(FORTUNE Magazine) – BEHIND General Electric's factory automation headquarters in Charlottesville, Virginia, rises a hill that employees used to call America's answer to Mount Fuji. The locals call it Piney Mountain. It was a symbol of GE's ambition to unseat Fanuc, the Japanese company that dominates the world market in industrial robots and machine tool controls, and be king of the hill in factory automation. More recently, Piney Mountain is where GE lost a mound of money after getting carried away by intoxicating -- and wildly unrealistic -- forecasts of the new factory automation industry and GE's place in it. % GE bet heavily on the so-called factory of the future, where automation systems would improve quality and productivity even if the plant was old. Five years ago the company invested $500 million to pioneer the business. GE grandly predicted that by 1990 it would have 20% of a market many forecasters thought would be worth $25 billion a year. The factory automation group it had assembled, GE said in a dramatic public announcement on April 2, 1981, could already be counted on for some $1 billion in annual revenues. The company would set itself up as America's factory-of-the-future supermarket, where industry could one-stop-shop for all the computerized gear and automation advice it needed. That heady vision never materialized. The project has cost GE over $120 million in losses in the past three years, more than twice what it anticipated. The company prefers not even to talk about 1990 anymore. ''We don't want to pie-in-the-sky forecast where the business is going,'' says Executive Vice President James Baker, who calls GE's original predictions ''flaky.'' In the past 14 months GE has gradually reorganized its factory automation group into three separate, quieter operations. Baker says the company has reduced its long-term projections of the overall business and its own market share and has lowered its sights to concentrate on ambitious two- or three-year plans for each factory automation unit. But he still thinks GE will be No. 1 or No. 2 in the business. He insists: ''It isn't going to be fiction.'' Automating a factory requires high-tech gadgets in seemingly endless variety -- drives for managing electric motors, controllers to turn on machines and robots and tell them what to do, sensors and cameras to monitor their work, computers and software, and communications networks to link it all together. When GE started its factory-of-the-future push, company units were already making some of these; the company shifted those units to the automation group. The biggest, at $400 million a year in revenues, was the so-called drive systems division, which sells specialized control systems for running metal and paper mills. Profits from it and other established businesses, supplemented with cash from GE, were supposed to finance the automation group's fledgling product lines. GE got hold of automation products it didn't make through acquisitions and licensing agreements. For $150 million GE bought Calma, a fast-growing Silicon Valley maker of computer-aided-design (CAD) equipment in 1981. CAD ; enables engineers to design parts and buildings on computer screens faster and with fewer mistakes than on paper. When linked with an automated shop floor, CAD systems can transmit designs directly to the machines that build the new parts. GE got an overnight start in robotics by licensing an assortment of robots from Japanese, German, and Italian companies. With more than 15 product lines, GE could boast that it had more pieces of the factory-of-the-future puzzle than any other company. It quickly gained a reputation as the flashiest exhibitor at industry trade shows. Donald Grierson, a flamboyant and fast-rising executive with 24 years at GE, headed the division. GE's bold intent was to capitalize on the fervor for reindustrialization that swept America in the early months of Ronald Reagan's presidency. New tax incentives fostered investment in plant and equipment; a fast start for what GE Chairman John F. Welch Jr. called ''the emerging megamarket of factory automation'' seemed inevitable. Welch, who declined to be interviewed for this article, has a Ph.D. in chemical engineering and is a ferocious advocate of high-technology opportunities for GE. He became chairman on April 1, 1981, and it was no coincidence that the inaugural factory-of-the-future announcement took place the next day. Charlottesville quickly turned into the company's most desirable post. Spirits ran high at the new $30-million headquarters, and work at Piney Mountain took on Silicon Valley airs. Engineers developing new machine controllers labored long hours and threw Friday afternoon beer parties. Employees sported T-shirts that read ''Automate, emigrate, or evaporate,'' the slogan Baker used on the banquet circuit to jolt prospective customers into modernizing their factories. The factory of the future was the most dramatic and visible element in Welch's campaign to create a high-technology image for GE. The project also exemplified the new chairman's wish to encourage entrepreneurship within a company well known for highly structured ways. Says Frank Curtin, a former factory automation manager who, like many of the GE alumni interviewed by FORTUNE, admires Jack Welch: ''He wants guys to take big swings without fear of getting penalized if they strike out.'' Donald Grierson swung for the fences, and GE's fortunes in factory automation are largely reflected in his rise and fall. Grierson left the company in July and is a Charlottesville-based private investor and director of a number of small, fast-growing high-technology companies. He has a personal publicist and likes to bill himself as an ''extrapreneur'' -- someone who helps start several companies at once. Grierson declined to be interviewed for this article. When former GE colleagues reminisce about him, Grierson seems larger than life. ''He makes corporations do things they otherwise wouldn't,'' says one. GRIERSON, whose background was in manufacturing jet engines, began his rise in 1974, when insiders say Welch and other senior executives first noticed his gift for strategy. GE set Grierson on an uphill roll that carried him through five assignments in five years. He progressed so quickly that on his arrival in Charlottesville, former colleagues recall, he owned three houses in different cities that he hadn't had time to sell. A charismatic dealmaker and bold promoter, Grierson fell short on ''the details, the profit-making part'' of business, according to former boss Baker. Many whose own careers were derailed at Charlottesville blame GE for promoting Grierson too fast and beyond his competence: ''the most perfect case of the Peter Principle'' is one former colleague's acid assessment. Another likens him to the honorable schoolboy, after the John le Carre spy novel in which a brilliant agent provocateur is sacrificed by his bosses. Grierson's group was a loosely organized confederacy of semiautonomous businesses, encompassing eight separate sales forces and perhaps six development programs. The day GE unveiled its factory automation strategy, Grierson jokingly described the group as a ''polygamous marriage.'' But within a year he had a more realistic assessment of the odds against quick success. In a talk to reporters and security analysts, Grierson called the group GE's attempt ''to fill an inside straight'' in factory automation. The challenge, for GE and customers alike, was to make the automation pieces work together on actual, disorganized factory floors. GE offered to lend customers its engineers to help with planning and installation. To persuade customers it could do the job, GE pointed to ambitious projects then under way in its factories, including a $300-million locomotive plant renovation in Erie, Pennsylvania, and a $38-million dishwasher plant remake in Louisville, Kentucky. Grierson often flew sales prospects to see these showcases, which FORTUNE has classed among America's ten best-managed factories (May 28, 1984). The products in GE's lineup did not plug together readily. In 1982 GE dusted off a communications network developed in one of its military labs and told prospective customers it could be used on the factory floor. Yet to hook GE's CAD equipment, robots, and controllers together for a working trade show exhibit in late 1982 cost the company $1.6 million. Many engineers in Charlottesville were galled by the company's claims. ''We had our neck stuck out so far,'' remembers one. ''We said we were going to tie it all together -- but it was impossible. You always sell a lot you don't have and then find some way to deliver it. But this was all hype.'' Perhaps fortunately for GE, customers hesitated to sign up for automation systems. U.S. capital spending slumped in 1982. In addition, managers were skeptical that the factory of the future might disrupt their production of the present, making it less, not more, efficient. A GE consulting team, which specialized in planning automation systems, had discussions with scores of companies in 1982 but ended the year with only nine projects, half of them within GE. No outside projects approached GE's internal projects in scope. WITH THE SALE of systems stalled, GE's product lines had to compete separately in the market. Robotics was all elbows from the start. GE had rushed to get into the business, hurriedly licensing 11 different robots developed by other companies because it feared the market was about to take off. It set the new division up in a disused Orlando, Florida, flashcube plant. Conservative forecasts at the time called for robot industry sales to hit $1 billion a year in 1988, and GE wanted at least 20% of the action. Robots, however, turned out to be hard to sell and still harder to install properly. Even the lowliest robots had to be expensively customized to fit customers' factories so they would not go fumbling around crippling productivity. Robot industry sales undershot the forecasts by more than half. GE, whose licensed robots had little unique to offer customers, was caught in a ruinous price squeeze with dozens of competitors. It ended up selling robots for less than it thought it needed to charge, and spending far more than it expected on installation and service. Until GE sharply curtailed the business in 1983, robot losses of some $10 million a year were nearly as large as robot revenues. Even costlier problems were brewing in the business of numerical controls, the specialized computers that guide big machine tools in cutting, drilling, and shaping parts. Historically the biggest seller of numerical controls in the U.S., GE failed to keep products up to date and lost its dominance in the late 1970s to Fanuc, which supplied controls for the Japanese machine tools that invaded the U.S. market. GE had promised customers when it assembled the factory automation group that it would leapfrog Fanuc with a dazzling new numerical control device. Due in 1983, the new product would have more features than Fanuc's advanced but simple designs. It boasted, among other things, a computer screen on which it would simulate machine motions as they were programmed. It would warn of accidents, such as collisions of tools, before they happened. But GE ran into trouble when it hired a flock of computer whizzes to build the ambitious new device and rushed the product to market. Naive about machine tools, the whizzes created a product so fraught with software problems that GE is still ironing them out. ''They were not machine people,'' Baker says. ''It was a mess.'' Annual sales of its numerical controls sank 30% to around $60 million, and estimated annual losses mounted to nearly $10 million. A frustrated Welch alluded to the trouble in 1984's annual report, describing it as a ''management execution miss.'' The biggest calamity of all is Calma, GE's CAD acquisition. A profitable supplier of a broad range of software and hardware for parts designers, circuit designers, and architects, Calma was the second-largest CAD company in 1981, the year GE acquired it. (Computervision was No. 1.) Since then, according to Daratech, a Cambridge, Massachusetts, market-watching firm, Calma has lost nearly half its market share and slumped to fourth place. One industry analyst calls it ''GE's little bundle of disaster.'' GE showered Calma with cash, eager to prove that it could acquire and successfully manage Silicon Valley growth companies. It cut prices and poured money into marketing in an effort to boost market share. ''We were driving the business,'' Baker says. ''We paid no attention to losses.'' The tactic worked at first -- in 1982 Calma's market share reached a high of around 12%. But problems latent when GE acquired the business gradually emerged. Calma's business was split among three distinct product lines. Its R&D units which had been widely known in the industry as ''sandboxes'' where engineers with new ideas had free play, lacked the planning and discipline to keep all three up to date. Marketing-minded GE didn't address the problem. New products showed up on the market late and sometimes performed unreliably. In addition, although GE boasted about its incentive compensation plans, it failed to keep key Calma sales executives and engineers from defecting to rival start-ups. Political struggles broke out between remaining Calma employees and GE newcomers. To head Calma, GE named Robert Smuland, a jet engine executive, and surrounded him with other GE veterans who knew little about CAD. One vice president, now head of a Silicon Valley company, described his own appointment as a case study in bad management selection. When Calma started getting badly beaten in the market and revenue growth slackened abruptly at the end of 1983, Smuland was unable to find out why. He was also unable to construct a realistic business plan for 1984, which ended in unexpectedly heavy losses -- an estimated $40 million on revenues of just over $200 million. Smuland wasn't around to see these year-end figures. Before 1984 was over GE transferred him back to jet-engine operations in Evendale, Ohio, and fired 15% of the Calma work force. DURING 1983, just as losses at Calma and the numerical controls operation began to mount, GE's effort to revive sales of complete factory automation systems caused a crisis. It was precipitated by the arrival of general manager Frank Curtin. A charming, ambitious veteran of the machine tool industry, Curtin was recruited by Welch and Baker to get systems sales on track. Despite early disappointments, GE still wanted to make it big in complete systems. The business plan Curtin produced -- approved by Welch -- was of almost Wagnerian proportions. It called for GE, still starting virtually from scratch, to bring in $600 million annually from factory automation contracts by 1988. GE set the stage with a big play in advertising. A splashy 15-page color supplement in trade magazines urged customers to ''stop waiting and start automating.'' What touched off the crisis was the ad's centerpiece: an announcement that GE would guarantee the performance of factory projects it took on. This guarantee, conventional when applied to, say, a dishwasher, raised eyebrows among factory managers. The costliest components of most complete, or turnkey, factory automation projects are big machine tools and computers, neither of which GE makes. To meet its ambitious sales goal, GE was offering, in effect, to guarantee the performance of other manufacturers' products as well as of its own. Curtin found he didn't have the corporate backing he thought he had. ''As we got into proposals that were really turnkey,'' a former executive says, ''faces at GE began to pucker.'' In one embarrassing episode shortly after Curtin resigned last year, GE headquarters, apparently having second thoughts about the extravagant guarantee, ordered the withdrawal of a bid for an $11- million General Dynamics contract. The job called for seven machine tools, worth over $8 million and not made by GE, and some $2.5 million of GE controls and consulting. Westinghouse got the contract instead. ''GE got cold feet about the systems business,'' says a former executive. ''We were already losing money, and systems looked like an opportunity to lose a lot more.'' GE denies it ever wanted responsibility for large turnkey projects. In 1984 the company quietly took back its offer of performance guarantees. In the closing months of 1983, morale in Charlottesville skidded. ''It went from being GE's glamour spot to a leper colony,'' a former executive recalls. Says another: ''People came to see this business as a people eater.'' From 1981 to 1985 GE's robotics and numerical controls operations each went through three general managers. Grierson, under increasing pressure from his bosses to control losses -- $40 million after tax in 1983 -- groped for solutions. ''He knew he had a problem, but he didn't know what it was,'' says a former subordinate. ''People were giving him answers he wanted to hear.'' He curtailed R&D spending, demoralizing the engineers. A senior engineering manager recalls long hours spent in budget revisions. Says he: ''It got so bad I gave up making budgets because there was never enough money. I just spent what I had to.'' General Electric gave Grierson less than a year to bring the business under control. As losses widened in 1984, the company removed Calma from Grierson's group. Calma now reports to a board headed by GE Vice Chairman Edward Hood. Calma's new president, a veteran recruited from IBM, is spending more heavily than ever on product development and pursuing a turnaround plan that calls for the company to start making money again by 1988. Early this year GE formally disbanded Grierson's organization. Grierson's post was abolished, along with the entire layer of management that served as his staff. The organization's one big cash cow, the drive systems operation, was split off. By consolidating sales forces and research laboratories, Marion ''Rick'' Richardson, the no-nonsense vice president now responsible for Charlottesville, has cut the losses of the remaining businesses from around $40 million in 1984 to less than $10 million this year. GE says the operation has been breaking even since July. GE's humbled factory automation project is building toward a redefined, downsized future. With no chance of becoming the mammoth success once imagined, it has a good chance of becoming a more modest winner. Revenues this year should total $600 million; by 1988, says GE, they'll hit $1 billion. One apparent success is GE's line of programmable controls -- simple computers used in factories to switch lights and machines on and off in prearranged sequences. Nurtured under Grierson, the business grew from a laboratory project into a $100-million business this year. It has earned GE a place as a scrappy contender in a $600-million-a-year subindustry dominated by two old- line giants, Allen-Bradley and Gould. Another Grierson project that may pay off soon is Industrial Networking Inc., a joint venture formed in 1984 to manufacture factory communication gear. Its products will replace GE's original factory network, which sold poorly. The products are positioned to cash in on factory communication standards being pushed by General Motors (FORTUNE, October 28). GE even has guarded hopes in robotics. It no longer dreams of unseating Fanuc, but plans instead to make money as a niche player. GE recently began marketing a line of welding robots manufactured by a Japanese company and fitted with unique GE sensors. Unlike existing robots, GE's new line can weld dirty or rusty parts, enabling customers to skip expensive cleaning and buffing steps. Perhaps the most surprising winner is GE's automation systems business. The company still has systems salesmen, who now look for what its indefatigable marketers, ever ready with a new slogan, dub ''automation partners.'' These are savvy customers willing to rely heavily on GE products and -- in sharp contrast to the arrangement under the notorious guarantee -- to share responsibility for the outcome of their automation projects. According to general manager Duane Shull, GE has more such partners than it can handle. The systems business could bring in a sturdy $40 million this year. That is far behind the $150 million Curtin projected for it, but Shull expects the operation to grow at a steady 15% rate, pulling a growing volume of product sales behind it. Until the road toward the factory of the future takes its next unpredictable turn, GE's newly realistic automation effort seems secure. General Electric's board, which kept close tabs on the factory-of-the-future push, claims to be satisfied with the Charlottesville denouement. ''Management said to us, it's a big market, and it is,'' says director Walter Wriston. ''They said, we have the skills to do it, and that was true. They also said, we'll lose money, and they did. I suspect we'll look back (someday) at the beginnings of this business and say it was a success. How big, I don't know.'' The factory of the future at GE may never attain the summit imagined by Don Grierson. But out back, Piney Mountain no longer threatens to become General Electric's Heartbreak Hill. BOX: INVESTOR'S SNAPSHOT GENERAL ELECTRIC SALES (LATEST FOUR QUARTERS) $28.0 BILLION CHANGE FROM YEAR EARLIER UP 0.5% NET PROFIT $2.3 BILLION CHANGE UP 5% RETURN ON COMMON STOCKHOLDERS' EQUITY 17% FIVE-YEAR AVERAGE 18% RECENT SHARE PRICE $58.50 PRICE/EARNINGS MULTIPLE 11 TOTAL RETURN TO INVESTORS (12 MONTHS TO 10/11) 10% PRINCIPAL MARKET NYSE Explanatory notes: page 194