[LWN Logo]

To: linux-arm@vger.rutgers.edu, general@netwinder.org
Subject: Intel, StrongARM, Linux, WinCE, Red Hat, Corel
From: Jim Pick <jim@jimpick.com>
Date: 	03 Oct 1998 20:03:34 -0700


Corporate giants are stirring.  Here's one possible scenario...

A few years ago, Intel and Microsoft were really freaked out when Java
appeared out of nowhere, and Larry Ellison let loose with the "Network
Computer" idea.  Although NCs haven't really become a reality yet, the
idea did represent a fundamental change in direction for the industry
- and a huge potential threat to the Wintel market.  Intel and
Microsoft weren't prepared for that.  Since then, the $1000 Wintel
machine has gained dominance at the retail level, Linux is starting to
look like a threat to Microsoft, and Intel has dropped from 85% to 45%
market share.

But Intel and Microsoft aren't stupid - since their initial surprise,
they have quietly made moves to have new products in place that can
cannibalize their existing products.  Think of it as "insurance", in
the event that they start losing market share fast.

The sub-$500 computer segment is going to happen - but it won't be
dominated by larger-die-size iX86 compatible chips like the Celeron or
highly profitable per-unit OS's like Win98.  Intel quietly moved and
purchased the StrongARM design and fabs from Digital (now Compaq).
Microsoft sunk a lot of money into a complete ground-up rewrite of
their OS for low-end machines - the result being the Windows CE
operating system.

What's Intel's next move?  Well, Intel makes a lot of motherboards -
so I wouldn't be surprised to see them come out with some ARM-based
motherboards for the clone market (perhaps using the NLX form factor,
or smaller).

There isn't much of a demand for these yet, but there definitely could
be.  With Intel's support for Linux (via their investment in Red Hat),
plus Microsoft's WinCE - there are now "safe" OS's available that
would make general-purpose sub-$500 ARM-based machines a viable
product.

What about Microsoft?  Well, they really need to kill Win98, which has
too much backwards-compatibility baggage and crud that has accumulated
over the years.  They've invested relatively little in it recently -
it's just a cash cow which they are going to take out to pasture and
shoot fairly soon.

They've got two modern OS's positioned to succeed it - WinNT and
WinCE.  WinNT is the safe bet - it's the heavy and bloated solution
for high-end iX86 machines (with a large measure of Win98
compatibility).  WinCE is the lightweight, clean, completely
incompatible OS which will replace Win98 on the extreme low end.  When
Linux inevitably begins to challenge Microsoft, WinCE will be their
extreme low-cost answer.  They'll try to postpone that day as long as
possible so they can sell as much WinNT as possible, which is a much
more profitable product for them.

I think that in 2 years, these sub-$500 machines will become
mainstream, and LCD screens will get really cheap as well.  You'll
start seeing Linux available pre-installed on these machines, as well
as WinCE, which the manufacturers will be able to license for
negligable cost.  There will still be a healthy market for iX86
machines running Win98/NT though.

IMHO, Intel has the low-cost manufacturing and iX86-leveraged
marketing muscle to ensure the StrongARM is *THE CHIP* that the
cloners use for the low end machines.  I believe Microsoft will jump
on the bandwagon here as well - pushing the StrongARM as the volume
platform for running WinCE on the low-end.  After long last, the
StrongARM is a RISC design that Microsoft can push without angering
Intel.

WinCE will only become a viable replacement for Win98 when you can buy
shrink-wrapped apps for it.  The big problem with WinCE right now is
that it runs on so many processors.  Currently, if you buy a WinCE
pocket PC, handheld PC, set-top box (ie. WebTV), or one of the
upcoming Auto PCs, typically the only place you can get additional
WinCE applications for your specific, proprietary architecture is from
the manufacturer who made your particular model of hardware.

But when Intel uses it's muscle to make the StrongARM the dominant
WinCE platform used by the cloners for low end clones - you are going
to see Microsoft pushing it's ISV community to market shrink-wrapped
WinCE apps for the StrongARM.

There is a bit of a chicken-and-egg aspect to all this.  Intel and
Microsoft don't want to cannibalize their highly profitable iX86/Win98
lines until they have to.  But they will, before the competition does
it for them.  With the StrongARM and WinCE, they are well positioned
to maintain their market share.  WinCE's main competitor looks like it
will be Linux.  The StrongARM will be challenged by a wide array of
low-end RISC designs (ie. MIPS, Sparc, Alpha, PowerPC, or maybe even
TransMeta).

Corel's NetWinder may be the first product to usher in a new era of
low-cost PC designs.  Although it's way above the sub-$500 price
point, it's still a viable StrongARM-based Linux platform.  Intel,
Corel and Red Hat will work together to make it a real platform.  Good
software (including commercial stuff, like WordPerfect) will become
widely available for it.

Intel may follow this up by pushing out StrongARM motherboards to it's
current iX86 manufacturers who will be able to sell viable Linux-based
products.  This will likely coincide with a big Microsoft push for a
WinCE port that runs on the Intel StrongARM motherboards.  Microsoft
will also push it's ISV community to support the StrongARM/WinCE and
produce shrink-wrapped apps.

If this comes to be, it seems likely that Linux could become extremely
popular on the StrongARM in sub-$500 PCs available through retail
channels (sharing the market with WinCE).  This would be great for
Corel, who would have a good suite of Linux/StrongARM applications
ready-to-bundle with these cheap Linux machines.  I'm sure Microsoft
would also bundle MS-Office "lite" with the WinCE boxes.  In short,
customers will buy a $500 office suite, and get the hardware for free.

The more profitable Windows 98 and NT will still be the dominant
players in the traditional iX86-based PC segment.  Linux could be very
popular at the low-end, where Microsoft will have to give away WinCE
for next to nothing.  It also means that the StrongARM port of Linux
could become even more popular than the iX86 port.  Of course, this is
all contingent on how seriously Intel is going to push the StrongARM.

What does everyone else think of this scenario?  Likely?  Or
impossible?

Cheers,

 - Jim



unsubscribe: body of `unsubscribe linux-arm' to majordomo@vger.rutgers.edu