高塔半導體 (TSEM) 2002 Q3 法說會逐字稿

Good morning and welcome, ladies and gentlemen, to Tower Semiconductor third quarter and nine-month conference call. At this time I would like to inform you that this conference is being recorded and our participants are in a listen-only mode. We will open up the conference for questions and answers after the presentation. I will now turn over the conference to Yoav Nissan-Cohen of Tower Semiconductor. Go ahead, sir.

Thank you. Good day and welcome to the Tower Semiconductor conference call. I am Yoav Nissan-Cohen. I am very pleased with the progress we have made this quarter. We cannot change the market situation but we can continue to operate and execute --. [ Inaudible ] Let us start fab 1. [ inaudible ] Compared to 11 -- in the previous quarter. Revenue was $2 million which is 5% increase the previous quarter. The other revenue performance --. [ Inaudible ] Our fab 1 continues to. [ Inaudible ] Third quarter. Increased by 5%, decreased by only 1%. [ Inaudible ] Moreover, fab 1 generates. [ Inaudible ] Compared to. [ Inaudible ] In the previous quarter.

Having said that, our net result. [ Inaudible ] By increased -- the noncapitalized fab 2 expenses increased from. [ Inaudible ] In q 2 to 11.7 million dollars in Q3 and we expect the expenses to continue to increase the coming quarters. As a result of this, the. [ Inaudible ] Increased from 11.3 until in q 2 to 14.5 million in Q3. The other leading foundries we expect wafer sales to increase up to 10%. In the next quarter or in Q1, 2003 as the basis, we also expect to have additional revenues of approximately $4 million related to the second development milestone of the agreement. Our visibility for next year is currently limited. However, the commercial production in fab 2 which is scheduled for early 2003 should have a significant positive impact on our overall revenues during the year, primarily in the second half. Our target is to reach cash break-even by the end of 2003 and turn to profitability in 2004.

As you all know, tower focuses on specialized technologies such as the nonvolatile memory and image sensor or what is called in the industry CIS for digital imaging. I'd like to discuss recent exciting developments in both areas. Let me start from the C.I.S., image sensor area. We recently started in fab 1 production of 14 megapixel image sensor which is used by Kodak for its most advanced professional digital camera. To develop this leading product, we have cooperated with a leading C.I.S. company from Belgium with excellent design capabilities, as well as with Kodak as the end customer.

This offers higher resolution and better performance than any other 35 millimeter format imager in the industry. Other high-volume C.I.S. products are currently in different stages every prototyping, and we expect C.I.S. production volume to increase during 2003. We continue to enhance the performance of fab 1 imaging technology as well as to advance the microtechnology for fab 2. Regarding the non volatile imaging technology there is also significant progress with our development efforts in this area. In fab 1, we see significant increase in microflash product sales.

We start add joint project with our Japanese partner to develop the embedded microflash memories in .18 micron process technology. The project continues according to plan. As mentioned earlier, last quarter we received $4 million from our Japanese partner after achieving the sales development milestone. We expect to achieve the second milestone late in this quarter or Q1, 2003 at the latest. To the best of our knowledge, no foundry supplier is able to provide these products in micron technology. This has the potential to position us as the leader in the embedded flash market. We continue to move forward on underbudget targets with prototyping the first two products for customers of partners and achieved very good results.

We expect commercial production as we said earlier, to begin in early 2003. Meanwhile we continue to install production tools in preparation for the production commencement in fab 2. Another major recent development in the technology agreement that we have signed with Motorola, which enables us to keep up with the high-end technology. It will enable us leverage on Motorola's expensive -- extensive experience in this technology. The development activities have already begun and we expect to offer the microprocess by the end of 2003. Prototyping services will begin as early as the first half of 2003 initially in the Motorola fab. By the end of this year our investment will reach $650 million and we plan to continue the investment in 2003.

As you know, some of the dominant players in the foundry market have cut their extension plans for the next year. However, they have limited capacity for leading edge technologies. It's already tight and this should create capacity shortage which may provide a great opportunity. On the financial side we met our fundraising goals. During the month of October we raised approximately $80 million from the exercise of the investment agreement result of strategic partners, the private placement and the right offering. We greatly appreciate all the investors who continue to believe in our success and invest in our future.

In summary, we are excited about the major progress we made this quarter and look forward to the expected growth of the next year. I would now like to turn the call over to Amir Harel he will for his report on our financial condition and performance. Amir, please?

Hi, everyone. I would like to share with you some of the financial highlights of the third quarter. Fab 1, 12.2 million versus 11.6 in the previous quarter, representing a 5% increase. With regard to revenue, $4 million relating to a previously-announced technology deal. The gross and operating loss in fab 1 have decreased for the fifth sequential quarter due to continuous major cost reduction measures executed by the fab 1 team, as well as improvement in sales.

The operating loss in fab 1 was 2.6 versus $9 million in Q3, '01 although sales increased by $2.2 million only, implying a very significant cost reduction effort. The consolidated net loss was $14.5 million versus $14.5 million previously. The loss includes $11.7 million fab 2 expenses of which $4.75 million are expenses related to the .13 micron technology agreement.

Cash flow from operations in fab 1 was positive 1.4 million in Q3 '02 after break-even in the previous quarter. The consolidated cash closing balance as of September 30th was $65 million versus $54 as of June and $33 million as of December. This balance has gone up significantly in October due to the money raising of $80 million. The total balance sheet increased to $630 in September from $590 million in June and $470 million in December.

The increase is primarily as a result of increase in bank loans and the profits from equity and convertible deals. The cash in amounts described about were invested in the third quarter and the first nine months primarily in fab 2 equipment and the construction project. The shareholder equity as of September 30th is approximately $250 million. As of September 30, the company had $30.2 million outstanding shares and $4.7 million options, including options to employees, directors, and banks.

On top of that, up to 5 million shares might result from the related options deal which was completed in June '02. After September 30th, we raised $80 million, 45 from milestone 4, 20 .5 from the rights offering and 60 million from O.T.P.P. As of today after the October money-rising activities, the number of shares outstanding is 3 3.4 million and additional 3.2 million were added p. I would like to remind you that some statements made during this call may be forward-looking and that actual results may be different from what is currently expected. The risk factors which may influence or result are listed in the form two prospectus as filed with the S.E.C. Yoav?

Thank you, Amir. This ends our opening statement and we are now ready to answer questions, please.

Thank you. The question-and-answer session will begin at this time. If you are using a speakerphone, please pick up the handset before pressing any numbers. For U.S. participants to ask a question, please press 1 followed by 4. If you wish to withdraw your question, please press 1 followed by 3. For participants in Israel to ask a question, please press star followed by 1. Please stand by for your first question. The first question comes from Quinn Bolton. Please state your affiliation followed by your question.

Hi, guys. a Couple of questions for you. First, with the recent fundraising of $80 million, I was wondering, how much is still required by the end of the year under your bank agreements then I've got a couple of follow-up questions.

Okay. Thank you, Quinn. We are -- as you know, we have covenants under the agreement with the bank, and the next covenant is for $27 million, which we are supposed to raise until the end of the year. I'd like to mention to you that in the past, a few times we updated the agreement with the been and changed them to be in better alignment with the reality or better alignment with the exact requirement of the company. And as of now, we are -- we don't need so much cash, and we don't have cash demand for the beginning of next year, so we may be able to discuss it with the bank and adjust the agreement, if needed once again.

Okay. Thank you for that. The second question, just with the signing of the licensing agreement with Motorola for their .13 copper process, how does that change, if it changes at all, the agreement with Toshiba, because I believe you had also scienced a license for Toshiba's .13 process. Was that an aluminum processor if you could give us details. Okay.

In generally speaking, it has not changed the agreement that we had with Toshiba. The agreement we had with Toshiba was for either two or three-year generations of technologies. We are very successful with our relations with Toshiba on the .18 micron aluminum generation and managed to bring up the process in record time, if you ask me, with a very good result.

On the .13 micron, we still have the option to install the technology and it will be done if we'll see substantial business coming from Japan, but we had other reasons related to compatibility with standard technology and so on, which suggested to us that working at this stage with Motorola would -- and bring the Motorola .13 micron technology would make more sense at this stage, mostly because of the compatibility with the standard of the industry.

Okay. So if you don't use the .13 process from Toshiba, does that leave you the option to, say, use a .09 process from them in the future or some even more advanced process?

We have the option for them, basically next generation in both directions, okay, and I -- I'm to the exactly sure at this stage with whom we'll go with the 19 millimeter, and it will depend to a great extent on the compatibility with the standard found Dree.

I'd like to pull your attention to the point that Motorola has an agreement of compatibility with the S.E.C., basically an agreement that has Phillips, Thompson and Motorola to create some kind of an industry standard. So maybe to make -- it would make sense for us also to align to this standard. And we certainly have the option to do so.

I understand. Okay. Last question, can you just remind us of the planned wafer start ramp for fab 2 in 2003?

Okay. We plan to end the capacity over 2003 and to reach a capacity of at least 10,000 wafers by more or less Q3 time frame. Of course, the actual production rate would depend on some other parameters such as the market situation and so on.

Right now we are very lucky to have what I call friendly customers, which is those companies or those customers who have invested and have given us incentive to succeed in the markets. They are the customers to start production in early '03. There are also a small number of customers that I call friendly because of either past relations in fab 1 or some other reasons, and these are going to be the type of customers that we are going to start renting the facility. But I have to tell you that this is not enough.

And basically, the company has to bring more customers in order to enable us to run the capacity in the way we want for high utilization of the facility.

And this is -- with the current market situation, this goal is more challenging because right now it is more difficult to bring other customers than what we wanted, and I believe that next year by the time that we'll be up and running in volume production with initial set of customers, and we'll be able to show good results in a very objective manner, and hopefully also the market will start turning in the positive direction, I think it's going to be substantially easier for us to bring other customers as well.

Thanks and good luck bringing on those additional customers. I know it's a tough market.

Thank you, Quinn.

Our next question comes from Ali Irani. Please state your affiliation followed by your question.

Yes. Good morning, gentlemen. I was hoping you could share with us the mix by technology currently in fab 1, and also what your third quarter capacity was at that facility, and how you see some of the rationalization of the costs that you were talking about placed through for that fab over the next couple of quarters, if you could add some color on that, please.

Okay. Thank you for the question. Let me refer to first of all, you asked about the technology segment. So I would say about 25% of the facility is still running what I call legacy products. About a little bit over 25% is running a different type of mixing of products, mix and a will go digital. All together we have about 35% of the facility, which is running none volatile memory, divided equally between the microflash and other technologies, and the rest which is in the order of 10% is going into the other technology. This is overall sales for the quarter.

I can tell you that if you look at our current situation, then the percentage of image sensors has gone up already quite significantly and I'm comparing now the current run rate in the facility relative to the average sales during Q3.

So you could expect it in q 4 you will see more activity in the image sensor, and in the next few quarters, this is the key segment that we'll grow on the expense mostly of the mixing of technology and the standard legacy products. With respect to the capacity, we are running today still at a little bit less than 50% utilization of the facility.

This is more or less, I would say, in line with what we can see with the leading foundries, if you look at their utilization rates and the older technologies. Of course, what you see with the leading foundries is that they get substantially higher utilization rate at the leading edge technology, which is a great sign for us to expect much the better situation in fab 2, which is going to start from .18 micron technology.

Now, we are really focusing -- focused our efforts during the last year on -- on the cash situation, and we didn't like the situation that fab 1 could become a drain of cash. And we are very much pleased with their results in fab 1. Right now we have about 1.4 million dollars positive cash from fab 1, so at least we generate some cash.

And we expect to stay positive in Q4. When it comes to 2003, then I think that you would get a similar answer from some other foundries. It's really limited visibility. We expect that our image sensor business will grow, as I said before, but it is quite did you have difficult to predict what's going to be the loading or demand on the legacy product and the mixing products.

In terms of overall capacity in fab 1, in terms of wafer starts, could you remind us of that in the third quarter and also if you could be kind enough to give a breakdown by line width in the facility, the capability?

We -- our capacity -- you know that capacity is a function of the mix, of the technology mix, so right now...

Sure.

The capacity that we have on the current mix is at the highest level, I would say, 18,000 wafers per month. We had a capacity of 20,000 wafers per month previously but that was with a little bit -- let's say different mix, which was more into the area of higher geometries. When we move now more into the .5 and .35 then the overall capacity went down to 18 or even below. I would say that this is divided more or less or somewhat less than half of it can do .5 and below, and the other half can do .55 or .6 and above. And the .35 micron capacity is limited to the order of 4 or 5,000 wafers per month.

Great. Thank you very much.

Thank you.

Our next question comes from David Stelereck. Please state your affiliation followed by your question.

Good morning. It's can you give us some more color onto embedded flash at .818 micron? What exactly are you doing and how bug the -- potentially the market is? -- how big the market is?

The potential is huge. This is really one of the key technologies where we can differentiate from the competition. As I said before, at least today there is no foundry which offers an embedded .18 micron, embedded flash .18 micron technology. And therefore, it provides a great opportunity for power to position ourself as a leader in this area. Now, I have to be careful because at the same time what I have to tell you is we still don't offer it to our customers, and currently we are co-developing this technology based on our know how together with our Japanese partner.

There is a team of 15 engineers who are currently residing in Japan for about nine months, to bring up this process and technology together with our Japanese partner, and more or less concurrently to bring up the process here in Israel and no more than a delay of, let's say, a couple of months after the process will be up and running in Japan.

It's going to be up and running here. We would like to be in a position to start working with the first customer on this technology at some point by let's say late 2003. The key is that we will be able to offer a high density embedded flash module. High density, I mean going from 1 megabit to 32 megabit of embedded flash modules. You will not find it in the industry today. Even people who are contemplating developing .18 micron embedded flash, typically they speak about much smaller densities because the architecture of the microflash enables us to go to high density while other architectures do not allow it. So I hope that this will be significant. Now, there are many applications for our Japanese -- applications for that. Our Japanese partner today has a business line of $600 million of microcontrollers for different applications starting from mobile, automotive, appliances, what have you. a Consumer market.

These microcontrollers, even today, they have nonvolatile memory, but today, most people are using lead-only memory which is not giving the -- is not making possible for the customers to reprogram the device either late in the manufacturing cycle or in the field, and all these customers, because of today trying to market requirements, looking for a solution that will give them reprogram ability so they can change the operating cost of the microcontroller in the field. And for that, they need embedded flash. This was the main reason for our Japanese partner to work with us, and I hope that this will just be a sign of demand that we'll see for this technology.

Right. Thank you.

Thank you.

Ladies and gentlemen, once again as a reminder, for U.S. participants to ask a question, please press 1 followed by 4. For all Israel participants if you would like to ask a question, please press star followed by 1. The next question comes again from Ali Irani. Please state your question.

Given the direction of wafer starts in the fourth quarter and limited visibility into the first quarter can you comment on pricing trends in the marketplace? Are you saying those continue to come down? Including the higher nodes and the higher technologies or do you see your C.I.S. technology one that provides you a better mix as it strengthens into the next two quarters?

Okay. We said that Q4 results will show up to 10 -- up to 10% decline in sales. When we look forward, as I said before, it's quite difficult to give the forecast because of the limited visibility. But if you ask weighs our plan or what seems to be the highest likelihood scenario, is that if you want to be conservative and say there will continue to be a decline in legacy and some mixing of our products, which is going to be compensated by the C.I.S. demand or C.I.S.product, so all in all right now, we look forward q 1, we say we don't right now see a significant change from q 4. Again, it's hard to give accurate forecasts. I certainly hope that the market today is overrule pessimistic because of different reasons and that people, because of this, are actually completing inventories. And we pay see higher demand in the legacy and mixed signal than what we are currently forecasting.

In terms of the pricing trends, though, per wafer, when you look at the 10% revenue decline, how much of that do you see coming interest volume and how much of that do you see coming from blended A.S.B.s?

I would say most of it is volume-dependent. In particular if you look at our technology, then if you have a legacy product, typically you don't see major price pressure because the switching cost is very high and people are not -- are not pushing very hard to change. It is different when you look at newer technologies than typically you see much faster price decline. Typically if you look at the curves, you see that prices are saturating after a while, and that all the technologies do not continue to go down.

Great. And if you could give us a general idea, a ballpark, what kind of premium price per wafer does your C.I.S. technology provide tower relative to standard .35 micron wafer?

We have to be careful. Definitely you get higher price per wafer but you also invest more in the wafer so I wouldn't say that when it comes to margins, you really make substantially better performance than what you get on standard wafers, but on the other hand, at the same time, what I can say is that prices on the standard technologies are more susceptible to pressures and you may see more changes. And so they are less stable with that respect. But -- and when it comes to prices, the prices are substantially higher than what you get from standard wafer. It can be 50% higher than standard wafer.

Great. I appreciate that color.

Our next question comes from Quinn Bolton. Please state your question.

I just wanted to follow-up on the microflash technology. I know that A.M.D. and Siphon semiconductor recently signed a joint development product to develop next generation of microflash and I think some of that may revolve around multi-bit per cell, sort of a quad bit per cell technology. To the extent that Siphon introduces new generations of microflash, what are Tower's rights to manufacture in your foundry the new generations of microflash? Thanks.

Okay. Thank you. Quinn, we have the rights to use the microflash patents, the key patents which are required to -- patents which are required to any of these derivatives. Now, you're right. Right now there is a development on quad bit per cell, which I believe is important for commodity products, okay.

If one can do quad per cell then of course he's much better off than what anybody can do today. I have to tell you that we looked into this very carefully, and we concluded that at least for the embedded applications, this is not very likely to happen, and it's not going to provide a significant advantage because basically, when you go to quad from two bit% to four bit%, you pay. Everything becomes more complicated.

The circuit technology becomes more complicated, and you pay in the module area. Now, that makes sense, or may make sense for very high density. If you go to 1 gigabit products, then that would make sense.

You are willing to pay the penalty, sort of, say, and give that area in order to get the very high density that you get from quad 2 per cell. But onto embedded, typically you still would talk about much smaller modules, and for this, the penalty in terms of the area is just going to be too high to pay.

Also in terms of product sensitivity, embedded requires robustness to work in different environments and different designs t has -- the model has to work on different kind of a logic designs, and therefore, it has to be much more robust than what you can do in commodity where you can spend all the time in the world to perfect your design. So all in all, we don't see any of these quad as practical for the embedded space.

With respect to area per bit, for embedded applications, what we also with the microflash is by far so much better than any other architecture today in the embedded space, that we feel right now that what we need to do is just focus on that direction. Let me even tell you that for the lower module density, we even are doing some work on the one bit per cell just in order to reduce even furthermore, to reduce the periphery area than you have a point where one bit of cell becoming more economic.

So there is -- for embedded space, the solutions and the directions are different than what people should be doing for the commodity space, and as you know, we are focusing on the embedded space area.

Right. Okay. Thank you.

Thank you.

Ladies and gentlemen, if there are no further questions, I will turn the conference back to Yoav Nissan-Cohen to conclude.

Okay. I will close the conference call. I would like to thank you for the participating in the call this morning. I look forward to continuing to share with you our progress in the future. Have a good day.

Ladies and gentlemen, this concludes our conference for today. Thank you all for participating and have a nice day. All parties may now disconnect.