WHY THE FASTEST CHIP DIDN'T WINDigital's superspeedy Alpha outraces the Pentium. But so far, it can't crack the mass market
Last summer, executives from Digital Equipment Corp. fanned out across the country. They traveled light, carrying the latest samples of the company's powerful Alpha computer chip encased in plastic sleeves the size of credit cards. Their mission: To talk computer makers, including Compaq Computer Corp. and IBM, into using the chip as the engine for speedy new machines.
But after 10 months on the road, it's beginning to look more like mission impossible. While Alpha boasts bragging rights as the world's fastest microprocessor--twice the speed of Intel Corp.'s popular Pentium chip--Compaq and IBM just aren't interested. Alpha's raw computing power, they say, isn't enough to make up for its minuscule market share and lack of software. ''What's the gain for the pain?'' asks Mike Perez, vice-president of Compaq's server-products division. The leading PC maker estimates it would have to spend $50 million just to retool its manufacturing plant for Alpha. ''Searching for absolute performance may be an interesting technical exercise,'' he says, ''but it's not where we want to be from a broad business standpoint.''
WHITTLED AWAY. After five years, Digital executives are still knocking on doors and still coming away almost empty-handed. They insist that Alpha's big break is just around the corner. But the chances of that happening are getting slimmer by the day. Despite spending an estimated $2.5 billion to develop and promote Alpha, the chip ranks dead last in market share with less than 1% of the $18 billion microprocessor market, vs. Intel's 92%. And with new generations of lower-cost chips due this year, including Intel's Pentium II in May, Alpha's performance advantage is being whittled away.
It couldn't happen at a worse time. The early sales spurt of Digital's own Alpha computers has slowed to a crawl. On Apr. 17, when Digital reported revenues for the third quarter ended Mar. 31, sales of Alpha computers grew an anemic 3%--the third straight quarter of single-digit increases. Digital stock is taking a beating, down 32%, from 38 on Jan. 30 to 25 1/2 on Apr. 15. Says analyst George Weiss of the Gartner Group Inc.: ''If Alpha does not generate volume soon, the question arises: What is the Digital advantage and why are they blowing it?''
Good question. In an industry so thoroughly shaped by bigger-better-faster technology, how could a superior computer chip fail to catch on? Clearly, Digital takes the rap for its own missteps. But behind Alpha's poor market showing lies a broader business lesson: New technologies, no matter how whizzy, have an increasingly tough time finding acceptance against well-entrenched competitors, like Intel. Despite the quick-change nature of the computer industry, a huge installed base can be an impenetrable fortress.
HESITATION. Not that Digital didn't contribute to its weak Alpha showing. BUSINESS WEEK interviews with current and former Alpha executives show the company hesitated backing the chip in its early days--a crucial error--before charging full speed ahead. And the company squandered a performance lead through a series of other blunders, ranging from a lack of software to take advantage of the chip's speed to internal haggling over pricing.
Digital sees it differently. CEO Robert B. Palmer admits Alpha has not lived up to the company's original hopes, but he insists its best days are ahead. As a sign of the chip's viability, he points to the $7 billion worth of Alpha computers that Digital has sold since their introduction in 1992. What's more, Palmer is proud of the fact he has continued investing in Alpha, keeping it technically tip-top for five years, despite company losses totaling $5.4 billion over the same period.
There is no question Alpha came at a critical time. When Digital launched the chip on Feb. 25, 1992, the company was losing $8 million a day. Its reputation as a provider of world-class technology was sinking fast. So Digital execs seized on Alpha as the core of their comeback plan for a wide range of machines, from powerful PCs to large-scale computers.
But even before Alpha hit the market, Digital fumbled. The company had shown off early versions of the chip at an industry conference in February, 1991, and engineers at Apple Computer Inc. were impressed. Apple was in the market for a new chip supplier, and Alpha looked promising.
In late June, John Sculley, then Apple's CEO, invited Kenneth H. Olsen, Digital's founder and president, to dinner. Sculley had a proposition: Apple's Macintosh computers were starting to run out of gas, and he wanted to do a complete redesign with Alpha at the heart of the new Macs.
But Olsen had doubts about Alpha. His unshakable faith in the VAX computer, which had turned Digital into IBM's most formidable competitor in the 1980s, made him reluctant to phase it out too soon in favor of Alpha. Olsen asked a team of Digital's top engineers to extend the computer's design for another generation--and he rejected Sculley's proposal.
A few months later, Apple announced that its new Macs would run on the PowerPC chip, a competing design by IBM and Motorola Inc. Sculley says one Digital director later told him that Digital's board was ''distressed that nothing came of these discussions and that Digital lost a great opportunity.'' The Alpha faction at Digital was crestfallen. ''Ken did not want the future of the company riding on Alpha,'' says William R. Demmer, a former vice-president of Digital's Alpha and VAX businesses who retired in 1995. Too bad. With Apple as a customer, Digital would have had 3.4% of the microprocessor market, although a distant No.2 to Intel. Olsen did not respond to repeated requests for an interview.
MEMORY HOG. By late July, 1992, Olsen's misgivings about Alpha ceased to be an issue. The company had ended its fiscal year with a $2.1 billion loss, and Digital's board asked the 66-year-old founder to step down. His replacement was Bob Palmer, a Texan who had spent the bulk of his career in computer chips, most recently as head of manufacturing.
With Olsen gone, Digital's senior management quickly settled on a strategy of trying to license the Alpha design. ''We had to get a major chip player to accept Alpha and drive it into the marketplace,'' says John F. Smith, then Digital's senior vice-president of operations and now president of PerSeptive Biosystems Inc., a pharmaceutical equipment maker. Smith and other Digital execs didn't think they had a chance persuading computer companies to use Alpha--unless a big chipmaker was behind it.
So Smith went to the biggest maker of chips in the business--Intel. In the fall of 1992, Smith called Andrew S. Grove, CEO and president of Intel. Grove was well aware of the technical merits of Alpha, but it did not take him long to pass on licensing the chip. Adopting Alpha, even as a high-end addition to Intel's commodity business, would require a costly overhaul of Intel's design and manufacturing processes. ''We didn't want to branch out to an incompatible deal,'' Grove says.
Nor did Texas Instruments, Motorola, and NEC. So Digital set out for even bigger game--Microsoft. Digital executives were pushing for Microsoft to adapt its next-generation software, Windows NT, to Alpha. Palmer met with Microsoft CEO William H. Gates III and agreed to make NT a central part of Digital's computer business. In return, Microsoft agreed to release a version of NT for Alpha at the same time it came out with versions for Intel and the MIPS chip designed by Silicon Graphics Inc.
Digital executives were elated. They constructed their business plans for fiscal 1993 around the assumption that NT would quickly drive Alpha into high-volume markets, including corporate PCs. That meant pouring tremendous resources into fine-tuning Alpha to work with Microsoft's software.
But in early 1993, when Digital engineers eagerly loaded test versions of NT onto their computers, it quickly became apparent that the software needed far too much computer memory to run on a typical PC--or even a $5,000 Alpha machine. That put NT beyond the reach of the mass market that Digital had been counting on for Alpha.
A mad scramble erupted inside Digital. Palmer ordered a crash program to bring out Alpha computers running a stripped-down version of Digital's Unix operating system, a rival to NT. But many customers doubted Digital's commitment to Unix would last once Microsoft solved its NT problems. Revenue fell, and Digital posted a $224 million loss that year. By focusing Alpha on NT, Digital ''lost three years in the market,'' says Edward J. Zander, president of Sun Microsystems Computer Corp.
Still, the future of Alpha--and Digital--increasingly depended on NT. And Digital's relationship with Microsoft was growing more complex. In mid-1993, Digital engineers looking into early versions of NT noticed that some portions of the program bore a striking resemblance to an advanced operating system called Mica that Digital had developed, but canceled in 1988. Mica was the brainchild of Dave Cutler, a former Digital software star who joined Microsoft in 1988 and was now the chief architect of NT.
Palmer decided that Digital had a legal claim against Microsoft. But, insiders say, instead of filing suit, Palmer chose to use the threat of legal action to spur Microsoft into improving Alpha's prospects. Microsoft execs won't comment, saying it concerns legal matters. Still, Palmer's gambit appears to have worked. By the spring of 1995 the two companies hammered out details of a broad agreement for Digital to provide NT network installation services for Microsoft. Announced with great fanfare by Digital in August, 1995, the alliance included payments by Microsoft estimated at $65 million to $100 million to help train Digital NT technicians.
On the surface, that appeared to help boost Digital into a prime spot for milking NT's strong growth. Since October, Microsoft has also dropped its support of two rival chips--the PowerPC, made by IBM and Motorola, and MIPS. But this crucial alliance may now be fraying. On Mar. 19, Microsoft announced a joint marketing and services agreement with Digital's archrival, Hewlett-Packard Co. The Palo Alto (Calif.) company will hawk NT in large corporations--the role Digital once held exclusively.
In retrospect, former Digital executives say, the company would have had more success if it had lowered the price of Alpha computers. By late 1994, engineers had found a way to deliver a $4,995 ''AlphaStation'' aimed at the fast-growing workstation market. But a clash broke out among Digital's senior management because of fears that this could damage the 50%-plus gross margins Digital enjoyed on high-end computers, one of the few bright spots in its gloomy financial outlook.
''SPLIT PERSONALITY.'' Digital's PC Business Unit was none too happy about the prospect of low-cost Alpha machines, either. Execs there lobbied against it, saying it would undercut prospects for building a big business around Intel-based PCs. When the Alpha- Station finally hit the market, it bore a $7,995 price tag. ''Digital has always had a split personality,'' says Enrico Pesatori, who left Digital last summer after the PC Business Unit he headed suffered continuing losses. ''Every Intel server sold means an Alpha server that is not sold.''
Today, Digital's Intel-based machines running NT outsell its Alpha computers 9 to 1 in the high-volume market for low-cost servers, says the Gartner Group. That's because software for NT has to be specifically tuned for Digital's Alpha chip, a task few software developers are interested in taking on given the small volume of Alpha-based computers. Even longtime Digital customers, such as Toys 'R' Us Inc., are building NT networks linking thousands of locations on Intel computers instead of Alphas. ''The biggest reason is the lack of software,'' says Matthew J. Lombardi, vice-president for information technology at Toys 'R' Us.
Of course, Digital is more than Alpha and Intel-based computers. The company, for example, is expected to post revenues of $5.8 billion in services this year, while networking and computer storage systems should contribute roughly $1.4 billion. But it is Alpha that distinguishes Digital from its competitors, offering the company its best hope for returning to star status.
And Alpha's best opportunity, Palmer insists, lies in the future. In mid-March, the company introduced a low-cost version of the Alpha chip priced from $295 to $495--some $100 less than its previous minimum price. That makes Alpha affordable, for the first time, in PCs costing less than $2,600. Digital is hoping that this, combined with a Samsung Electronics Co. alliance to make and market Alpha, will help the chip crack the volume computer market. ''Alpha will carry Digital for 20 years while other chip technologies have fallen by the wayside,'' Palmer says.
A couple of years ago, he predicted one in five computers running Microsoft's Windows NT software would be powered by Alpha. The number today is closer to 1 in 20, according to the Gartner Group. So far, though, only 27 small companies have plans for low-cost Alpha computers, the largest being German PC maker Vobis Microcomputer. Vobis tried to launch an Alpha workstation in the German market four years ago but abandoned the campaign after selling fewer than 500 machines in two months. Vobis CEO Gert Huegler, though, is now willing to give Alpha a second chance because, he says, it delivers better performance than Intel. Also key: Digital has an answer to the software shortage--so-called ''translator'' software that enables Alpha to run programs written for Intel chips.
Even so, Huegler figures that Vobis will be lucky if sales of the $4,000 Alpha machines amount to 5% of the company's overall revenues of $1.3 billion. And if it doesn't--so what, he says. ''We can switch it on or switch it off,'' Huegler adds.
Unless more computer makers start flipping the switch on Alpha, Digital may wind up with the world's fastest microprocessor and the smallest installed base of customers.
By Paul C. Judge in Boston, with Andy Reinhardt in San Francisco and Gary McWilliams in Houston
Updated June 15, 1997 by bwwebmaster
Copyright 1997, Bloomberg L.P.