The Hindu Business Line : When the icon speaks...

Taipei 101 is the world's tallest building. It certainly occupies pride of place in Taipei's skyline, at a height of 509 metres. It is also the first building in the world to cross the half-kilometre mark in height.

Dr Morris Chang is revered as an icon in Taiwan's semiconductor industry. This is his 51st year in the industry. The Taiwan Semiconductor Manufacturing Company, of which he is the founding chairman, began in 1987 and was the first to offer `dedicated silicon foundry' services to the world. (In other words, a concept similar to offshore outsourcing in software for which India is famous.)

The two have several firsts to their credit. The commonality stops there. It is going to be easy for newer constructions across the world to surpass Taipei 101 in height. But not many in the semiconductor industry might find it easy to surpass the kind of awe that Dr Chang commands among his industry peers and countrymen.

Those in the semiconductor industry in Taiwan and visitors from India alike seemed overawed by Dr Chang's mere presence at the inaugural session of SemiTech Taipei 2006, where industry folks gathered. eWorld, present there to hear Dr Chang speak, thought those who missed the event would like to know what Dr Chang said of his over 50 years' association with the industry and the disturbing patterns that he sees now. Of particular interest are his comments on the China-Taiwan semiconductor equation. Inferences for India's chances to get a foot in the door (for both software and hardware opportunities) abound in those statements.

Excerpts from the talk:

Back then

Semiconductors have been my life. I started in a job here in 1955. This is my 51st year. I am not done yet. (When I started), the industry itself was only three years old. The transistor was invented in 1948 and took only five years to become commercialised. I came here as a production line engineer. Texas Instruments (where I worked) had started to make silicon transistors instead of using Germanium. My job was to develop a technology by which the operator could solder the wires of the transistor even as the heat generated was not passed on to the transistor. This required a combination of heat engineering and metallurgy. The objective was to maximise the speed of soldering and not affect the metallurgy of the transistor.

I came up with a technique to improve productivity. The yields improved. Even back then, yields of 10-15 per cent were considered good.

Now, I am both envious and sorry at the same time, when I look at engineers' jobs. Envious because there is a lot more equipment and a lot more theories (to apply). Sorry, because they only do a small job and do not look at the whole thing. The overall view is helpful and is stimulating.

Growth concerns

From 1953 through to 2000, the semiconductor industry has enjoyed a compounded annual growth rate (CAGR) of 16 per cent. It has been a remarkably long period of growth. It has generated employment of millions of workers. And good jobs at that, not just another job. It has brought food and comfortable living to millions.

Since 2000, growth has been erratic. The six-year CAGR has been less than 10 per cent. Something has changed. It takes a very courageous and unwise person to predict that the age of 16 per cent growth would return. My thinking is that in the next 5-10 years, we would see growth rates in high single digits... possibly 8-9 per cent.

Gordon Moore (of Intel) once famously said that if growth of 16 per cent year went on forever, then the semiconductor industry would become bigger than the world economy. So, we should expect high only single-digit growth in the next 10-15 years.

All along, the main engine of industrial growth has been technological improvement and density improvement (Moore's law). The doubling of density every two years drives new applications - in computers, telecom (mobile phones) and the like. The laptop is an example of the result of improving density. Doubling of density on chips allows newer applications. That could go on for another 10 years. (In an aside, Dr Chang recalled his conversation with Intel's current CEO, Dr Craig Barrett. "Ten years ago, I felt that doubling of density in chips would continue for another 10 years. When I asked Craig for an opinion, he thought about it for a while and grimly said "Yes, another 10 years." Now, I feel that it could go on for another 10 years. When I recently asked him about this, he thought about it as he did before and again said, "Yes, another 10 years."!!)

That is not the point. Technically, density would double every two years. The cost of production would go up and not as many applications would remain economical. It would be an economic road block, not a technical road block.

(That is why), growth rates would come into single digits.

Now, the cost of building a fab is about $3 billion. The cost of research and development of newer technologies x now 65 nm and later 45 nm x is about $1 billion. The design cost to use 90 nm or 65 nm in applications is about $60 million. There will be less and less applications that can afford to use such technology.

Every company in the space thinks that double-digit growths are possible. Not all of them can achieve that. Some believe 15 per cent growth is possible. I would call that the ‘Jack Welch standard’. That is his record.

The China-Taiwan equation

China is ahead of Taiwan now in this space. The two have similarities. China will also face the limits that Taiwan once did. Taiwan is in a reasonable position in semiconductors now. But there is a gap between that and being world class. What is it? Three things help explain it:

Creativity and originality. Technology design. We need to build more creativity and reach world class.

Co-operation. We need it on a grand scale. Alliances are not strengths of the Taiwanese or Chinese. They see it as a zero-sum game. There is too much of "I win, you lose" kind of attitude. That is obstructive.

Strategic thinking. We need to think about the war and not the battle. The battle is irrelevant to winning.

If Taiwan and China can do those three things, then they can win the long-term game.

bharatk@thehindu.co.in

More Stories on : Interview | Events | Hardware

Article E-Mail :: Comment :: Syndication :: Printer Friendly Page