CTC Re: Dell does very little R&D, relying mainly on the
work of its value chain partners such as Intel.
Releative to Intel/MSFT, Dell could be viewed as a very
low R&D company. Actually they sit somewhere between the
extremes of CPQ and GTW. Most of my friends work in various
R&D capacities at Dell so I have a good sense of what they
do and what their challenges are.
PG (The Product Group) employs several hundred people at a
cost of >$120MM a year. I'd bet there are many hi-tech
companies which people consider "R&D intensive" that don't
even generate 120MM in revenue.
The misconception by many about Dell being just a "box maker"
really discounts the true effort required to design and test
these systems.
Yes, Dell leverages heavily off of Intel and they don't design
the silicon. Dell buys parts and integrates them into custom
solutions. There are a tremendous number of decisions that must
be made in order to design a system with wide market acceptance
at a profitable cost structure. This can make or break you as
these permutations can be overwhelming. You are at a distinct
disadvantage if you try and buy a "solution compromise" from
someone else and slap your name on it.
A good analogy would be home builders and architects. They
don't design the components that make up a house (lumber, cement, carpet, fixtures etc.) they select from available materials and
make the creative and business decisions required
to assemble the final product.
Why does Dell design most of their own systems? Because they
can't buy them from anybody else. Custom designs can be highly
leveragable, scalable and re-useable. You don't get that with
Intel or Tiawanese motherboards... not only that, you are limited
to ATX form factors and the like.
Take the Optiplex Gn for instance. One motherboard,
three chassis. A low profile desktop, midsize desktop and
Mini-Tower. This is accomplished by having all PCI/ISA
expansion slots on a separate riser card. The Video,
NIC and sound are all on-board population options. Replace
the 300Mhz PentiumII with a Celeron and you have a product
matrix that can scale price points from sub-$1000 to >$3000
without additional R&D effort. The list is much bigger but
you get the idea.
I am purposely staying away from the electrical engineering
work, the stuff that's more closely associated with "true"
R&D because of the broad audience here. But I can assure
you it's not just a connect the dots effort. Ever designed
an source synchronous interconnect interface that can
only tolerate 100 pico-seconds of skew between any two signals??
It can cause more than a few sleepless nights<gggg>
MEATHEAD |
