[maemo-users] Nokia: Linux Needs to Learn Business

From: David Dyer-Bennet dd-b at dd-b.net
Date: Sun Jun 15 18:28:02 EEST 2008
Mark wrote:
> On Fri, Jun 13, 2008 at 3:36 PM, Theodore Tso <tytso at mit.edu> wrote:
>>  Do you really think it is OBVIOUS that it is
>> always cheaper to make your own and never to buy from another
>> company's product?
> Because it *is* obvious. The fundamental assumption is that the given
> company has the size and resources to manufacture chips. Give that
> capability, yes, it is *always* cheaper for them to manufacture their
> own.

Wow; this demonstrates ignorance of a depth I had not believed existed 
in the universe.  I'm impressed.

Let's go through the reasons why it's terribly expensive to manufacture 
custom-designed chips in relatively small quantities:

You can't keep the people and support infrastructure working at 
capacity, which raises the cost for what you get.

You probably need about the same number of really *top* (expensive) 
people to make things work as somebody doing 10 times as much designing 
and fabricating (at least, there's a minimum number; a big chip designer 
may need *more* than the minimum, but even your small company just 
designing a few needs that minimum).

It takes time to set up a fab for a particular chip; that's down-time 
when it's non-productive.  So short runs have added costs.

If you don't need enough chips built to keep your fab working full-time, 
that's even more expensive (you paid the same price for it as the guy 
who *does* work his full-time).

There won't be the support tools to support design and debug with your 
custom chips (unless you build them yourself).

The design and setup costs have to be amortized across however many 
chips you build; the fewer chips you build, the more each has to 
contribute towards those costs.

I'm sure I'm missing some things; while I've worked around semi-custom 
VLSI design (I wrote behavioral modeling code and developed test 
patterns), I've never worked near a fab or around full-blown custom chip 
design.   But generally, most of this comes under "economies of scale", 
a very well-known concept in economics.  You should look into it.

David Dyer-Bennet, dd-b at dd-b.net; http://dd-b.net/
Snapshots: http://dd-b.net/dd-b/SnapshotAlbum/data/
Photos: http://dd-b.net/photography/gallery/
Dragaera: http://dragaera.info

More information about the maemo-users mailing list