使用警語:中文譯文來源為 Google 翻譯,僅供參考,實際內容請以英文原文為主
Operator
Welcome to Tower Semiconductor's second-quarter 2007 results conference call.
All participants are at present in a listen-only mode.
Following managements' statements, instructions will be given for the question-and-answer session.
As a reminder this conference is being recorded, August 15, 2007.
With us online today are Mr.
Russell Ellwanger, Tower's CEO and Mr.
Oren Shirazi, CFO.
I would like to hand over the call to Noit Levy, Director of Investor Relations and Communications.
Noit, please go ahead.
Noit Levi - IR
Thank you and welcome to Tower Semiconductor's conference call for the second quarter 2007.
Our agenda for today's call is as follows: Russell will start with remarks about the quarter's highlights.
Oren will then provide an analysis of the second-quarter financial results.
After managements' prepared remarks, we will begin the question-and-answer session.
Before we begin, I would like to remind you that some statements made during this call may be forward-looking and are subject to uncertainties and risk factors that could cause actual results to be different from those currently expected.
These uncertainties and risk factors are fully disclosed in our Form 20F and 6-K as well as filed with the SEC and the Israeli securities authority.
They are also available on our website as well in our Forms F-1 and F-3.
Tower assumes no obligation to update any such forward-looking statements.
Now I'd like to turn the call to our CEO, Russell Ellwanger.
Please go ahead.
Russell Ellwanger - CEO
Thank you, Noit.
Welcome and thank you very much for joining us today.
I am pleased to report that for the second-quarter of 2007, we reported record quarterly revenue of $57.1 million representing a 28% year-over-year growth and meeting our previously stated guidance.
Also we generated positive EBITDA for the seventh consecutive quarter and positive cash from operations for the third consecutive quarter.
We expect to continue to report positive results in EBITDA and cash flow from operations throughout 2007.
Revenue for the first half of 2007 was $113 million, representing a 40% growth when compared to the first half of 2006.
This is to be compared to the foundry industry weighted average decline of approximately 5%.
Driving our financial results is our continued focus to be aligned to our customer needs and the result in strong customer demand specifically within Fab 2.
While Fab 2 is experiencing very strong increase in demand, we are experiencing a softened demand in Fab 1.
Predominately as a result of this softening in Fab 1, which we see as short-term, we are guiding Q3 to a mid-range guidance of $58 million representing modest growth over Q2'07 and an approximate 12.5% year-over-year growth which according to current third-quarter guidance from other foundry participants significantly exceeds year-over-year foundry industry average.
With respect to Fab 1 demand, Fab 1's large customers serve specific segments of the consumer market.
These segments are subject to industry cycles, market share shifts as well as product generation cycles.
Due to these factors, several of the product families that Tower produces in Fab 1 are experiencing a reduction in demand.
We believe that this situation is temporary and that during Q4'07, we shall return to the high utilization rates that we experienced in Fab 1 for all of 2006 through Q1'07.
There will be a result of Tower increasing its market share with existing large customers as a result of multiple new tapeouts which we have shipped and then these tapeouts becoming qualified and reaching production volumes as well as the projected need for inventory replenishment of existing high-volume products.
As you may recall, we are organized into three product-lines along distinct business and technology segments.
I'd like now to share with you our market perspective and Tower's Q2 achievements and upcoming quarter activities for each of these product-line segments.
The CMOS product-line serves several market segments primarily in the consumer, general-purpose logic, and analog switches markets.
During Q2, we continued the steep ramp of the 0.13 micron volume production and we're shipping several thousands of wafers each month.
In Q3, we forecast to ship approximately 10,000 0.13 micron wafers and still our customers' demand exceeds our present capacity.
We will speak to the capacity expansion plans a bit later in this call.
As stated in prior quarters, we continued the volume manufacturing of customer products in the standard 0.16 micron process which provides a 10% linear shrink or 19% area reduction in comparison to the standard 0.18 micron process.
To help enable a Tier 1 customer in the area of mobile applications, we further developed our advanced 0.16 microns by providing a low leakage version that consumes about 20 times less current than the standard 0.16 platform.
The first product on this new platform is already taped out and we expect to ramp to high-volume production in the second half of 2007.
We also developed a low leakage variant of the standard 0.18 micron platform to address the need for portable application.
The first products will ramp in Q3'07.
In Fab 1, we released to production several voltage translators products that were designed and manufactured by Tower's team from data sheet through silicon verification on Tower's advanced 4.5 volt 0.35 micron process.
These products attain very fast propagation delays (approximately 2 nanoseconds), with up to 200 MilliHertz data rate, very low power consumption and a wide operating voltage range.
We expect very strong volume from this platform.
Through collaboration with the industry leader in ESD protection, we will prove 20KV ESD performance with very low capacitance on our advanced 0.35 micron process.
Several of our customers require such ambitious ESD protection performance.
In the CMOS image sensor market, Tower served primarily three segments, cell phone cameras, high-end cameras and dental and medical x-ray sensors.
This is a highly specialized, high to very high average wafer selling price segment.
The cell phone camera market is continuously growing and the supplier market is consolidating to several large IDMs.
Tower, however, has the unique position in this market due to superb technology especially in the Taiwanese and Chinese market.
Therefore we have seen nice growth of the CIS business in the cell phone camera market.
During the second quarter, we had a major technological breakthrough that enabled us to reduce the dark current by a factor of 20 as well as to drastically reduce the number of defected bright pixels and thereby attain excellent picture quality surpassing most all of our competitors and putting us on par with the leading IDM on pixel performance.
This achievement has allowed us to ramp to mass production with one of our major customers in China and to compete well within the Chinese market.
The high-end cameras consisting of digital single lens reflex for still image photography, cinematography, and high-end HD video, a 1080 pixel format, experienced nice growth for Tower.
These products require excellent performance and advanced technologies.
Tower, again, with its unique technology and especially with low dark current performance, is well positioned to serve them.
During the last quarter, we started our steady ramp to production with one of our major customers with really excellent results.
In the x-ray CMOS sensor market where Tower has a large market share in the dental x-ray segment, we have also seen steady growth and a technological shift to smaller pixels.
The higher resolutions are supported by our new development with stitching and pixel technologies in Fab 2 0.18 micron technology.
We see great opportunity in the medical market where very large sensors are needed.
Our CMOS sensor technology has performance and cost advantage over other technologies currently used in the medical market.
We see a trend of moving to digital imaging from the film x-ray which can bring substantial amounts of business to Tower.
Our stitching technology development in Fab2 allowed us to prototype three different new products with major customers and additional ones are lined for taping out their new products, mainly for dental x-ray in Fab 2, in Q3 and Q4 this year.
Looking into the third and fourth quarters of 2007, we see the following developments in our image sensor business.
We will continue shipping thousands of wafers to the cell phone market in China.
Our lead customer is projected to introduce more ambitious products based on our 2.2 micron pixel technology as well combined with excellent dark current performance.
Our approach of supplying our pixel IP to our selected customers is proving itself and we expect to see production of the 2.2 micron pixel in the fourth quarter of this year.
In the high-end market, we continued the steady ramp of very high-end video sensors, 12 mega-pixels, and in parallel are developing a future generation of pixels that will allow global shutter with very low noise.
In addition, we started a joint development program with one of our major customers for the Next Generation pixel development for high-end DSLR market.
In the x-ray market, we already had three tapeouts of new products in Fab 2, two of which utilize our unique stitching technology allowing us to produce dies much larger than the reticle field.
We're expecting at least three more products to tape out in the second half of this year which we expect will ramp to mass production in the first half of 2008.
Moving forward to the mixed signal and RF domains, where Tower specializes in RFCMOS, RFID and power management segments, the wireless market is growing fast with RF-CMOS overtaking Sage.
0.18 micron RFCMOS demand is very strong.
We observed some migration of 0.13 micron for lower cost.
We expect Tower's revenues from RF products to more than double in the second half of 2007 when compared to the first half.
During the last quarter, we continued to ship at rates of several thousands of wafers per month to the two largest RF customers in Fab 2; several additional customers are ramping nicely.
In the RFID domain, Tower has proven its leadership by being the sole supplier for a leader in this industry.
There is tremendous interest in RFID in China with several customers in the first stages of our customer design funnel.
This past quarter Tower delivered a keynote presentation at the 2007 ASID conference in China, acknowledging technical expertise in this emerging growth segment.
The power management market is very diverse and growing fast.
Tower is focusing on applications that require voltages below 55 volts that may be served by its extended 0.18 micron platform.
For Fab 1, Tower engaged in a new mode of operation of assuming responsibility for the product porting design.
This resulted in fast and smooth migration and facilitated prompt ramp to production.
Looking into the third and fourth quarters of '07, within this RF and mixed signal product-line, Tower will launch its 0.18 micron power management kit for breakdown voltage of about 55 volts or less in Q3 '07.
The kit includes state-of-the-art offering limos devices (on non-Epi wafers) with 23 mom mm which competes with the best-of-breed Epi based offerings of other foundries and provides the lower cost of being non-Epi.
The platform also offers a unique combination of three micron copper interconnect layer on our 0.18 aluminum back-end to allow for extremely low resistivity for power management application.
In the past year, Tower recognized the extreme importance of technology and products transfers from integrated device makers that have a major drive to outsource some or all of their products outside of their own product-lines.
Hence, we established a special product-line to focus on and work with customers in the area of IDM transfer.
The international rectifier program was one such program that we engaged with and executed on in 2006.
The program is progressing according to the agreed plan.
To date, all technical results meet or exceed all expectations and should result in production volume ramp in the fourth quarter of '07.
As observed from the above discussions, the demand for our products is increasing in Fab 2.
In Q2, Fab 2 shipped on the average about 17,000 wafers per month achieving in June a peak month with an overall shipment exceeding 21,000 wafers.
During Q3, we expect at least an additional 10% increase in shipments.
In order to support the growing demand of Fab 2 products, we are in the middle of efforts to increase the Fab 2 capacity.
The prior announced tool acquisition is almost complete.
We're currently at 23,000 wafer starts per month, actually having achieved in the present month over 24,000 wafer starts in Fab 2.
We're focused presently on increasing the capacity beyond the 24,000 and to do this by going after used tools within the market.
There is presently several 8-inch copper fabs that are downscaling or closing or converting to 300 millimeter.
This is putting into the market much lower cost advanced technologies tooling capability that we could bring into Tower at about one-third the price of new tools.
We're actively pursuing the acquisition of these tools for advanced technologies of 130 nanometer and 90 nanometer aiming at a capacity of above 30,000 wafer starts per month by mid 2008.
Our current customer lineup and product pipeline supports the demand for the increased capacity.
As a result of the tool low cost, we expect a high return on investment and an 18-month pay back period on the equipment.
We anticipate that the ongoing increase in capacity backed by present customer demand will result in significant increases in our sales, cash flow and overall operational results.
In order to fund the additional expansion, we have signed letters of intent with our lender Banks & Israel Corporation to provide credit lines up to $60 million to secure the funding for the purchase and in addition, we raised approximately $40 million of long-term bonds from Israeli institutions.
The financing approach is made with the target of not diluting our current shareholder equity.
In summary, driven by high utilization rates, ongoing increases in our wafer shipment capacity and the growth of our customer base in particular large customers, Tower has consistently reported revenue growth over the past two years which outpaced the foundry industry.
From a bigger picture perspective, Q2 2007 marked the two-year anniversary of the management team at Tower.
During this time, we tripled our revenue having posted eight consecutive quarters of revenue growth, seven quarters of positive EBITDA and three consecutive quarters of positive cash from operations.
In Q3, we reached 24,000 wafer starts in Fab 2 with additional funding at hand to continue the ramp to beyond 30,000 wafers per month of advanced technology node capacity with customer demand in place to consume the added capacity.
We hence target continued growth with a 2008 revenue target of above $300 million.
With that, I'll turn the call over to Oren.
Oren?
Oren Shirazi - CFO
Thank you, Russell, and hello everyone.
Let me begin by stating that similar to the last quarter we added non-GAAP results to our press release and to this analysis which excludes depreciation and amortization costs as well as stock-based compensation costs.
We believe this additional format will be useful for investors and shareholders to better analyze our financial metrics including our significantly improved growth and operating margin in recent years which excludes GAAP expenses.
These exclusions, which mainly relate to depreciation of Fab 2 which began operations over four years ago, are fully reconciled in the release itself.
For the second quarter ended June 30, 2007, we reported sales of $57.1 million, meeting our previously stated guidance and representing year-over-year growth of 28%.
For the first six months of 2007, sales totaled $112.7 million, representing 40% growth when compared to the first six months of 2006.
Non-GAAP gross profit for the second quarter of 2007 was $19.2 million representing a gross margin of 34%.
Non-GAAP operating profit for the second quarter of 2007 was $11.5 million, representing a 20% operating margin.
Non-GAAP profit for the first six months of 2007 was $40 million and non-GAAP operating profit for the first half of 2007 was $23.4 million.
In the second quarter of 2007, we improved our loss in accordance with GAAP by $9.2 million to $34.4 million or $0.28 per share.
This is compared to GAAP loss of $43.6 million or $0.55 per share for the second quarter of 2006.
For the third consecutive quarter we achieved positive cash flow from operating activities and for the seventh consecutive quarter, we achieved positive EBITDA of approximately $13 million.
This means that excluding depreciation and amortization, we have a positive P&L.
Other indicators for the efficiency of our cost structure are the following metrics: The ratio of our total non-GAAP operating expenses to our sales was reduced year-over-year from 18% to 13%.
For the second quarter of 2007, our operating expenses on a non-GAAP basis were $7.6 million, representing 13% of the revenue.
This is compared to $7.8 million or 18% of the revenue in the second quarter of 2006.
Turning to the balance sheet, we are starting the third quarter of 2007 with approximately $60 million of cash on hand including the proceeds from our recently announced long-term bond fund raising.
Our $60 million in cash is compared favorably to $41 million at the end of 2006.
Shareholder's equity was a positive $92 million in the end of the second quarter as compared to negative number in the end of June 2006.
Our current ratio which is our current assets divided by our current liabilities as of the end of the second quarter of 2007 was 1.30 versus 1.36 in the end of 2006.
And now I will discuss sales by major customers.
During the first half of 2007, we continued to expand our customer base.
We had total of over 50 customers to whom we ship wafers, out of which seven customers that each contributed at least 4% of our consolidated sales, as compared to having only three such customers in the comparable period in 2005.
Our leading and largest customer in 2007 continues to be SanDisk Corporation to whom we shipped 0.18 and 0.13 micron wafers.
And in addition we had six customers in the first half of 2007 to whom we recognized revenues of between 4% to 12% of our consolidated revenue.
In Fab 1, it was Vishay Siliconix and ON-Semiconductor and in Fab 2, Zoran Corporation, Macronix International, including its affiliate , Sitel and Atheros Communications.
Now we would be happy to take your questions.
Operator
(OPERATOR INSTRUCTIONS) Ramesh Misra, C.E.
Unterberg Tobin.
Ramesh Misra - Analyst
Good evening, gentlemen.
My first question was in regards to pricing for 0.13 wafers, pricing and profitability.
Roughly can you talk about how that compares to product out of Fab 1 and also out of your 0.18 micron products?
Russell Ellwanger - CEO
One more time, how it compares to products out of --?
Ramesh Misra - Analyst
Fab 1 in terms of profitability and also in compared to 0.18 micron products.
Russell Ellwanger - CEO
Why doesn't Oren speak to the gross margin there?
Ramesh Misra - Analyst
Okay.
Oren Shirazi - CFO
So basically we didn't refer directly to selling prices of the wafer but generally in the market the 0.13 micron wafers are stable at around $1100, $1200 per wafer and the gross margin on that business approximately 75%.
I mean selling price over the variable expenses is 75% gross margin which is of course better than margins in Fab 1 or in the 0.18 which generally are between 50% to 70%.
Ramesh Misra - Analyst
Okay.
In regards to the capacity expansion, of this additional 6000 wafer starts per month, can you provide some kind of breakout how much will be additional 0.13, how much will be 90 nanometer and what is the plan on 90 nanometers are beyond?
Russell Ellwanger - CEO
Everything that we buy would initially be started at 0.13.
So the tool set with 0.13 and 90 is predominantly the same with the exception that at the 90 we would need to add to it the low k dielectric.
But the initial tool that will come in will be started up as 0.13 as we have present very strong demand at 0.13.
And that is where we would be bringing it up at.
It's not 6000 though.
The initial target of what we're looking at right now is at least I would say it's more on the order of 8000.
I know that we -- I said in there greater than 30 but we'd be looking at bringing in probably about 8000 wafer starts at a minimum.
Ramesh Misra - Analyst
Okay.
And roughly when do you expect the sales equipment to start showing up in Fab 2?
Russell Ellwanger - CEO
We are presently in negotiations with suppliers of the used equipment and what our target is is to be bringing it in really in the first half of Q4.
What we'd like to target is to have installed capacity on the order of several thousands of wafers going into the additional of 0.13 going into the first quarter of '08.
Ramesh Misra - Analyst
Okay.
And so by Q2 of '08, do you expect your initial pilot production runs or at that point you actually expect production wafers to be running?
Russell Ellwanger - CEO
In Q1 we'd expect to already start incremental so it's not the 6000 or 8000 wafer starts, whatever, the actual amount turns out to be, it's not that it would be digital.
It will be ramped up tool by tool in areas of about 2000 wafer starts.
But I would expect that in Q1 we'd already start seeing additional revenue from the capacity increase that we'd be bringing in in Q4.
Ramesh Misra - Analyst
Okay.
In terms of --
Russell Ellwanger - CEO
You would expect that in Q3 -- the end of Q3 we would see the full impact of it as far as production qualified wafer starts.
Ramesh Misra - Analyst
Okay.
So in terms of profitability, Russell, what does this do to the timeline for that?
Russell Ellwanger - CEO
It probably pushes it off about two quarters whereas we might have targeted previously the end of 2008.
The fact of the incremental additional depreciation and that you have on the books one to two quarters of depreciation before you'd see the revenue, so it would push it off probably to the beginning of 2009.
But the degree of profitability obviously will be higher because of having the increase in the depreciation.
And the return on the depreciation is quite good.
The tool costs as we had mentioned for the -- what we're targeting from several of the sources of the used tools is on the order of 30% to 35% of that of a new tool.
So on whatever capacity is that we would be installing, the period of return is about 18 months from the first production run.
Ramesh Misra - Analyst
Okay.
In regards to the competitive landscape --
Russell Ellwanger - CEO
Ramesh, the second part of that -- depending on ultimately how much we bring on board, if we bring on 8000 wafer starts, what we would see is somewhere close to a doubling of the cash flow that we presently have.
Ramesh Misra - Analyst
Okay.
In regards to the competitive landscape, how would you view your 130 nanometer product portfolio versus the rest of the products?
So another way to put it, is that there's lots of capacity on the 0.13 and does increasing your efforts on 0.13 and smaller geometries actually put you on a head-on collision course with the larger players in the market?
Russell Ellwanger - CEO
What we mentioned in the call is that as well as from the 15 to 24,000 and now by adding additional 0.13 capacity, it's customer driven.
So if it wasn't purely customer driven, I would say yes we would be competing with some of the larger foundries at the advanced technology node to try to get this.
But the fact that it's driven through customer contracts that we have or will shortly have, I would say that it's a very secure avenue for Tower to go with as you have a higher selling price.
And the tool costs of bringing the used tools at this technology node is not so different than bringing used tools of 0.18 micron technology.
So the return on the capital investment is substantially better because of the ASP at the wafer.
There's two ways that we can get there.
When I say ASPs, one is by going to the more advanced technology node on a core CMOS platform; the other is through the specialization for example that we have in the 0.18 micron when it comes to the image sensor products and some of the power management products that we are driving.
So I think what we're doing here with the capacity expansion has a very good return but really driven from the fact that its lined up with customers.
Ramesh Misra - Analyst
Okay.
So can you talk about these customer commitments?
Over a longer-term, what kind of a lock do you have with these customers that are driving this capacity expansion?
Russell Ellwanger - CEO
What type of a lock?
So the capacity expansion that we're looking at right now is not off of a take-or-pay agreement, if that is what your question is.
It's off of customers that we have strong relationships with that have given us within their forecast that if we have X amount of wafer incremental capacity, they would buy that wafer capacity from us because of [pleasure] with the performance that we have with them to date.
So that is what the lock is but I think that that is actually a much stronger lock than a contract because ultimately the only thing that you can really go on in any business relationship is trust between supplier and customer.
Ramesh Misra - Analyst
Right.
Russell Ellwanger - CEO
Do I think that there is every strong lock in it?
Yes.
But is the lock is something that is on paper?
No.
And would not expect it's going to get there.
Now that is on the immediate capacity.
We have several very, very interesting projects that maybe will be being announced within the next month that could maybe also clear up why it is that we are focused on some 0.13 capacity expansion.
But that is not yet released.
Ramesh Misra - Analyst
Okay.
And the processes that you will be -- you are potentially looking at going forward, like your Fab 1 processors, are these going to be specialized?
Are these going to be tailored to the customer that makes it difficult for them to transfer to other customers?
Russell Ellwanger - CEO
There is really two avenues that make it very difficult for customers to transfer to somebody else.
One is as you say, technology platforms that are very tailored to your customer that are very unique to the customer.
A lot of that is what we have again, within the image sensor portfolio, the high-end video cameras, the high-end -- just cellphone cameras to where the customers used our pixel IP.
They have very, very high barriers for exit.
The other area where there is a very high barrier for exit is the IDM transfer itself to where the customers put quite a bit of energy into working with you to take on their process flow.
In that case, the customer is an integrated device maker.
And the IDMs themselves also have a very high barrier for exit because there's tremendous costs involved and commitments for them to do the transfer with you.
Once you run their flow in your fab, it is very difficult for them to go elsewhere with that same flow and it's difficult for them to do that in multiple, multiple different foundry sites just because of the cost of trying to transfer that process.
So in the case of the IDM, I think there's also a very, very high barrier for exit.
And the nice things about IDMs is that their long-term contractual agreements on pricing so you really know what you are going to be at on a volume.
Ramesh Misra - Analyst
Okay.
Russell Ellwanger - CEO
Does that answer your question, Ramesh?
Ramesh Misra - Analyst
Yes it does, Russell, thanks very much.
I have one last question and then I will get off and let others get in as well.
In Fab 1, Oren, you mentioned Vishay and ON are your largest customers right now.
Where does International Rectifier figure on that list and where do you anticipate them to kind of move in to top [three] position in the next quarters?
Russell Ellwanger - CEO
So International Rectifier is a Fab 2 customer, it's not a Fab 1.
It's a 0.35 micron flow more or less that we brought into the Fab 2 process line.
For IR, had mentioned that that should be going into volume production in Q4 and we would see them being in the top several handfuls of customers as far as revenue in 2008.
Ramesh Misra - Analyst
Okay.
Thanks very much Russell and Oren.
Russell Ellwanger - CEO
Thank you.
Operator
Robert Maire, Needham & Company.
Robert Maire - Analyst
A couple of questions.
In ramping up the additional capacity you had mentioned about being able to get secondhand tools on the market and such.
It seems like a fair amount of memory capacity though not necessarily copper and 8-inch maybe coming available soon.
Have you identified sources of the equipment and can you give us a little more granularity as to when you see the equipment rolling in and time to ramp it up?
And I thought I heard beginning of next year but perhaps you could give us a little more granularity on where you are in that process?
Russell Ellwanger - CEO
Sure.
To begin with you are absolutely correct, there is a quite a bit of 8-inch memory capacity that is coming into the used tool market.
For the front end of the process line, the 8-inch tools -- it doesn't matter if it's a copper flow or if it's a memory flow, an aluminum flow, the front end tools for the advanced technology nodes are pretty much compatible in the same including the lithography.
So for that -- all of the capacity that is coming into the market is beneficial for Tower or for anyone else that wants to expand capacity in 8-inch.
The back end right now as far as copper, there is two specific large volume manufacturers that have put their copper back end tools into the market at 8-inch.
And both of these are very large manufacturers.
Where are we within the process?
Actually, really as we speak, the VP for which procurement rolls up under is in the U.S.
in negotiations with the tool suppliers of these use tools.
What we expect to have is, as I had mentioned, about 2000 wafer start capacity installed and qualified for the end of Q4, an additional 2000 for copper.
Now, that may be very aggressive, but certainly installed and well underway of being qualified.
And we would expect to see incremental revenue from that in Q1 of 2008.
And then we would believe that the entire package will be fully installed and in route to qualification by the end of Q2 2008.
Robert Maire - Analyst
So in terms of the purchasing effort, you are just south with the negotiations and once that's concluded, you would probably be in receipt of the equipment a month or two down the road?
Russell Ellwanger - CEO
Some of it will be staged over some period of time.
One of the companies that is moving away from the 8-inch line is defacilitizing piece by piece over several quarters.
And the other is coming in more or less in one shot.
Robert Maire - Analyst
Are there any other facility changes that you have to make in terms of cleanroom space or other stuff, or it is that essentially set to go currently?
Russell Ellwanger - CEO
The cleanroom space itself is there; that we don't have to do any work to.
There is certainly some facilities work that you always have to do if you bring in new tools, but it's not a prohibitive amount of money.
Without doing any upgrades, we can take on -- any additional filtering, we can take on about another 5000 or 6000 wafer starts.
And beyond that, we have some minimal amount of money that has to be spent on facilities upgrade.
But it is upgrading systems; it's not scrubber systems etc.
It's not making more cleanroom space itself.
Robert Maire - Analyst
Understood.
In terms of the demand side, it seems that just generally in terms of consumer and wireless devices and cellphones and such that inventory remains relatively low, and demand is in pretty decent shape, and most of the larger foundries are reporting fairly significant utilization rates.
Do you have any guidance from your customer base going out beyond Q3 and Q4 or a sense as to how sustainable the current demand cycle is or if some of this is obviously seasonally driven, or any further input from your customers as to longer-term views?
Russell Ellwanger - CEO
We certainly do.
I mean most of our customers will give us their one-year plan, their 18-month plan.
How close do they hit the plans, that is -- some are very good at it; some are less good at it.
Certainly we've seen very, very big growth with our large customers, and probably one of the benefits that Tower has been able to realize over the past let's say six to eight quarters is that some of these large customers that we've engaged with are very large.
And even if the industry is weak, we have had the ability to grow market share.
So that is why our Q4, Q1, Q2 of this year, we're very strong compared to the industry.
It was, let's say, half to two-thirds the fact that we were growing market share with customers, and the other 40% of the business was because we are engaged in noncyclical markets as well.
Robert Maire - Analyst
And that's --
Russell Ellwanger - CEO
Did that answer your question, Robert?
Robert Maire - Analyst
Yes it does, that was great.
Excellent.
And in terms of pricing out there, any comments as to ASPs?
I'm assuming because of your answer that ASPs have been relatively steady and given the utilization rates, any uptick in ASPs or any other changes that you've seen over the last quarter?
Russell Ellwanger - CEO
We do not see and uptick in ASP.
Since I have been at Tower, I've never seen there be an increase in ASP independent of utilization or capacity needs.
So I think ASP will always decline.
It's a question of getting into the technology node at the right time so that you can continually reduce your manufacturing costs at a rate faster than ASP reduction.
Whoever is the market leader of entering any technology nodes will realize huge decreases in ASP.
Now they have the initial benefit of an 18-month very, very high ASP but most ASPs if you look at any of the charts no matter what the technology node is, they all more or less over time will go down to the same asymptote -- you know, within a 15%, 20% band and --
Robert Maire - Analyst
Yes, I guess I shouldn't say increasing but depending upon the season we've seen utilization of ASP (multiple speakers)
Russell Ellwanger - CEO
The rate of decrease does depend on the degree of utilization in the market, right?
If you have people being put on allocation then they don't really drive for lower ASP.
So -- but within our models, I will say that we have seen some decrease in ASP within Fab 1.
Our ASP reduction has been very stable and that deals with the fact that we have a good amount of specialty products and customers that are very tight into the fab.
So in Fab 1, let's say we averaged over the past eight years a 2% or 3% ASP decline per year, we've been able to reduce the operational costs at a rate higher than the ASP decline.
We are seeing at 0.18 an ASP decline.
If I look at 18 month ago versus now, there has been a not overly significant but certainly not insignificant decline in ASPs at 0.18.
But where we have I think the benefit that comes into this is the 0.18 platform continually drives into more advanced technologies that demand very, very high wafer prices.
Now we have some wafers at 0.18 micron that are selling for $2600 and that deals with having very specialized products and that does help the average ASP at the fab.
So no matter where you are within the technology lifetime cycle, the more specialty you can add to the platform the more you can move away from that -- approaching the asymptote on ASP decline.
Robert Maire - Analyst
Okay and one last question if I might.
In terms of pixel count on image sensors, we've obviously seen a fairly steady increase in both in cellphones and other devices.
Can you tell us where you are currently in terms of your average device that is going out the door and where (inaudible) going and how does that impact pricing?
Russell Ellwanger - CEO
As far as cellphone right now, we still sell a reasonable amount of VGA but we also have a nice split of 3 megapixel.
But even the VGA is high-end as far as performance.
It is very low dark current.
So we have very nice ASPs there.
As far as the high-end cameras, it is 12 megapixel that we are shipping now.
Robert Maire - Analyst
Okay.
So average continues to go up in a relatively constant basis here?
Russell Ellwanger - CEO
On the high end, yes.
And on the cellphone I think that 3 megapixel right now is becoming the standard of the good manufacturers that they are more or less all starting to produce a 3 megapixel and there is a big demand at that.
But VGA is still a very strong demand.
Most cellphones now have two cameras.
They have the VGA facing the user that is being used for the videoconferencing and then you have the 3 megapixel that is being used for the still frame pictures.
Robert Maire - Analyst
And how has that impacted ASPs?
It seems almost as if megapixel count has gone up faster than ASPs have declined on a per bid basis.
Is that accurate or not really?
Russell Ellwanger - CEO
We see a premium on the 3 megapixel versus VGA but both are still selling for very good prices.
But I wouldn't say that you see a ratio.
If you go from VGA to a 1.8 or a 2.2 to a 3, it doesn't ratio as far as the ASP.
You do get a premium for having more advanced megapixel but it doesn't ratio directly.
But VGA, if it's a very good VGA chip it still demands a very nice price.
Robert Maire - Analyst
Okay, great.
Thank you very much.
Operator
Edwin Mok, Needham & Company.
Edwin Mok - Analyst
Hi, Russell.
Thanks for taking my question.
There was a report that SanDisk invested in you guys during the quarter.
Any comments on that?
Russell Ellwanger - CEO
There were rumors that SanDisk was going to be investing, there are still are rumors that SanDisk would be investing in Tower.
I don't think there was any report that they actually did invest.
Because that is very easy to track.
But we had stated previously and I'd state the same now, SanDisk is really -- SanDisk is every good customer for Tower.
We have very strong relationships with them.
We value them as a customer and if and when this becomes public, everyone will be the first to know.
Until then, we really don't have any comment.
Edwin Mok - Analyst
Great.
Thanks.
In terms of SanDisk, my understanding is you have been using 0.13 micron to make batch control for them, is that correct?
Or are you guys looking for other option (inaudible)?
Russell Ellwanger - CEO
I'm sorry, I didn't hear the second part of the question please.
Edwin Mok - Analyst
Sorry, the second part of the question, are you guys looking at other options to -- with SanDisk besides flash control?
Russell Ellwanger - CEO
No, everything we make with SanDisk are controllers and we make 0.18 and put 0.13 controllers for them.
Or we make controllers at 0.18 technology not [n 0.13] I should say.
Edwin Mok - Analyst
Great, that is all I have.
Thanks.
Operator
(OPERATOR INSTRUCTIONS) (inaudible) Lehman Brothers.
Unidentified Participant
Thanks for taking my question.
I had a quick question on the gross margins in Q2.
It seems when I look at your results it seems that the gross margin went down from 38% to 34%.
Can you walk me through what caused the decline especially given that Fab 1 has lower incremental gross margin?
And it's my understanding is that Q2 was skewed more -- the revenues in Q2 more of the revenues in Q2 were coming from Fab 2 than from Fab 1.
Thanks.
Russell Ellwanger - CEO
Oren?
Oren Shirazi - CFO
Yes correct, basically as Russell mentioned, we had a very good quarter in Fab2 and in Fab 1 we had a slight decrease in utilization where in our industry both fabs are characterized with the fact that we have very low variable expenses.
This is why mentioned before that we have between [50]% to 70% margins on any incremental revenue.
So when we had this softening in Fab 1, it reduced the utilization rate for Fab 1 and that is why it caused this reduction, the slight reduction that you mentioned.
Still this 34% is of course much better than we used to have in the past which was a negative numbers or last year it was around 20% gross margins, so this 34% is still very nice.
Also in Q1, we had some specific sales with a high gross margin, so the 38% was indeed a specific achievement.
But the 34% looks very much reasonable now.
Unidentified Participant
Okay, great.
Operator
Thank you.
There are no further questions at this time.
Mr.
Ellwanger, would you like to make a concluding statement?
Russell Ellwanger - CEO
Sure.
So firstly, thank you all very much for participating on the conference call.
Overall our long-term outlook is stronger than it was a quarter ago with substantial large customer demand, new projects that have both been announced and will be announced shortly and a ramp up plan that we just presented.
We look very forward to meeting with you, speaking with you individually.
We will be presenting at the Citigroup technology conference in New York on September 6 and invite you all to attend.
And as I mentioned, we look very forward to sitting down with you one on one at that time if you have further questions or any time in the interim, please contact us and we'd love to get into a dialog.
So, thank you again very much for the interest in Tower and for your support.
Operator
Thank you.
This concludes Tower Semiconductor's second-quarter 2007 results conference call.
Thank you for your participation.
You may go ahead and disconnect.