By Michael Brody RESEARCH ASSOCIATE Wilton Woods

(FORTUNE Magazine) – SIERRA: IBM'S NEW MONEY MACHINE The Armonk giant is counting on its most advanced line of mainframes for a lot. The hurry-up introduction of a basic model may save the company from turning in its first no-growth year since 1979. Once established, the line could earn a quarter of IBM's profits. IBM'S DEADLIEST ENEMY in the computer business is IBM ''ghostware.'' Those specters -- rumors of impending new and better models -- frighten customers away from the company's current machinery. The phenomenon has bedeviled IBM since the Sixties and was especially spooky this year when the company introduced an important new product: its most advanced mainframe, the model 3090, known as the Sierra. But this time Big Blue proved a remarkably skillful ghostbuster. It came out of worse than usual circumstances in better than usual shape. The problem was on John F. Akers's desk when he took over as chief executive in February. Although IBM was predicting ''solid'' sales and earnings growth in 1985, with a projected increase in product shipments of about 30%, those forecasts were looking shaky. The mighty dollar was slashing foreign sales and shrinking profits translated from foreign currencies. U.S. economic growth was slowing and dragging much of the computer market down with it. And sales of IBM's most advanced mainframe computers, known in the industry as the 3080 series (officially the 308X series), had been hit hard since the end of last year by rumors that the next generation was about to be introduced. IBM calls that new series the 3090; Sierra was once its code name. Despite all the talk about personal computers and desktop work stations, mainframes are IBM's most important business by far, accounting for a quarter of sales and half of profits. The company's top line of mainframes accounts for up to a quarter of earnings, so IBM has special reason to worry when an anticipated new model starts squelching sales of its predecessor. This time the computer industry's rumor mills began grinding themselves into a frenzy over Sierra even earlier than usual, long before IBM was ready to unveil the machine. By last fall whispers abounded that prototypes of the new mainframe were being tested at so-called beta test sites, on the premises of two or three major customers sworn to secrecy. IBM insists those rumors were wrong, but they had their effect. With Sierra seemingly imminent, orders for the 3080 series began to dry up. Reports Richard Kazan, president of Capital Associates International, a large computer-leasing firm: ''In the last six weeks of the year, when usually you see some of your heaviest volume, we did almost no volume in the 3080 series.'' The bad news was that IBM didn't plan to start shipping the Sierra in quantity before November 1985. This meant that for as much as a year, customers might avoid the 3080 series yet be unable to get a 3090. Those who needed immediate computing power might simply upgrade the machines they had or buy a used model. Worse, they might turn from IBM to the plug-compatible manufacturers, Amdahl (48% owned by Fujitsu) and National Advanced Systems (a subsidiary of National Semiconductor and a sales agent for Hitachi), whose machines mimic IBM's but cost less. Desperate IBM users might even go to the Bunch: Burroughs, Univac (now Sperry Corp.), NCR, Control Data, and Honeywell. These companies produce computer systems that don't harmonize with IBM's but compete with them. IBM began to strike back on February 12, when it formally announced the introduction of the Sierra. As usual, the company treated this major announcement like a ho-hum affair. It was buried in a list of about 40 new products, which led off with ''enhancements'' of the 3080-series computers and included price cuts on certain 3080-series processors and support systems -- and incidentally announced the new machine the whole industry had been waiting for. While IBM has remained customarily taciturn about its ghostbusting strategy, a consensus of former employees, customers, and industry analysts suggests that the plan had several elements. IBM's first aim was to save the market for the 3080 series. The company stressed that the 3090 was strictly ''an evolutionary approach to giving customers an expanded processor capability,'' according to an IBM marketing vice president, William J. Filip. The plain- vanilla 3090-200, as it's known, is scarcely more powerful than the top model in the older series, the 3084-Q. Industry consultants rate both at about 28 MIPS -- that is, capable of executing 28 million instructions per second, the standard unit of computer horsepower. The 3090-200 costs slightly more than the 3084-Q, which IBM discounted when the 3090 was announced, so the crucial ratio of price to performance isn't strikingly better for the 3090. IBM's implicit message: maybe it still made sense to buy from the 3080 series. That message left a lot of corporate data-processing executives who had plenty of 3080-series power on hand wondering why they should move over to the new machine. So while trying to persuade mainly middle-market customers that the 3080 still had a lot of virtues, Big Blue's salesmen were telling their biggest clients, according to industry analysts, that there was much more to the Sierra than the public announcement revealed. Some curiosity-provoking innovations in the machine's internal architecture suggest that the computer can be upgraded in ways that haven't been announced. Salesmen reportedly made it clear that peripheral equipment and performance kickers to come later would leave anyone without the Sierra at a hopeless disadvantage. The strategy worked. Many major customers ordered the machine. The lure of its advanced technology was too strong to resist -- even though IBM has priced the Sierra high. Prospective Sierra customers have been grumbling about the absence of any dramatic drop in the price-performance ratio with the model 200 and the difficulty of getting discounts on the new machine. ''It's been more difficult to deal with IBM,'' says American Airlines senior vice president Robert W. Baker. IBM can get away with high prices, industry analysts say, because it dominates the U.S. mainframe market -- its share is close to 80%. New computer lines are always most expensive when they're introduced; after that, manufacturers customarily cut prices 10% to 15% a year. Veteran analyst Ulric Weil of Morgan Stanley says of the Sierra introduction: ''IBM is very smart on pricing. They are very clever about fine-tuning the pricing curve.'' Frank Gens, an analyst with International Data Corp., a Framingham, Massachusetts, consulting firm, points out that neither the plug-compatible manufacturers nor the Bunch have profit margins that would allow them to win a price war against their giant competitor. So IBM has lots of pricing flexibility. IBM's tactics, carried out through the late winter and spring, produced plenty of orders for the 3090. But sales of the 3080 continued to languish. That made the company's biggest problem getting the new machines out the door faster than originally planned. In the tough economy, the company faced sagging 1985 sales and earnings unless it could start delivering 3090-200s in quantity before November. In mid-June, Akers was forced to admit to security analysts that, contrary to earlier and rosier forecasts, earnings and sales might not grow at all in 1985. For a company that venerates steady growth, that prospect was horrifying. Akers also announced that shipping of the Sierra would begin in October, but that news wasn't good enough to overcome the shock of the financial forecast. IBM's shares plunged amid a general sell-off in technology stocks that knocked the entire stock market for a loss. A few weeks after Akers's talk, IBM began telling its biggest customers privately that delivery of Sierra machines would be moved up much further. The company is tight-lipped about how it managed the feat, though an industry analyst who follows IBM closely says flatly, ''Production and engineering at IBM really had their arms twisted to promise those things as early as they're sending them.'' In any case the assembly lines in Poughkeepsie, New York, which turn out the Sierra, got the job done. IBM's second-quarter results were weak, as expected -- sales and earnings, haunted by the looming Sierra introduction, were down from a year earlier. But on the same day that IBM released those figures, it announced news that completely overshadowed them: the Sierra would start being shipped in late August. The stock market quickly appreciated the impact that the accelerated shipments will likely have. IBM's shares, which had hit an all-time high of $138.25 in February, bottomed out at $117.375 after Akers's grim statement in June. Since the announcement of the new shipping schedule, they have rebounded to more than $130. Revenues and earnings should start to recover in the third quarter -- which one security analyst had previously expected to be ''as soft as puppy poop.'' By the end of the fourth quarter, likely shipments of around 500 new mainframes, costing at least $5 million apiece, will give IBM a shot at some increase in revenues and earnings for 1985 over 1984. Profits then were $6.6 billion, or $10.77 per share, on revenues of $45.9 billion. WITH ITS DIFFICULT birth over, the Sierra looks robust. IBM has a 15-month order backlog, and analysts Richard Imershein and Marc Butlein of the Gartner Group, a Stamford, Connecticut, technology research firm, say their surveys show that more than 2,000 machines were ordered for U.S. delivery during the first two weeks IBM accepted orders. Some of those orders may evaporate, since the computer industry has never required deposits from customers. But some may also turn into orders for the more powerful 3090-400, priced at $9.3 million for an estimated 50 MIPS of computing power. Though the model 400 isn't officially due until the second quarter of 1987, Butlein and other analysts expect it to be delivered earlier, possibly in the third quarter of 1986. Frank Gens of International Data Corp. concludes that ''IBM made the best of a bad situation and handled the Sierra introduction about as skillfully as one could hope for.'' But IBM is conjuring new ghosts -- in the form of new technology -- faster than ever. That makes ghostware a problem that won't go away. Though the painful launch of the 3090-200 is over, that machine is already haunted by the specter of the 3090-400. And many in the industry feel sure that a 3090-600, so far unannounced, looms just a few years away.