DSP Technology Is Old Story!
A lot of non-engineers are convinced that DSPs are the
wave of the future. But TI's been making them since 1982.
Here's the URL for TI developments during the decade of
the 80s, from their own web page:
TI 80s: ti.com
Click on that link and read it through. Print it. Think about it. It
has some implications that run completely counter to the current
DSP story.
First of all, it is clear that DSPs are already very old. Even TI
says they've been around since 1982. DSPs are a very old
story. They were more evolutionary than revolutionary back
in '82, and they will remain more evolutionary than revolutionary.
TI has been successful with them so far, but there are some
fundamental reasons why the future changes in their evolution
are not in TI's favor. These reasons include canibalism of TI's
other product lines, the introduction of a lot more players into
the DSP market, the fact that DSPs are becoming easier and
easier to design (for all companies), and the fact that some of
those other companies have the inside track to supplying
those DSPs. I will expound on these in future postings, I
intend this as only a teaser. Watch this space.
But first lets analyze those TI developments of the 80s. I'll start
at the beginning of the decade, so go to the bottom of the sheets
you (should of) printed.
In 1980, TI's "Net revenues $4.1 billion". Net revenues have
grown a lot since then, but so has the number of shares.
Consequently the increase in sales per share has been pretty
small. Less than the auto stocks. TXN is not a growth stock.
1982: "first single chip digital signal processor." They've had
the DSP for 15 years and their growth rate is still bad.
1983: "TMS320 Digital Signal Processor" introduced. This still
provides the backbone of TI's DSP business. It's ancient by
semiconductor standards. One generation in semiconductors
is equal to 18 months. Ten generations have gone on since
then. In human terms, TI's DSP line is 270 years old.
1984: "first multiport video random-access memory chip."
These were revolutionary. Until recently, they were the standard
for high-end graphics systems. Still no growth. I'll talk more
about this when I post my observations on the memory market,
and TI's part in it.
1985: "TI Japan facility in Hiji first U.S. wholly owned company
to win Deming Prize." This is a prize for excellence in quality (I
believe). TI's quality has always been pretty good, though the
last time I worked with the MVP it was still buggy as hell. But I
suppose that's the case with all "TMX" (i.e. pre-production)
parts. Anyway, TI has had excellent quality for more than 12
years, still no growth.
1988: "quantum-effect transistor." It's been out 9 years. What
is it? Why hasn't it brought any growth? Somebody post a
scientific explanation, please.
As you can see, the TI story is long. Their products are sometimes
revolutionary. But. They haven't been able to grow. And they're
talking about a development from 15 years ago. What is going
on?
First of all, I don't think TI is badly managed. Far from it. Running
a big company like that is incredibly difficult. I also think that GM
is well run, and TI hasn't done that much worse than GM. But
growing a company who's sales are already many billion dollars
per year is very hard. Hanging onto large market shares in a
commodity market is impossible. Look at GM. And small
companies are constantly springing up and taking a bite out of
you. All the other big semiconductor houses have the same
problem, except Intel. It's not a matter of management, and
making management changes doesn't solve it. For very big
companies, just staying alive is success. TXN is not a growth
stock. It is a great company, but not at this price.
-- Carl |
