Conference ricks::dechips

641.0. "Making money from FAB6; can it be done?" by PCBUOA::KRATZ () Tue Jan 28 1997 17:25

From Investor's Business Daily, 1/28/97
    
Headline: Computers & Technology Making Money In Chips: Can It Be Done At DEC?

======================================================================
   Some companies churn out profits. Some just churn out break-even
forecasts.
   Digital Equipment Corp., glamorous computer giant of the '80s, has
been the '90s leader in "we'll be even by next year" projections.
Chief Executive Robert Palmer's latest prediction: break-even in
Digital's semiconductor operations by '98.
   Break-even in chips would be an enormous feat for Digital. In
'92, Digital began building a $450 million fab, or chip-making plant,
in Hudson, Mass. Its chip operations have been in the red ever
since.
   The problem, simply put, is that Digital has never been able to
fill the factory. Its chief product, the speedy but costly Alpha
microprocessor, never really caught on beyond its use in Digital's
own PCs, workstations and servers.
   But Digital, based in Maynard, Mass., now seems to be taking a
different tack. "Fill the fab" is the rallying cry in Hudson these
days. And Digital is moving on several tracks toward that end.
   * Late last year, Alpha prices were slashed. Speedy 300-megahertz
chips that sold (or more precisely, were not sold) for $695 were
reduced to $395.
   * Alpha licensees Samsung Group of Korea and Mitsubishi Electric
Corp. of Japan should soon help broaden the market and lower the
prices of Alpha chips. A new Alpha, developed with Mitsubishi, will
compete with Santa Clara, Calif.-based Intel Corp.'s much-hyped
Klamath multimedia processor, due out this spring.
   * Meanwhile, Digital has quietly been scoring agreements with
companies to design products around its low-power Strongarm
processor, developed under a license from Advanced Risc Machines Ltd.
of the U.K.
   Troubled companies tend to retreat behind a fortress mentality,
and Digital is no exception. Ed Caldwell, vice president of
semiconductors, is tight-lipped about his chip operations. But when
asked if it would be correct to say the Hudson plant is still just
half-equipped, Caldwell did allow that the statement was "fair."
   Caldwell, however, would not reveal the capacity use for the half
of the plant that is equipped. It's easy to see why a fab producing
at less than half its capacity would not yield any return on
investment. But how far might it be from breakeven?
   To break even, chip plants generally must match their construction
costs in annual revenue, noted Will Strauss, president of Forward
Concepts, a market research firm in Tempe, Ariz. Having made some
incremental equipment purchases beyond its original $450 million
construction cost, it's likely Digital has spent more than $500
million on its Hudson fab.
   Caldwell shed no light on what Digital might have realized in
Alpha revenue last year. But back-of-the-envelope calculations based
on Alpha systems volumes suggest Alpha sales probably did not top
$100 million.
   But Digital has had some success selling a variety of data
communications chips. In all, Digital likely produces about $187
million in annual chip revenue, estimates analyst Mark Kirstein at
In-Stat Inc. That's still a long way from $500 million.
   Digital's best fab-filler hope may be Strongarm, the 32-bit RISC
processor developed under license from Advanced Risc. With its high
speed and low power use, Strongarm is a natural for portable devices.
   Apple Computer Inc.'s MessagePad 2000, a new personal digital
assistant, will be built around Strongarm. Five makers of network
computers, including closely held Wyse Technology Inc. of San Jose,
Calif., have announced machines built around Strongarm.
   Smart mobile phones and set-top boxes used to deliver video on
demand are also likely markets for the chips. And Strongarm chips
may find their way into TVs designed to double as Internet browsers,
notes Terry Shannon, publisher of the newsletter Shannon Knows DEC in
Ashland, Mass.
   "It's my belief that Strongarm is the solution to the 'fill the
fab' equation," Shannon said.
   But many of the major Strongarm products - like Apple's Message
Pad 2000 - are not yet in production. Indeed, of the makers of the
five network computers to be built around Strongarm, only Wyse has
had any significant sales.
   Keep in mind that there are less than 300,000 Alpha computer
systems installed worldwide -hence, the importance of Strongarm and
lower-priced Alphas.
   As Intel has shown, volume production is the name of the game in
chips. High volumes lead to higher numbers of good chips per wafer,
ultimately expanding profit margins.
   The slashing of Alpha prices is clearly designed to jack up
volumes. But is this simply a desperation move?
   No way, insists Caldwell, arguing that the timing is right for
price cuts. Digital, he notes, recently released its FX!32 software,
which lets Alpha run 32-bit Windows programs at what Caldwell calls
"interesting" speeds. That may not be as fast as they run on
Pentiums, but a lot faster than they've run until now.
   "That provides an opportunity," said Caldwell. "As long as there
was little opportunity to grow volume, you tend to keep prices up."
   Of course, that's a policy that's failed. Without a pricing
advantage, and lacking the breadth of software to compete with WinTel
machines, Alpha has been spurned by major PC makers. But Alpha's
superior performance has never been questioned.
   "(Digital) does have the fastest off-the shelf chips you can buy,"
said Forward Concepts' Strauss. "I don't see anyone throwing rocks
at their performance."
   With Strongarm, Digital has once again come up with a best-of-
breed in processor performance. There are few chips to rival
Strongarm for embedded applications in portable devices that require
digital signal processing. Certainly, there's no entrenched rival
like Intel to unseat. So Strongarm already has won far broader
backing among systems makers than Alpha ever achieved.
   So although break-even is no sure thing for Digital's chip
business, it's a forecast that doesn't seem entirely unrealistic.
    
    

    Many of the numbers in this article are wrong, some (such as revenue) are 
    way off which is to be expected since we don't give these numbers out.   

    -Bruce 

    Fine, while we sit here with "better" numbers, the industry reads
    editorials such as this and believes them.  I would in the absense
    of anything else.
    
    				-John

    The same is true with all of Digital's Business units. We do not
    publish numbers per units so people like the Gartner group guess and
    others publish there guesses.  
    
    -Bruce  

    OK.  So short of publishing real numbers that we don't seem to want, what
    can we say to help consulting groups like Gartner make better "guesses"
    about us.
    
    				-John

    re: .3
    The comparison to other Digital Business Units may be accurate, but our
    real comparison benchmark should be to the competition.  Digital Semi
    is usually pretty good at this  (especially compared to other Digital 
    business units ;^)  )

    re: .4 & .5
    
    My reason for making my comment was to inform people that read the note
    that the numbers are inaccurate and hopefully prevent people from using
    them. I was not trying to make any other point. 
    
    -Bruce  

    Hopefully Caldwell was misquoted as well.
    
    If not, you have to wonder how the SBU feels about the there-was-
    little-opportunity-to-grow-volume-so-we-kept-the-prices-up policy
    as they were struggling to make numbers with Alpha XL, etc.
    K

      The Alpha XL was our first generation of Personal Workstations.
      
      They where priced against percived competition, not from production cost 
      really. The Alpha CPU costing 500$, 1000$ or 1500$ ment different 
      grossmargins, but these are where not at the level where that was 
      influceing the price conciderably.
      
      This has changed and the PWSxxxa is more dependant of the component 
      prices as the NT workstation market has expanded with Compaq, HP and soon 
      IBM jumping the bandwagon. Also with the clones increasing in numbers and 
      price competitiveness (Enorex etc) it will be more PC style pricing that 
      reigns.
      
      Fx32 and PCA56 will make life more interesting when it get boxes to 
      market.
      
      
      Bj�rn