Technology at a crossroads, or, Why hasn’t the tech industry performed better? (I)–Hardware

Three posts to come…

I mentioned in a post on Recent market Action in late May that the tech industry had been a laggard during the current market rise.  I said that the IT index is a mix of large-cap companies that have little growth in prospect and some smaller, more exciting names.  I’d like to elaborate on that post.  I’m going to do this in three parts:  hardware, business software and games.

Hardware:  prohibitive cost of chip production

Take hardware first.  To build a new state-of-the-art semiconductor fabrication plant from scratch costs at least $3 billion.  Two or three companies are unlikely to band together to share the expense, for fear that their design and process technology secrets would be stolen by the others if they were to operate in a single plant (Intel told me once that they won’t sell an obsolete fab to anyone, but dismantle it and sell the components piece by piece instead, for security reasons).

Getting the capital to built a new plant is a huge hurdle for almost anyone except Intel.  But that’s not the only stumbling block.

What would you do with the plant once it was completed?  Potential output is maybe $8 billion in sales annually, and it has to be run at a high usage rate to be profitable.  That would mean either having to get a chip into every cellphone sold everywhere around the world, or getting one into at least a third of the PCs.

That isn’t going to happen.  Device assemblers aren’t going to allow any supplier to have so much market power.  One Intel is bad enough!  So almost no one has the sales generating potential to justify building a new semiconductor plant.

Foundries are the answer

Increasing capital intensity of the semiconductor production business is a key reason for the increasing prominence over the past ten or fifteen years of the independent chip fabrication, or foundry, business.  This is not simply overflow capacity for integrated semiconductor firms.  The existence of foundries has also allowed the flowering of pure “fabless” semiconductor design firms as independent entities,  using foundries exclusively to produce output for them.  In today’s world, the top foundries, like the cream of the crop, Taiwan Semiconductor Manufacturing Company (TSMC), use cutting-edge processes to produce the most advanced semiconductor devices.

Although the foundry business is more sensitive to the business cycle than integrated semiconductor companies, foundries have been steadily increasing their market share as fewer and fewer design companies can afford their own production capacity.  So if we want to see the future of the semiconductor industry, we should look to the foundries.

A mature business

What are they saying?  TSMC has just announced a reshuffling of top management.  Dr. Morris Chang, the founder of TSMC, returns as CEO.  Dr. Rick Tsai moves from CEO to head of new business development, targeting solar and LED (light emitting diodes).  Employee unhappiness with recent layoffs that Dr. Tsai oversaw may have something to do with the timing of the move, but the main reason is that Dr. Chang foresees only about 2% annual growth for TSMC from its semiconductor manufacturing operations over the next 10 years.

True, mainland China has designated semiconductor manufacturing as a national industrial development priority, so TSMC may be facing some competition from emerging foundries there.  On the other hand, TSMC has over half of the global foundry market and by far the deepest relationships with fabless design firms worldwide.  And the legitimacy of a fabless firm and the reliability of its output implied by its being a customer of TSMC mean the fabless firm needs TSMC at least as much as TSMC needs them.  So the 2% growth is probably indicative of the semiconductor market as a whole.

If so, then the recent newspaper quote, attributed to a California IT executive,  that if it weren’t for Steve Jobs, Silicon Valley would be Detroit, is more than just a good one-liner.  The IT industry as we have come to know it over the past thirty years is mature.  That’s what the stock prices of the big-cap It names are showing.

The case of netbooks

Netbooks are an interesting case in point.   About two years ago, Asustek (ASUS), the Taiwanese company that invented the netbook, decided to not just do “the usual” with its laptops–“the usual” being a 5-6 lb unit with a brighter (maybe bigger, too) screen and a faster Intel chip than the previous year, at about the same $1500-$2000 price.  Asustek did make some of them, but it also introduced a 2 lb unit with an 8″ screen, a less-powerful microprocessor and 30 Gb of flash memory storage (with more available online).  It had 4-5 hour battery life and was good for email, Web browsing, internet video phone.  It was also about $300.

The rest of the industry laughed at ASUS, but the netbook has been a phenomenal success so far.  If sales hit the optimistic side of projections, netbooks could approach 20% of total laptop sales this year.

The sharp decline in product quality and service from Dell and the inability of Microsoft Vista to function effectively may have opened the door for the netbook.  But the fact that business users adopted the netbook so quickly shows how little they care for “the usual”–and, implicitly, for the Intel-Microsoft PC architecture that lies behind it.

In addition to the less-is-more form factor, the netbook also introduces a much wider audience to Linux.  The early netbooks had 20Gb of flash memory storage.  The buyer had two choices–Linux, which takes up 2Gb or Windows XP, which takes up 8.  (In addition, ASUS  produces conventional laptop motherboards configured so the user can either boot up using Windows, which takes a minute or so, or Linux, which is almost instantaneous.)

True, netbook makers quickly switched to 160Gb hard disk drives, to cater to customers who want the familiarity of Windows.  But the choice illustrates how much space the Microsoft products use up and the loss of speed that using Windows forces.  But this story hasn’t been fully written, either.

1.  New netbooks powered by non-Intel processors using technology from ARM will soon becoming out.  ARM, a British company, provides the designs behind the chips in many mobile devices.

2. Microsoft has (so far) decided not to support ARM processors in Windows 7, its newest operating system.

3. Intel has decided not to support Windows XP–but only Vista (ugh!) and Windows 7–in its newest Atom chips for netbooks.

So:  netbook manufacturers are trying to make the devices operate more like smartphones.  Microsoft and Intel are trying to force them back into the traditional laptop pigeonhole.  It seems to me that if Linux were a more familiar operating system, this would be no contest–Wintel would lose.  As it is, we’ll have to see how the situation develops.


One of the characteristics of recession is that it accelerates the pace of change.  Customers, both individual and corporate, reexamine their spending habits in a way they don’t do when they are feeling well off.  Investors also look more carefully at where they’re putting their money.

In the technology area, this reexamination is highlighting the maturity of many well-known large-cap tech names.  They are a lot like Detroit–though without the massive government protection that the domestic car industry has long enjoyed.  This realization is being reflected in stock price action.

Leave a Reply

%d bloggers like this: