Kash,
First having your own chipset ASSURES quality and performance.
Very true, but requires additional resources with added risks.
Pretty DUMB to have a BILLION dollar fab investment wasted by crappy MB's and chipsets.
Motherboards were an issue in Q3, but not in Q4. So, I believe FAB25
over capacity was the main reason behind the idling Dresden fab.
Secondly it helps on the margin side. One can bundle the CPU,MB just like INTEL did to kill AMD's K6 margins mid last year.
That was a one time tactic used by Intel. It was effective when
there weren't any Athlons and AMD had difficulty of producing high
speed K6-Xs.
Thirdly, VIA is a predator. AMD is getting pretty NAIVE if they think these guys are a valuable LONG term partner they should nurture.
This is very true :^(
Fourthly, clearly chipset makes a real world difference in performance. 3rd parties have alway suffered against Intels chipsets in performance. AMD own seems like a pretty good solution. Now VIA comes along with a crappy chipset. Intel must be rolling on the floor laffing their assess off.
And finally there is history. Many folks hated the Super 7 MB's due to crappy chipsets and MB's. This definately limited theiracceptance. Admittedly 3rd parties have gotten better - just not as good as intel.
Perhaps you can come up with a real good reason - besides motherhood and apple pie.
IMHO, only option AMD had left at the time of Athlon introduction was to
focus on the Athlon chip + a basic chip set as a backup plan + a reference mobo,
and leave everything else to the partners.
You were forgetting AMD's situation last year - dwindling cash reserves, heavy
losses, decreasing market share, manufacturing problems, Intel the perfect
machine, RAMBUS + i820 + Coppermine threat (similar to Willamet propaganda
that's going on now ) without the problems that were surfaced much later,
law suits, low stock price, etc.
At the time of Athlon release, AMD was not in any shape or form to take any
additional risks other than the Athlon itself. Growth in business requires
capitol and AMD was not in a position to garner any additional capitol
required to manage chipsets and motherboards' growth.
Also, AMD couldn't have foreseen everything that we see now - the
manufacturability and scalability of the Athlons, the success of Athlons
in the market, and most importantly the problems experienced by Intel. The
success of Athlon would have been a lot tougher if it were not for the
glitches experienced by Intel in the past 5 months.
Given their situation in the first half of 1999 - IMHO, the decision by AMD
to leave the chipset and motherboard business to its partners was a
wise and a good business decision.
I also see many strategic advantages behind this - effective way (combined
strength) to compete against a formidable competitor with more than $300B
capitalization, provides great resistance against proprietary standards
by the competitor that could obsolete your technology in no time (this business
is 30% architecture and the rest is marchitecture), empowered(great incentive)
partners to fuel the proliferation of the new architecture and any future
architectures, etc.
As per the recent CC and comments from Via, my understanding is that - AMD will concentrate on the high-end chipsets for the Athlon, while the Taiwanese chipset manufacturers concentrate on the low end solution for the Athlon platform. So AMD is not leaving the entire chipset business to its partners.
Best regards,
Goutama 
