decoration decoration
Stories

GROKLAW
When you want to know more...
decoration
For layout only
Home
Archives
Site Map
Search
About Groklaw
Awards
Legal Research
Timelines
ApplevSamsung
ApplevSamsung p.2
ArchiveExplorer
Autozone
Bilski
Cases
Cast: Lawyers
Comes v. MS
Contracts/Documents
Courts
DRM
Gordon v MS
GPL
Grokdoc
HTML How To
IPI v RH
IV v. Google
Legal Docs
Lodsys
MS Litigations
MSvB&N
News Picks
Novell v. MS
Novell-MS Deal
ODF/OOXML
OOXML Appeals
OraclevGoogle
Patents
ProjectMonterey
Psystar
Quote Database
Red Hat v SCO
Salus Book
SCEA v Hotz
SCO Appeals
SCO Bankruptcy
SCO Financials
SCO Overview
SCO v IBM
SCO v Novell
SCO:Soup2Nuts
SCOsource
Sean Daly
Software Patents
Switch to Linux
Transcripts
Unix Books

Gear

Groklaw Gear

Click here to send an email to the editor of this weblog.


You won't find me on Facebook


Donate

Donate Paypal


No Legal Advice

The information on Groklaw is not intended to constitute legal advice. While Mark is a lawyer and he has asked other lawyers and law students to contribute articles, all of these articles are offered to help educate, not to provide specific legal advice. They are not your lawyers.

Here's Groklaw's comments policy.


What's New

STORIES
No new stories

COMMENTS last 48 hrs
No new comments


Sponsors

Hosting:
hosted by ibiblio

On servers donated to ibiblio by AMD.

Webmaster
1248 ("Responses to the Compaq SPARC Threat") | 397 comments | Create New Account
Comments belong to whoever posts them. Please notify us of inappropriate comments.
411A (SPARC, MIPS & Compaq)
Authored by: Anonymous on Saturday, November 17 2012 @ 03:32 PM EST
Oops. I only saw the excerpts before I transcribed this so
there was some duplication of effort. It's related to 1247
and 1248 which are posted again after this comment.

[ Reply to This | Parent | # ]

1247 ("A Tale of Two Markets - Workstations vs PCs")
Authored by: Anonymous on Saturday, November 17 2012 @ 03:33 PM EST
http://groklawstatic.ibiblio.org/pdf/iowa/www.iowaconsumercase.org/011607/1000/PX01247.pdf


<p>
PLAINTIFF'S EXHIBIT 1247<br />
Comes v. Microsoft
</p>

<p>MICROSOFT RISC PC TASK FORCE REPORT</p>

<h1>A Tale of Two Markets - Workstations vs PCs</h1>

<p>ADVANCED TECHNOLOGY AND BUSINESS DEVELOPMENT</p>

<p>MICROSOFT CONFIDENTIAL</p>

<p>MS-PCA 1113854<br />
CONFIDENTIAL</p>

<p>
<b>TABLE OF CONTENTS</b>
</p>

<p>
TABLE OF CONTENTS<br />
INTRODUCTION<br />
MARKET COMPARISON<br />
<span style="margin-left: 4em">Sun's Strategy</span><br
/>
<span style="margin-left: 4em">Fundamental
Issues</span><br />
<span style="margin-left: 4em">Second Order
Effects</span><br />
WHAT TO DO<br />
<span style="margin-left: 4em">"Business Workstation"

Market</span><br />
<span style="margin-left: 4em">Technical Workstation
Market</span><br />
<span style="margin-left: 4em">Power PC</span>
</p>

<p>
2/26/92 11:13 AM MICROSOFT CONFIDENITAL PAGE 2
</p>

<em>[Ed: Page footer retained for date information. All
other headers and
footers omitted.]</em>

<h2>1. INTRODUCTION</h2>

<p>
We are facing a threat from Sun and the SPARC world which
has generated lots of concern both here
and among our OEM customers. One key issue which everyone
who looks at this problem discovers
at some point or another is that Sun's present business is
not the traditional PC market. At one level
this is obvious, but when it comes down to deciding how to
compete with Sun, the differences in their
current approach generate a lot of confusion. Don't we
need to have a workstation offering? Is RISC
what is important or is it UNIX? Do our OEMs need direct
sales?
</p>

<p>
The closest example was Carl's mail the other day that
started by saying that we are facing a
workstation threat rather than a RISC threat. Although I
largely agree with Carl's proposed actions,
and believe the he understands this issue, I think that
the reality of the situation is not that simple -
there is more than one threat, and it is easy for us or
our OEMs to get confused. We need to be very
clear on the distinction between the markets and what the
real threat is.
</p>

<p>
The challenge is all about Sun - they are the real enemy.
Unfortunately that has become a point of
confusion. In particular, we must distinguish between the
following goals:
</p>

<ol type="A">
<li><b>Prevent Sun from attracting our present end user
customers.</b> Sun clearly has a goal of
coming after our present customer base, or at least in
having the SPARC world as a whole do
so. We must take steps internally to our market to make
it immune to this thread. Although
this is phrased in terms of the final result with end
users, our OEM customers and ISVs will
fall long before end users do.</li>

<li><b>Compete with Sun for their <i>present</i> and
near
term customers.</b> This means going after a
particular set of customers in a particular fashion, and
is driven by looking at Sun's <i>current</i>
tactics.</li>
</ol>

<p>
These are not mutually exclusive, and it might even be
possible to achieve both of them with one
strategy, but in general they are utterly different and we
have to keep them straight.
</p>

<p>
Our efforts to create a RISC PC have been focussed
primarily on plan A. Compaq, on the other
hand, has gotten interested in B, in part because they
are worried about how to do plan A without
hurting their present market. Recent email here at
Microsoft about promoting 486 machines as
"business workstations" is aimed at plan B (but with an
eye toward assisting A).
</p>

<p>
We need to have a strategy in <i>both</i> cases. This is
true both in planning our own product line up, and
also in being able to articulate both strategies to our
OEM customers. As events progress they will
become increasingly confused over what to do, and this
only plays into Sun's hands.
</p>

<h2>2. MARKET COMPARISON</h2>

<p>
The most salient points of distinction between the two
markets are as follows:
</p>

<table border="1" cellpadding="1">
<tr><th>Workstations</th><th>PCs</th></tr>
<tr><td><b>175K machine/year run rate.</b> This is for

the
largest binary compatible standard.</td>
<td><b>12 million machine/year run rate</b> - of binary
compatible machines.</td></tr>
<tr><td><b>Direct sales</b> - Workstations today are
a "top
down" decision.</td>
<td><b>Retail/indirect distribution.</b> PCs have been a
"bottom up" phenomena.</td></tr>
<tr><td><b>High margins</b> - There is no clone
competition
yet.</td>
<td><b>Low margins</b> - enforced by competition from 3rd
tier clone vendors.</td></tr>
<tr><td><b>High growth.</b> Buyers are installing
workstations
for new tasks.</td>
<td><b>Lower growth.</b> PCs are more mature, more
upgrades than new adoption.</td></tr>
<tr><td><b>Bought as system</b> - Workstations sales
almost
always include servers and desktops.</td>
<td><b>Bought piecemeal.</b> The Lan is usually bought
from a separate vendor.</td></tr>
<tr><td><b>Small set of ISVs with niche products.</b>
This is
mainly focussed on engineering and technical
areas, with few horizontal apps.</td>
<td><b>Broad set of ISVs with horizontal applications.</b>
This is focussed on personal productivity and
individuals.</td></tr>
<tr><td><b>No good software distribution channel.</b>
Today
it is a mix of direct sales and VARs and what
amounts to mail order from the Catalyst catalog.
Plans for CD Rom distribution in the future.</td>
<td><b>Strong indirect distribution channel for binary
application software.</b> Application vendors in the
PC market do not typically have to have direct
sales forces, and can easily get mass
distribution.</td></tr>
<tr><td><b>UNIX</b> - This is perceived as a "real
os",
with
virtual memory, multitasking etc. In some cases
it also benefits from a "standards" and openness
message.</td>
<td><b>Windows &amp; Dos</b> - This mean no virtual
memory,
multitasking. Windows 3.0 actually corrects
most of these (NT fixes the rest) but there is
perception of weakness.</td></tr>
<tr><td><b>One primary vendor.</b> Sun is working hard
to
create the infrastructure to allow clones.</td>
<td><b>Large base of manufacturers</b> - This is enabled
by a well established infrastructure which
includes companies like Intel, Chips &amp;
Technologies and of course Microsoft.</td></tr>
<tr><td><b>RISC price/performance.</b> This is due to a

mix
of newer technology and the open processor
business model.</td>
<td><b>x86 price/performance.</b> The PC world is based
on an aging, closed processor which is technically
obsolete.</td></tr>
<tr><td><b>Powerful desktop machine.</b> Like a PC,
with
hi-
res graphics, a big monitor &amp; more RAM.</td>
<td><b>Nice desktop machine.</b> Like a workstation with
worse graphics &amp; less RAM.</td></tr>
</table>

<h3>2.1. Sun's Strategy</h3>

<p>
Our present understanding of Sun's overall strategy is as
follows:
</p>

<ul>
<li><b>Stay the course with direct sales and the
conventional UNIX workstation market.</b> They
will rely on their present approach as the mainstream core
of their business for the next two
years. The growth rate is large, and they can comfortably
use this to finance their assault on
other markets, and give them enough time to get the pieces
in place. The key niche markets
that they will exploit are technical workstations and
software development within
corporations. This will broaden to cover an increasingly
large set of customers, but it is a
mistake to think that this approach us the limit of Sun's
ambition. They will gladly take on
retail distributin when they have the pieces in place to
be successful at it. The market they
milk today is a convenient stepping stone, but not an end
unto itself.</li>

<li><b>Experiment with other channels.</b> This will
occur through limited test cases such as their
deal with Micro Age, and through SPARC clone companies
such as Northgate, Compuadd
and others who are expendable missionaries in new
markets.</li>

<li><b>Build an arsenal of ISV support.</b> They do not
have a sufficient set of desktop productivity
applications to really threaten PCs for control of
mainstream office computing, but they get
more support every dey. Their present growth rate even
<i>without</i> having such apps will take
them to 300K-500K units/year run rate within the next year
to 18 months, which looks
attractive to ISVs. In addition, there is the effect of
being the number two platform.
Everybody does some development on the number two
platform, for incremental revenue,
strategic hedge and so forth. Dos has been number one for
a long time. For a while, the
number two spot was taken by the Macintosh, then for a
while it was OS/2. We are in an
enormous danger right now of having SPARC be number two
and Windows be number one.
Since ISV support is cumulative and Sun is looking for
critical mass, having them in number
two is terrible for us - we must try to get one of our
platforms (such as Windows on RISC) to
be number two, and SPARC to be third or less.</li>

<li><b>Serve as a focal point for rancor with Intel and
Microsoft/</b> Many people in the industry
are jealous of our success - this includes ISVs, OEMs and
others. They also hate Intel for
much the same set of reasons. IBM's role in the market is
also a point against the PC world,
although less direct than Intel and us. Sun is the
beneficiary of all of this because they offer
a brand new world. The old saying about the "devil you
know being better than the devil
you don't" does not seem to impress many people.</li>

<li><b>Once they are ready, make a major push on the PC
market.</b> Within the next two years
they will be in an excellent position to directly assail
PCs with widespread retail distribution
of both machines and binary software packages. In early
1992 they will be selling at the
300K - 500K machines per year run rate, will have a
critial mass set of major applications,
and they will have a commanding price/performance lead on
the x86. Note that Sun is
playing a long term bet, and they can afford to wait until
the time is ripe.</li>
</ul>

<p>
This all assumes their present level of industry support.
The endorsement of major players in the PC
market might accelerate things.
</p>

<p>
Note that the strategy above is NOT aimed at bringing RISC
into the PC market - rather it is trying to
grow the workstation market to the same volume levels, and
the same distribution methods as PCs.
This is a critical distinction. Sun is creating a
parallel world to the PC industry, in much the same
way as the Macintosh is a parallel world to IBM compatible
PCs. It is not a high end PC, but rather a
new beast which has some key differentiating features.
</p>

<p>
The other key thing to note is that as far as Sun is
concerned, the die is cast - they are committed to a
strategy of rapid expansion and an assault on the PC
customer base. They have taken the step to build
a clone market, and this will <i>not</i> let them continue
as a technical niche oriented, workstation only
company. Leading a flotilla of clone vendors cuts into
your market share and your margins and the
only way that you can come out ahead is to have it
simulate extraordinary market expansion to
compensate. They have taken this bet, and there really is
no turning back - we can count on them to
see it through to the death.
</p>

<h3>2.2. Fundamental Issues</h3>

<p>
The comparison table covers many points that people raise
about the two markets. If you step back
and look at the really fundamental issues it boils down to
some very simple points.
</p>

<ul>
<li><b>We have the ISVs and a binary software distribution
channel.</b> This is tremendously
important if you want to address the mainstream office
productivity market. Sun is working
as hard as they can to fix this shortcoming. It is the
single largest thing holding them back
from the mass market - today you just can't use a Sun
workstation to do most of the things a
PC is used for. This is true regardless of price, who
sold you the machine, or other factors.
Although distribution, OEM support and other things are
crucial, ISVs are key. Sun is
already building all of the other infrastructure necessary
(clone kits, retail etc) - in part to be
ready and in part to influence ISVs.</li>

<li><b>We have the bulk of the end users.</b> This is a
vitally important thing to remember, but
people seem to forget this all the time - from a raw
numbers point of view, Sun wants our
market a hell of a lot more than we want theirs. If
tomorrow every existing Sun user threw
out his machine and bought a PC with Windows, we would
scarcely be able to tell from our
own sales figures - nor could our OEMs tell unless all of
the volume went to one of them.
Our kind of volume is the pot of gold that motivates Sun
to make some very risky moves,
such as bringing on clones (which could hurt margins). PC
OEMs that look at the
workstation market as an opportunity should keep in mind
that Sun itself is trying to shift
their market away from where it is today in order to
attain PC-like volume levels.</li>

<li><b>RISC gives them a superior technical base.</b>
Over time, ISVs can take the 2X or more
performance advantage that Sun will soon have and turn it
into compelling end user features
that we wil be unable to match with the x86 world. The
truly dangerous thing is that it
involves a change to the binary software compatibility
standard. It is tricky for us to take
advantage of RISC, and transfer our existing momentum
because it means we cannot rely on
compatibility - which has been our biggest strength. This
is true for our software, and also
true for our OEMs who have existing businesses to
protect.</li>
</ul>

<p>
In the "first order" approximation, <i>nothing</i> else
matters. In this sense we <i>are</i> facing a "RISC
threat" -
it is the only one of the points listed in the comparison
table above which in the long run is dramatic
enough that it could break our current momentum with ISVs.
This may sound extreme, but I think it
is the accurate long term view. There are many other
factors which will need to be considered once
you dive into the problem, but this is the key.
</p>

<p>
At the same time, it is important to recognize that it is
a long term issue. If we lose badly enough in
the short term, we may never get to roll out the long term
solution. We certainly must work to
address the short term issues, but without a long term
strategy we are sunk. Since the long term plan
takes a very long time to reach fruition we must start it
at once.
</p>

<p>
Given this view, the primary strategy for us is very
clear - find a way to transfer our existing strength
to a platform which includes RISC, and thus deny Sun the
one crucial point of differentiation which
cannot be addressed with incremental near term moves.
This is another way to phrase plan A
discussed in the introduction - prevent Sun from taking
over our market. The plan of record for
doing this is based on the following ideas:
</p>

<ul>
<li><b>Leverage Windows.</b> We will use the Windows API
implemented on top of NY to be the
bridge for ISVs, and make it easy to port a Windows
application to RISC.</li>

<li><b>Position RISC as a high end personal computer.</b>
This is one of the key goals of the Power
PC approach - unify the RISC platform with the x86 product
line and make them appear to
be part of a continuous spectrum.</li>

<li><b>Tantalize ISVs.</b> The other key thrust in Power
PC is to create a platform (for both x86 and
RISC) which gives them some very interesting new
capabilities. The goal is to make Power
PCs (of both instruction sets) the favored platform for
innovation.</li>

<li><b>Use our existing OEMs to create a standard.</b>
This has had very mixed success because
some of our OEMs, like Compaq are not comfortable in the
inherently risky role of an
innovator. Nevertheless, we must try to use our influence
in the industry as a weapon to
offset the fact that Sun is <i>way</i> ahead of us in many
other areas (reference platform, clone kits
...).</li>
</ul>

<p>
Note that this says nothing about competing with Sun in
their <i>present</i> market, or adopting their
present tactics because that is not our primary goal. It
is a fine secondary goal, but we must first
strengthen the PC industry against an impending assault
from Sun. This means changing the
instruction set and establishing a new binary standard,
which is a big enough undertaking that we
must move on it at once.
</p>

<p>
There are several immediate corollaries to the point that
our main strategy is not an attempt to go after
Sun's present market or tactics:
</p>

<ul>
<li><b>Market strength by itself is not enough.</b> We
cannot keep Sun away by just making
Windows on x86 even more popular. That would still leave
Sun with enough points of
differentiation (particularly RISC) that they can continue
to gain momentum. The analogy
we've used in the past is that no amount of strength in
the character mode PC market could
have stopped the Mac. The only thing which might have
worked would have been the ability
to deny the Mac it's key point of differentiation from
other PCs - which was GUI.</li>

<li><b>We do not have to stop Sun ISV activity in order to
win.</b> Some people (including those at
Compaq) get confused that we have to shut off ISVs from
going to Sun. We don't have to
do that, because we are ahead and will win a tie. In the
short term we must make sure that
ISVs funnel some of the Window 3 momentum toward RISC by
making it a small
incremental investment. In the long term we must make
sure that the next set of "killer
apps" that really exploits RISC performance is not a Sun
only phenomena. Every new
successful machine has been established on the strength of
one (or a mere handful) or killer
apps, and we must make sure that SPARC does not
<i>uniquely</i> get an edge over us in this way.</li>

<li><b>We do not have to be point for point
competitive.</b> One comment that has come up a lot is
whether we can "compete against UNIX with Windows" or
similar direct comparisons. In
this phase of our strategy we are not trying to compete
directly at all - instead we are trying
to end run the competition by making our own market
stronger. It is asking the question the
wrong way around - it is they that are coming after us.
We want to put <i>them</i> in the position
of saying "how can we compete with Windows using UNIX,
especially now that Windows
includes RISC!". The answer we want them to come up with
is that they probably can beat
us in the small, niche oriented, UNIX-loving market they
own today and not in our mass
market.</li>

<li><b>Hurting Sun sales is not enough.</b> One dangerous
illusion is that through efforts like plan B,
we can slow down or stop Sun. This ignores the fact that
Sun is making a long term bet on
the validity of RISC technology, and short term issues are
not going to effect them, unless
they are severe enough to actually bankrupt them. They
will stick to their current market,
keep making some money, and gather ISVs under the banner
of platform neutrality and RISC
performance. As the performance gap with x86 widens, and
some killer apps are written
which exploit it, they will have their chance. A good
analogy is Microsoft application
strategy to bet on GUI - it was a long term move which bet
on using a fundamental paradigm
shift in the technology to advance against entrenched
competition. Hurting Sun's present
market is a lot like Lotus beating up on Multiplan - it
was irrelevent to Jazz (their early shot
at GUI) and later on ws irrelevent in fighting Excel.</li>
</ul>

<p>
Obviously we would like to make our existing market super
strong, stop all Sun ISV activity and beat
them across the board on every feature. It is important
to remember that our strategy is not so fragile
that we need each of these in order to win.
</p>

<p>
The time scale on which we do this is also important.
Although the fundamental issue is a long term
one, that does not belie any lack of urgency. We need to
establish a new binary standard, which is a
hell of a hard thing to do. Precisely because of its long
term importance, it also takes a long time to
take effect. We are also way behind Sun, who put their
platform out four years ago.
</p>

<h3>2.3. Second Order Effects</h3>

<p>
Assuming that we have a strategy to deal with the
fundamental issues, there remains the question of
what to do about the shorter term tactics, and how we or
our OEMs can "take the battle to the enemy"
by adopting one level or another of plan B - competing
with Sun. There are several reasons for doing
this:
</p>

<ul>
<li><b>Give Sun some grief in their home territory.</b>
This is mainly interesting as an indirect way
to slow them down and delay or prevent their entry into
our market.</li>

<li><b>Make UNIX continue to look confused.</b> One of
their points which is anathema to us is
UNIX. This has been a mixed blessing for them because
UNIX is plagued with
fragmentation and confusion. Anything that we can do to
continue that state of affairs helps
us.</li>

<li><b>Learn any interesting tricks for our own
market.</b> The fact that Sun can sell 175K
units/year without a lot of the key benefits PC have is an
interesting fact that might help our
OEMs with incremental sales.</li>

<li><b>Prevent Sun from gaining general mindshare.</b> A
much more important issue than hurting
their sales in their home market is stopping the stream of
good news which fosters a general
impression of wild, monotonic success. As an example, the
PC industry must be seen as
having a long term future, and not simply be a interim
stage until we can all afford powerful
SPARCstations.</li>
</ul>

<p>
It cannot be stressed enough that these goals, however
nice, are not sufficient by themselves. Unless
we could <i>kill</i> Sun or <i>kill</i> UNIX outright
there is <i>no</i> way for Microsoft or the PC industry to
win by
taking these approaches alone. It is highly unlikely that
we or any of our OEMs could kill Sun at this
point by trying to compete on their home turf. Even if
one or more of our customers could do so, if
they didn't do it with a platform that we are involved in
(such as some kind of UNIX), we would just
as bad off. We'd replace "Sun" with another name. If a
PC company out-SPARCed Sun, or MIPS
suddenly took all of Sun's market, or IBM RIOS came out of
nowhere and wiped them out, then
pretty soon we'd be writing memos like this one talking
about how we faced a threat from the new
victor. The only long term hope is to get a platform which
we control to win, and that means NT
Windows.
</p>

<p>
Looking at the market comparison, there are a number of
points in Sun's favor which we, or the PC
industry, could exploit:
</p>

<ul>
<li><b>Direct vs indirect sales.</b> Sun is able to reach
some corporate customers which cannot buy a
similar package of goods and services from PC vendors
because they have a direct sales
force. One can make an argument that as corporate buying
decisions become more "top
down" and centralized, and systems increasing involve
networks and total systems sales, that
one needs a direct sales force to compete.</li>

<li><b>Selling the whole system, including the
network.</b> This is another area where the bulk of
the PC industry has not done a good job to date, but Sun
has. Part of this is a packaging
issue (UNIX includes most networking features), part is
sales and distribution and part of it
is product line (good support for diskless desktop
machines, admin features, security
features).</li>

<li><b>UNIX is a requirement for a segment of the
market.</b> This is because there are a number of
specialized applications which are only found there, or it
is the preferred development
environment for many people because of training and
background.</li>

<li><b>Sun machines are a good value versus PCs.</b> This
is particularly true in having a lot of
RAM, performance and graphics resolution. Most PC
configurations do not compete in this
domain, and if you do set up a PC which does, it is much
more expensive. We could
certainly put our best foot forward and create both price
competition and correct the
perception that PCs are weak compared to
SPARCstations.</li>
</ul>

<p>
The first two points really mean that IBM has not been
holding up their end. This is exactly the area
where they have traditionally been strong, and the market
to which OS/2 Extended Edition was
aimed. There are a number of obvious flaws with their
execution (such as OS/2 itself, too much
reliance on host based systems etc), but there are number
of similarities in the goals.
</p>

<p>
It is certainly possible to take an set of x86 based PCs
and servers (including the network) and
compete head to head with Sun. This means using a direct
sales force, and having the right product
pieces - but you could do this using existing products off
the shelf. It is just a marketing and
packaging issue. Ideally IBM and others will start to do
this.
</p>

<p>
It is the height of folly to believe that this is Sun's
secret, or the main thing we have to fear. The
direct sales force was necessary when they sold $40K
systems which had little third party software.
Now that they sell $5K systems it still works as long as
you sell a bunch of them at a time. The
traditional minicomputer and mainframe companies all do
this - direct sales is still a powerful tool and
Sun has used it like others have in the past - no better
or worse. If this was the sole issue, IBM would
have flattened the whole issue years ago, and they are
still the ones to emulate if you want to go about
this - not Sun. Sun will continue to do the familiar
thing and rely on direct sales until they have their
act together - which means ease of use, ease of
installation and most of all enough third party ISV
support that their systems appeal to the broad set of
people who buy their "solutions" in shrinkwrap at
Egghead rather than needing to have the Sun salesman set
up a turn key system. At that point Sun
will do retail, and what ever other channels make sense.
</p>

<p>
UNIX is another issue which can be a red herring. It is
the universal choice for machines that have
no applications - people in school learn how to write for
UNIX, academics (an important part of the
workstation market, especially in the early days), and it
was the obvious approach for hardware start
ups like Sun which founded the workstation market -
porting UNIX was easier than writing a
proprietary system from scratch. It is very easy to
forget that the company that lead the workstation
industry for many years, and who only recently slipped
into second place (largely because they did
not have RISC and lost to the SPARC's price/performance)
did <i>not</i> use UNIX, and <i>did</i> in fact write a
proprietary operating system (Apollo Domain). Apollo has
a UNIX mapping layer that works pretty
well, but they still use Domain as their base.
</p>

<p>
This is not an attempt to ridicule UNIX, but it is
important to kepp things in perspective. There are
customers that really do need UNIX, either because they
need features that it has, or for other
reasons. Nevertheless, it is not the source of Sun's
strength, and OEMs who think that they need it in
order to compete with Sun are almost certainly fooling
themselves. The market that really needs
UNIX already has it - Sun's market expansion is largely to
companies and customers which are
outside of the traditional scientific and engineering
user, or government user who really does want
UNIX. Unless you are after the SUN <i>installed base</i>,
or the installed base of other random
workstations (which in toto is far smaller than Sun
today), then UNIX in and of itself is not really the
hard requirement that it appears to be. It is very ironic
that at the very time that Sun is trying to get
ISVs from the PC and Mac worlds so that they can <i>break
out</i> of the traditional UNIX mold, other
people look from the outside and think that UNIX is a key
ingredient.
</p>

<p>
In general, trying to enter the market at this stage to
try and beat Sun by competing head to head with
them with UNIX is a sucker's game. Sun is too far ahead,
and too much in control of the key UNIX
standards to make it viable. Mini companies with an
existing sales force might sell some to loyal
accounts as a defensive action to keep Sun out, but they
are not going to take a significant chunk out
of Sun's hide. In flailing about, endorsing UNIX,
endorsing RISC etc, they will only play into Sun's
hands.
</p>

<p>
Finally, there is the issues of the hardware
capabilities. Sun has achieved the image of power and
technology. Part of this is based on the fact that they
used to be much more expensive, and part is a
real edge - the cheapest Sun machine is arguably (but not
dramatically) faster than the fastest machine
in many of our OEMs product line ups. In the short term
this is almost entirely a marketing issue
rather than a technical one - we need to get more
visibility for high end PCs, get the price of the 486
down, and create the impression that high end Windows
machines are cool. One thing that is
interesting is that to date Sun has <i>not</i> tried to
compete in the realm of the imagination by having a
machine which tantalizes ISVs in the way that Power PC
will, or the NeXT machine did when it was
introduced (it was pale in comparison). SPARCstations are
fast and have more RAM and resolution,
but they are not very interesting in other ways. This
could well change in the future, but for now Sun
means power and technology - not innovation.
</p>

<h2>3. WHAT TO DO</h2>

<p>
Our direct action plans are contained elsewhere, and I do
not want to just repeat everything here.
Here are some brief summaries which condense the key
steps.
</p>

<h3>3.1. "Business Workstation" Market</h3>

<p>
This is name that Intel has come up with to describe a
tactical short term response to Sun using x86
machines. Lots of email has been sent about this
recently, but here are the key steps:
</p>

<ul>
<li><b>Define a "flagship" 486 machine configuration for
1991.</b> This must be something which
OEMs can rapidly assemble and market to end users as a Sun
competitor, and has a hot
Windows machine.</li>

<li><b>Create a catch all name and slogan for these
machines.</b> The recent suggestion of the "4
Plus" machine is a good one (486 + 4 meg RAM + 40 meg HD
+ "4th" generation
display...). The name <i>must</i> be applicable to
machines from Compaq and IBM set for fall
release which will not carry the logo or name, so a self
describing name like 4 Plus is good.</li>

<li><b>Evangelize these to OEMs at our October Technical
Briefing.</b> We want to tell people that
the 386SX is doing well as the minimum Windows 3 machine,
but there is a real opportunity
to push the <i>cool</i> Windows machine.</li>

<li><b>Ask Intel to be aggressive about pricing the 486 to
OEMs that sign up.</b> This helps get
price competition with Sun, and makes the 4 Plus machines
appear mainstream rather than
the extravagant image they have today because of
ridiculous prices.</li>

<li><b>Cooperate with Intel and OEMs in a marketing
campaign to corporate end users.</b> This
pushes the notion of a scalable family of Windows 3
machines, perhaps called "the business
workstations" lead by the 4 plus machine.</li>

<li><b>Consider putting together products to
support "system sales".</b> This means LAN oriented
products. An example is the Workgroup Windows product
suggested a while back which is
optimized for diskless Windows machines. This is much
more of a packaging and marketing
issue than technology, although some work might have to be
done.</li>

<li><b>Encourage some OEMs to combine these machines with
direct sales.</b> A feature wise
competitive platform based on x86 machines with Windows
and LanMan (or Novell for that
matter) could easily be put together and sold to many of
the business customers that Sun is
attracting. There are plenty of OEMs who could see this
as an opportunity, and we should
support them if that is their goal.</li>
</ul>

<h3>3.2. Technical Workstation Market</h3>

<p>
This is largely a market which Sun is expanding away from,
but in the meantime there are companies
like DEC, HP, IBM and others who are committed to slugging
it out in the shadow of the Sun
juggernaut. Some OEMs see it as an incremental revenue
opportunity. In general this is an illusion,
Sun's strategy does not leave a lot of room for other
large players, but there are some for which it
could work out. The mail order companies who are planning
SPARC clones (Northgate and
Compuadd) for example, have little investment at stake,
and might have value to add in pioneering a
new channel for this type of machine. They can piggyback
on Sun's efforts and sell incremental
machines to universities or other places which have a Sun
network installed but need some more cheap
machines.
</p>

<p>
The key thing to do in this market is to prevent Sun from
achieving a total clean sweep. Unless they
organize very quickly, the anti-Sun forces will find
themselves swamped in the next year. Headlines
in the industry press saying that Sun has whupped DEC, HP,
IBM and others do not help our cause.
From the other perspective, most of these companies are
potential Power PC OEMs and we should
channel their Sun hatred toward this end.
</p>

<p>
Many of these companies are seeing the proverbial hand
writing on the wall, and see some or all of
the issues in the Trends on the Microprocessor Industry
document occurring. This is causing them to
organize, and as long as they are at it, they may as well
organize around a platform which does us
some good. The primary effort which effects us is the
DEC/MIPS/SCO project to establish a
standard UNIX version and clonable hardware platform based
on MIPS.
</p>

<p>
The steps that we should take are:
</p>

<ul>
<li><b>Negotiate a role in the MIPS/SCO version of
UNIX.</b> They are open to this, and we have
several potential approaches. It is important that we get
a stake and some control in this.</li>

<li><b>Work with DEC to merge their hardware reference
platform with ours.</b> Indications are
that they will be open to this. This would prevent a
needless fracturing of the MIPS market
and give chip companies a near term target for
making "clone kits" from our ASICs (and
those of DEC). It also helps de-emphasize our role in
creating a hardware design.</li>

<li><b>Encourage the non-Sun workstation world to sign
up.</b> Most of this effort will be done by
MIPS (with some help from DEC) and by the chip companies
who want to sell clone chips.
Our private support will be important to tip the balance
at many key OEMs, such as HP.
This would be announced in the next couple of months,
letting the world know that there was
a strong standard other than Sun.</li>

<li><b>Fold the MIPS UNIX based companies into Power PC
when it is available.</b> The basic
hardware platform would the same as Power PC (some
advanced features might need to be
added). Many Power PC supporters will not come from the
ranks of the MIPS workstation
world, but there is no reason not to work the other way
around and let the workstation guys
switch to Power PC when the time comes.</li>
</ul>

<h3>3.3. Power PC</h3>

<p>
This is the core strategy to lock Sun out of the
mainstream PC market. It is covered elsewhere, but
here is a brief description:
</p>

<ul>
<li><b>Define a set of innovative system features.</b>
These include advanced graphics, audio,
minimum RAM configurations etc which are designed for the
next generation of
applications.</li>

<li><b>This set is implemented with both R4000 and 486
processors.</b> We supply system software
for each, using the 32 bit Windows API. This includes new
APIs and libraries which use the
advanced hardware features. Other system features like
the user interface bitmaps and "look"
are tuned to the graphics (photorealistic shading
etc).</li>

<li><b>Power PC is the high end, ultimate Windows
machine.</b> This is its initial positioning - the
top 5% - 10% of the PC industry. Power PC with RISC is
the hot platform for those people
who have made the move to Windows apps exclusively and
don't need old binaries. It is
important to have Power PC with RISC appear as the newest
member of a broad Windows
family, which spans the range of computing from handhelds
to fast RISC machines.</li>

<li><b>Leverage Windows ISV support.</b> There is full
source compatibility between 32 bit
Windows on x86 and RISC, so that addressing RISC is a
small incremental investment. We
also provide a set of tools (pseudo 32 bit support) to
make it easy to port from 16 bit
Windows, and maintain common source code.</li>
</ul>

[ Reply to This | Parent | # ]

1248 ("Responses to the Compaq SPARC Threat")
Authored by: Anonymous on Saturday, November 17 2012 @ 03:35 PM EST
http://groklawstatic.ibiblio.org/pdf/iowa/www.iowaconsumercase.org/011607/1000/PX01248.pdf


<p>
PLAINTIFF'S EXHIBIT 1248<br />
Comes v. Microsoft
</p>

<p>MICROSOFT RISCPC TASK FORCE REPORT</p>

<h1>Responses to the Compaq SPARC Threat</h1>

<p>HIGHLY CONFIDENTIAL</p>

<p>NOTE: This memo is very confidential and discretion
should be used in copying
or distributing it - even within the company.</p>

<p>ADVANCED TECHNOLOGY AND BUSINESS DEVELOPMENT</p>

<p>MICROSOFT CONFIDENTIAL</p>
<p>MS 5012188<br />
CONFIDENTIAL</p>

<em>[Ed: Duplicate page omitted.]</em>

<p>
<b>TABLE OF CONTENTS</b>
</p>

<p>
TABLE OF CONTENTS<br />
INTRODUCTION<br />
<span style="margin-left: 4em">Sun's Strategy</span><br
/>
<span style="margin-left: 4em">Probable Compaq
Plan</span><br />
<span style="margin-left: 4em">Compaq's
Motivation</span><br />
<span style="margin-left: 4em">Impact on
Microsoft</span><br />
<span style="margin-left: 8em">PC Industry
OEMs</span><br />
<span style="margin-left: 8em">PC Industry
ISVs</span><br />
<span style="margin-left: 8em">Networking
Business</span><br />
LIVING IN A SPARC WORLD<br />
<span style="margin-left: 4em">Windows layer for
SunOS</span><br />
<span style="margin-left: 4em">NT Windows for
SPARC</span><br />
<span style="margin-left: 4em">NT as a base for
SunOS</span><br />
<span style="margin-left: 4em">Applications on
SPARC</span><br />
<span style="margin-left: 4em">Conclusions</span><br
/>
FOSTERING SPARC ALTERNATIVES<br />
<span style="margin-left: 4em">Uniting thc MIPS
Community</span><br />
<span style="margin-left: 4em">Managing PC
ISVs</span><br />
<span style="margin-left: 4em">Wild Ideas</span><br
/>
<span style="margin-left: 4em">Conclusions</span><br
/>
RECOMMENDATIONS
</p>

<p>
2/26/92 12:43 PM MICROSOFT CONFIDENTIAL PAGE 2
</p>

<em>[Ed: Page footer retained for date information. All
other headers and
footers omitted.]</em>

<h2>1. INTRODUCTION</h2>

<p>
We have recently learned that Compaq is seriously
considering a project to enter the workstation
business with a SPARC and UNIX based machine. This memo
covers some initial thinking on what
this means to Microsoft, and how we should respond.
</p>

<p>
Obviously the best thing would be to have them change
their minds. The question of how to cause
this to happen will be covered elsewhere. This memo
assumes that we will immediately begin a
dialog with Compaq to try convince them to do something
else (nearly anything else would be an
improvement), but in parallel we must start to plan, and
act, on a response which assumes the
worst. It is fairly safe to assume that Compaq will wind
up in one of a couple of modes:
</p>

<ul>
<li><b>Rapidly reach a formal decision in favor of
SPARC.</b> Once their consensus driven process
locks in on a decision it is hard to change, and we can
assume that it will be months before
we could turn them around. Even then, we will need
dramatic new data to change their
minds, and there is only a small chance that we could win
even then. A further
complication is that to do SPARC they are likely to enter
into some commitments early on
which are tough to reverse - it is not just engineering
work.</li>

<li><b>Return to confusion on the RISC issue.</b> This is
the state that they were in for the last
several months (although some of that was cover for their
SPARC investigation), and it is
possible that we could get them back into that mode of
operation. They could wind up
right back in the SPARC camp, or do something else
random.</li>

<li><b>Run down multiple paths.</b> They could start
working on the SPARC plan without a formal
decision, just as they worked on the MCA bus machines
while thinking about EISA. This
would still be a very dangerous situation, but it would
increase the likelihood that we could
get them to tum around later. The problem with this is
that the likely scenario for their
SPARC project involves building or buying a sales force
and other activities which can't
easily be cancelled late in the game.</li>
</ul>

<p>
Another way to put this is that we will either lose very
fast, or we will be in a long holding pattern
from which we could still lose. In either event we must
proceed at once with our response - there is
no point at all in waiting for Compaq to turn around, or
in hoping that they will quickly come to
their senses and join our present RISC plan.
</p>

<p>
Finally, I should note that the ideas below are the result
of conversations with many people.
</p>

<h3>1.1. Sun's Strategy</h3>

<p>
Our present understanding of Sun's overall strategy is as
follows:
</p>

<ul>
<li><b>Stay the course with direct sales and the
conventional UNIX workstation market.</b> They
will rely on their present approach as the mainstream core
of their business for the next
two years. The growth rate is large, and they can
comfortably use this to finance their
assault on other markets, and give them enough time to get
the pieces in place. The key
niche markets that they will exploit are technical
workstations and software development
within corporations. This will broaden to cover an
increasingly large set of customers.</li>

<li><b>Experiment with other channels.</b> This will occur
through limited test cases such as their
deal with Micro Age, and through SPARC clone companies
such as Northgate, Compuadd
and others who are expendable missionaries in new
markets.</li>

<li><b>Build an arsenal of ISV support.</b> They do not
have a sufficient set of desktop productivity
applications to really threaten PCs for control of
mainstream ofiice computing, but they get
more support every day. Their present growth rate even
without having such apps will take
them to 300K - 500K units/year run rate within the next
year to 18 months.</li>

<li><b>Once they are ready, make a major push on the PC
market.</b> Within the next two years
they will be in an excellent position to directly assail
PCs with widespread retail distribution
of both machines and binary software packages. In early
1992 they will be selling at the
300K - 500K machines per year run rate, will have a
critical mass set of major applications,
and they will have a commanding price/performance lead on
the x86.</li>
</ul>

<p>
This all assumes their present level of industry support.
Even though Compaq is not likely to ship
any appreciable volume in SPARC until late1992, their
endorsement will clearly help.
</p>

<p>
Note that the strategy above is NOT aimed at bringing RISC
into the PC market - rather it is trying
to grow the workstation market to the same volume levels,
and the same distribution methods as
PCs. This is a critical distinction. Sun is creating a
parallel world to the PC industry, in much the
same way as the Macintosh is a parallel world to IBM
compatible PCs. It is not a high end PC, but
rather a new beast which has some key differentiating
features:
</p>

<ul>
<li><b>RISC.</b> This means three direct benefits over
the x86 - 32 bits, better price/performance,
and higher absolute levels of performance. Indirectly,
the open proccmor model will
ensure that SPARC's lead over the x86 and closed
processors will increase over time.</li>

<li><b>UNIX.</b> This is a mixed blessing, but their
positioning is to try milk as much as possible out
of "open systems", and the supposed technical quality of
UNIX.</li>

<li><b>Networking and connectivity.</b> Nobody would
accuse UNIX of having an elegant or even
very good networking story, but since it has been put into
place over a number of years it
does work, is mature, and has been ported to all manner of
machines and networks.</li>

<li><b>The perception of being high tech.</b> Sun and
other workstations have the cachet of being
sophisticated, powerful state of the art machines, and
this aura helps set them apart from
the PC industry.</li>
</ul>

<p>
Despite the enormous momentum behind Windows 3, as long as
Sun can position themselves in a
significantly different market, they are largely immune
from assault and can continue on the strategy
above. An analogy that we've used in the past is the case
of the Macintosh, which was introduced at
a point of unprecedented strength for the character mode
IBM PC. This did not eliminate the
Mac, and in fact it is hard to imagine that any amount of
increased volume in character mode PCs
would have stopped it. The Mac used GUI to put itself in
a class by itself, and this have the
breathing space to grow, and as long as the SPARC world
uniquely enjoys the features above it will
have the same kind of chance.
</p>

<p>
Our strategy has been to deny key points of
differentiation to SPARC by broadening the Windows
work to include to include them. The most dramatic and
most important is RISC, and that has
been the item which has been discussed the most. The high
tech image is addressed by the Power
PC features. Networking is also being addressed.
Although not unique to RISC, it is a key part of
our long term systems strategy.
</p>

<h3>1.2. Probable Compaq Plan</h3>

<p>
Our present understanding of the Compaq strategy is as
follows:
</p>

<ul>
<li><b>Enter the UNIX workstation market with a SPARC
based machine.</b> This will probably be
of their own design rather than a SPARCstation clone.
This would be positioned very
clearly at workstations, and at Sun's present and near
future market. They have an explicit
goal to keep it very separate from the x86 PC market to
avoid any negative impact on their
current systems, and they think that this separation is
pretty easy to achieve.</li>

<li><b>Use a Compaq direct sales force to distribute the
machine.</b> This choice is determined by
three factors - their desire to compete head to head with
Sun in the traditional workstation
market, the need to keep this activity separate from their
present distribution channel and
finally they have a long term goal to build a real direct
sales force, and this project provides
a convenient opportunity. There are hints that they may
acquire a mini or workstation
company such as DG or Wang to get an established sales
force in a hurry, but they may just
build one from scratch.</li>

<li><b>Create a limited consortium.</b> Compaq will
attempt to balance the necessity of having
industry support to attain critical mass with their large
desire to have a proprietary
advantage. They do not be vulnerable to cheap
SPARCstation clone kits (although exactly
how is not clear to us yet). The data to date suggest a
group of 3-4 companies with sales &gt;
$2 billion, and with little enough clout that Compaq
gets "51 % of the votes" (in Gary
Stimac's words).</li>

<li><b>Rely on the Compaq name and prestige.</b> They do
not seem to mind us doing something
else for RISC which does not involve them, because they do
not feel that it will amount to
much. They think that the Compaq name and support is
critical to the success of any such
machine, so other efforts will be just so much noise.</li>

<li><b>Out execute Sun and the SPARC clones.</b> In
attempting this strategy they are clearly
counting on superior execution to win them a good slot in
the SPARC pantheon - i.e. that
there is room for two major players (themselves and
Sun).</li>
</ul>

<p>
The time frame for this is uncertain, but Stimac said that
they could have sample machines within 9
months, and ship within 12 months. This would be
consistent with building a machine from scratch
(you can manufacture an existing design such as the LSI
Logic SPARCkit in much less time - say 3
months).
</p>

<p>
The Compaq strategy outlined above is much more of a pure
workstation approach than the one
that Sun is on. This is because Compaq has a huge PC
business to protect. Ironically, by endorsing
SPARC they are giving Sun and others a powerful means to
attack the PC business even if Compaq
itself achieves a way to isolate their own workstation
business from PCs. This is a very key
distinction between what they are planning to do and the
IBM RS/6000 strategy which they admire
so much. In IBM's case the RS/6000 has no life of its
own - IBM controls it completely and also
benefits completely if it is successful.
</p>

<p>
In Compaq's case, SPARC <i>does</i> have a life of its
own, and Compaq is far from being the sole
beneficiary if SPARC wins. Sun and others can directly
attack the PC industry, and in doing so
compel Compaq to either follow suit and cannibalize their
PC business, or resist the trend and lose
in the SPARC market. For example, if Sun continues to up
the ante in aggressive pricing,
marketing aimed at PC end users, ISV evangelism and other
anti PC activities, Compaq will have to
match Sun to remain competitive in the SPARC market.
</p>

<h3>1.3. Compaq's Motivation</h3>

<p>
One view of Compaq's interest in SPARC is that it is
simply an extension of their desire to ape
IBM, and more generally to be a quality implementor of
other people's strategies (i.e. superb
knock-off artists). They need to copy any strategy that
IBM has, ergo they need to have an answer
to the RS/6000. Given that there are some nearly
insurmountable barriers to directly cloning the
RIOS chip set, they looked around to see who is the next
biggest player (Sun) and set off to copy
them and win through superior execution. Compaq has
recently taken the same approach in the
printer business.
</p>

<p>
Stepping back, there are several likely factors which are
motivating Compaq:
</p>

<ul>
<li><b>Cloning the strategy of moving into
workstations.</b> This was discussed above.</li>

<li><b>Obtaining incremental revenue and market share.</b>
The workstation market has a faster
growth rate than the PC business, and they perceive that
they have sufficient skills to beat
the present competition.</li>

<li><b>Get involved with RISC without jeopardizing their
PC businss.</b> They understand that
RISC will be important long term, but they remain
terrified of anything which introduces
RISC in a fashion which might in any way reflect on the PC
market.</li>

<li><b>Get involved with UNIX.</b> They are getting a
much more favorable view of UNIX than in
the past and this gives them a way to hedge their bets. I
do not believe that they have an
explicit goal to get out of the Microsoft dominated world,
but having a strategic hedge is
sensible.</li>

<li><b>Build a direct sales force without disrupting their
present channel.</b> The workstation
business gives them an opportunity to do this without
upsetting their dealers.</li>
</ul>

<p>
One interesting question is the degree to which we have
accelerated Compaq's concerns in these
areas, or done things to cause them to move to SPARC. It
is likely that our efforts to involve them
in the RISC PC, and our emphasis of the Sun threat has
highlighted the importance of this area, but
in general I think that at most we got them thinking about
this six months sooner than they might
have otherwise.
</p>

<p>
During the meeting with Stimac there were a number of
comments that indicated that they were
none too pleased with Microsoft's power in the industry.
One concern that we had during the
meeting, was that Compaq was upset with Microsoft taking
an active role in trying to define a
hardware platform and pushing them and the rest of the
industry to RISC. After thinking about it,
and considering their SPARC strategy in more detail, it
appears to me that anti-MS feelings,
including fear about our hardware prototyping efforts and
the role we have intended to play in
RISC PC have essentially nothing to do directly with their
basic strategy.
</p>

<p>
They appear to be doing exactly what one would expect from
Compaq if we had never told them a
thing about RISC, and instead they had discovered it on
their own. Their natural tendency is to
clone the winner rather than to innovate with a bold new
approach. They would never undertake an
initiative like setting a new RISC PC standard by
themselves - there are too many unknowns for
them. If you set aside the notion of doing their own
standard, what else could they do except clone
an existing one? Sun is their choice over our RISC PC
because they are more established whereas
our plans are still on paper, and there is less perceived
risk to their existing business.
</p>

<p>
Certain individuals at Compaq probably do hate us, or are
uncomfortable with our power, and we
should never get lulled into thinking that they are not
just as envious and jealous as the rest of the
industry. Probably the biggest sin that we have committed
in their eyes is that we ship products
before they are ready - this offends their perfectionist
sensibilities, and in the case of RISC they are
worried that our haphazard approach could hurt them very
badly by impacting their present market.
These personal factors help grease the wheels for any plan
that shifts power away from us, or
reduces Compaq's reliance on Microsoft, but there is no
evidence to support the contention that
they are on an anti-Microsoft jihad as a matter of company
strategy. Nevertheless we should not be
fooled into thinking that they like us, or will cut us a
break in any way - if the long term effect of the
strategy is to put us in our place, so much the better
although it is not an explicit goal.
</p>

<h3>1.4. Impact on Microsoft</h3>

<p>
Sun's clear goal is to be the system software provider and
the leading hardware company in the
SPARC market, and have that market become the key high
volume segment of the computer
industry. From a technology standpoint they have a
fundamental advantage over the x86 based
world, and there is every reason to believe that in the
long term RISC based machines will
completely replace the x86 for all desktop personal
computers. If Sun succeeds, our systems
business will die along with the x86.
</p>

<p>
That is the big picture view, and it shows why we need to
be excited - billions of dollars, and
Microsoft's identity as a company that is important in
systems software - hang in the balance.
</p>

<p>
Narrowing down our focus a bit, it is interesting to
project the near term impact of a Compaq
SPARC project on our part of the world.
</p>

<h4>1.4.1. PC Industry OEMs</h4>

<p>
One can argue whether Compaq is just following a clone
strategy, or has some deeper plan, but
when it comes to the rest of our OEM customers, there is
little doubt, especially among those who
are focussed primarily on the PC market. They are not so
ambitious as to think that they need to
emulate IBM in every way, but by the time Compaq is doing
it too, it will get their attention. The
rush of interest in MP and server machines following the
SystemPro introduction is ample testimony
to this fact - despite the fact that nobody (even Compaq)
is burning up the sales charts in this
market. The minute that Compaq announces a RISC strategy,
every other OEM is going to give it
some serious thought. Some will wait, some will rush in,
but they will all think about it.
</p>

<p>
The reaction to a Compaq SPARC annotmcement will group
OEMs in three basic categories:
</p>

<ul>
<li><b>Pure PC companies.</b> This includes big ones like
Zenith and Tandy, as well as the second
and third tier people. SPARC will be the leading contender
for most of the them, because
it is the simplest thing to do - just copy Compaq. The
only thing to consider is whether
their distribution channel can handle it, but many of them
will invert this problem into a
perceived opportunity - i.e. that their <i>advantage</i>
is that they're in a different channel than
Compaq with the same kind of machine.</li>

<li><b>PC companies which are already in the workstation
and mini business.</b> These guys will
be caught in a tough position, because Sun is their
nemesis, and they will be loathe to
support them. The canonical example here is HP - they are
the second largest workstation
company, and also have a PC business.</li>

<li><b>Workstation and minicomputer companies.</b> People
without a substantial PC business
include companies such as DEC, Silicon Graphics etc.
These people would not normally
care at all about what Compaq does (or at any rate it is a
second order phenomena for
them), and are unrelenting in their opposition to Sun.
They will see an incredible boost of
momentum for Sun and will be very interested in doing
nearly anything else as a matter of
survival.</li>
</ul>

<p>
The last two categories of companies are the simplest ones
to predict - they are directly threatened
by SPARC (even more than we are) and they will react
sharply in the opposite direction.
Ultimately they may give up and make SPARC clones, but
they will almost certainly make one last
attempt to beat them first. The motivation is pride,
inertia and the fact that Compaq (at number
two) will have taken last really desirable spot in the
SPARC line up (not to mention Compaq's little
consortium). This gives them nothing to lose and
everything to gain by going in any direction other
than SPARC. The primary distinction between the two is
that workstation companies in the PC
industry will naturally look to Microsoft for a solution
to this problem, wheres those who are not
big customers of ours wouldn't normally think of that. In
either case, they are very approachable
for a counter SPARC strategy.
</p>

<p>
The pure PC people are another matter entirely. They will
have a much more direct reason to just
sign up with SPARC. They are especially vulnerable to the
same pitch that Compaq itself is falling
for - enter the workstation market as a way get
incremental revenue without effecting your PC
business. The keys to convincing these people not to go
with SPARC are to play on three things:
</p>

<ul>
<li><b>It will hurt your PC businss.</b> SPARC is the
natural enemy of PCs, and it is too late to
side with SPARC in this fight because a1l_of the good
seats are already taken.</li>

<li><b>You can adopt a RISC strategy which actually
benefits your PC business.</b> This is the
Power PC message. Since most pure PC companies do not
understand the UNIX market,
they would be much better off with a way to address RISC
which leverages the thing they
do understand, and their present users and distribution
channel.</li>

<li><b>It's time to turn the tables on Compaq.</b> This
only works for the larger PC companies, but
it can be effective. These guys are clearly in third
place behind IBM and Compaq in the
PC world, and they would still be third place (or worse
yet) in the SPARC world. If they
are offered a chance to move up in the hierarchy, and
still have the comfort of having major
forces in the industry supporting them, they will see this
as an opportunity.</li>
</ul>

<p>
This covers the near term reaction to a Compaq SPARC. The
long term consequences depend on
the course that we follow. In the absence of any MS lead
RISC project, I expect that we will see
over half of the PC only crowd offering SPARC machines
within a year of Compaq actually
shipping. The only reason that I do not say all of them
is that the SPARC market can only absorb a
certain amount of growth until mass market binary
applications start to appmr.
</p>

<h4>1.4.2. PC Industry ISVs</h4>

<p>
The effect of a Compaq SPARC announcement on PC industry
ISVs will depend a lot on our
strategy and how it relates to Windows. It will also
depend a great deal on how active Compaq is
getting ISV support. If they are really hard core about
the notion of separating the UNIX and Dos
parts of their world, then they will not embark on a big
ISV program unique to the machine and will
instead let Sun's existing efforts handle it. On the
other hand, if Compaq has some differentiating
feature over Sun (for example they decide to push Motif
rather than OpenLook) then they will have
to go to ISVs.
</p>

<p>
If we do not have a compelling way to tie RISC into
Windows, and execute on it very well, then we
will see a lot of PC industry ISVs look at SPARC as being
enormously more interesting after
Compaq than before.
</p>

<h4>1.4.3. Networking Business</h4>

<p>
One of the primary attractions to UNIX from Compaq's point
of view is that it has mature
networking. We can expect that their direct sales force
will be pushing this very strongly as part of
selling the machines. This suggests that networks will
become one of the first areas where we will
conflict develop between the x86 and SPARC sides of
Compaq's business - which machines get
pushed as the departmental network? What is the server
OS? If you buy one kind of Compaq
machine the sales rep will help you out, but the other
kind you have to call this dealer.
</p>

<p>
This has a lot of potential to negatively impact our
network business, and any attempts to establish
OS/2 as a server OS. A natural separation for Compaq to
make is to say that simple file and print
services belong on x86 PC based servers up to the
SystemPro, since the issue is mainly one of I/O
throughput rather than processor speed. The dominant
software here would be Novell. When you
need to do database operations, or any other kind of
compute intensive server task, it will be hard
for them to avoid selling the SPARC machine.
</p>

<h2>2. LIVING IN A SPARC WORLD</h2>

<p>
One way to view this developmmt is that we should learn
very fast how to live in a SPARC
dominated world, because that is going to be a reality.
There are several degrees of emphasis that
you can place on this. At the very least, Sun and SPARC
will continue to be a viable system for a
number of years, and we could view this as an incremental
revenue situation (much like Compaq).
At the other extreme we could just admit that SPARC is
likely to take over from the x86 within the
next five years and we should jump on board as hard and
fast as we can. In any event, the sections
below describe some of the opportunities that we could
approach assuming that SPARC is going to
be important.
</p>

<p>
There are three different approaches for distribution that
we could take in working with SPARC -
one would be to deal directly with Sun, the other would be
to work with Compaq and the third is to
try retail. In the discusion below we will assume that
Compaq will have the same system software
base that Sun has, and this means that for all intents and
purposes we have to work with Sun. Retail
is a poor choice for the present SPARC market because it
is all done by direct sales. By the time
that retail is effective, it will be too late for most of
the approaches below (except applications).
</p>

<h3>2.1. Windows layer for SunOS</h3>

<p>
The idea here is to take the portable version of Windows
being developed for NT, and make it work
on top of SunOS instead of the NT kernel. There is
considerable precedent for such a project - we
negotiated a deal just like this with Sun several years
ago, but it fell through at the last moment
when they were feeling their oats and thought they didn't
need us (that was right at the moment
when they first signed up with AT&amp;T to control UNIX).
</p>

<p>
If there wasn't an existing window manager and look and
feel for UNIX this would be
straightforward - it would be to UNIX like Windows is to
Dos. The proposed deal we had with Sun
would have had a portable version of PM be the only GUI
interface with SunOS, and be bundled
with every copy. Now that there is OpenLook in the
default position, a number of problems arise:
</p>

<ul>
<li><b>Sun won't bundle it.</b> Obviously we can try to
get this, but it is hard to see why they would
want to at this stage. They can sell at the rate of 200K+
machines per year without this
today, and there is no present sign of their growth
slowing down. They are getting a slow
but steady series of defections from the PC ISVs, and with
Compaq on board they have a
good shot at the rest. As a company, they are a lot less
enamored with the "big deal
syndrome" than we are and they are likely to just go out
and compete with their product.
There is also the strategic cost of letting us get into a
strong position in their system
software business.</li>

<li><b>System utilities and UNIX apps will follow
OpenLook.</b> They are too far down the road to
pull back now. The fact that there would be two different
looks, and two different APIs is a
pain to both end users and developers. Over time people
would tend to view one of the
approaches as being a second class citizen - if that
stigma fell on Windows (and Sun would
have every reason to make that so), then ISVs would
migrate to Openbook and we would
be eliminated.</li>

<li><b>Windows on SunOS could be clumsy.</b> Although the
core kernel issues are not difficult,
there are a lot of other things that would need to be done
to make a slick system.
Examples include how you share the screen with Openbook
apps, sharing clipboard
formats, whether you can do DDE with OpenLook apps etc.
This has much of the flavor of
the problems that arise in trying to run Windows and PM at
the same time in a nice
fashion, but the details are worse because OpenLook
differs a lot more from Windows than
PM does. It is not so simple to add</li>

<li><b>Sun will continue to evolve OpenLook.</b> They
will add new features in an attempt to be
competitive in general, but this will cause a direct
challenge to Windows. Unless there is a
high degree of cooperation between us and Sun this will
make life difficult.</li>

<li><b>Windows will evolve in directions that are hard to
support.</b> Our advanced data storage
initiative is a good example of something that will be
very difficult to implement on top of
SunOS, and there are likely to be more of these in the
future.</li>

<li><b>We would be vulnerable to direct 0penLook ports, or
SMK approaches.</b> The primary
value of the Windows layer is to make it easy for an ISV
to address both SPARC and
Windows on x86 with the same source base. To the extent
that this is not smooth because
of the evolution issues or clumsiness issues above, or
that the resulting app looks like a
second class citizen in an OpenLook world, ISVs would be
motivated to go directly to
OpenLook. If SPARC remains a minor phenomena, then they
would use our layer, but
once it is important they will look hard at other
approaches. The most viable alternative is
a software migration kit which Sun or third parties could
provide, which makes it easier to
port from Windows, but yields a more OpenLook-ish app as
the final result.</li>
</ul>

<p>
This does not mean that a Windows layer for SunOS is a
terrible idea, but it does raise a lot of
questions. It is not a simple project and we would have
to overcome these issues in order to make it
viable.
</p>

<h3>2.2. NT Windows for SPARC</h3>

<p>
Another obvious approach is to build essentially the same
NT Windows product that we are
presently targeting for a MIPS based Power PC, but offer
it on SPARC instead as an alterative to
SunOS for the native operating system. This raises its
own set of challenges:
</p>

<ul>
<li><b>Sun's strategy is based on controlling the
operating system.</b> The operating system is key
to any plan to dominate an instruction set - they need
control of the operating system
software in order to be able to get an advantage over the
SPARC clones. This comes up in
a huge variety of areas - multiprocessor machines, new
system level instruction set changes,
moving to a 64 bit address space, making handheld
machines, adding multimedia
capabilities - the list goes on and on. The operating
system sits at the critical crossroads
between the hardware and applications. To first
approximation, the only software that is
really visible to the hardware, or that is visible to
ISVs - is the operating system. Sun
understands this, and they are not very likely to give up
their system software business, or
to let us get any kind of serious position in it.</li>

<li><b>We are not well suited for their preent market.</b>
Although we are working on making a
compelling state of the art product, NT Windows is not
particularly well suited for the
traditional workstation market. Given that Sun, and
Compaq, are both working the
workstation market first and foremost in the initial
phase, we are bound to get a fairly small
penetration. Our product will shine when it is marketed
as a Power PC - not as a weird
kind option in the workstation catalog.</li>

<li><b>We would have to plug and play with their network
strategy.</b> The most direct example of
us not being well suited for their present market is that
we would need to work on a Sun
network. Essentially all of their sales are machines on a
network , and our situation would
be hopeless unless we could have individual machines
running NT plug into a Sun network.
This means we would have to support NFS and their entire
net strategy (directory, mail,
security...). Although it is possible we could do this by
licensing software from Sun, it
would still be a lot of work, and it would constrain us in
doing our own networking vision.</li>

<li><b>They control the customer via a direct sales
force.</b> It is very difficult for us to come in with
a different operating system when Sun has dedicated people
on site, and is selling a
complete solution rather than a retail machine. This does
not give our system any room to
grow and build up momentum. This will change at some
point when they go retail, but by
then it will be too late.</li>
</ul>

<p>
One of the themes that runs through the problems above is
that there is already a strong operating
system strategy for SPARC, namely SunOS. Furthermore, the
marketing environment for SPARC
is dominated by Sun0S and matched to its natural
constituency and feature set. Compaq is not
considering doing a RISC PC or Power PC which just happens
to have a SPARC CPU inside - they
are planning on going into the Sun clone business and
going after the same customers. We are ill
suited to competing in that environment, and without some
room to grow, we would never get
critical mass and succeed.
</p>

<p>
It is interesting to contrast this with the MIPS based
Power PC plan that we have had to date. In
this case we are being marketed as a high end Windows
machine, and the feature set would he
attuned to that need. The Power PC is designed to mesh
well with a network of Windows machines
- it has the same apps (upon recompilation) , the same
look and feel, the right network support
(LanMan and Novell) etc.
</p>

<h3>2.3. NT as a base for SunOS</h3>

<p>
Another idea is to try use NT as the base for building an
entire UNIX version, probably by turning
SunOS into a subsystem. This is similar to the Mach
approach, and from a technical standpoint the
NT kernel would be great for that. It is difficult to
imagine that Sun would let us do this, or would
be interested themselves. They have the capability to do
this themselves, and there is little reason to
let us into the revenue stream for this reason.
</p>

<h3>2.4. Applications on SPARC</h3>

<p>
Finally, our applications group is a way that we could
profit on SPARC even if we do not get any
systems revenue. The issues here are fairly
straightforward - we would simply port our apps to
0penLook. One could imagine using some portable Windows
code as an internal porting aid, but
that is just an implementation detail.
</p>

<p>
This case differs substantially from the historical
example of our success on the Macintosh:
</p>

<ul>
<li><b>Apple never had the potential to kill our systems
business.</b> The Mac was destined to be
popular, but because it was proprietary, it was a self
limiting and could not threaten our
systems business. The Mac offered incremental revenue
with no strategic consequences.
Lending support to SPARC is a different matter
entirely.</li>

<li><b>SPARC does not offer a unique technology.</b> The
Mac gave us a unique chance to do GUI,
and thus be the foundation for a long term applications
strategy. SPARC has no such offer
- RISC is not unique (indeed MIPS has better performance),
and the rest of the system
definition is boring and not substantially different than
today's PCs. Power PC on the other
hand does represents an opportunity to raise the bar on
the minimum system and take
advantage of technological synergy.</li>

<li><b>Apple had a much better distribution strategy.</b>
The current Sun direct sales force
approach is not conducive to selling our applications. We
really need to have retail
software distribution. SPARC will have that eventually,
but until that point it makes the
business case of doing SPARC applications much tougher.
Although there is a big
advantage in being first on a new platform, there is also
the phenomena of being "all
dressed up with nowhere to go" - it doesn't count if you
are so early that you are gaited by
having an immature infrastructure such as the distribution
channel.</li>

<li><b>The prment SPARC market is niche oriented.</b>
There is an interesting chicken and egg
problem - SPARC has a very poor selection of mainstream
office productivity applications
today - which means that its present set of users clearly
do not place these in very high
regard. SPARC sells today to people in niche markets
(electronic CAD etc) - although
they may have an interest in word processing, spreadsheets
etc, the per capita demand will
be much less than the PC market. This reduces the
effective installed base and sales
volume that we can look at to project applications
revenues.</li>

<li><b>Our opportunity cost is higher at present.</b> Our
apps group has a historic opportunity to
reap the advantages of having bet on GUI, and on
Windows.</li>

<li><b>SPARC app sales will impact sales on other
platforms.</b> In the early days of the Macintosh
we had a very small market share in applications, so the
Mac represented good incremental
revenue. Someone who bought a Mac and our software was
almost certainly a customer
that we would not be able to reach with other products.
This is no longer true, and many
business customers that buy SPARC and our apps would be
likely to have bought our apps
on Macintosh or under Windows. In encouraging the growth
of the SPARC market we
have to recognize that the customers we attract will not
all be incremental additions.</li>
</ul>

<p>
Many of these points apply only to the early stages of
SPARC. Obviously if SPARC is destined to
succeed and be the high volume office platform, then we
should do applications for it. The points
above mainly speak to the issue of when we should jump on
the SPARC bandwagon. The net result
is that the business case for doing SPARC apps is much
tougher than it would appear at first
because the SPARC market is les lucrative than it appears
(niche market, impact on other
platforms) and because there is a tradeoff against our
systems business.
</p>

<p>
This would suggest that we should not do SPARC
applications until the point where we think that
SPARC's success is a foregone conclusion, and the SPARC
infrastructure and channel is in place to
make our entry meaningful.
</p>

<h3>2.5. Conclusions</h3>

<p>
The fundamental problem that we face in looking at SPARC
as an opportunity is that they don't
need us. The very reason we are discussing SPARC is our
fear that Compaq's support lets them
achieve critical mass. This naturally limits Sun's
interest in doing a special favor for us, yet without
some kind of edge or unique advantage it is tough to
compete with their own system software which
has many built in advantages.
</p>

<p>
This is another way of saying that Microsoft is not much
like Compaq - it is difficult to work up
enthusiasm for buying into somebody else's game and their
rules and still beating them through
superior execution. Compaq only has experience in
succeeding at this strategy against IBM, which
is a very slow moving company that does not know how to
execute all that well, and does not even
understand how to press their advantage. With a couple of
minor changes in strategy, IBM could
have eliminated Compaq's big claims to fame - for example
if IBM had wanted to do the 386 first ,
Intel would have put the fix in for them and the Deskpro
386 never would have been. With the
right licensing terms and up front negotiations, IBM could
have had Compaq and the rest of the
industry locked into the MCA bus and there wouldn't have
been an EISA. IBM is not ruthless,
innovative or even all that ambitious, and Compaq may
discover that they need new tactics against
an opponent like Sun which is all of these.
</p>

<p>
Compaq's view of Sun as a company that can be easily out
maneuvered in their home court, on a
game that they invented is not one that I share. That
goes for their system software as well as for
their systems - either one is a very tough nut to crack.
</p>

<p>
If in fact we have to do this, the key will come through
putting enough pressure on them that they
need us, getting some initial breaks in this way, and then
pressing our advantage with flawless
execution. That is utterly different from what we are
used to doing.
</p>

<h2>3. FOSTERING SPARC ALTERNATIVES</h2>

<p>
Julian Schwinger, a Nobel prize winning physicist, was
often at odds with the rest of the physics
community. One of his books starts with the quote
</p>

<p style="text-align: center"><i>If you can't join 'em,
beat 'em!</i></p>

<p>
This inversion of the usual homily is appropriate here -
it behooves us to consider how we can beat,
or at least impede SPARC, because joining the SPARC
movement is going to be very difficult.
</p>

<p>
Note that this is a good idea even if we think that it is
likely that in the long term we will join. There
is no reasonable scenario in which it benefits us to sit
back and let SPARC momentum continue.
Anything which reduces SPARC's power and momentum is
positive for us - it is either the opening
that we need to compete with them, or it provides the
leverage we need to negotiate a graceful
entrance into the SPARC business.
</p>

<p>
Although the specter of Compaq throwing in with Sun is
daunting, we should not forget that we
have been dealt some very good cards:
</p>

<ul>
<li><b>Windows is emerging as the key API for PC ISVs.</b>
This gives us the ability, if we are
careful about it, to deliver those ISVs, and the attendant
momentum of their support to
Power PC.</li>

<li><b>We have a great deal of influence.</b> Much of the
computer industry looks to us for guidance.
They do not always like this and they can be resentful,
but this doesn't mean that they won't
do what we say.</li>

<li><b>SPARC has powerful enemies.</b> Many of the
world's largest computer companies are
committed to fighting SPARC, or die trying. This is a
potent resource which we can tap
and channel. The one thing that the anti-SPARC forces of
the world lack today is
leadership and a shared mission. Encouraging them to them
is a high leverage role for us
to play.</li>

<li><b>Compaq will not compete directly with Power PC.</b>
They are sincere about avoiding their
present channel and overlapping with their present market.
Their support of SPARC is a
strong endorsement to the industry, but they will not have
anything which directly confronts
Power PC as far as end users are concerned.</li>

<li><b>SPARC is not ready to compete in the PC
industry.</b> Although they have their act together
in the workstation arena, they do not have sufficient ISV
support to mount a credible
launch in the PC industry. Unless we mess up in a big way
they will not attain this in the
time between now and the shipment of Power PC for MIPS.
As long as we can get a
reasonable number of Windows applications to port, we can
easily dominate them in a
direct showdown.</li>
</ul>

<p>
If we play our cards right, we can parlay these advantages
into a pretty complete victory. This will
not be easy by any stretch - but it is possible.
</p>

<h3>3.1. Uniting the MIPS Community</h3>

<p>
The only RISC chip that has a hope of beating (or even
slowing down) SPARC is MIPS, so the
clear thing to do is to strengthen the MIPS camp. Our
plan of record has basically ignored the
MIPS based workstation market, and we have focussed only
on a very elite group of PC
manufacturers who build the reference platform primarily
for NT Windows.
</p>

<p>
This strategy does nothing to slow down SPARC in the
workstation business, and it gaits the
availability of the reference platform to OEMs, chip
vendors etc on the availability of our software.
In a world where SPARC is getting a powerful edge it
suggests that we revise this to give more near
term benefit to the MIPS world:
</p>

<ul>
<li><b>Cause the MIPS community to unite behind
standards.</b> We would use our influence with
key OEMs to get every major Sun opponent to endorse a
common MIPS based platform
standard and a common UNIX standard. The key elements are
an R4000 reference
platform system (with associated chips) which we can
supply from our own effort, and a
single version of UNIX which we should cause to be
knighted as the standard (and maybe
participate in it business wise). We would approach the
standard in such a way that some
companies (such as DEC) could have incompatible hardware
below the OS level (and thus
do a lot of porting work) but there must be application
level binary standard.</li>

<li><b>NT Windows is the carrot, and Sun is the stick.</b>
We would tell these companies privately
that we will support the MIPS standard with NT Windows and
Power PC, and make an
appeal against SPARC. We may or may not choose to make
our support public early in the
game. Our goal is that we get them all to support NT
Windows - either as primary or in
some cases as the secondary operating system. The key
difference between this approach
and our current plan in that we do allow/encourage them to
go ahead and ship UNIX on
our machine design - especially if the hardware is ready
before our software.</li>

<li><b>Allow a much broader initial group.</b> We would
target a much larger number of OEMs - at
least for the agreement on the reference platform and the
UNIX standard. We might
restrict the initial availability of NT Windows to give a
smaller set of OEMs an initial time
advantage on shipping, or we might not.</li>

<li><b>Announce early.</b> We would make sure that the
MIPS unification announcements happened
as soon as there were enough signatories. We would also
consider pre-announcing NT
Windows &amp; Power PC early as well. It is important to
break the monotonic stream of
good news about SPARC.</li>

<li><b>Get Windows ISVs to support Power PC/4000.</b> The
one key card that we hold is Windows
3 and control of the Windows API. We must use this to
ensure that Windows ISVs port
their applications to Power PC, and to MIPS.</li>

<li><b>Get Power PC/4000 to ship ASAP.</b> The UNIX
oriented standard is just a placeholder to
slow Sun down and to give Compaq something to worry about
in the UNIX market they
are so eager to join. The real vehicle for blocking Sun
is to get the PC industry to ship
MIPS based Windows machines. We have to make NT Windows a
priority like none
before and get it done.</li>

<li><b>Promote the hell out of NT Windows on MIPS.</b> We
will have to spend a lot of money and
effort in promoting Windows on MIPS, even though it is
likely to come installed on the
hard disk rather than be a retail product. We would also
require the OEMs to push very
hard. There are a number of creative things that could be
done to help establish the
machine.</li>
</ul>

<p>
Once there is a strong unified appearance to the MIPS
workstation market, Sun's momentum will
lessen. For example, the UNIX ISVs are pretty much all
platform neutral today in the sense that
their apps are available on more than one machine - where
the number 1 platform is SPARC and
numbers 2 through N are random with no one platform
getting a decisive margin. Once the market
consolidates into just two mainstream platforms (plus the
RS/6000 as a random wild card), it would
be crazy not to support both of them. This helps to deny
Sun a lock on the applications and slows
the growth rate of SPARC unique applications.
</p>

<p>
This also helps the industry infrastructure gear up to
make systems. Having the architecture spec
and ASIC designs go out to chip manufacturers early on
will bootstrap the process of getting good
support chips, having third parties supply add ons etc.
</p>

<p>
One big plus is that achieving unification of the MIPS
world is relatively cheap in terms of the
commitment that we need to make. The primary activity
would be flying around the world
convincing people. All that Microsoft would really need
to commit is that we would in fact make a
version of NT Windows for MIPS and offer it for sale -
along the lines of what we have to commit in
the MIPS contract when we exercise the option. I do not
think that we will need to commit to doing
this exclusively by any means, so if we later need to try
SPARC we will be able to do so. The reason
we can get away with this is that the biggest companies,
who are most likely to try extract such a
promise, already hate SPARC and already want to compete in
the UNIX market so much that they
will pursue this course independently of whether we ever
ship NT Windows.
</p>

<p>
The companies which are likely to join up include:
</p>

<table border="1" cellpadding="1">
<tr><th>Company</th><th>Market
Area</th><th>MIPS
already?</th><th>Probability</th></tr>
<tr><td>Olivetti</td><td>Europe</td><td>No&l
t;/td><td>90%</td></tr>
<tr><td>Siemans/Nixdorf</td><td>Europe</td><td&
gt;Yes</td><td>90%</td></tr>
<tr><td>Bull/Zenith</td><td>Europe &amp;
US</td><td>Yes</td><td>90%</td></tr>
<tr><td>Apricot/Mitsubishi</td><td>Europe</td><
td>No</td><td>70%</td></tr>
<tr><td>Nokia</td><td>Europe</td><td>No</
td><td>70%</td></tr>
<tr><td>NEC</td><td>Japan</td><td>Yes</td
><td>90%</td></tr>
<tr><td>Sony</td><td>Japan</td><td>Yes</t
d><td>90%</td></tr>
<tr><td>Acer</td><td>Taiwan &amp;
US</td><td>No</td><td>70%</td></tr>
<tr><td>Daewoo</td><td>Korea &amp;
US</td><td>Yes</td><td>90%</td></tr>
<tr><td>DEC</td><td>US &amp;
World</td><td>Yes</td><td>80% ??</td></tr>
<tr><td>HP</td><td>US &amp;
World</td><td>No</td><td>50%
(90% long term)</td></tr>
<tr><td>MIPS</td><td>US</td><td>Yes!</td&
gt;<td>100%</td></tr>
<tr><td>Misc workstation (Silicon
Graphics...)</td><td>World</td><td>Mixed Yes &amp;
No</td><td>90%</td></tr>
<tr><td>Misc mini &amp; mf (Amdahl,
Tandem...)</td><td>World</td><td>Mainly
Yes</td><td>90%
(will support but few products)</td></tr>
<tr><td>Misc 2nd &amp; 3rd tier PC OEMs
(AST,Dell...)</td><td>World</td><td>No</td><td&
gt;50% -
90%</td></tr>
</table>

<p>
This list omits the more questionable companies, although
we might want to give them a try. It also
omits companies that would be very nice to have, but whose
likelihood of joining is unknown. This
category includes companies such as NCR (we should be able
to get their Tower &amp; mini division
even if the PC side does not do Power PC right away) and
Tandy (not normally high end enough for
Power PC, but might lend support). AT&amp;T is another
random case - they might be worth bringing
on and might not.
</p>

<p>
Europe is clearly the best geographical area - we should
get a clean sweep of the major companies,
because they are either already signed up for MIPS
(Siemans, Bull) or could easily be influenced by
us (Olivetti). Japan is also strong if we can get NEC,
because they dominate the market so much.
The US is actually the worst area for large name brand
companies. Zenith comes along with Bull,
which helps a lot.
</p>

<p>
HP and DEC are the most interesting since they have the
biggest reputations and would help the
image. They could be hard to convince. Each of them
would join in an instant if they knew
Compaq was going SPARC, but might drag their heels
otherwise. One interesting thing about both
of them is that their PC businesses are not doing as well
as their workstation &amp; mini businesses.
We could position Power PC to them as a way to get synergy
between their workstations and PCs
to help fix this problem.
</p>

<p>
Note that this is the list for the general support of the
strategy including Power PC, the reference
platform, and the UNIX standard. We would select a subset
of these companies for the Power PC
consortium - which is omits the companies from outside the
PC industry like Tandem or Amdahl.
Their support is important image wise for confronting the
SPARC armada.
</p>

<h3>3.2. Managing PC ISVs</h3>

<p>
The biggest single asset that we can bring to bear on the
problem is our control of the Windows
API, and therefore the PC ISVs. This will require careful
management, since they basically hate us.
On the other hand, they love to hate us - despite the
bitching, they do make money and are not
stupid enough to hurt their own businesses just to spite
us.
</p>

<p>
The general idea for how we can manage the ISVs onto RISC
is to work as hard as humanly
possible at getting 32 bit Windows to be a reality on the
x86, and make the recompile over to RISC
seem like a compelling incremental investment. The basic
plan goes as follows:
</p>

<ul>
<li><b>Continue to encourage the rush to Windows.</b>
This hardly needs to be repeated here, but it
is essential to do whatever we can to keep ISVs moving to
Windows. Making clear
statements about the future of Windows will help.</li>

<li><b>Head off platform neutrality.</b> We do NOT want
the next priority after an initial Win 3 app
to be a move toward platform independence. We obviously
cannot stop people from being
sensible about organizing their source code, but it is
essential that we keep the Windows
API moving toward our goal, and give ISVs something
meaningful to do that helps our
strategy.</li>

<li><b>Promote the Win 32 API.</b> The first race to pick
is taking advantage of the 386 fully with 32
bits - at first via the thunk toolkit, and later with a
fully 32 bit system. The large increase in
386 sales due to Windows means that there will be a
substantial installed base of Win 3 on
386 so this is a natural and compelling thing for ISVs to
do. We should promote this with
big seminars like the original OS/2 seminars - held
worldwide etc. Depending on timing
this might be x86 only, or we might want to promote our
RISC strategy as well.</li>

<li><b>Provide good tools for 32 bits and for RISC.</b>
We have to make this transition as smooth as
possible. This is especially true of the move between 32
bit x86 and RISC - it must be very
easy to do. In the case of moving from 16 bits to 32 bits
on x86 we will have a lot of pull
from the 386 installed base to help out, but we still need
good tools. Our tool strategy
should include evangelizing third party tool vendors as
well as doing internal work.</li>

<li><b>Evangelize Power PC.</b> The dual nature of the
standard makes this a particularly good way
to get ISVs involved with RISC. Many horizontal
applications will not have to do a great
deal of work to support Power PC in an opportunistic way
(i.e. they'll be nicer when on it,
but not have a dedicated version), but it is still
valuable to promote it. We need to get a
few exciting high profile apps to show the way.</li>

<li><b>Come up with ways to encourage ISVs to support
RISC.</b> This includes financial support
and a variety of other inducements. Once we have gotten
them to do a Win 32 app and
made the port to RISC reasonably easy to do, it is only a
question of the incremental
investment that is necessary to make the release. The
OEMs can help position this as an
industry wide phenomena rather than something done at the
bequest of Microsoft.</li>
</ul>

<p>
Every step is quite solid and independent of the Compaq
SPARC issue until you get to the last
point. It should be relatively straightforward to get
ISVs to support the Win 32 API in one way or
another no matter what Compaq does. The tricky point
comes when we get the ISV to actually
commit to releasing a version for our RISC version - that
is the point where Compaq's support, or
lack thereof, will make some of them balk, or at the very
least take a wait and see attitude. The way
to overcome that is primarily through good marketing to
the ISVs and good promotion. We must
also make sure that there is a solid perception that the
machine has sufficient support from OEMs
to be successful.
</p>

<h3>3.3. Wild Ideas</h3>

<p>
Here are some sample ideas about dramatic (or just crazy)
things that could be done to enhance the
plan discussed above. Be forewamed that they are not full
proposals at this point - just
</p>

<ul>
<li><b>Distribute apps (and/or working models)
pre-installed on the hard disk.</b> This is one of
the ideas that has been discussed already for Power PC.
This helps make the case to ISVs
because they get very high visibility if their app is
pre-installed. Ideally this is done via a
machine serial number scheme so that you can buy the app
by phoning an 800 number with
a credit card and get a key. We can use our own
applications to help force the issue - for
example we could say that two apps in each category will
be included (to be fair), and
bootstrap the process with Excel and Win Word. This would
put the pressure on
Wordperfect and Lotus to either join up or see us get a
real leg up.</li>

<li><b>Convince someone to bet the farm.</b> There are no
end of companies, especially in Japan,
that seem to want to spend insane amounts of money to
break into the computer business.
Some recent examples are Kubota (with Ardent), Matsushita
(with Solbourne) and Canon
(with NeXT) - each has sunk around $100M so far, and don't
have much to show for it.
We could consider telling another such company that Power
PC is enough of a paradigm
shift that it is an excellent opportunity for them,
especially if Compaq and IBM are sitting
this round out. One obvious candidate is NEC - we could
get them fired up about using
this as their big entree into the international PC market.
The total investment is not
necessarily as large as the ones mentioned above - the key
is to get them to be very active in
promoting the machine.</li>

<li><b>Attract software (or other) vendors which have a
direct sales force or special distribution
network.</b> A good example here is Oracle (but Novell
may also apply). If they had a very
good position on RISC PC, they could be very effective at
helping to establish the machine.
This does not mean selling the hardware, but there are a
variety of ways they could make it
appear like a well supported mainstream choice to their
customers. This ranges from just
having their sales force push it, to offering special
service and support services for
configurations including the machine.</li>

<li><b>Create a large marketing war chest.</b> This can
be funded through contributions from our
OEMs etc.</li>

<

[ Reply to This | Parent | # ]

Groklaw © Copyright 2003-2013 Pamela Jones.
All trademarks and copyrights on this page are owned by their respective owners.
Comments are owned by the individual posters.

PJ's articles are licensed under a Creative Commons License. ( Details )