GENE AMDAHL FIGHTS TO SALVAGE A WRECK His failed supercomputer swallowed most of the $230 million that investors bet on Trilogy Ltd. Now the wizard of high tech is trying a new strategy.
By Myron Magnet

(FORTUNE Magazine) – BUSINESS TROUBLE comes in all sizes, rarely bigger for one man than Gene M. Amdahl's quarter-billion-dollar debacle with Trilogy, a company he founded to beat IBM at its own game. Trilogy's 1984 downfall plunged Amdahl into guilt and recrimination. But as misery loves company, his violent, front-page, heroic disaster has joined hands with the ordinary, grinding, slow-death trouble at an outfit called Elxsi. Amdahl hopes to transform two negatives into a positive. While this isn't an unmitigatedly happy ending to a story of failure and disappointment, it is at least the sort of interesting, guardedly optimistic turn of events that happens in real life. The two companies, linked since October in a merger as hardheaded as the most unromantic Victorian marriage of convenience, now have a chance of winning some measure of the success that eluded them separately. Yet Amdahl and the officers of Elxsi must at times feel chained to each other like escaped prisoners, enjoying a welcome but qualified deliverance. And, however promising, the merged company's prospects are far from what Amdahl, 63, had envisioned when he founded Trilogy in 1980. The company's main hope is a just-announced technical breakthrough that hitches the high-powered Elxsi (pronounced Alexi) mini-supercomputer to the success of Digital Equipment Corp.'s triumphant VAX minicomputer. Users of the 30,000 VAXes chugging away on scientific and engineering problems face a tough choice when they reach the limits of VAX computing power. They can hook up a second VAX, but because of the machine's design they get a lot less than twice the bang for twice the bucks. They can buy a more powerful machine from various DEC competitors, but it won't readily run the expensive software bought or developed for the VAX. The new Elxsi machine, says the company, is more powerful than the VAX but no more expensive, and it accepts most VAX software without a single change -- claims made for no other machine so far. Though worthy, this effort is a comedown for Amdahl. When he founded Trilogy, it was as if he proposed to create the world in five days rather than the previous record of six. But why not, given his reputation for near omnipotence? The architect of IBM's celebrated, unprecedentedly successful System 360 and ultimately IBM's director of advanced computer systems in the Sixties, Amdahl had broken away in 1970 to form his own company, Amdahl Corp., which competed directly and brilliantly -- for a while -- with his former employer.

A classic of business strategy, his 1970s campaign aimed straight for a chink in IBM's pricing structure. The prices of the various IBM machines were based on their computing power, not on their production costs. IBM didn't build a bigger computer than it did, Amdahl observed, because the high price it would have to bear would make the market too small to be worth Big Blue's while. So Amdahl Corp. built the machine itself -- an ultrapowerful computer so perfectly compatible with IBM's software that it was virtually an extension of IBM's line. Yet its price was the same as IBM's top machine, meaning that IBM couldn't respond unless it cut prices all down the line, wounding itself with the same bullet that would rid it of Amdahl. Big Blue finally pulled the trigger after Amdahl had won 22% of the large-systems market and posted a 30% pretax profit for 1977. In 1979 Amdahl's profits plunged and its founder left. But Amdahl wasn't finished with IBM. With Trilogy he aimed to sink his teeth so deep into Big Blue's leg he would never be shaken off. He set to work creating an advanced technology of such overweening ambition it takes your breath away even now. To boost speed, the basic measure of computer horsepower, he would develop a densely integrated superchip, filling an entire wafer of silicon that normally gets chopped into about 100 slower chips. Not content with only one technical revolution, he would then turn from semiconductors to computers themselves -- a different field of engineering -- and embody his coaster-size megachip in an innovative system twice as fast as IBM's Sierra, the advanced machine IBM was developing but had not yet introduced. The Trilogy prospectus warned in big letters that the company was risky, no place to put your retirement fund. But investors brushed past the warning signs. Perceiving Amdahl to be one of the trailblazing geniuses of the computer age, they endowed Trilogy with the biggest start-up kitty ever: $230 million. ''They thought that with me there,'' says Amdahl, ''it wasn't really risky.'' But it was. ''It was risky for reasons I hadn't originally known how to evaluate, and in retrospect can't fully evaluate either,'' he says. ''It wasn't that the technology couldn't be done; it was that we had bitten off more than we could chew. The technology was so advanced it required something new in almost every area of supporting technology.'' The problems were nowhere near as dramatic as later press reports, which told how a prototype chip had overheated to a merry red glow and self-destructed before astonished Trilogy executives. The problems were more mundane, like getting a layer of wires to stick to a layer of plastic so the wires wouldn't ultimately curl up like the clenched claw of a dying beast. Finally the wires stuck. ''There wasn't any single technological problem that wasn't solved,'' says Amdahl. He could do it all, given time -- but in the fast-moving computer business you don't have much time. The project kept missing deadlines, then revised deadlines. Harder to fix, the company wasn't working quite right either. The parts of the organization never seemed to mesh precisely. Engineers didn't develop that second-nature reliance on one another that for Amdahl marks a successful technology company. Instead employees galloped off in all directions, painting their own projects as essential and demanding reinforcements. The organization grew bigger and more unwieldy by the month. Like tunnels started from both ends that fail to meet in the middle, the separate projects never cohered into a whole. ''We were off only a little,'' Amdahl says in his slow, shy manner, ''but it was enough.'' AN EARLIER AGE would have said that the Fates did not smile on Amdahl. Just after Trilogy was started, driving the forest-green Rolls he had bought when Rolls-Royce's jet engine company chose an Amdahl computer, Amdahl collided with a motorcycle and killed its driver. The emotional turmoil and the need to defend a civil damage suit kept his mind focused on more than Trilogy. Amdahl pleaded nolo contendere to a charge of misdemeanor manslaughter, and his insurance companies settled the civil suit out of court for an undisclosed sum. Then flood struck: The worst rainy season in recent California history turned Trilogy's building site into an equipment-devouring sea of mud, delaying construction of the chip-making plant for months. Worse, water had seeped into the bearings of the air-conditioning system as it was being installed. They rusted out once the plant was operating, the system stopped filtering air, and dust ruined semiconductors that had to be kept superclean. Figuring out what was wrong took five months. Two years into the project, a brain tumor deprived Trilogy of its president and Amdahl of his old friend, Clifford J. Madden. By March 1984 Amdahl and his top colleagues knew the project was in trouble. Trilogy couldn't produce satisfactory chips in quantity, and Amdahl thought it might take two years, and much more money than the carloads Trilogy had already raised, to achieve that. In late May he abandoned the idea of building a computer. In August Trilogy officers persuaded Amdahl to admit total defeat and fold the superchip. The failure tore him to pieces. A rigid, rectitudinous South Dakota farm boy whose intense manner, thick wavy hair, and stubborn dimpled chin give him the look of an evangelical Kirk Douglas, he couldn't come to terms with it. This, after all, is a man who makes a science even of his pleasures, with a wine connoisseurship as precise as analytical chemistry. His personal loss of nearly $2.25 million was a substantial portion of his net worth, diminished over the years by generous gifts to South Dakota schools. But that loss isn't what troubles him most. ''So many people invested because I had my name on it. I feel responsible,'' he says. ''I get haunted by guilt feelings. I sort of cringe when I see my acquaintances. That was a bruise -- a real bruise that hasn't healed yet. And I don't know that it ever will.'' As he speaks of ''the disgrace of the Trilogy failure,'' his hands shake. What could be salvaged from the wreckage? The decision wasn't Amdahl's alone, for in late 1984 he stepped down as chief executive (remaining chairman), and the board of directors put in Henry Montgomery, a no-nonsense turnaround man with battle ribbons from the Memorex and Fairchild Camera & Instrument campaigns. Montgomery set engineers to doing contract research for other computer makers. Trilogy sold this business to DEC for $10 million this June, along with a license to manufacture technology it had developed for connecting groups of chips. That left interminable, prosaic mopping up: settling a shareholder suit, buying out limited partners for stock, leasing equipment and facilities piecemeal. Amid the ruins, Trilogy retained one other asset: cash, some $70 million of it, unspent in the grandiose debacle but melting away steadily. Like getting an organ transplant, Trilogy could use that remnant to acquire a company with a developed product and so stay in business, if only as a holding company. But where to find a cash-hungry candidate? The semiconductor business? Too crushed by the slump. Work stations? ''That was going to be too crowded,'' says Amdahl. What about Elxsi, the company with the odd name patched together from ''electronics + systems integration''? Started in 1979 by a group of friends led by Joseph Rizzi, a founder of the successful Intersil semiconductor company, Elxsi was based on technological ambitions that, if not as awesome as Gene Amdahl's, were big enough to give it trouble and a need for cash. Now that computers have become so fast, holds the emerging engineering orthodoxy, the trick to getting the next really big leap forward is to design machines made up of several processing units that can rip a complex job into parts and work on them simultaneously, like a school of piranhas processing a goat. Elxsi proposed to build the first commercial parallel processor, as such a computer is called. Always anxiously peering into the future for every possible development, the Elxsi team engineered with an obsessive thoroughness that delayed their entry into the market a costly year. When they did get to the marketplace in 1983, they expected a throng of technical and scientific users to clamor at their San Jose door for this computing panacea. No need to market so irresistible a machine; they just started producing Elxsis by the dozen and hired staff and leased space to produce even more. Result: Elxsi lost $6.8 million in 1984 and $17 million last year. As his money ran out, Rizzi thought he could raise only $10 million or $15 million more from the venture capitalists who had been backing him. ''That wouldn't have been enough money to build a serious computer company in today's market,'' he says. Introduced through the San Francisco investment banking firm Hambrecht & Quist, Rizzi and Amdahl agreed in May 1985 to merge. When the deal closed last October, Elxsi, whose performance had weakened further, became Trilogy's subsidiary and only significant business. In January, after forcing out Montgomery and getting Amdahl named C.E.O. once more, Rizzi and Amdahl sent in a brisk, tough new president, 48-year-old Lancashireman Peter Appleton-Jones, to make Elxsi shine its boots and march along smartly. An ex-senior marketer at IBM and Cray Research, he was not at a loss for what to do. He slashed staff and shut down manufacturing until the company had sold its inventory. He ordered his sales troops to stop substituting perfumed fantasies for plain truth in their projections. He also told salesmen to de- emphasize Elxsi's smallest units and to focus on selling bigger, more profitable systems. Appleton-Jones says the losses are declining, ''and if we're able to continue that trend, which I'm hopeful of doing, then we'll be profitable in the fourth quarter.'' Fueled by Amdahl's advice and exhortation and his company's cash, Elxsi engineers brought to the boil new developments that had been simmering slowly. Several weeks ago it brought out new central processing units, up to 12 of which can be plugged right into a single computer. The new units are twice as powerful as the previous model and have four times the internal memory. And now comes the ability to run VAX programs, allowing Elxsi to present its machines as extensions of the formidable DEC computers rather than as direct competitors. At first glance this strategy seems to have Gene Amdahl's fingerprints all over it. Isn't this what he did to IBM, building a compatible machine more powerful than Big Blue's but no costlier? The similarities are obvious, but while Amdahl ardently supports this strategy, he didn't originate it. Everyone connected with Elxsi seems to have a different account of who did. So this story does not end with Gene Amdahl -- late in his career but with his ambition stoked up as hot as ever -- planning to beat one more titan at its own game for the sheer glory of the challenge. He is only trying to save what he can for the Trilogy shareholders. In that effort he has spent much of the last two years on tasks that don't need his talents, like selling equipment and subleasing property. His fame has helped Elxsi's marketing effort; some of his ideas are incorporated in the recent new products and in more still to come. But he is Elxsi's elder statesman, not its chief designer, and now that Appleton-Jones is in place, he says, ''there isn't a full-time job for somebody above him.'' He talks about retirement in a few years. Overseeing his investments in high-tech start-up companies, he says, will give him ''many ways to be an entrepreneur vicariously.'' It's too soon to tell how much of the wealth of Trilogy shareholders Amdahl will salvage with the Elxsi move. Though competing mini-supercomputers have emerged with capabilities Elxsi lacks, the company argues that it has a much more flexible general purpose machine than any competitor. ''I think we have a basis for becoming a significant company,'' says Amdahl, but a lot depends, he adds, on whether Elxsi can grow quickly. The quicker the better, for his peace of mind. Even with Elxsi folded in, Amdahl's Trilogy, launched with a quarter of a billion of mostly other people's dollars, now has a market value of about $40 million.