital technical services at Arthur Andersen & Co. "They've traded too long on their networking advantage."

Of course, the IBM model has advantages, too. Growing software investments keep DEC's 27,000 customers from easily abandoning their VAXes. That means lots of future add-on and upgrade business -- and relative freedom to set prices. Like IBM, DEC need not always have the latest widget to be profitable. For example, it entered the engineering workstation business four years after Apollo and Sun. But it expects to sell $ 1.3 billion worth of the machines this year -- and be the market leader in 1989.

Short of an acquisition or merger -- both of which Olsen continues to rule out -- few observers expect any dramatic changes from DEC. And that's what worries investors. As IBM learned long ago, simply catering to a big customer base can be a game of diminishing returns. The cost of computing is falling so rapidly -- by more than 20% a year, some analysts estimate -- that it is imperative to win market share to show healthy revenue gains. At this point, DEC's VAX strategy is still working. But it may get tougher to achieve the 20% annual growth that the company plans for.

Naturally, Olsen disagrees. He was asked recently what DEC would do with its $ 2.7 billion cash hoard. Analysts have suggested that DEC should start paying shareholders a dividend or perhaps buy up shares to bolster the stock price. Olsen rejects both ideas. "If we ever get our message across, we are going to need [the cash] just to supply products." No doubt he's right. Still, DEC's message may be getting well-worn -- just as its competitors are improving theirs.

GRAPHIC: Cover Photograph, no caption, RESNICK/PICTURE GROUP, Picture 1, OLSEN HOLDING FORTH IN A MEETING, Digital's president embodies the New England virtues of thrift and honesty and a single-mindedness that is DEC's strength -- and perhaps its problem, RESNICK/PICTURE GROUP; Picture 2, no caption, RESNICK/PICTURE GROUP; Chart 1, DEC'S PROBLEM, ITS SALES CONTINUE TO GROW . . . BUT ITS MARGINS ARE SLIMMING . . . AND ITS PROFIT GROWTH IS SLOWING, DATA: COMPANY REPORTS, GOLDMAN, SACHS & CO. ERIC HOFFMANN; Picture 3 and 4, TWO WHO ENGINEERED DEC'S PUSH INTO IBM TERRITORY, Sales boss John Shields has introduced some flashy marketing for the machines that master architect William Strecker designs, RESNICK/PICTURE GROUP; Picture 5, NETWORK TERMINALS AT BANKERS TRUST, Digital has expanded into the heart of commercial data processing, KEN KERBS; Picture 6, DIGITAL VICE-PRESIDENT JOHN SMITH, "We are in a better position than anyone to be the No. 1 integrator of desktop devices", RESNICK/PICTURE GROUP; Chart 2, DEC'S PLUNGE, DATA: BRIDGE INFORMATION SYSTEMS, INC. ERIC HOFFMANN

Copyright 1988 McGraw-Hill, Inc.