Mentor's Lessons In The School Of Hard Knocks

Back in the summer of 1989, the livin' was easy at Mentor Graphics Corp. The eight-year-old Oregon company was the world leader in one of the electronics industry's most vital technologies--software used to simplify the fiendishly difficult task of designing advanced computer chips. Loyal customers included a high-tech who's who, from Apple Computer and Boeing to Samsung. At Mentor's headquarters in Wilsonville near Portland, a masseuse roamed the cubicles, providing neck massages, and work had begun on a new campus with a 24-hour gym and a $1 million child-care center. Meanwhile, Mentor's best engineers were developing what they thought would be a quantum leap in software. Executives expected sales to triple to $1 billion by 1991.

Today, the masseuse is gone--and so are the gigabuck dreams. Mentor is more than two years late with its huge software project and had sales of only about $355 million in 1992--17% below 1989's. Its losses during the past two years total $115 million, and it has surrendered its market lead to Cadence Design Systems Inc., whose 1992 sales rose 11%, to about $435 million. The company is even vulnerable to upstarts such as Synopsys Inc., which are taking the technological lead. Mentor still has a cushion--lots of cash, plenty of loyal customers, and promising products. But its plunge highlights a hazard facing high-tech industries: Leaders of one technology generation often fail to catch the next one. Says Mentor Chief Executive and cofounder Thomas H. Bruggere: "We made all the classic mistakes."

From its founding in 1981, Mentor had been a pioneer in its field. Along with Daisy Systems Corp. and Valid Logic Systems Inc., it developed a new generation of electronic design automation software, the critical tool that lets engineers design ever-more-complex chips and computers. In the 1970s, designers had laboriously laid out each of the thousands of transistors and wires on a chip using a minicomputer terminal with crude graphics. With the new software, engineers could assemble simple circuit diagrams that represent clusters of transistors--and design far more complex chips containing hundreds of thousands of components. For the first time, they could also simulate an entire chip's performance--all on a more powerful desktop workstation. Thanks to superior software, and its decision to use industry-standard workstations instead mf proprietary hardware, Mentor took the market lead in 1985.

Then, Mentor tried to reach too far too fast. "We got stars in our eyes," concedes cofounder Kenneth G. Willett. First, it tried to expand into four new businesses, including design systems for software engineering. When those didn't take off quickly, management came up with a more sweeping vision: "Changing the way the world designs." Instead of giving engineers a slightly improved tool with the next version of its core software, Mentor wanted to help them achieve giant advances in productivity. "We tried to build the ultimate dream machine," says former President Gerard H. Langeler, who left last June to become a venture capitalist.

faltering. Mentor's dream was a $30,000 to $80,000 bundle of 50 carefully coordinated programs. A so-called software framework would let customers move data quickly among the programs--normally a lengthy or impossible task. And it would let design teams work on a new chip or circuit board at the same time, saving months or years.

To achieve this vision, engineers kept adding nifty new timesaving features, such as an on-screen temperature gauge to warn of design flaws that would mean the finished chip would generate too much heat. Moreover, Mentor was the first to tackle a major project using a new computer language that incorporates so-called object-oriented programming, which makes it easier to upgrade software. But as its complexity grew, the project fell months behind. In early 1990, Mentor began drafting engineers from other projects and served free lunches on weekends to spur them on. The coordination required to manage all the new bodies just slowed down the product, known as Release 8.0, or "eight-dot-oh" to Mentor gearheads. Customers began calling it "late-dot-oh."

As Mentor faltered, Cadence, a five-year-old rival in San Jose, Calif., moved up. Cadence had design programs that could handle more complex chips and that ran on a wider variety of workstations. By mid-1990, Mentor's delays, coupled with the poor economy, hit home. The company's sales declined as some customers stopped buying older programs and waited for 8.0. In November of 1990, Bruggere, who had left day-to-day operations largely to Langeler and was campaigning for Oregon Governor Neil Goldschmidt, returned to get 8.0 moving.

Mentor's darkest days were still to come. In April, 1991, it reported its first quarterly loss as a public company. In August, Bruggere shut down three of the new businesses, all of which were losing money, and laid off 435 employees. Morale was so low that only 20 of Mentor's 1,300 Portland-area employees attended the company's annual picnic. By year's end, Cadence passed Mentor in software revenues. Worse, when 8.0 was finally shipped in early 1992, it was slow and riddled with bugs. "We completely botched the product-development process," says Bruggere, who had been a software-engineering manager at Tektronix Inc. before co-founding Mentor.

Meanwhile, hardware sales trends were also conspiring against Mentor, which had always sold its software bundled with workstations. Since most customers now had workstations, computer sales--once half the company's revenues--dropped off faster than expected. As its losses mounted, Mentor's stock hit an all-time low of 5 1/4 last Mctober.

Mentor's tailspin has fueled speculation that it may be acquired--possibly by Electronic Data Systems Corp., a partner on a pending U.S. Navy contract. Mentor and EDS say that's not likely. And Mentor denies speculation that CEO Bruggere's days are numbered. Director David R. Hathaway says a chief operating officer will be hired soon to help Bruggere run the company.

Despite the problems, Bruggere insists Mentor can regain its lead. Some 40% of its customers now use at least some of its new software. A speedier, cleaner version of 8.0, slated for release last December, is due out in this year's first quarter. Customers seem impressed: Daimler Benz's chip operation recently placed a $15 million order. Mentor still has $134 million in cash, nearly three times its long-term debt. And it has 12% of the fragmented $1.2 billion industry, trailing only Cadence with 15%.

But in an industry that moves so quickly, Mentor will have a tough time catching up. Analyst John C. Levinson of Goldman, Sachs & Co., for one, expects its sales to sag again this year, to $323 million. Moreover, startups such as Synopsys are trying to leap ahead with next-generation design tools. Bruggere contends that Mentor's broader product line--and strategic vision--will win out. But he doesn't talk anymore about changing the way the world designs chips and circuit boards. Instead of creating the ultimate dream machine, Bruggere and his team are just trying to build products people will buy.

    Before it's here, it's on the Bloomberg Terminal.