Hi robert w fain; Some (long) scattered thoughts:
I think that DSP is here to stay, just like digital logic came and
is still here. TI has always had its share of home runs. If the
company ever quits making home runs it will melt away as
technology moves away from it. If you want a grand slam home
run, go look for a tiny company that currently has no product
at all, but might someday have a couple percent of the DSP
market.
In other words, any semiconductor company has to continually
develop new products, and DSP is just the natural extension of
digital logic.
But DSP really doesn't have a lot to do with TI. DSP is just a
technique of manipulating what used to be analog data in
a digital format. That is something that any guy can do. I have
a system on my home computer (about $2000 worth of Xilinx
software) that allows the design of 10 billion add/second DSP
designs into field programmable gate arrays. No big deal. Its
just the current technology, and nobody has a lock on it.
What I'm trying to say is that DSP is a technique, you can't
take over the business like you can the PC processor business,
and get a grand slam home run. Instead, as it becomes more
larger, more companies enter the market and keep prices low.
There was a market similar to DSP that TI once had a very good
position in. That was digital logic. I still have my TI TTL hard
bound manual, but the chips are completely obsolete. I keep it
for some sort of sentimental reason, and because the part numbers
never did become obsolete, but are now sort of standard
nomenclature. Anyway, the fact that TI had a lead in TTL did not
prevent the market from being completely commoditized. In fact
there was no team that hit a grand slam home run in digital.
Reason was because it was a market that was too big and was
not naturally susceptible to monopoly. If you wanted to make big
money in that technology explosion, you put your money into little
companies with no piece of the pie that would grow into larger
pieces. Companies that extended digital into alternate technology
like Xilinx, Altera, IDT, etc. The expansion of the field of digital
logic was not kind to the market leaders, and I don't think it likely
that the expansion of DSP will be particularly kind to the market
leaders there, either, and for the same reasons.
The PC processor business has been more susceptible to monopoly
because of software compatiblity requirements that just don't apply
to DSP. DSP is about specialized algorithms, not general code.
The end-user isn't going to try to run "Doom" on his digital signal
processor. The same thing applied to digital logic, and that is why,
even though both design techniques are revolutionary, the big
boys were only able to maintain their sales not grow them much
in the fields.
It is true that TI has increased sales in DSPs, but you should note
that as designs get moved into DSPs, TI's sales of older chips
will corresponding reduce. In other words, just like integrated
circuits cannibalized transistor sales so will DSP sales cannibalize
digital and other semiconductor sales. Consequently, the growth
rate to watch with TI is not the DSP growth rate, but instead the
growth rate of their total semiconductor sales. Otherwise you
are only looking at the good news, not making a realistic
investment analysis.
TI currently has the fastest DSP processor on the market, I
believe. (You should probably note that the company that won
the microprocessor wars never had the fastest microprocessor.)
This is great but the majority of DSP solutions (in terms of unit
shipments) will not involve digital signal processors. The reason
is similar to the reason why such a small percentage of digital
logic designs involve standard digital logic chips. ASICs are
just too cheap. Think about this. The digital "market" expanded
incredibly, but the market for (standard) digital integrated circuits
collapsed. The DSP "market" has the same problem. Fact is
that TI's digital signal processors are not optimal for any given
problem. They are like standard digital integrated circuits, and
will not make an appreciable percentage of the DSP "market"
in 2010. Instead, designers will design the equivalents of current
ASICs. How much income do the old companies that designed
standard TTL logic get from the ASIC houses? Not much. What
happened is that the technology was just too broad for any one
company to keep a hold of it. DSP is similarly broad.
And about intellectual property. TI has processors that are good
at implementing DSP algorithms. They are the fastest in the
world, as far as general purpose DSP processors go. Of course,
special purpose DSPs will whip general purpose DSPs, just like
ASICs whipped standard logic. But lets take a look at that
intellectual property.
Intellectual property comes in two major flavours. Designs and
patents. Patents are useful, but I don't see TI as having patents
that will prevent other companies from implementing DSP hardware.
(You are no doubt aware of TI's famous "Patent attorney full
employment act" that they executed some years ago.)
The basic problem with patent coverage is that there are just
too many different DSP algorithms and uses. Most designers
just ignore prior patents, and since other companies never get
to look into the details of the millions of designs done each year,
TI is unable to obtain much advantage with a patent. Patents
are more useful when TI wants to go after competitors like ADI.
But these are competitors selling similar standard DSP solutions,
and I just don't see them keeping a major part of the market.
The other type of intellectual property is the actual design. But
you and I both know that the value of a design decreases rapidly
with time. After 5-10 years just about any design is pretty much
worthless. So the enduring value comes from the designers who
keep up with times and come up with new designs.
The historical problem with "design" intellectual property in the
semiconductor industry is the fact that companies can't prevent
their designers from changing firms. A new start-up can and does
hire away key employees with attractive stock options and
sometimes higher salaries. In addition, employees take their
stock option profits and start up new competitors themselves.
The upshot of all this is that the small companies in semiconductors
are the ones that grow fast, not the big guys. The only exception
has been Intel, and they have had an effective monopoly on the
definition of PC microprocessors. That monopoly has only kept
them ahead of their competitors by a time that is measured in
months rather than years, and that slim advantage seems to be
dissolving as consumers and businesses seem to be beginning
to buy cheaper machines rather than the latest and greatest.
This monopoly situation just does not apply to TI. The designers
can design TI out of a high volume product with a low unit cost,
and this keeps unit profits low for large volumes - therefore no
monopoly pricing power.
Anyway, I'll be looking to buy TXN if it gets cheap enough, I think
it is a good company. I'm very cheap, and hate to pay too much
for anything, especially stocks. The thing to remember in stock
investments is patience.
One thing that was negative for TXN relative to the other semis
was the high insitutional ownership. The smart money (i.e. big
guys) have been net sellers of stock for months now, and TXN
being a favorite of theirs, its stock has dropped badly. If the
big guys buying back into the market (maybe it will happen
sometime next year), that could drive TXN back up.
I hope I haven't been too negative here. I know you guys don't
want me telling you these things. I am really not against TI as a
company at all. I think they will continue to thrive, its just that I
don't expect their growth rate to suddenly depart radically from
its long term trend. I intend on continuing to design DSP hardware
for the next 20 years, and I expect TI to come up with some
amazing things over that period. I don't think this is a junk stock.
TI is a real company with real products. It's just that I think its
future prospects have been exaggeraged by its investors and,
especially, by its management.
-- Carl |
