The Survivor
She fought off cancer, then turned a struggling maker of design software into an industry giant. Is there anything that tough-talking Autodesk CEO Carol Bartz can't do?
By G. Pascal Zachary

(Business 2.0) – When Autodesk named Carol Bartz as its new chief executive in 1992, the then 43-year-old former Sun Microsystems sales manager was destined for corporate stardom. She was taking over one of the nation's fastest-growing software companies—maker of a groundbreaking program that allowed architects and engineers to abandon their T squares and pencils and design anything from a backyard arbor to a skyscraper on a personal computer. Her ascension instantly made her the most prominent female executive in technology—and one of the most buzzed-about up-and-comers in all of business.

But the storybook script ended there. Her second day on the job, Bartz learned she had breast cancer and began an arduous battle with the disease. Meanwhile, she had inherited a cabal of rebellious Autodesk programmers—led by a co-founder with a cultlike following—bent on humiliating the outsider. Worse, Bartz's first major product introduction was a dud, bringing Autodesk to the brink of collapse and raising questions about her competency.

A lot of people would have quit. Not Bartz. Between chemotherapy sessions, she worked doggedly to professionalize the company's amateurish business practices and to recruit desperately needed management depth. Ultimately, Bartz survived the cancer, and by the late 1990s it looked as though she had pumped new life into Autodesk too—only to see her world nearly collapse again. With Autodesk products slow to adapt to the Internet, Bartz endured a falling stock price, an exodus of talent, and a chorus of taunts about the company's future. At one meeting with stock analysts, she unloaded: "You'd be happier if we were selling plastic-wrapped fruit baskets over the Internet?"

Defying calls from analysts for her resignation, the tough-talking Bartz often lashed out at her executive team, opening some meetings with a taunt of her own: "Tell me why I shouldn't fire the whole lot of you." Her job, however, was never in serious jeopardy. During Autodesk's dog days, "we always asked what can be done differently," recalls board member Hal Dawson, an Autodesk director since 1988. "We always concluded we couldn't find anyone better than Carol."

Their faith has been rewarded. Every Fortune 100 company today uses Autodesk programs, which are as dominant among architects and engineers as Microsoft Word is with writers. The company ranks as the No. 1 performer in the S&P 500 for the past 12 months and has tripled in value since 2002. Sales in fiscal year 2005 (which ends on Jan. 31) will top $1 billion for the first time, and profits are expected to rise about 50 percent.

Just as important, Autodesk's push into three-dimensional design products—programs that let users create richer design simulations and tweak them onscreen—is making inroads with companies interested in managing huge projects with software that runs on most PCs. The market for 3-D is potentially vastly larger than today's 2-D market, and Autodesk is already making headway. Architects working on the 72-story Freedom Tower planned for Ground Zero in New York, for example, are using a new Autodesk product called Revit to tinker with countless 3-D models before official plans are due in 2006. "Autodesk is hot," says Sasa Zorovic, an analyst with Oppenheimer in Boston, "and has room to run."

The question, of course, is how far and for how long. Bartz rebuilt Autodesk by spiffing up its bread-and-butter moneymaker, expanding the product line, and imposing adult management on a former wild child of a company. But now Autodesk, once just a software supplier for relatively small shops with PCs, is increasingly trying to build products able to tackle chores heretofore handled by much more expensive systems from relentless competitors like Oracle and IBM. That's uncharted competitive terrain for Autodesk. And for Bartz, perhaps the ultimate corporate survivor, it presents a whole new array of perils.

Bartz's tenacity wasn't forged solely by her struggles with disease and rebellious nerds. Her mother died when she was 8, and she was raised by her grandmother in a small Wisconsin town. She worked her way through high school as an assistant to the president of the local bank. Bartz earned a computer science degree at the University of Wisconsin—one of only two women to do so in 1971. That planted the seeds of a career that took her first to Digital Equipment Corp. and then to Sun. "Failure is not in her vocabulary," says Tony Peach, an Autodesk product manager.

Bartz excelled on the sales side at both DEC and Sun, and at Autodesk she has forged an effective leadership style out of her no-nonsense charm. Tech is full of swaggering CEOs; Bartz prefers poor-mouthing to pomposity. "She doesn't have an ego button to be pushed, thank God," Dawson says. Her periodic blue-streak outbursts are softened by an instinct for diplomacy and a folksy style that connects her with people of all sorts.

It's all on display one recent October morning at the company's headquarters in San Rafael, Calif., during an informal "coffee with Carol" gathering of about 250 employees. Autodesk's stock hit an all-time high this morning, and Bartz opens the meeting with trademark humor. Apologizing for arriving five minutes late, she quips, "My staff will tell you, never have a breakfast with Carol. That's when people tend to get fired."

The room erupts in laughter, and Bartz launches into an explanation of the company's recent success. Then she takes questions, the first of which is in her face: How much longer will she stay as CEO? Bartz, standing with a wireless mike, swings into her answer like a game-show host. "There was a rumor I'd bail out when we got to a billion in annual sales," she says. "Comes from something I said 12 years ago. Well, it's taken longer to reach a billion, but I don't see any change happening. As long as I'm learning, and I am, I want to stay."

And why not? Autodesk is a vintage software company that, improbably in an era of pressure for quick success and instant loss-cutting, seems to get better with age. It was born in 1982, the brainchild of a gifted, iconoclastic programmer named John Walker. One of a generation of daring and often self-taught programmers that included the likes of Bill Gates, Gary Kildall (the first PC operating system), and Wayne Ratcliffe (the first PC database), Walker was adept at taking concepts popular in the esoteric world of mainframe computing and applying them to the then-nascent PC. He formed Autodesk around a 2-D drawing application called AutoCad that he'd shrewdly purchased, Gates-like, from an independent programmer.

The program appealed instantly to architects and designers of physical objects, and Autodesk became a PC standard in the 1980s. "The computer transformed architecture," says Mario Guttman, who entered the field in the 1980s and now directs the computer-aided-design operations of HOK Group, one of the world's biggest architectural firms. Because of Autodesk, he says, "the profession is becoming an information management profession. I couldn't work without AutoCad."

Success came in spite of a party atmosphere at Autodesk that prefigured some of the excesses of the later Internet boom. The company's offices were in trendy, bucolic Marin County. Employees brought their pets to work (they still do), and consensus ruled (no longer). Walker's favorite beer was Miller, which he served by the keg at Friday afternoon bashes for the staff.

As the company grew, though, the eccentric co-founder tired of management chores and eventually retreated to Switzerland in the late 1980s. Yet he remained the power behind the throne, with a fanatical following among a group of programmers who called themselves "Core." By the early 1990s, as Microsoft took command of the software industry, Autodesk still hadn't released a Windows version of its flagship product—and quickly stumbled into crisis. And that's when Walker stepped back into the fray.

From his Swiss perch, he unleashed a terrifying 44-page letter to employees. The screed accused then-CEO Alvar Green, a former company bookkeeper handpicked by Walker to run the company, of neglecting product development and thus inviting ruin. Autodesk, Walker railed, had become "stuck in the past, mired in bureaucracy, paralyzed by unwarranted caution." Green was soon out, and Walker agreed to the board's selection of Bartz to be the company's first professional manager. Not a programmer, she was instantly regarded as an alien presence by some of Autodesk's coding purists. But she came armored with a take-charge mentality. "I'm not here to appease anyone," she told employees shortly after she started. "I'm here to build a business."

That began, literally, from her hospital bed, where tending to office duties helped Bartz keep her mind off her cancer. Just hours after having surgery to remove a breast, she called a promising job candidate who was mulling an offer to become head of HR. "I was just awake from surgery," Bartz says. "I said, 'Do we have a deal or not?' What was he supposed to say? He just said, 'I've got to work for this woman.' Then, of course, he tried to quit two weeks later when he found out what a mess the company was."

At first, Bartz did her diplomatic best to keep Walker involved—he was still a hero to the engineering troops—but he refused any role that required him to fit into an organization or follow rules. Bartz, however, had the backing of the board, and over time she outmaneuvered Walker, dispersing his power base within the company and gradually bleeding off his influence. He left for good in 1994, followed eventually by several of his devoted coders.

But by then Bartz had already begun to prove her operations mettle. She lined up AutoCad squarely behind Microsoft's Windows platform and imposed formal processes and clear timetables on the once hippy-dippy programming group. Though she could not free Autodesk from its scary reliance on AutoCad upgrades to generate new revenue, which meant financial performance fluctuated with the ebb of each product cycle, she at least was keeping the firm afloat—no small feat in the 1990s, a decade that became a graveyard for many of Autodesk's fellow software pioneers. Storied companies like Ashton-Tate, Borland, Digital Research, Lotus, and WordPerfect all hit the wall and either died or slid into irrelevance.

As the '90s drew to a close, however, Autodesk began to fade again. Revenue was climbing steadily, but Bartz was still struggling to broaden the product line beyond AutoCad. She had pushed to beef up Autodesk's efforts in digital animation and 3-D software, but the Internet boom was siphoning talent away from the company. Bartz says, "I'd go to investor conferences—with standing room only at presentations by Used-Fucking-Golfballs.com—and I'd get four shareholders listening to me."

The talent exodus particularly galled Bartz, who turned down several juicy job offers to stay at Autodesk. "I just can't stand people that walk," she says. "I'm not built that way." Bartz even kept a list of the "most preposterous" dotcoms and showed it to her staff every week to keep the recruiting effort up. "I had to keep kicking them in the balls, because if we couldn't beat GolfBalls.com"—a real company at the time—"then we'd have to get out of here. I couldn't attract anyone to come to Autodesk. It was an awful time."

More painful than the boom, of course, was the bust. The collapse of IT spending hammered Autodesk along with everybody else. Sales fell from a peak of $947 million in fiscal year 2002 to $825 million the next year. Profit margins cratered, and the stock collapsed to a two-year low of $12.

That, it turned out, was the darkness before the dawn. Defying some advisers who counseled reining in her diversification push, Bartz barreled ahead, figuring that the bust opened up opportunities to buy her way into growth at bargain-basement prices. In February 2002, Autodesk paid $133 million for Revit, gaining a hot 3-D product for designing and managing construction projects; it also got the people who created the program. "We held on to the founders and nearly all the developers," says Jay Bhatt, the Autodesk vice president for corporate strategy who oversaw the purchase. Bartz presided over half a dozen other acquisitions from 2002 on.

Equally significant, Bartz began yet another management overhaul. Two years ago she brought in a new CFO, Alfred Castino, to impose badly needed cost controls. Castino, who had worked at HP and PeopleSoft, discovered "gigantic gaps between what we were doing and best practice." In his finance department, Castino recalls, "we had an organization I hadn't seen since the 1980s." Headcount in finance alone has dropped from 225 to 150 on Castino's watch, and he expects more trims to come.

Another coup was the return of Carl Bass. A decade ago Bass was a fiery Autodesk coder who tangled with Bartz during a software redesign. Bass loudly predicted that the project would fail; when it did, Bartz asked him for a proposal on how to fix it. When he declined, she fired him. A few months later, she rehired him to manage the next version of AutoCad, which was a stunning success and made Bass's reputation at the firm. Then Bass left during the dotcom boom. Bartz wanted him back—and bought his company in 2001 to get him. Now he's COO.

Bass is credited with bringing a new discipline to the company's product development and sales efforts—and is seen as Bartz's heir apparent. All told, initiatives from the likes of Castino and Bass have improved Autodesk's operating margins from 3 percent at the end of 2002 to 18 percent in the latest quarter.

The embrace of 3-D, however, has been Bartz's most important move, and it required her trademark perseverance. Consider the company's push into the 3-D software used to design machine tools. AutoCad was a 2-D drawing product, but it was so dominant in its sectors that Autodesk had long shied away from the riskier 3-D world, a highly fragmented market where numerous small software outfits like SolidWorks and Autosys claw for business. By 1996, though, Bartz had decided that Autodesk customers would eventually demand a 3-D offering and that Autodesk needed to provide one. But developing its first 3-D program, called Inventor, took four years; then the program stumbled out of the gate, at least partly because the company's sales team feared that it would cannibalize existing AutoCad sales. "We had a lot to protect," says Robert Kross, who oversees the Inventor line.

Only this year, after a total investment of $500 million, has Bartz's gamble begun to show a return. Westinghouse, for example, uses Inventor to design nuclear reactor parts. Utica Enterprises, a supplier to GM, uses it to design the assembly line for the Hummer. Leatherman Tool Group uses it to mock up new versions of its multipurpose gadgets. Inventor fetches a higher price than AutoCad—about $4,995 per copy vs. $3,750—but also supports the older app by making it easy to swap data back and forth. "It's essential the two programs work well together because we use both," says Richard Carraway, a manager at Utica.

The move into 3-D doesn't just ally Autodesk with cutting-edge designers. It's allowing the company to extend its reach to the factory floor by helping to manage product data. Until recently, product-design data had been difficult to share electronically. Software giants like Oracle have attacked the problem with soup-to-nuts database systems that integrate manufacturing information into a company's entire computer network, which usually requires a top-to-bottom redo. Bartz wants to solve the problem from the designer's desktop with a program called ProductStream, released in July. ProductStream lets engineers share the underlying data of their designs—material, specs, and the like—with managers on the factory floor or outside suppliers. Autodesk hopes to line up enough customers to let it iron out any real-world bugs, then shoot for a wide release in 2005. "Our philosophy is to make sure it works," says Kross, a former manager with General Electric.

Bartz often reminds Kross that he'll have to run flat out if he wants to beat companies like Oracle. To motivate him, she calls him Seabiscuit. "I passionately hate that nickname," Kross says. "But this is Carol at her best. She knows ProductStream is like a horse running without a competitor and that I'll move heaven and earth to prove we're moving fast enough."

The new product initiatives are weaning Autodesk from its dependence on 2-D AutoCad sales, which in the latest quarter provided roughly half the company's $279.6 million in sales. And there are plenty of other new business lines for the company to target. A contest over how best to share design documents, for example, is pitting Autodesk against another venerable software giant, Adobe. Both are trying to do for design documents what Adobe's PDF format does for books and articles: allow PC users to swap them easily without altering the original. That's no easy feat with design documents. Their masses of data—mathematical expressions, ratios, specs—are often lost or reduced when shipped electronically. Adobe product manager Pam Deziel says she sees a lucrative market "that we'll continue to focus on." Bartz agrees about the potential but doesn't think much of Adobe's efforts, saying, "What the fuck does Adobe know about engineering drawings?"

Competitors might have a question too: What does Bartz know about reaching beyond the desktop and selling products that purport to do some of the same chores as ubiquitous enterprise systems from Oracle or IBM? Of all Bartz's efforts to diversify her product line, this stands as the most logical—and the riskiest. Should they fail, offerings like ProductStream would undermine Autodesk's credibility with huge customers or, worse, drag the company back into stagnation.

Everything that Bartz has been through has toughened her for the battles to come. And there's no questioning that toughness: She delayed the start of her seven-course chemotherapy treatment by a month, against her doctor's advice, so she could chart an initial course for Autodesk. Besides, no one believes more in the Autodesk mission than she does. At a recent employee meeting, she insists, "We are not a fluke." The company's current run, she says, is the start of a great new phase, not a final act. Then, raising her voice, she says, "We've got to believe in ourselves. Because if we believe, there's no stopping us." The room grows silent. Bartz's smile fades. She's getting serious. "So you'll hear this in the coming months: Do you believe? Over and over. Until you come up to me and say, 'I believe.'"

Building Out

FY 2004 SALES BY PRODUCT CATEGORY