It now seems likely that the year 2000 represented the peak of the
desktop SPARC workstation market, which started in 1989 with the launch of the
SPARCstation 1. And I predict that, by the end of 2003, there will be only one
supplier of desktop SPARC workstations left to service a market that's already
declined significantly in revenue as a result of the recession in 2001, and the
overlap in capability between Unix workstations and dekstop PCs. And there are
no prizes for guessing that will be
The workstation market has outlasted its usefulness, but it's
interesting to see where it's been.
In the early 1990's SPARC
workstations provided significant time to market benefits for a wide variety of
technical developers in markets as far apart as publishing, software development
- SPARC processors were faster than those used in PCs
or competing Unix workstations
- Unix workstations allowed users to work with larger
memory models which couldn't be accomodated on PC operating systems or (then)
PC hardware busses
- multiprocessing, appeared in SPARCstations years
before PCs, and enabled users to get significant computing speedups at economic
The market was also good for wannabe SPARC workstation oems, because a
typical SPARC workstation could be sold at an ASP which was an order of
magnitude higher than a PC, with a cost which was only fractionally more. The
profit potential in this market attracted dozens of competitors. The peak of
the SPARC workstation market (number of competing oems) occurred in 1994, when
oems could choose competing 32 bit SPARC chips from Sun, Ross Technology,
Fujitsu Microelectronics and Weitek. In fact HAL Computer launched their own 64
bit workstation using their own design of SPARC chips, before Sun's own Ultra 1.
But Sun had several major advantages over its SPARC rivals at that
- Sun controlled the operating system. The FUD factor
(Fear, Uncertainty and Doubt) worked as well for Sun as it ever did for IBM in
its glory days of the 1970's. (In the PC market, Intel's rival AMD prints an
endorsement from Micosoft on its chips, to reassure users that the AMD processor
will in fact run Windows.)
- Sun had a well developed VAR friendly reseller
- and most important of all... Sun's competitors
were good at designing products but mostly useless at marketing.
There isn't the same range of competing systems
makers, or chip suppliers today when at the and of 2001, the competitive market
for RISC workstations is vastly different.
- PC's no longer have the same memory restrictions as
in earlier generations.
- PC's overlap SPARC workstations in performance.
- Users have grown more sophisticated, and after 20
years of living with a PC market, user companies are no longer willing to pay
much of a premium price for a RISC workstation, so the workstation "goldrush"
If we take a dispassionate look at the SPARC workstation market of today,
then it is no longer a driving force for technology in the computer market. That
role has been taken up by the fast growing high availability rackmount server
segment, which we'll return to in a future article.
In fact, if Sun
were a PC company, its total annual output of desktop units, which is about the
same as one day's production at Dell Computer, means that Sun wouldn't even
appear as a named company in most market research reports. It would just appear
in the anonymous pie chart labelled "others". This may come as a
surprise to some readers who mainly rely on Sun's own web site for their news,
because for most of this year we've seen a succession of announcements from Sun
claiming that it's been increasing its market share in the Unix workstation
market. If instead, readers had been seeing their news in the SPARC Product
Directory, they would have seen through this Sun doublespeak. The "Unix"
workstation market is mostly "Sun" anyway, since other Unix
workstation makers have long switched switched to making Windows workstations.
So when Sun has been claiming year on year market share increases, it's often in
quarters when Sun's own year on year volume shipments have actually declined.
you look at the ads in Sun focused publications, you'll probably reach the
conclusion that Sun's desktop SPARC competitors have already given up the battle
of promoting themselves against Sun. Most of these companies now sell systems
which use Sun motherboards, and as the market share of SPARC workstations
continues to decline, we're going to see just one company left standing, the
company which started the whole thing back in the 1980's - Sun Microsystems as
the only company left actively marketing commercial desktop SPARC workstations.
There'll also be a handful of companies making workstations for specialised
markets such as the military... and SPARC portables are a different market
again, which haven't yet peaked, and like of the PC portable segment, are doing
a lot better than their desktop cousins.
Will there still be a need
for SPARC workstations in the future?
Surprisingly, I think Yes.
Although most applications will surrender to the Intel architecture PC, there
are two types of application for which SPARC workstations will continue to be
needed for many years.
- semiconductor design - which used to be the prerogative of a handful of
major oems in the 1970s, is now used as a tool by thousands of companies and is
accessible even to the smallest startup company. SPARC workstations still
provide the raw horsepower for much of the Electronic Design Automation market,
and there seems no compelling reason for current models written around the
SPARC/Solaris platform to be rewritten for other targets. In the early days of
the SPARCstation market, Sun's workstations could even be seen in the chip
design cubicles at Intel Corp. (Do they still use Sun now? Well I don't get out
so much nowadays, so maybe a kind reader could fill me in here and let everyone
- software design - as long as the SPARC server market continues to be a
major segment in the server market. there will be productivity advantages for
some software developers to use a compatible development platform. Also from the
personnel point of view, it's easier for companies to train new software people
to become conversant with one target platform and use the same platform for
development, especially when there's an ever present shortage of skilled
Conclusion... As far back as the mid 1990's I was telling many of
Sun's new partners that the workstation was no longer a critical factor in the
SPARC market. It had long been overtaken in strategic significance by Sun's
server business. I believe we've now seen the peak of the desktop SPARC
workstation market, and that while volume shipments may rise in the next few
years, they will recover at a slower rate than the PC market as a whole. We can
look across to the Apple Mac market to see that some users still prefer
alternatives to the Wintel PC in specialised vertical markets. That's even more
true in SPARC, which is based around synergy with a strong server business.
So as long as Sun keeps bringing out new workstation models, there
will be users who want to buy them, but it's likely that most of Sun's desktop
SPARC competitors today will either withdraw from that market, or just become