台積電 ADR (TSM) 2004 Q3 法說會逐字稿

Welcome to Taiwan Semiconductor Manufacturing Company's third-quarter 2004 results webcast conference call.

Today's event is being chaired by Ms. Lora Ho, Vice President and Chief Financial Officer, and Dr. Rick Tsai, President and Chief Operating Officer.

This conference call is being webcast live via the TSMC web site at www.TSMC.com and only in audio mode.

The dial-in lines are also in a listen-only mode.

At the conclusion of the management presentation, we will be opening the floor for questions.

At that time, further instructions will be provided for the procedure to follow if you would like to ask any questions.

Please be advised, for those participants who do not yet have a copy of the press release, you may download it from the TSMC Web site at www.TSMC.com.

Please also download the summary slides in relation to today's quarterly review presentation.

Once again, that URL is www.TSMC.com.

I would now like to turn the conference over to Dr. Elizabeth Sun, TSMC's head of Investor Relations, for the cautionary statement before the main presentation by Ms. Ho and Dr. Tsai.

Please go ahead, Dr. Sun.

Good morning and good evening to all participants.

This is Elizabeth Sun, head of Investor Relations for TSMC.

Before we begin, I would like to say that management's comments about TSMC's current expectations made during this conference call are forward-looking statements subject to significant risks and uncertainties and that actual results may differ materially from those contained in the forward-looking statements.

Information as to those factors that could cause actual results to differ materially from TSMC's forward-looking statements may be found in TSMC's annual report on Form 20-F (ph) filed with the United States Securities and Exchange Commission on May 28, 2004 and such other documents that TSMC may file with or submit to the SEC from time to time, except as required by law.

We undertake no obligation to update any forward-looking statement whether as a result of new information, future events or otherwise.

Now, I would like to turn the conference call over to Ms. Lora Ho, our Vice President and Chief Financial Officer.

Welcome to TSMC's third-quarter conference call and to share our positive results of our successful quarter.

I am pleased to say that (indiscernible) five consecutive quarters of strong financial performance, TSMC has delivered another set of record-breaking financial results for both the top line and the bottom line.

We have prepared some slides for you to look at as you listen to the call; these slides are available on TSMC's Web site.

During the call, I will refer to the slides by their individual number.

Please be reminded that all the dollar figures that I will be referring to today will be denominated in our local currency; that is NT$.

To start with, let me take a look -- let's take a look at some highlights of our third-quarter, 2004 operating results.

These figures are shown on slide number four.

Our third-quarter revenue amounted to 69.7 billion.

Net income after-tax totaled 27.9 billion.

Both of these are record high numbers for the Company.

Meanwhile, Earnings Per Share reached $1.20, an increase of 20 percent sequentially.

TSMC ended the third quarter with 124.4 billion in cash and short-term investments.

Shareholders equity was 379.3 billion, 28.2 billion higher than that of the second quarter '04.

During the third quarter, we shipped a total of 1.333 million 8-inch equivalent wafers, 4 percent higher than 1.28 million pieces shipped in second quarter '04.

Our overall utilization rate, which included TSMC fabs, wafer tab and SSMC was 103 percent for the third quarter.

Overall, I'm pleased to tell you that our ROE for the quarter has reached 30.6 percent, 3.8 percentage points higher than the 26.8 achieved in second quarter.

Let's go to slide number five.

Compared with the second quarter this year, revenue grew about 7.5 percent and gross profit margin improved to 2.7 percentage points to 46.1 percent for this quarter.

Despite slightly lower utilization, our gross margin kept going up due to lower product costs in 8-inch fabs, continued improvement in yield in 12-inch fabs and the better product mix.

Operating expenses were at a similar level with that of last quarter.

As a percent of total net sales, operating expenses took up 8.6 percent of the total quarter revenue, slightly lower than that of the previous quarter, which was 9.3 percent. (indiscernible) an expense of .3 billion for the second quarter, net nonoperating items for the quarter was an income of 0.6 billion mainly due to higher interest income, foreign exchange translation gains and increased loyalty (ph) income.

Investment income increased by 10 percent to 1.4 billion in this quarter, mainly due to continued improvement of wafer fabs (indiscernible) our results.

Net income before tax improved 22 percent to 28.1 billion.

Income after-tax came to 27.9 billion, 19 percent up in comparison to the previous quarter, while EPS improved 20 percent to $1.20 per share.

You can see from slide number six that, on a year-over-year basis, our revenue increased by 27 percent and gross margin was 7 percentage points higher than the quarter last year.

Especially, revenue waiting for our advanced (ph) technology went from 11 percent last year of to 30 percent this year.

We are also pleased to notice that there has been significant breakthroughs in production efficiency for our (indiscernible).

Total operating expense increased by 1 billion, mainly due to operating expenses for (indiscernible) fab-14.

However, as a percentage of net sales, operating expenses only took up 8.6 percent, lower than the 9.1 percent for the third quarter of 2003.

Nonoperating income increased (indiscernible) mostly due to higher interest income, foreign exchange translation gain on (indiscernible) marketing instruments and higher loyalty income.

Our net investment income was up 1 billion, mainly due to revenue improvements from wafer fabs, SSMC and Vanguard by about 1.7 billion, partially offset by higher expenses incurred at TSMC Shanghai of 0.3 billion.

Pretax income was 28.1 billion; this is 68 percent over the 16.7 billion achieved in the third quarter last year.

Our capital expenditures this year were much higher than last year, so was the related tax credit granted.

Our profit after-tax was up 84 percent (indiscernible) the third quarter 2003.

EPS also increased 85 percent from 0.65 to $1.20 per share.

Slide number seven will provide you with a few important items from TSMC's balance sheet and some key financial ratios.

First of all, our cash and short-term investments for the quarter was 124.4 billion, 1.1 billion higher than the previous-quarter position of NT123.3 billion.

Accounts Receivable balance has increased from 30.8 billion to 33.4 billion, reflecting higher sales activity in September.

Accounts Receivable turnover increased by 1 day to 43 days.

Current liability at the end of the quarter was 44.2 billion, which is 4.8 billion lower than -- 49 billion at the end of the second quarter.

The difference was mainly the paydown of cash dividends and employee bonus for the total amount of 12.8 billion, partially offset by increase in payables due to equipment vendors for a total amount of 6.6 (ph) billion.

Inventory-wise, our Q3 inventory came to 12.6 billion, 0.9 billion higher than that of the previous quarter, mainly due to an increase of higher-value advanced technology inventory caused by overall demand migrating towards 12-inch advanced (ph) products.

Overall inventory turnover remained healthy at 38 days for the quarter.

Current ratio for the quarter was 4 times, as opposed to 3.5 times in the second quarter.

TSMC continues to maintain a very solid balance sheet.

Slide number eight summarizes TSMC's cash flow for the quarter.

During the third quarter, with net profit of 27.9 billion and depreciation amortization of 15.4 billion, cash flow generated from operating activities totaled 38.2 billion.

On the capital expenditure front, our capital expenditure for the third quarter reduced to 16.5 billion, equivalent to US$490 million.

Altogether, our capital expenditures for the first three quarters of the year totaled US$1.7 billion and our projection of Capital expenditure for 2004 remains unchanged at US$2.4 billion.

In terms of cash investments, we have increased our short-term investment by 4.9 billion in monetary market funds.

Long-term investments also went up by 7.3 billion, mainly due to a 4.4 billion capital injection (ph) to TSMC Shanghai and the further investment in government and corporate bonds by 2.8 billion.

In respect of financial activities, TSMC has paid out cash dividends and employee profit-sharing of 12.8 billion.

Overall speaking, our cash balance reduced by 3.9 million for the quarter.

Cash and short-term investments together was up by 1.1 billion.

Now, let's look at some business analysis.

Slide number nine provides a quick look at our revenue breakdown by technology.

In quarter three, TSMC achieved significant growth in our advanced technology segment, now defined as 1.3 micron and below, which now accounts for 30 percent of our total wafer sales, up from 25 percent in the second quarter.

Due to stable yield improvement of our 12-inch fabs, we expect a waiting (ph) of advanced technology products to continue going out.

Slide number ten provides a few by application.

Due to seasonal demand, computer sector revenue contributed to 31 percent of total revenue, or 4 percent up from that of last quarter.

Communications revenue remains flat, while (indiscernible) was down 3 percent.

The consumer segment declined by 2 percent after a strong second quarter to 20 percent of total revenue because of some inventory adjustments.

Slide number 11 shows the geographical breakdown of our revenue.

Revenue from North America, Asia-Pacific, Europe and Japan now accounts for 75 percent, 11 percent, 7 percent and 7 percent (ph) of our total revenue for the third quarter, respectively.

Compared to the last quarter, North America went up slightly while Asia and Japan came down a little.

Slide number 12 provides the revenue breakdown by customer type.

In the third quarter, revenue generated from fabless customers account for 69 percent of total sales, and IDM 31 percent when system (ph) accounts for less than 1 percent.

Now, let's move to slide number 13.

This slide shows our fab utilization and ASP trend.

For four quarters in a row, TSMC has achieved utilization higher than 100 percent.

Our utilization rate for the quarter reached 103 percent.

AFC increased by 1.4 percent this quarter, mainly due to better product mix.

Let's turn to slide number 14.

This slide summarizes our install capacity by fab.

Our total capacity during the third quarter was 1.259 million 8-inch equivalent wafers, 10 percent up from the second-quarter actual.

Due to increase from our 8-inch and 12-inch fabs, we expect (indiscernible) for the fourth quarter to (indiscernible) to be 1.320 million 8-inch equivalent wafers.

This is about 5 percent higher than that of the first quarter.

It is mainly from the 12-inch capacity extension.

Our projection for total 8-inch wafer equivalent capacity for 2004 remains at 4.8 million pieces, unchanged from our previous indication.

The last slide, number 15, provides a recap of major events during our third quarter.

I will leave that for your own reference.

Overall, this has been a successful quarter for us.

However, there has been some near-term softening in customer demand and based on our current business outlook, management's expectation for the fourth-quarter 2004 performance are the following -- wafer shipments to decrease by a single digit percentage point sequentially, overall utilization rate to the (inaudible) mid '80s percentage level, gross profit margin to be in the range of 42 percent to 44 percent, ASP to remain at about the same level sequentially.

This ends up my presentation today.

Thank you very much.

Operator, please open the floor for questions.

Thank you. (OPERATOR INSTRUCTIONS).

Robert Maire with Needham and Company.

Yes, your guiding utilization to be in the mid '80s in the current quarter -- I'm assuming that's based upon the first few weeks of the quarter and where you've seen utilization rates trail off to at the end of the third quarter.

Would you have any rough initial idea about first quarter of next year?

Would we expect a flattening or a further decline from mid '80s percentage or any comments on that currently?

It's too early to comment on the first quarter 2005; it's a couple of months from now.

So, we will not comment (indiscernible) at this moment.

Okay, thank you.

Mark Fitzgerald with Banc of America Securities.

Thank you.

The utilization rates and the technology mix at 30 percent, I'm a little curious why ASPs are so flat or just up increasing modestly here.

Is this an overall pricing problem for the industry?

No, as the Chairman said in the (indiscernible), we haven't seen any unusual pricing situation up to now.

We don't expect to see an unusual pricing situation also in the near future.

Would you expect pricing then just be stable through '05?

No, we are not saying that.

We are saying we're not seeing an unusual pricing situation.

There's always the pricing pressure from the end market.

I would say, roughly speaking, we're holding our price quite well in general.

Okay.

Then just one quick follow-on -- can you make a quick comment, given the guidance from Morris about the track growth for the industry next year and a modest growth above that for the foundry business?

I'm a bit curious about your capital spending outlook being flattish to just slightly down, especially given the low utilization rates, why you wouldn't pull your CapEx down -- (technical difficulty).

Our next question come from --.

No, we haven't gotten a chance to answer; we haven't answered the question yet.

Okay, you are asking about 2005 and CapEx 2005.

I think, this year, the total CapEx is US$2.4 billion.

It's too early to talk about 2005 CapEx, but we do see a very strong demand on the advanced technology, especially in 90 nanometers, which we're going to continue to invest.

So with that in mind, although I cannot give you an exact figure but we think that 2005 -- (technical difficulty) -- will be comparable to 2004.

Ali Irani with CIBC World Markets.

Good morning, thank you.

Could you talk a little bit about your 65 nm pilot efforts and some of the advanced technologies like SOI, where you are seeing demand at the bleeding edge?

I'm hoping you could update for us the number of customers and projects you have going on right now at 90 nm and to what extent your 90 nm processing is on 300 mm versus 200 mm wafers.

Thank you.

On 55 nm, basically our own test vehicles is our own design SRAM. (indiscernible) progress is on schedule.

We plan to have already production in, I think, late at the end of next year, early 2006.

We are working with several key customers on some early tape-out (ph) in the first half of next year.

So, the major R&D effort now is on 55 nm, and as you know that we also signed an SOI JVP with (indiscernible) so we will be able to developing 65 nm and later on 45 nm together.

They are also transferring the 90 nm to us.

On SOI, I think the demand up to now is still more in a niche type of a market, mainly microprocessor and not all microprocessors, as you know.

We do expect, however, the SOI to become more and more toward the mainstream as we go down from 55 to 45 nm.

For 55, we expect to see a stronger demand also in the very high-speed and also in some other cases low-power applications.

We are very excited about this development program and we expect to gain -- to get extra market share from this effort in the next few years.

As a clarification, if I could, would you be exploring both the bonded and the implanted SOI processes, given that your spearheading efforts for both the 65 and 45 nm node keeping all the options open or have you chosen a process of record at this point?

I don't think we've made a decision yet but I'm also not the best person to give you a direct answer to that.

If you are interested, we can refer you to our R&D people.

Thank you.

On a 90 nm customer base, could you provide us with an update on the expansion of projects and how you see that going?

You speak of strong demand -- maybe some color on the quantity.

We're now in the bottom production for the quarter for both and the demand continued to -- and also, the important thing is that we are now having very strong demand for different applications, all the way from (indiscernible) based on (indiscernible) graphics to PODs (ph) and the consumer applications.

So that's why we are very bullish on 90 nm.

We believe 90 nm is the technology (indiscernible) with a very strong economical advantage.

We are running, by the way, we are running 90 nm port (ph) and 300 mm wafers right from the development stage.

We have never run any -- we are not running any wafers in the 8-inch wafer, so all the demand and the yield improvements are in 300 mm wafers.

It's 300 mm wafers that we believe that will give us -- that has given us also a strong edge against some of our competitors.

So I suppose this is a big reason why you see your CapEx at this point for next year being driven to a flat number despite your utilization, because demand is going at 300 mm where you are investing.

Yes, yes.

Michael McConnell with Pacific Crest Securities.

Just a point of clarification here -- we're just reading stories over here and it looks like the Chairman talked about CapEx being potentially down 10 percent next year.

Could you clarify what his message was on the earlier conference call?

I think what Chairman said this afternoon is the CapEx for 2005 will be comparable to 2004 within a plus or minus 10 percent range.

Shifting to the end markets, in terms of the strength, obviously Q3 looks like it was going to term early, driven by your PC customer base.

In terms of the decline in Q4, could you just kind of stack up the end markets in terms of what's relatively the strongest versus what looks like it's going to be the weakest, please?

I think that we now all know that inventory correction is the main culprit for the slowdown.

What we're seeing, I think, overall PC market -- the PC market is, in comparison, doing better than our content communications, or it is also -- it is slowing down somewhat but the power (indiscernible) is the PC chip makers and I guess also their customers have been maintaining a better inventory strategy and they are probably also very experienced with many previous cycles.

So, you're seeing that the ROPC (ph) revenue ratio is moving up.

Communication is more volatile.

Handsets I think is -- handsets, in some cases, in some customers I think that it's affected by China's overall policy of their tightened economy.

I think that also creates some problems for the consumer chips.

That's some of the color that I can think of.

Okay, so in terms of I guess Q4, if we look at the orderbook right now, would you expect your ratio, in terms of sales to PC customers, going higher, sequentially -- (technical difficulty) -- basis?

I don't know.

I believe so.

Let me see.

Yes, I think we will somewhat, but the (indiscernible) moderate.

Medhi Hosseini with Friedman, Billings, Ramsey.

Yes, going back to the Chairman's commentary regarding a flattish semiconductor industry next year, assuming that units would actually grow, I'm thinking that, in the backdrop of your capacity increasing by 20 percent and even if you look to keep the CapEx level next year, I would expect your capacity to be up 10,1 5, even 20 percent next year.

So, what's the disconnect here?

You are looking to '05; units could actually grow at a lesser degree compared to owned internal capacity.

What gives you the confidence to actually keep the CapEx relatively flat to '04?

Well, I think the demand, of course, as you said, may not grow as rapidly in the year 2004 and also, you said that the (indiscernible) units too but units will still grow.

The question, the issue there is they will grow in different technology nodes.

In some technology nodes, they will grow much more slowly or even be stagnant.

In other area, you will grow it much faster.

That's what we're seeing, especially as we said earlier, in advanced technologies, 90 nm and .13, .11.

We continue to see strong demand for the products because most of them are new products.

You have to remember, usually during a time like this, everyone almost without exception is trying to come up with new products to grab market share rather than stick with the existing, older products.

So, that's I think -- we absolutely as we have done through the years, we absolutely will get the highest market share for those advanced technologies.

If I hear you correctly, even if units were to grow at a lesser degree compared to your capacity, you expect utilization rates to at least be 85, high 80s, because of the mix.

We have not forecast any utilization rates for next year yet.

Just as a follow-up, as IDMs make the proper changes to their inventory, have you seen a significant pull-in by some of your IDM-type customers pulling up capacity, doing it in-house?

We are seeing a little bit but really not significant.

The inventory control -- I mean, there seems to be less inventory buildup in comparison.

So, not (indiscernible) from IDMs in terms of their putting it (ph) into their own fabs.

Neal Kantmishro (ph) with Credit Suisse First Boston.

Yes, this was a question asked during the afternoon as well.

Could you please explain how the utilization could go down significantly if the shipments are down only in single digits?

Can we take this to mean that Q1 '05 will be significantly weaker?

Let me try to explain your question.

You are asking the mismatch of utilization versus shipment quality (ph)?

Yes.

Our utilization is calculated based on wafer output, not the wafer shipments.

You know there is some timelag between the two.

Also, we do not separate the 8-inch and 12-inch when we are counting the utilization, so one 12-inch wafer we also count as one wafer and 1-inch we also count (indiscernible) wafer.

So actually the revenue component is quite different.

Another one is the utilization on wafer output and (indiscernible) engineering wafer, which was excluded from our utilization calculation but(indiscernible) wafer may be achievable.

Also, the Vanguard product is excluding (indiscernible) wafers without the calculations (ph) that it will be included in our revenue, so that's several places that you can reconcile the difference between (indiscernible) and wafer shipments.

Thank you.

Two very maintenance questions -- what is the depreciation which you'd modeled for 2003?

What are your ramp-up plans in China, and how does the fab-9 ramp up in 2005?

Let me explain your depreciation question first.

You know, we are going to spend 2.4 billion this year, so naturally the depreciation is going to increase somewhat in 2005 (inaudible) because we have more 8-inch capacity coming down from depreciation, so even with the CapEx, the depreciation increase is not going to be significant.

There will be some increase but not a big increase.

Can I press you to quantify that little bit -- 5 percent, 10 percent, 15 percent -- (Multiple Speakers)?

I will say 10 percent is probably all right, in that range.

Ashish Kumar with Credit Suisse First Boston.

Thank you.

Earlier, you spoke about 90 nm and strong demand.

My first question was that we expected I think about 1 percent revenue in $Q '04 from 90.

Does that hold.

What is the outlook for a best case for 90 nm revenue in fourth quarter '05?

I can tell you that, on fourth quarter, our 90 nm will have 1 percent revenue.

In terms of '05, we're going to increase that percentage very significantly, but I cannot show you a percentage at this moment.

Okay.

Based on your earlier comment on utilization being calculated on wafer out -- I'm sorry, your wafer output (indiscernible) does that mean, in fourth quarter '04, our shipment is going to exceed -- that is wafer shipment will be more than wafer in production?

So in other words, work in progress (indiscernible)?

We are -- actually every quarter our wafer output does not equal to wafer shipment, because certain of our -- a portion of our wafer output will need testing and so there is always a (indiscernible) time for that.

With the current business environment, I think the work in progress in the fourth quarter has decreased compared to that of the third quarter.

You also -- when you spoke about 90 nm, you said that there are strong economical advantages in 90 nm.

Could you elaborate on that?

I think the 90 nm, being a generation following the 0.13 micron, which I'm talking about 0.13, which has been more difficult to (indiscernible) generation for the industry because I believe (indiscernible) really the first of production generation for the (indiscernible) that from (indiscernible) point of view that (indiscernible) a change from aluminum to copper did impact the cost structure.

But the 90 nm, again being following the 0.13 (indiscernible), we have -- we mastered the copper technology (indiscernible).

At least with TSMP (ph), we also have done quite well with the (indiscernible) technologies, so I think our cost structure for 90 nm is quite good, quite good.

In the (indiscernible) I think it was suggested that '05 would be a flat year for semiconductor revenue count.

Also, I believe '06 will be flat.

Is there any reason for a bit of pessimism on '06 or this is just a false start and obviously things might change?

Or there is there something you see in '06 that makes you take a view to be another flat year?

Well, I think, basically, '05 we are seeing a flattish year. '06, as Chairman also noted when he answered the question that year, it was a bit too early really to predict for '06.

I think he said -- he made the comment on '06 being also flattish, I think mainly based on the longer-term semiconductor industry cycle.

Of course, there can always the certain accreditation to that cycle.

You know, these are very early comments on '06, and we're not -- right now, I think we're probably focused more -- we definitely are focusing more on 2005.

One just last question -- in your 2004 (indiscernible) management, the comment that was on the CapEx (indiscernible) you maintain a high transition (ph) rate.

You maintain your ROE forecast.

I believe you will increase cash dividend payout.

Just in case '05 turns out to be a bit weaker, might you change your outlook for '05 CapEx in the last three months?

You could change that outlook again and right from currently being flattish, you would not hesitate in cutting it down?

In other words, maintaining your transition rate and fee cash flow (indiscernible) market share gains.

Would that be fair?

Again, as we said earlier, of course most of the CapEx for '05 will be on 300mm wafers or 90 nm technology and the (indiscernible) 0.13 microns, .13/.11 micron technologies.

These are -- I think that the product, mainly in those technologies, are really new products from our customers and we, again, we intend to have the highest market share in those technology nodes because they will have long-lasting impact for our growth in the years ahead.

However, what we really need to do, in addition to the advanced technology area, is of course to work on more demand generation with our technology portfolio, more applications (indiscernible) in large quantities yet.

So, we have both.

We will wrap the higher fee market share with our advanced technology, but we also try to gain more shares in the mature technology.

(inaudible)

Chang Quai (ph) with Orun (ph) Technology.

good evening.

Could you comment on the ASP trend, going forward?

In the last three quarters, you did very well.

In the fourth quarter and going forward, what do you see?

You are asking a question on fourth quarter?

Fourth quarter I know you said it's flat.

What about for '05?

Supposedly, '05 is a down the year and we can anticipate that we will have some price erosion.

But as far as average ASP, I think we're going to ramp up our 90 nm and (indiscernible) .13 microns (indiscernible), so I think the (indiscernible) mix is going to be very significant.

So, by saying that, I think we can readily maintain our price position quite well compared to all the other competitors.

Okay.

Also, can you comment or give us some more color regarding your capacity for 2005, especially fab-12 and fab-14?

Okay, Rick just said that we're going to view capacity on 90 nm, which many (indiscernible) 12and 14 very aggressively.

You know, fab-12 is a very big fab.

If you (indiscernible) it up, it can accommodate 60K.

Right now, we're loading only 20, just in 20 K now, so I think that, for (indiscernible), we are going to increase gradually up to maybe more than 30 K in 2005.

For fab-14, we have started to produce product already and we're going to have 6 K in store in fab-14 and we will continue ramping up to meet our demand on the 90 nm.

Right, so by the end of '05, what's your projection for the capacity for these two fabs?

We have not really made a final decision yet.

We do know for fab-12 we will -- in addition to roughly 24, 25,000 (indiscernible) capacity, we will definitely add another 12 to 15 K to 90 nm.

We certainly will go up, at least I think maybe 15 K, 16 K. We figure we will continue evaluating the demand and make a decision on a more real-time basis.

Okay.

A quick question on the operating expense side -- in the last quarter, you are actually quite flat with the quarter.

Going forward, do you think it will be flat again or will be down a little bit, if the revenue is down?

I think we have maintained quite stable in terms of percentage to revenue.

We don't anticipate that percentage is going to increase much, even on 2005, that is a flattish year.

Matt Gable with Calypso Capital.

Two quick things -- could you put some more color on your sequential unit guidance for Q4? (technical difficulty).

Are we talking -- (technical difficulty) -- percent or down sequentially or 9 percent down sequentially?

Also, one more on the 2005 CapEx -- you said comparable to 2004 plus or minus 10 percent, then you said a couple of times that 2005 demand may not grow as rapidly as 2004.

Do you think 2005 CapEx then has a slight downward bias or are you still going to take it up slightly just because you need to spend on 300 mm?

You're asking about the fourth-quarter shipment quantities with regards to single digits.

It's not to the high end, and it's not to the low end either. (LAUGHTER).

Okay.

As to the 2005 CapEx, we have said many times -- basically you heard what we've been saying.

We also have not made any final decision.

I think, for 2005, we will really have to make some decision as we move along.

So, the answer you heard so far is really the best that we can provide you.

Great, thank you very much.

Christina Osmena with Jefferies.

A couple of follow-ups here on CapEx.

If your decisions haven't been made yet on which particular projects you're going to go with, does that mean that that needs to go -- that CapEx budget still leads to go through the Board approval process, or has it already been approved?

Yes, we proposed to Board for the approval of CapEx based on stage.

I would say it is not 100 percent passing approved by the Board yet, but we will look at the market demand and propose based on different stage.

At what utilization rate would you consider ratcheting back your CapEx plans?

If you're going to drop to the mid '80s, are you looking for something lower than that to start considering adding capacity at a slower pace, or --?

As we just described to you, the CapEx are mainly for advanced technologies, which we build to meet a demand, so you can imagine those capacity is going to be very full (sic).

Okay, a question here on your 90 nm -- are you charging -- for 90 nm 300 mm product -- (technical difficulty) -- by the die or are you charging by the wafer now?

(indiscernible) both way of doing business as we have with the other technology nodes also, so it's not just the 90 nm.

Okay.

Finally, do you tend to negotiate your wafer prices through long-term contracts?

Is there any period of time where those contracts are -- (indiscernible) the time where we should start looking for those contracts to start to expire or is it a year-end thing or how should we think about that?

We're mainly talking (inaudible) we've many -- we have various ways of doing business with them.

Pricing negotiations can be very different for different customers, so I'm not sure -- you know, really your question does not really apply to what we do with our customers.

Fayad Abbasi with Prudential Equity Group.

I have a question regarding some of your customers that have spoken of starts decreasing around the mid-August timeframe to you and for those that are IDMs internally as well through the end of the Q3 timeframe.

But then I had noticed that Ted (ph) noted that some of their business had increased or stabilized at least for the first couple weeks of October.

So I was wondering, in your wafer starts, is that what you're seeing reflected from your customers?

In other words, are things kind of stabilizing for the last couple of weeks or are there any notable trends there?

It depends.

I think, with some customers, we kind of put on the brakes early.

We're seeing some ordering moving slower, but it's not very clear yet.

We're seeing some (inaudible).

Does that also kind of play into the decrease in WIPs that you are expecting in Q4 as well?

Right now, for (indiscernible) WIPs, yes, it's lower than that of Q3.

Yes, I guess you can say that, yes.

Maybe a question on the Q time -- have you seen any changes there, or how does that change from Q3 to Q4, on average?

You mean the Q time for production costs?

Right.

Yes.

Obviously, for the certain technologies such as (indiscernible), Q time has come down quite a bit.

For .13, and .11, however, I don't have a number but I think (indiscernible) Q time there.

We have our demand quite strong versus our capacity.

Okay, and one question on your margins -- you had noted a couple of items resulting in the margin improvement in Q3.

Looking at your Q4 margin guidance and despite the sharp decline in utilization, the margins aren't coming down quite as much.

So I was wondering maybe if you could at least maybe rank at least what factors are having the most impact on the margin improvement in Q3 and then moving into Q4.

I can tell you there's two things that help us to improve the margin.

Number one is the 12-inch performance has come up quite significantly; the other one is the margin efficiency on 8-inch and the depreciation reduction compared to previous years.

It also (indiscernible).

You say that those are probably like the two largest and then after that, it would be a mix of shipments?

Yes.

Ben Lynch with Deutsche Bank.

Yes, hi.

Just sort of in the same vein as the last question -- the Q4 gross margin drop is I think relatively small given the significant drop in utilization.

Can you help me understand what impact inventories play on this?

Should we assume that a high proportion of your Q4 sales includes product currently in inventories which was produced at a higher utilization rate and therefore a lower unit depreciation cost? and that, as we go forward, what you're producing in Q4, low utilization/high unit depreciation, will flow through the P&L let's say in Q1 and beyond?

I think you're trying to figure out why -- how the utilization will affect our margins in the fourth quarter.

In the fourth quarter and beyond even, just because it's all a matter of when it flows through your P&L versus when you actually produced it.

Let me explain to you how we accounted (ph) inventories first.

We try to figure out whether the inventory carries a higher or lower costs.

Actually, we neutralize (ph) the inventory calculation based on our predetermined utilization, so you would not fluctuate up and down.

So, we have different utilization among quarters.

So the fact that you just described, it would not exist.

Can you explain what you mean by neutralized?

We usually use 85 percent (indiscernible) as our standard cost now when we calculate inventory.

Okay, great.

Thank you.

Shailesh Jaitly with Nomura Securities.

I wanted to understand a bit about your capacity expansion.

You mentioned a few times that all of the capacity expansion is going to be at the higher end, and the same thing you had mentioned in the previous calls.

Now, if I look at some of the numbers, overall capacity increased by about 11 percent and your higher-end -- the higher end of the processed inventories have gone up by 5 percentage points, despite being higher ASPs.

So should we view that the majority of the utilization decline is coming at the higher end?

(technical difficulty) -- I think Chairman has explained, this afternoon, the utilization weakening is coming especially from .18 and .15 microns, and some from .35 (ph) microns.

On the high-end side, utilization is quite high.

How do you reconcile that?

Does that mean that capacity which has been added is not really at the high end, or does that mean, at the high end, your yields are not as good?

It is in the high end and yield is also very good.

In that case, the revenue as a proportion for the higher end should increase in the same proportion as your capacity expansion.

I think you're trying to figure out the increase in the advanced technology.

The revenue percentage versus capacity increased, but if you look at our numbers, our revenue coming from advanced technology increased quite significantly.

In the second quarter, it was 25 percent, third quarter it was 30 percent and the fourth quarter and going forward, it's going to continue to increase.

Okay.

On the CapEx, what would be the linearity of CapEx next year?

Would it be back-end loaded or would you be spending an equal proportion in the first half and the second half?

We are expanding capacity in the first half, for sure, so -- but we do not really have a precise breakup but as I said, we are moving in equipment, .90 nm in fab-12 and the equipment for fab-14; we are moving them in as we speak.

We are installing them -- expediting the (indiscernible) also.

It's front-end loaded, basically.

One final question on your -- (technical difficulty) -- the fab-14, what would be the breakeven capacity for fab-14 and the breakeven capacity for your Shanghai fab?

Breakeven.

Breakeven.

To reach breakeven, what kind of capacity would you need to have in place?

Well, this is something I cannot tell you.

You know, it always depends on what kind of price you get and what kind of costs and your ramp up speed.

So there is no definite answer.

Is the fab-12 profitable at this juncture?

Say again?

Sorry?

fab-12, is it profitable?

More than profitable!

It's very profitable! (LAUGHTER).

Because talking to the industry people, I get a feeling that 11,000 wafers per month of capacity is what is needed for breakeven actual.

Would that be a correct understanding?

No.

I can tell you that cost (indiscernible) (indiscernible) reached in second quarter this year and we are ramping 27 K now, so it is more than profitable.

Our next question comes Terrence Chan (ph) from Overlook (ph) Investments.

Thanks for the presentation.

I think you had a great quarter.

My question relates to a whole host of a capital allocation questions.

I would first like to ask you about the dividend.

I have been led to believe that there are some (indiscernible) in the article which restricts the payout ratio to 50 percent.

I am just wondering whether that is the case.

If that is so, are you working to break that cap?

Okay, there is a constraint on the payout ratio but we're going to change that very soon, so I think management -- (multiple speakers).

(Multiple Speakers) -- timetable, timeline for that?

Because I think TSMC has gone from a high-growth company to one that's spreading out a lot of cash flow and to keep shareholders in, the dividend is certainly a way to show that there is a high level of concern for minority shareholders.

So, you're working towards that.

I would like to know what is the timetable.

Okay, well, making a proposal to a Board would change the article, so we hope we can change that before the end of this year so we can issue more cash dividend next year.

That's the plan.

Okay, so a higher payout ratio next year.

Related to that, there is about NT$3.8 of cash per share on the books.

I'm wondering if you're planning to pay that out as well.

If so, can you commit yourself to a timetable?

I cannot realistically see that number falling at all, because you're operating cash flow will cover your CapEx for a long time to come.

All right, I see what you are asking.

You know, in timeline, we don't pay dividends based on how much cash we have.

You have to see how much profit you have, but I can tell you that we are fully aware of this and we're looking forward to pay more dividends -- (multiple speakers).

(Multiple Speakers) -- financial practice, first that if you have too much cash on the books, you pay a special dividend, so you don't have (indiscernible) earnings; you pay out of your reserves.

So -- (multiple speakers) -- is something that you should consider.

You know, what's being done in Taiwan is certainly something to respect but I think you operate in an international capital market.

I think I will leave it at that but I think you should certainly think about paying out from your cash reserve.

My follow-up question would be, what are your plans regarding free shares to employees?

Next year, the U.S. is going to expense options in the P&L and I think, one way or the other, analysts will do that for your P&Ls anyway, so it's going to burn a hole in your P&L.

Would you consider increasing the cash element -- (multiple speakers)?

Yes, we are aware of the situation, and we are considering to increase the cash portion of the employees.

Is that a marginal increase or a substantial increase you have in mind?

We're preparing a proposal for our Board, and before that, we cannot really comment further.

We are preparing a proposal for the Board.

Well, you are always preparing proposals, and -- (Multiple Speakers).

No, we're not always preparing proposals!

When we're saying we're doing that, we are serious.

Please respect us for that.

Absolutely, absolutely, but I think a lot of long-term shareholders have been waiting for some concrete time on that front, so I don't think I'm speaking just for myself.

Finally, you have declared buybacks but then in the last quarter, you have not done so.

I'm just wondering why, because the stock is slow.

We think we will not (indiscernible) buyback.

We think paying more cash dividend has a higher priority than buyback.

I absolutely agree with you on that front.

I think, as a shareholder, I would prefer to have more dividends.

That would be great.

Okay, that's really all I have to say.

Thank you very much.

William Dong with UBS Securities.

I have just one quick question.

I noticed that (indiscernible) beginning the TSMC Shanghai.

Can you provide an update on the status of that?

Going forward, when should we sort of factor in that fab to reach some kind of volume production, or maybe -- (technical difficulty) -- what type of capacity should we expect by the end of 2005 out of that fab?

The fab is running pilot (indiscernible) with great success.

I think what the customers (indiscernible) came out with at least equal yields compared to the fabs in Taiwan.

But of course, by the end of the year, capacity is fairly low, but we will gradually be building the capacity up there for the local customer needs, as well as the other international customers who desire the production output in China.

Then the interest level and from the (indiscernible) customers has been high, so we are quite optimistic.

So we will build capacity there next year but yet, I think we will go above 10 to 15,000, but it's not -- probably around that range.

Ivan Goh with Dresner Kleinwort Wasserstein.

Just a couple of questions -- firstly, (indiscernible) say that he thinks that inventory correction could last one to two quarters beyond the end of the year.

I just want to get a sense of what you guys see in terms of where the rest (ph) of inventory belonging -- is it in communications or computer or PC area?

Where would the risk of inventory corrections belonging be?

As I made a comment earlier saying that the PC segment seems to control their inventory better, I think the other way really for you to look into that is to look at each individual semiconductor manufacturer (indiscernible) fourth-quarter guidance.

That's really pretty obvious which ones still have high inventory at hand.

Okay.

One financial question regarding the depreciation.

You guys seem to continue to benefit from a depreciation runoff at the 8-inch fabs.

Approximately how much reduction in 8-inch fab depreciation can we expect in 2005 versus 2004?

Well, I cannot tell you specifics but I can tell you more than half of our 8-inch fabs were coming down from depreciation.

This trend is going to continue through 2005 as well.

So what percentage of this depreciation would be 8-inch fabs?

Well, I don't have a number on top of my mind, but usually we don't release depreciation like that.

One question on the (technical difficulty) -- you guided for mid-80s for the fourth quarter.

If you just look at it on the monthly basis, what kind of utilization do you think you'll exit the fourth quarter rather than having an average?

What kind of utilization will you have exiting the year?

I think using a quarterly number should be good enough, and we don't normally talk about the monthly utilization.

It doesn't mean much.

Okay, and my final question -- you've talked about -- (technical difficulty) -- going into 2005 despite growth maybe slowing down next year.

Can you talk about the possibility of removing capacity from existing fabs, like something you did in the last down cycle?

No, we do not have such plans (indiscernible) as well, we expect to fill our load capacity in the future and it if there is one lesson we learned also back in 2002, it's -- remember, we were out of capacity for 0.25 micron, 0.3, 5.5 (ph) for a long period of time, so -- we do not plan to do that.

But I think, instead, the more progressive -- (technical difficulty) -- new applications to fill the fabs.

That's the way we want to go out, I think.

Operator, we will just open for the last question.

Pranab Kumar Sarmah with Daiwa Institute of Research.

Thank you.

My first question is, how many percentage of your revenue in third quarter has come from the 12-inch wafers?

Any color on that side?

We don't normally don't differentiate 12-inch or 8-inch.

As you know, the advanced technology will be produced from both 8-inch and 12-inch, (indiscernible) less than .13 microns accounts for 30 percent revenue in the third quarter.

This number is going to go up in the fourth quarter.

A significant percentage of .13 microns is from the 12-inch fab only?

Absolutely.

(technical difficulty) -- idea about the employees as a percentage of total expenses on last quarter?

Or some rough number (indiscernible) how many percentage of your total cost is on the employee side?

You are asking employee costs specifically?

Yes, that's true.

Employee costs account for a very small portion of our total cost structure, including the (indiscernible) sales and also in SG&A, so I cannot specifically answer your question on that.

In the last quarter, you have stopped giving the breakdown between on third quarter the (indiscernible) .15 micron and .1 micron technology.

For comparative purposes or to compare with the last quarter, could you give us a breakdown of your .15 and .18 micron technology?

I guess, on the second quarter, it was 13 percent and 29 percent, but in third quarter, you have it at -- (technical difficulty).

Yes, we started to change that because, to TSMC, .15 and .18 micron is considered matured technology -- (technical difficulty) -- normally represent by different technologies.

In last quarter, you said .15 had gone below 10 percent.

Did you just presume that?

No, we didn't say that.

No, it has not.

Okay, and my last question is on this one -- monthly revenue side -- like do you expect this fourth quarter probably starting with, say, front-end load sort of quarter or it will be a back-end loaded sort of quarter on a monthly revenue basis?

Well, we will not disclose monthly revenue forecasts but you'll see our monthly revenue announcement once we close the month.

Okay, thank you very much.

Operator, I think we should end our Q&A session for this evening.

Thank you.

Before we conclude the TSMC's third-quarter 2004 results webcast conference call today, please be advised that the replay of the conference call will only be available or accessible through the TSMC Web site at www.TSMC.com.

Thank you, all.