Information Processing

The Workstation Sweepstakes: It's IBM, Coming Up On The Outside

Errors by Industry Leader Apollo Have Left The Field Wide Open for Competitors

Gordon Bock in New York
With John W. Wilson in San Francisco, Alex Beam in Boston, and Marilyn A. Harris in Stamford, Conn.
Business Week

February 3, 1986

The founders of Sun Microsystems Inc. are a study in contrasts--and in the energy that drives a successful computer startup. Andreas Bechtolsheim, a soft-spoken, German-born computer scientist who designed Sun's first workstation as an academic exercise, is prone to kicking off his Birkenstock sandals while pondering a microchip question. William N. Joy, Sun's software guru, is an irrepressible extrovert whose bushy hair and beard give him the look of an erudite mountain man. Then there is President Scott McNealy, the buttoned-down yuppie with degrees from Harvard and Stanford, who yearns to play the world's top 100 golf courses.

All three are bachelors in their early 30s, married to Silicon Valley's currently hottest startup. They were brought together by a fourth co-founder, Vinod Khosla, an Indian-born entrepreneur who met his goal of retiring by age 30 and is no longer active in management.

The trio built Sun more on technical creativity than discipline, so it has had manufacturing and service problems. But revenues have more than doubled in the last 12 months, to $147 million, and McNealy calls $200 million a ''reasonable target'' for Sun's fourth fiscal year, ending in June. Hoping to catch rival Apollo Computer Inc. off guard, Sun on Jan. 20 brought out a new, low-cost line that it claims delivers 20% to 33% more processing power than comparable Apollo models--at half the price.

Apollo executives have tried to ignore Sun's gains and put it down as a price-cutting upstart. But the secret to Sun's success goes beyond the pricing of its machines--from $7,900 to $50,000. The Mountain View (Calif.) company has moved new technology to market quickly, with an almost fanatical commitment to ''open systems'' that make its machines adaptable to the industry's evolving technology. Apollo relies on proprietary designs. Sun uses off-the-shelf parts and urges buyers to customize the product.

A lot of customers like that. In early January, Schlumberger Ltd. chose Sun over Apollo for a $65 million contract--Sun's biggest competitive victory to date. The French electronics and oil-field equipment company wanted ''the flexibility of configuring systems any way we want to,'' says Richard A. Mohrman, president of its Applicon Inc. division. Another big win: a $35 million Toshiba Corp. deal.

Sun's reliance on industry standards is a natural outgrowth of its humble origins. Bechtolsheim was pursuing a doctorate at Stanford University when he set out to design a powerful computer using off-the-shelf components. But he needed money. In early 1982, before he could find a company that would pay to license the design, McNealy and Khosla, classmates at Stanford's business school, suggest-ed starting Sun. They recruited Joy, the architect of an important version of the Unix operating system, which was then gathering momentum as an industry standard for engineering computers. Within a year the four had $4.6 million in venture capital and were shipping $1 million worth of computers a month.


Since then, Sun got an additional $31 million in private financing, including $20 million from Eastman Kodak Co. in a deal that gave Kodak a 7% stake in Sun. Although Sun has been profitable from the start and still has more than $15 million in cash, its growth will demand another infusion of capital soon. ''Going public is a distinct possibility,'' McNealy says.

In preparation for major-league competition, Sun has been hiring better managers. Last year its workstation division recruited Robert Garrow, a co-founder of Convergent Technologies, as general manager. Wayne Rosing, ex-director of engineering for Apple Computer Inc.'s Apple II Group, became engineering manager. In January, Sun raided its archrival for Robert R. Lux, head of Apollo's North American customer-service operations. Lux plans to quadruple Sun's field-service force. This is crucial now that International Business Machines Corp. is entering Sun's market.

These moves should help allay customer concerns over product quality, the Achilles' heel of many startups. The dead-on-arrival rate of Sun equipment ''is more than we would like to see,'' says C. Shelton James, group vice-president of Gould Inc.'s Infor-mation Systems Business Section, which signed a $40 million purchase agreement with Sun in May. He adds that Sun ''is going through all the things companies go through when they start out.''

McNealy insists that quality problems are behind Sun. Now, he's ready to take on IBM. After looking over the giant's latest workstation announcement, he declares with gusto: ''We're hungry and looking for lunch.''

For five years, it was one of the computer industry's little-noticed fishing holes, where a few companies reeled in a lot of money while no one else seemed to care. But when sales of technical workstations zoomed by more than 50% last year as most other parts of the business hit a slump, major players plunged in. It was hard to resist a market segment likely to reach the $1 billion mark this year.

The latest contender is International Business Machines Corp., which announced on Jan. 21 that it will ship its 32-bit RT Personal Computer in March. ''IBM has identified this as a fairly substantial growth area,'' says Brian Jeffery, a market researcher with International Technology Group in Palo Alto, Calif. ''They want a piece.'' IBM's entry puts heavy pressure on the players with established beachheads. Analysts expect it to weigh in immediately as a solid No. 4--behind Apollo, Sun, and Digital Equipment. That would put IBM ahead of Data General, Hewlett-Packard, and two dozen or more startups that will have to fight for their lives. ''When IBM enters,'' acknowledges Sun Microsystems Inc. President Scott McNealy, ''the scramble for market share begins.''

Workstations--souped-up microcomputers used in computer-aided design and manufacturing--are in huge demand because they offer more speed, computing power, and graphics capabilities than standard 16-bit personal computers. They are used for tasks as diverse as designing circuit boards, modeling atoms, and checking the welds on sports cars--jobs formerly done on high-priced mainframes or painstakingly by hand. Even in a tight capital-spending environment, the productivity gains these machines deliver make them popular with budget-conscious managers. The market has also been boosted by rapid price declines: A machine that sold for $40,000 in 1981 sells for $10,000 today. Analyst Richard C. Edwards of Robertson, Colman & Stephens calls this ''the breakpoint in terms of putting one on the desk of every technical professional.'' Although there are an estimated 40,000 workstations in the U. S. today, IBM executives believe that only 10% of the 3 million scientists, engineers, architects, and university professors who could use such a machine now have access to one.


IBM's odds of attracting these customers are enhanced by the errors of Apollo Computer Inc., which pioneered the workstation market in 1980. It relied on only three third-party resellers for more than half its revenues and stuck too long with its own operating system when the industry was standardizing on AT&T's Unix operating system. The result: Apollo's share of total industry sales dropped from 50% to 39% last year, estimates Dataquest Inc., a San Jose (Calif.) market researcher. The $296 million Chelmsford (Mass.) company laid off 8% of its work force and posted a $1.5 million loss for the year ended Dec. 28, compared with a $24 million profit in 1984. Apollo's stock price hovers at 14, down from a 52-week high of 30. Concludes Craig S. Symons, an analyst with Gartner Securities Corp.: ''Apollo will remain a survivor but will no longer dominate the market.''

Sun, a Silicon Valley hotshot, is close on Apollo's tail. Its installed base of machines is just 1,000 shy of Apollo's 15,000, and its market share last year rose from 16% to 20%. But IBM will squeeze Sun, too, with a stripped-down box priced as low as $8,636 to compete with Sun's basic $7,900 machine. In third place is Digital Equipment Corp., which entered the market last May. It is expected to move fast to head off the IBM threat with a more powerful follow-on.

IBM will try to get into customers' doors with the RT PC, which was code-named Sailboat during its development. Then its salespeople can push larger, more profitable models. But this approach may hit a reef: Selling to technically sophisticated customers is a lot harder than buttonholing office managers already familiar with IBM products. Many engineers and scientists are partial to DEC, which has long sold to technical markets. Even so, IBM should join DEC, Apollo, and Sun as contenders for what analysts expect to be $300 million in workstation purchases by General Motors Corp. in the next few years.

Some competitors contend that the RT PC, like a lot of IBM products, doesn't break much technological ground. The machine uses a simplified design approach, known as Reduced Instruction Set Computer (RISC) architecture, in addition to a customized form of Unix. David L. Nelson, Apollo's vice-president for advanced technology, dismisses it as a ''relatively vanilla product, less than we expected.''

But it pleases customers who've tested it. ''The RT PC is the first IBM product that fits our needs as a true workstation,'' says Steven R. Lerman, director of Project Athena, a campus computer network at Massachusetts Institute of Technology. ''This is a real departure for IBM--it's much more innovative.'' But the machine's success, Lerman notes, depends on how much software IBM can line up. The company, working with third-party developers, expects to have as many as 300 packages by midyear.

Still, anytime IBM enters a new market, it signs in with a big advantage that can overcome many problems. Says market researcher Jeffery: ''The combination of IBM marketing, the IBM logo, and a decent product will be hard to beat.''


Copyright 1986 McGraw-Hill, Inc.