Financial Daily from THE HINDU group of publications
Monday, May 15, 2006


eWorld
Features
Stocks
Cross Currency
Shipping
Archives
Google

Group Sites

eWorld - Interview
Info-Tech - Events
Web Extras - Hardware


Balancing act

Bharat Kumar

UMC on the dynamics of the semiconductor business.


Lee Chung, vice-president at UMC.

Lee Chung looks like an unlikely candidate for vice-president at UMC, one of the top semiconductor foundries globally. He looks younger than his 55 years. His smile adds to the effect. eWorld chatted him up. We discovered that he, and the industry, might have several things to smile about in the coming two years. Snapshots:

We were led to understand that the semiconductor industry has been able to shrug off its cyclical nature thanks to the mobile phone boom in Asia. But, the quarter was soft. UMC has indicated that for both wired and wireless communications the recent quarter had been a soft period. Have orders not been to expectation? So, the cyclical nature has not gone away... ?

Definitely not. Actually, it's second nature. The time interval has become shorter (between cycles). It's now more dynamic than it used to be. Many years ago, we were thinking about seven-year cycles. Now, it's every couple of years that we see a cycle.

As far as the last quarter is concerned, end of 2005, there was a soft demand in the wireless sector. That lasted only for a few months before it started picking up really strong.

Where does that demand come from?

Softening in the wireless space recently had a lot to do with the China market. The China market is in correction (mode) in the wireless sector. (That had its impact.)

If the cycle has become more dynamic, how do you forecast demand and then plan for capital expenditure (capex)? We see that your capex for 2000 and 2001 was high, in excess of $1 billion. It fell in 2002 and 2003 going below that mark, sometimes significantly. It has risen again in 2004 and 2005... $1.1 billion and $1 billion, respectively.

Yes. That's true... Because, whether you are making money or not depends on what capacity you actually have. In UMC, we work with customers very closely to make our forecasts. This is hard to make. Even our customers don't know (how things would pan out). What we do is we try to get as much information as we can to make sure we have the right expectations. There could be a lot of faulty information. The most important piece is this: we can always build the building first. It does not cost a lot of money, the equipment does. For the equipment, we do not order all the equipment first. The fab is designed with bottlenecks in mind. The bottlenecks are all the tools - some tools could cost between $20 million and $30 million each. Even if you have them arrive early, you can have your desired throughput. Your productivity could improve. If you need additional capacity at that time, then the other tools could come in quickly.

We order those that have a long lead-time, say 10 months to 12 months, (We bide our time on) others with 3-6 month lead-times. So, our cashflow is in control. That's how we manage it. There is no easy way to do this.

But, is there a bench mark? A certain capex-to-revenue ratio that you want to stick to?

No, we do not have a set ratio. Capex is typically between 30 and 50 per cent of revenues. But it could fluctuate.

How do you see the next 2-5 years? Predictions of 2005 September seeing a slump in the semiconductor space have not been accurate...

Our industry sees problems when we build too much capacity. This time the industry has been a bit more conservative. But I still see over-capacity being built up by end of 2007. So, late 2007, we will see the (beginning) of a slump.

When addressing the Indian delegation, you talked about small companies also migrating to 90 nanometer...

The mass production cost in 90 nm is pretty expensive. So, initially, we thought, companies would have trouble moving to this technology. But however, today, almost everyone in this sector - Digital TV or cell-phones or MP3 - is moving to this and with high volumes. To be successful, they have to move, thanks to competition. The mass cost also has come down, now lower by 50 per cent.

You also talked of Japanese companies sticking to the IDM model... ?

They are more conservative. They treat manufacturing as core competence. They don't want to give up control. It's a culture issue.

We wish to know if you are beginning to see any market signals in the 45 nm space? Could you also comment on this in terms of applications as well as in the areas from which your prospective customers could come?

A handful of customers have started looking into 45nm. We believe that handset and graphics applications will remain the early adopters, especially the handset (folks).

What are the newer growth areas that you see in the semiconductor space? Do set top boxes and game comsoles (such as the Sony PlayStation or the MS XBOX), LCD displays, DSL modems show promise?

To what level, in terms of revenue contribution, could these rise to, compared to contribution from the telecom sector?

Based on several forecasts, the semiconductor market will grow from 2005's $250 billion to $300 billion in 2006 at a pace of about 10 per cent CAGR. We believe that consumer space will be the key contributors. Many applications such as games and DTV-related products will present significant opportunities.

We understand that orders from old customers for you such as TI and Xilinx are up sharply in the 90nm space. ATi and Freescale are also upping business with you. This is an indication that you are seeing an upturn in demand. For the second II quarter of 2006 too, would you see sustained demand-rise in the semiconductor space?

The second quarter of 2006 will be slightly better than the first quarter of 2006. We do see strong growth in Q3, and expect strong Q4.

In general, capacity utilisation in the 300 nm space is high at 80 per cent-plus. Is this a sign that bumper profits are around the corner and that in the near future, it would be easier to manage high fixed costs?

We do believe that fab utilisation will reach higher levels, which, in turn, should result in better profits. Foundry is a very capital-intensive industry. It is always critical to keep the utilisation as high as possible.

More Stories on : Interview | Events | Hardware

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



Stories in this Section
`Welcome relief for telecom industry, subscribers'


As fresh as they come...
Matters of the heart
Excelling at work
On a mission to train the trainers...
ntldr file missing
When the icon speaks...
Balancing act
On the job
Creative drive
Striking a chord
The stick for India
Quiz
In the beginning, telephones had no numbers
Cool companion
A new note
Go mobile blogging



The Hindu Group: Home | About Us | Copyright | Archives | Contacts | Subscription
Group Sites: The Hindu | Business Line | Sportstar | Frontline | The Hindu eBooks | The Hindu Images | Home |

Copyright © 2006, The Hindu Business Line. Republication or redissemination of the contents of this screen are expressly prohibited without the written consent of The Hindu Business Line