高塔半導體 (TSEM) 2019 Q1 法說會逐字稿

Ladies and gentlemen, thank you for standing by.

Welcome to TowerJazz First Quarter 2019 Results Conference Call.

(Operator Instructions) As a reminder, this conference is being recorded, May 15, 2019.

Joining us today are Mr. Russell Ellwanger, TowerJazz's CEO; and Mr. Oren Shirazi, CFO.

I would now like to turn the conference over to Ms. Noit Levi, Vice President of Investor Relations and Corporate Communications.

Ms. Levi, please go ahead.

Thank you, and welcome to TowerJazz financial results conference call for the first quarter of 2019.

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 20-F, F-4, F-3 and 6-K filed with the Securities and Exchange Commission as well as filings with the Israeli Securities Authority.

They are also available on our website.

TowerJazz assumes no obligation to update any such forward-looking statements.

Please note that the first quarter of 2019 financial results have been prepared in accordance with U.S. GAAP.

The financial tables and data in today's earnings release and in this earnings call also include certain adjusted financial information that may be considered non-GAAP financial measures under Regulation G and related reporting requirements as established with the Securities and Exchange Commission.

The financial tables include a full explanation of these measures and the reconciliation of these non-GAAP measures to the GAAP financial measures.

Now I'd like to turn the call to our CEO, Mr. Russell Ellwanger.

Russell, please go ahead.

Thank you, Noit.

Welcome to our first quarter of 2019 results conference call, and thank you for your interest in our business.

The revenue for the first quarter of 2019 was at our guidance at $310 million.

EBITDA was $79 million and net profit was $26 million, with our cash from operations for the quarter being $75 million, of which $33 million was free cash flow.

This profit and cash generation continue to strengthen our balance sheet, with record shareholders' equity of almost $1.3 billion remain contributing to be in the strongest financial situation we have ever been in with open doors that we are exploring, which were not available in years past.

We look forward to further leveraging our strength and potential to increase value at TowerJazz that will last.

Looking ahead through 2019, the semiconductor market remains soft.

Macroeconomic uncertainties have led our customers to drive tighter inventory management.

In particular, we have seen a pullback in high-end optical for data centers, some families of discrete components and also for industrial vision sensors used in high-volume manufacturing lines.

However, the indications we see in the broad market as well as from specific customers are for a stronger second half.

We see a broader set of growth drivers than in previous cycles.

This is especially true for the analog sectors of the semiconductor market that we are focused on with upcoming drivers, including the global 5G rollout with the accompanying increased demand of wireless and infrastructure content; ongoing increases in automotive analog content, including sensors, sensor systems and battery management; along with IoT and artificial intelligence application.

Therefore, while we remain short-term cautious, we are optimistic that TowerJazz will emerge from the current market correction, not only in the strongest financial position it has ever been in but exceptionally well positioned for accelerated and sustained growth on both the top and the bottom line.

I'd like to provide you with some color on the activities that our various business units are undergoing, focused on trends and our related offerings to ensure future growth.

Our business units continue to develop unique and disruptive technologies, supporting market and customers' needs.

Looking at our RF business.

In mobile, and specifically for RFSOI, we are experiencing both increased order rates and expedite demand for existing 200-millimeter production parts, along with new parts on advanced flows of 200-millimeter and additionally, an aggressive ramp of new products in our 300-millimeter factory.

To support the strong 300-millimeter RFSOI ramp and ensure smooth growth well into next year, we have made our first investment to relieve bottleneck capacity constraints in our Uozu 300-millimeter factory.

In addition to supporting growth in RFSOI, this will enable growth in our leading 65-nanometer BCD power and image sensor technologies for which we have strong design wins.

Back to the RF mobile, we continue to invest in breakthrough technologies that could one day augment our RFSOI offering such as MEMS with our continued strong partnership with Cavendish Kinetics and new material systems that promise revolutionary RF switch performance.

Our infrastructure business, largely dominated by silicon germanium products for data center optical fiber connections, continues to suffer from an industry-wide inventory correction attributed to expectations for higher tariffs that caused multiple over-ordering of components last year that are now being burned off.

In some cases, 100G product purchases are being delayed in the hope that 400G will arrive quickly enough and become the better solution.

And also, a delayed renewed investment seems to be related to the release of next-generation data center CPUs from the 2 leaders, which are sampling now and should ramp to volume production in the second half of 2019.

It is generally expected that this will have run its course within the second half of 2019 and the optical component market will then achieve a double-digit CAGR in the coming years.

The underlying fundamentals of our business remain very strong with excellent market share estimated at greater than 60% of optical fiber [floors] and long-term growth prospects unaltered as data traffic is expected to benefit from growth in cloud computing and 5G wireless.

Today, these optics connections are made at speeds of 100 gigabit per second.

In addition to looking forward to recovering this 100G product line, we are winning new designs in future 400 gigabit per second systems.

It is generally expected that the 400G systems will start deploying soon and will hit large volumes over the next couple of years.

Within 400G, in addition to silicon germanium content, we anticipate ramping our silicon photonic platform to high volumes.

Between short-term recovery at the 100G market and the eventual increased TAM of the 400G with both silicon germanium and SiPho products, remain very bullish on the prospect of this market segment to both top and bottom line growth as silicon germanium and silicon photonic represents some of the best margin wafers in the company.

Our power business continues to see good revenue stream but more importantly, an accelerated pipeline of design activity in the differentiated areas of high voltage and our 65-nanometer BCD flow.

The 65-nanometer BCD in our 300-millimeter factory was our first 65-nanometer platform that targets low-voltage power with best-in-class figures of merit.

This platform is attracting a large amount of prototyping activity along with early production volumes.

In high voltage, we released this quarter our first design kit for 140-volt technology meant to address the needs of the fast-growing automotive battery management market for electric vehicles.

Although automotive as a whole is expected to be flat or maybe have a slight decrease in 2019, electric cars remain strong and are expected to grow, especially in China where Q1 year-over-year revenue more than doubled in EV sales.

We are experiencing a strong ramp with one family of parts used for power management with the lithium-ion battery stacks.

This 140-volt technology is also used as gate drivers for a wide variety of application.

The technology is unique in the foundry industry, achieving a high voltage rating without the use of more expensive silicon-on-insulator substrates.

For higher voltages, 200 volts and beyond, we have an SOI technology now having entered volume production, and we anticipate developing further high-voltage nodes on both volt and SOI wafers to respond to market dynamics.

As mentioned at the onset, we are seeing decreases in certain families of discrete components.

However, our business remains strong with additional share increases and very strong customer relationships and long-term engagements.

We announced an extension of our long-term partnership with Vishay-Siliconix, expanding into next-generation automotive platforms, extremely important due to the increase of electric content of automotive.

And we received a supplier excellence award from Infineon with the highest supplier score for technical customer support, delivery, collaboration and commitment.

Moving into the sensors business unit, which includes both our visual CMOS image sensors and nonvisual sensors.

Within the image sensor activities, our focus remains, firstly, the facial recognition sensor market primarily in the mobile segment, which is mainly driven by mobile banking and payment applications; secondly, medical and dental market segments where the served market size continues to increase as x-ray image intensifier tubes and flat panels are replaced by large stitched die, including paneled single die per wafer CMOS sensors; thirdly, high-end imaging, namely high-end photography and also small pixel global shutter for industrial.

Many applications in these markets are expanding toward differentiated backside illumination and wafer stacking technologies.

In all of the above, we continue to serve market leader existing customers for present state-of-the-art market demands as well as customer partner developments with existing and additional new powerful customers, driving breakthrough technologies to capture, for example, among others, the mobile facial recognition platforms for the many suppliers moving past the fingerprint sensors to 3D biometric recognition.

In the non-imaging sensor market, major growth is driven by the IoT market, which requires more and more sophisticated sensors at the edge.

We have developed a variety of sensor technologies, including highly accurate and high-range temperature sensors, high sensitivity as well as high-temperature magnetic sensors, radiation sensors, radon detection sensors, UV sensors and also gas and humidity sensors.

We are now actively offering these technologies to customers.

Our recent investment in AIStorm with proprietary technology combined with our analog building blocks provides a breakthrough analog artificial intelligence solution with low-power and low-cost artificial intelligence to edge devices.

Our plan is eventually to combine this AI solution with our sensors, enabling smart edge devices for diversified market application.

In terms of utilization for our various fabs during the first quarter of 2019, we saw the following utilization rates: in Migdal Haemek, Israel, Fab 1, our systems factory was 84% compared with 90% in the prior quarter.

We were recently awarded a significant high-volume discrete product win used in a high-end communication platform that should start to ramp and feed this factory in 2020 and beyond and which continues to prove the longevity of our analog model.

Fab 2, our 8-inch factory, was 70% compared to 76% in the previous quarter.

However, there is a strong ramp that has already started at the end of the first quarter that should bring the utilization up during the second quarter somewhere between 80% to 85%.

Newport Beach California, Fab 3, another 8-inch factory, was at about 80% utilization versus 90% in the prior quarter.

This is mainly due to some decreases in data center demand.

Our San Antonio factory, Fab 9, was about 50% utilization, a 5 point reduction from the previous quarter mainly due to reduction in several discrete products.

Blended TPSCo factories had an average of about 50% utilization.

8-inch utilization decreased due to market softness, and 12-inch foundry utilization continued to increase due to the high demand of our 300-millimeter advanced platforms.

We have invested to increase capacity, relieving capacity constraints of certain bottleneck tools specific to our third-party flows, and we are in the evaluation process of an additional substantial photo layer increase in that factory.

Finally, with regard to our TowerJazz Panasonic Semiconductor partnership and our long-term supply agreements with Panasonic Semiconductor, we announced an extension of our previous business partnership with Panasonic Semiconductor Solutions, PSCS.

Under the agreement, PSCS will continue to utilize TPSCo's 3 manufacturing facilities in Japan for its semiconductor business under a new and adjusted price structure.

This will result in planned revenue reduction according to present run rates from Panasonic Semi of approximately $20 million per quarter, with the low revenue levels and margins targeted to be compensated through efficiencies and cost reduction activities as well as third-party revenue growth, including the present strong 300-millimeter utilization ramp.

Over the past 5 years and now looking ahead, this partnership has been and continues to be a win-win proposition for all involved, for TowerJazz, for Panasonic Semiconductor Systems, for TPSCo and for our many important customers.

Looking ahead into the second quarter, we are guiding $306 million with a range of plus or minus 5%.

Considering the approximate $20 million announced reduction in Panasonic revenue for the second quarter due to the new contractual pricing, this midrange revenue guidance represents about 10% of sequential organic growth, which we define as total revenue excluding revenues from Panasonic and the TPSCo fabs and revenues from Maxim and the San Antonio fab.

Thank you.

With that, I'd like to turn the call over to our CFO, Mr. Oren Shirazi.

Oren?

Thank you, Russell, and welcome, everyone.

Thank you for joining us today.

I will start by providing P&L highlights for the first quarter of 2019 and then discuss our balance sheet.

Revenues were $310 million.

Net profit was $26 million.

Diluted GAAP earnings per share was $0.25.

Adjusted non-GAAP earnings per share was $0.30.

And cash from operating activities was $75 million in the first quarter ended March 31, 2019.

Balance sheet as of March 31, 2019, was stronger than ever with a record shareholders' equity of $1.27 billion, a record cash position and a strong 4.5x current ratio.

Net profit for the first quarter of 2019 of $26.2 million is approximately $100,000 better than the net profit in the first quarter of 2018 despite revenues being $2.6 million lower mainly due to $4.5 million better financing income net.

Diluted earnings per share was $0.25 for the first quarter of 2019 as compared to $0.26 in the first quarter of 2018.

The $0.01 lower EPS, despite the $100,000 higher net profit, is due to the previously reported full conversion of Jazz bond into 6 million ordinary shares, which occurred during 2018.

As I will discuss later, the fully diluted share count remains unchanged at 108 million.

We'll now provide the cash flow highlights for the first quarter and the balance sheet analysis as of March 31, 2019.

During the Q1 '19, the company generated $75 million of cash from operations and invested $42 million in fixed assets.

Comparing to the first quarter of 2018, cash from operations was $75 million and investments in fixed assets were $40 million.

Net current assets, as presented on our balance sheet, namely current assets less current liabilities, was $781 million as of March 31, 2019, resulting in a current ratio of 4.5x as compared to $592 million net current assets and current ratio of 2.9x as of March 31, 2018, and as compared to $784 million net current assets and current ratio of 4.8x as of December 31, 2018.

Short-term and long-term debt presented in the balance sheet as of March 31, 2019, have increased from the following 2 reasons: one, the implementation, as required by GAAP, of accounting standard update ASU No.

2016-02, Leases effective January 1, 2019, with regards to lease right-of-use assets and lease liabilities, which implementation has also increased fixed assets; and two, first principal payment of $18 million scheduled to be paid in March 2020 for the Series G bonds, which were issued back in 2016, has been recorded in short-term debt for the first time.

As of March 31, 2019, our total cash, short-term deposits and marketable securities exceeded our total debt, as presented in our balance sheet, by the amount of $365 million, which is $9 million lower than the $374 million as of December 31, 2018, mainly due to $37 million lease contract value, which was included in short-term and long-term debt for the first time, offset by $33 million free cash flow generation.

Shareholders' equity as of March 31, 2019, reached a record of $1.27 billion as compared to $1.24 billion as of December 31, 2018.

On May 7, 2019, Standard & Poor's Ma'alot, an Israeli rating company which is fully owned by S&P Global Ratings, completed its annual business and financial review of the company and its Series G bonds and reaffirmed its ratings to be ilAA- with a stable horizon.

Moving on to elaborate on the tax line in the P&L, I would like to describe our applicable and effective tax rates.

Our U.S. affiliate, Jazz and TowerJazz Texas, which own our Newport Beach facility and San Antonio fab, respectively, were taxed at a 21% rate starting in 2018 following the U.S. tax reform as compared to 35% prior to that reform.

TPSCo's profits from the Japan operations are subject to an approximate 32% tax rate.

Our profits in Israel from Fab 1 and Fab 2 operations, while subject to a 7.5% statutory tax rate, are not expected to result in any tax payments in Israel for the foreseeable future since we have more than $1 billion in historical NOLs still to be utilized, which can be carried forward indefinitely.

Considering these, and since we have certain tax exemptions, discounts and credits, our all-in worldwide weighted average effective tax rate was 6% for the first quarter of 2019 and 4% for the year ended 2018.

(inaudible) activities.

In relation to the euro, we still have almost 0 business in euros, hence no exposure to the euro.

In relation to Japanese yen, since the majority of portion of TPSCo's revenue is denominated in yen and the vast majority of TPSCo's costs are in yen, we have a natural hedge over most of our Japanese business and operations.

In order to mitigate the remaining yen exposure, we execute zero-cost cylinder hedging transactions.

These zero-cost cylinder transactions hedge all currency fluctuations to be contained in a narrow range as compared to the spot exchange rate.

Hence, while the yen rate against the dollar may fluctuate, our margins are almost not impacted.

In addition, in relation to the Japanese yen impact on the balance sheet, we have a natural hedge on the cash and the loans balance since the loans and the cash are both yen denominated.

This helps us to protect from potential yen fluctuations.

Lastly, in relation to fluctuations in the Israeli shekel currency, we have no revenues in this currency.

And while less than 10% of our costs are denominated in the Israel currency, we also hedge a large portion of this currency risk using zero-cost cylinder transaction.

A last note on our share count.

As of March 31, 2019, we had 106 million ordinary shares.

The fully diluted share count remained unchanged at 108 million shares, similar to that of previous quarter, and included 2 million maximum possible shares to be issued comprised of ESOP-related options and RSUs.

This ends my summary.

And now I wish to turn the call back to the operator.

(Operator Instructions) The first question is from Cody Acree of Loop Capital.

Congrats on progress.

Russell, could you just talk a bit about your level of visibilities, some of the comments you made in your press release and in your prepared remarks about your optimism that the second half will be stronger, particularly on the organic side?

Can you talk about either by segments where you're hitting that visibility?

Or are you seeing this already in order rates?

Or are you just hopeful that you get through sort of the inventory excesses and that is replaced by growth in the second half?

It's a combination of all of what you said, Cody.

In the area of the optical components for -- predominantly, that is down in the data center.

It's what customers believe and what analysts have written as to the inventory burn-off as well as the other factors that I had mentioned, that the second half would be coming back with the data center.

The question, I suppose, with the silicon germanium and data center isn't when this rebounds and accelerates and goes back to double-digit, I mean it's not a question of if, it really is just a question of when, it definitely will.

And I believe that the very steady growth that -- or demand growth that we have over the past few years, what we see now is an aberration to that.

But this is according to analysts and customers, and it's pretty much every report that I've written would refer to a recovery in the second half.

If I look at our mobile business, in and of itself, our mobile business is actually stronger for the second quarter guidance than it was a year ago with some very strong customer forecasts coming in for the third and fourth quarter; in some cases, possibly even POs.

But that I'm quite confident on as well.

We continue the growth there that we have, and a lot of that is driven through the capabilities that we have in 300-millimeter RFSOI.

The demand that we're seeing is very, very real on 65-nanometer power management, which is a new platform; really, you could even say a new sort of market for us because this platform makes us very competitive in that market.

And that's also off of customer forecasts.

It's not existing POs.

For the most part, our business is a term business.

So we receive POs for the most part within a realistic time frame of when the parts should be shipping.

So we don't have at this point -- other than those long-term contracts that we have, the take-or-pay agreements some with the IDMs or the TOPS business and the Maxim and the Panasonic, we don't have orders at this point for Q4 that I could really talk about what is the Q4 backlog.

That's not the nature of our business.

But the customer forecasts for the most part are fairly accurate.

Did that answer your question, Cody?

Yes, it did, Russell.

And then just as my follow-up.

For the RFSOI business, you went through a period of about 3 years where you were gaining pretty significant market share, and then in 2018, when we had the inventory correction, you looked to kind of plateau that market share and just switching your capacity over and ported it to either of the fabs or chose to use the Newport Beach facility for more silicon germanium.

Is this transition to 5G or the movement to 300-millimeter, is this likely to -- are you expecting just dollar content increases?

Or do you expect you'd be gaining share through the remainder of this year in RFSOI?

Presently, pretty much everything that's happening in 300-millimeter is share gain.

And the other -- the 200-millimeter activities that are going on is really related to customer demand and new tape-outs that I think also relate to share gain because they're for customers that we didn't necessarily have high volumes with in the past.

Some of the previous strong customers have already given us new tape-outs as well.

So whereas I had mentioned that we have lost some share in 2018 off of decisions we had made in '17, actions have been done and activities happened that, that share is coming back.

Some of it we'll see at the end of this year and certainly in 2020.

Great.

I can give you more information.

No, please, please, I thought you were finished.

Now if I was to look year-over-year for the mobile area, RF mobile area, we're up about 15% year-over-year for it in the Q2 guidance.

The next question is from Mark Lipacis of Jefferies.

First one on the Japanese factories on use of the 300-millimeter.

Sounds like you're still -- utilization rates are going up.

Can you, Russell -- could you just kind of give us a sense by vertical market what is driving the increased utilization rates at the 300-millimeter factory and any quantification about how much it increased in points?

That's the first question, and I have a follow-up.

I don't have that right off the top of my head, the amount of points that increased.

I will get that to you by the end of the question.

But as far as the market that it's serving, the biggest growth at 300-millimeter right now is RFSOI and obviously then serve into front-end modules.

So that's the biggest portion of growth.

The other portion of growth is a variety of end applications for the 65-nanometer BCD.

And again, the end markets there are very, very diverse.

But power drivers and ultimately power management, I think the biggest opportunity that we have within that is power management going into data centers.

Okay.

And then on the previous call, you had mentioned that -- the previous earnings call, you had mentioned that in the third quarter this year, you might be at a point where you need to increase the CapEx in that factory.

It sounds like you started that already with a photo layer line.

Could you -- do I have that right?

Could you just provide a little color on that?

Certainly.

Indeed, we said that we would be making a decision because we thought that we would need to increase capacity.

The ramp is actually happening quicker than we had anticipated.

So the first step in increasing capacity, the capacity that had been set up in the factory really dealt with Panasonic flows.

The third-party flows aren't necessarily the same exact tool usage.

Well, they're definitely not the same tool usage off the same amount of layers.

So nominally -- and whenever we report utilization numbers, we report them against the total photolithography capacity.

Photo is typically the most expensive tool in the factory.

So you look at the maximum amount of photo layers that can be manufactured in the facility, which is some integral of all the photo tools plus how many layers they can do per hour minus whatever scheduled PM.

However, for most factories, you have multiple different products running, so there's a bottleneck tool for the product that isn't necessarily -- if you add all of the different flows you have in a factory with the bottlenecks of those flows, at any given time, it doesn't necessarily meet the 100% level that you would say would be 100% based on photo.

This is getting complicated and I apologize, but it's a very good question and I really do want to give it a very accurate answer.

So on the activities that we have now, the photo capacity of the 300-millimeter factory, we cannot meet that full photo capacity because there's other processing tools that become a bottleneck because we cannot get enough layers for those tools because the usage of the tool is different than it was for the Panasonic flow and/or our flow requires more steps of those specific tools.

So what we've invested in right now is relieving the bottlenecks so that we can maximize the output according to the photolithography that is already in the factory.

And that has been invested.

That wasn't a huge investment, but it was an investment that we have done.

One of the tools is in place.

Other tools are on way to coming in place right now.

The next phase, though, is actually increasing the photolithography capability, which isn't just adding photolithography.

It then has, in addition to the photolithography, additional processing tools.

And that is what we had talked about previously about the Q3 decision, which maybe we'll be making as bit earlier sometime this quarter, but we're involved right now in very strong investigation, even starting with supplier negotiations to understand exactly what the costs will be and what is the ROI of doing that increase.

That will be a substantial increase in capacity against what we talk now on photo layers.

When I talk utilization now, for example, in the Uozu factory on my third-party flows, I'm running on very, very high utilization at the moment, and that will increase in -- well, the Q2 starts, not what I talked about in Q1, there's a very high utilization.

By releasing with these bottleneck tools, I get higher.

Got you.

All right.

That's very helpful.

And then a follow-up question on -- at silicon germanium, at the end of last year, things seemed to be going very well -- or in the middle of last year, and then you were adding SiGe capacity at Newport Beach and San Antonio.

It sounds like maybe the most recent forecasts in the near term aren't as robust as you had expected.

Have you -- do you modulate your capacity addition here?

Or how do you think about the original capacity additions you had started?

Do you have to follow through with those?

And then would you expect those to get filled up by the end of the year?

That's another very good question.

You're 100% correct.

The data center-driven silicon germanium is right now at pretty much the same level as it was in revenue shipments in Q2 of 2018 before we increased the capacity or during that capacity ramp.

So you could think, was this an exercise in vain?

I don't think so at all.

And we're continuing actually with the qualification with San Antonio as I believe the demand, as we stated, the recovery of the 100G platforms and the entry into the 400G platforms, we will need every bit of the capacity that we've invested.

So at the moment, some of the -- or not some, a good portion of the silicon germanium capability that we built up and we fully utilized in the fourth quarter -- at least as far as Newport Beach, San Antonio wasn't yet fully qualified.

But in the fourth quarter, we really did use all of the silicon germanium capacity that we had built up.

Presently, the start rate, that is not being used, but I do expect, as stated, that, that will come back in Q3, Q4, and if anything, we're still considering further investment.

So I don't think there was a mistake made there at all, but for the very moment, it is unused capacity.

The next question is from Rajvindra Gill of Needham & Company.

So on the organic growth rate of about 10% sequentially, wondering how do we think about that on a year-over-year basis?

And as we progress throughout the year with some of the ramps in power and on RF mobile and recovery data center, how do we think about organic growth rate kind of at a high level?

So Q2 guidance reflects 10% sequential organic growth, and year-over-year, it's 1%, which is, of course, better than the industry.

And Q1 reflected, on a year-over-year, which is what you asked, 5% organic growth, which is also better than the industry, which was down.

Okay.

And then it will be fair to assume that, that growth rate could potentially accelerate organically with these new product ramps and a potential recovery in data center.

Is that fair without giving numbers?

Yes.

Mainly for the -- from the RFSOI in Uozu, which is what Russell mentioned, maybe less of data centers.

Okay.

Great.

And for my follow-up, on the data center side, just a question in terms of the 100 gig.

Is the kind of pullback related to transceivers?

Or is it more on the switch side?

Or is it both?

Is it kind of across all the customers as well?

Or are there specific customers where you're seeing a pullback?

Just want to get a clarity on the pullback.

I'm sorry.

One more time, please, the question.

On the data center pullback, is it -- I'm wondering in terms of what parts of that really around the transceivers?

Or the switches?

And is it across specific customers?

Or is it kind of broad-based with respect to the pullback in the data center that you're seeing?

It's pretty equal, actually, against all of the customers that serve data center, which isn't all of our customers that we serve within silicon germanium.

But -- so it pulls back.

What do we do with the silicon germanium, it's compilers, TIAs, et cetera, et cetera.

But -- so it's not any single customer, it's really the data center growth itself is down.

The data center build-out, I should say.

The next question is from Quang Le of Crédit Suisse.

Could you please update us on your silicon germanium revenue?

I think in Q4 '18, you said it was around $50 million, if I'm not mistaken.

How much was it in Q1?

And how much do you think it will be in Q2?

We don't typically segment in that way during the year, but I don't have a problem to give you the number.

The Q1, I don't have off the top of my head.

Q2, I do have in front of me because I have a Q2 versus Q2.

So Q2 total is about $34 million, and Q4 was -- so our guidance is about $34 million, and as you say Q4, I'd hit $50 million.

Got it.

And then can you please clarify also on inventories, I could see that inventories in this quarter was at rather higher level at the 14% of annualized sales, and this is versus 10% to 12% in several previous quarters.

Do you see this coming down in second quarter and second half?

Yes.

This is not -- we don't -- inventory, the ratio that you're calculating -- the major part of our inventory is not at all finished goods or things like that.

It's mainly raw materials and the other chemicals and others, and there are also silicon wafers.

The main reason for the growth is actually what we discussed in the past, which is that we wanted to secure inventory of RFSOI for the 300-millimeter fab and also for the 200-millimeter fab.

So basically, it's in order to secure the strong -- we have a very strong confidence on the growth of the Uozu RFSOI, like Russell mentioned.

We see there a huge growth.

We even consider, as he mentioned, CapEx expansion in that area.

So obviously, we needed to secure the capacity to supply chain.

And so we purchased enough inventory to be quiet for the year 2019 and even part of 2020, we committed to purchase so that we have all the wafers on hand because as you maybe know, there is not many vendors that produce the raw material silicon wafer for the RFSOI, certainly in 300-millimeter.

Sorry, if I can just go back and clarify one thing.

The number that I gave you wasn't 100% correct.

The SiGe revenue for Q2 was guided to be closer to $40 million.

There's also LNAs and PAs that we do within the SiGe.

I was just referring to the nonmobile.

The next question is from Richard Shannon of Craig-Hallum.

I'm going to ask a SiGe question as well.

I wasn't sure based on the response to the last customers -- or last questions, Russell, whether you're actually starting to see increased loadings in those parts that go into the data center or not.

Can you clarify that, please?

No, I am not seeing it.

That's what I -- I'm sorry if I didn't answer the question clearly.

I think I had handled it during the call.

But no, that's the area that we're seeing a big pullback in Q2 specifically.

Okay.

What's the manufacturing cycle time for those products?

They're very long-layer products.

The range is anywhere from 39 to 43 layers.

So if you're dealing 2.5 day per layer, 2.8 day per layer, then you can fill that up.

So it's a...

Okay.

So if we were to see a pickup by the end of the third quarter.

We ought to see those forecasts turn into POs here within the next few weeks then.

Is that fair to say then?

That's fair to say, or one could say that if you see a pickup in the third quarter, you would see the revenue in the fourth.

Okay.

All right.

Fair enough.

I want to jump over to the topic of SOI.

Your comments the last quarter and additional ones you said here, obviously you're seeing a strong pickup in, I guess, in the intersection of your higher-performing products as well as kind of spread over 300-millimeter and some 200-millimeter stuff.

So I want to ask you how you see your share position in these newer leading-edge products here with SOI.

I don't know if it's exiting this year or 2020 or whatever, any way you can help us understand where your share position exists there?

I don't have a very good answer for you on the share position.

We had, at our strongest point, been somewhere at about a 30%.

We dropped down a bit.

And I think we're recovering back to that.

So I would say that they were somewhere between a 20% to 30% level in share, I would believe.

But I'm not 100% sure on that.

I'm not necessarily as worried per se about share as I am with building the share at specific customers that are good customer partners for us.

And that's the -- where we have focused in regaining where we know that we had lost some share and then in driving very strong -- greater than 80% to 100% share of the customers that we think are very, very strong customers that are aligned with us on road map.

The big growth that we're having in 300-millimeter are with customers that we have predominant, if not all of their market.

Okay.

That is helpful.

One last question, I'll get in line here.

You talked about a few quarters here about silicon photonics, and it sounds like you're starting to see some good forecasts.

I'm wondering if you can kind of give us an update on when you expect good volume production to start and when we should see the revenues ramp there.

Is that something that could happen this year?

Or is it a 2020 story?

Or can you help us kind of narrow that down a little bit?

We have lead customers and a very critical program that should be prototyping for a big customer in the first quarter of 2020.

And it's a very, very involved, interesting, big program.

I don't think silicon photonics will hit major, major volumes really until 2021, 2022.

I think there will be some ramp throughout 2020, but I think in the 2021 time frame when 400G really gains big footing, then the 400G is really what the silicon photonics gives its big bang for the buck for.

So it's really a question of 400G rollout.

The next question is from Lisa Thompson of Zacks Investment Research.

I would like to go back to the topic of over ordering of inventory and tariffs and things.

So have you seen any change in behavior of customers thinking about or looking to change suppliers possibly in the long run if things keep up?

And how does that either benefit or hurt you?

I think there's certainly some concerns from some customers to be able to be sourced outside of the U.S. That's a reality.

And fortunately, we have our factories in Japan.

So that becomes a very good point of benefit for the TPSCo that we have.

Predominantly, to change suppliers, I don't think we've heard anything about someone wanting to change supplier from us.

But the ability to source from outside of the U.S. for some customers is important.

Okay.

So there hasn't been any talk yet, but you're in a good spot.

I believe so, yes.

The next question is from David Duley of Steelhead Securities.

Just a couple of questions.

What do you expect the CapEx rate to be for the balance of the year given all the moving pieces of your capacity expansions and whatnot?

So we expected to remain flat to current, meaning $42 million to $44 million a quarter.

With the footnote about what Russell mentioned before with the possible photo layer capacity expansion for Uozu, of course, if we want to increase the capacity node, so this will be a special press release, a special announcement, and we will -- this will, of course, be above the model.

But currently, this wasn't yet approved, but we are having evaluation.

When it -- we will decide on that, we will, of course, let you know how much money are we talking about to be -- in addition to the model.

But so far, we are -- even with the bottleneck CapEx that we already approved, which Russell updated for Uozu, this is still in the budget of the model, of the $42 million to $44 million a quarter.

And so at this point -- well, I guess I was asking, if you do make that expansion in Uozu, how much do you think that would increase the CapEx rate?

We're looking at different scenarios now.

I think a minimum would probably be -- for that one expansion, I don't know it's a CapEx rate, it's a onetime expenditure, but it would be anywhere from $60 million to maybe an $80 million, $90 million what we're looking at.

Okay.

Great.

And then what should we expect for gross margins in -- both in Q2?

And then if you do have a recovery in volumes in the second half of the year, how should the gross margin react?

Yes.

So we didn't give special guidance about the margin exactly going forward, but we did say that, of course, Q2 against Q1, the $20 million reduction of Panasonic on the pricing, this is, of course, going all the way down to the gross and operating EBITDA margin and the 51% of that to the net.

And then now the organic growth, let's say, in -- the guidance is $16 million, usually counted 40%, 50%, 60% incremental margin for the gross and operating.

So of course, if you do that back of the envelope, it's a reduction on the net, which was expected.

And -- but as we said, we are targeting to mitigate this margin reduction later this year from efficiencies and from the Uozu ramp.

So the Uozu ramp that Russell talked about, which really a lot of more starts in Q2, will be attributed to more revenues from Uozu.

And this, together with the cost reduction in TPSCo that we started and will bring more fruits towards the end of this year, will mitigate -- there may be a reduction in margin in Q2.

Okay.

Final question for me is you talked a lot about increasing growth in the RFSOI business, and I guess it's because you're taking in some advanced parts in your new factory -- or your 12-inch factory in Japan.

But maybe just clarify, why is that business all of a sudden starting to grow when it kind of went through a difficult time in 2018?

The 2018 was really related, and I said it many times, to repurposing the factory at Newport Beach to being able to do a lot of silicon germanium and hence moving away from some other capacities that had to be done in that factory and that were requests to be done at lower prices.

So with several specific customers, we've -- due to our technology and a good partnership mentality, have regained market share.

That space has also seen a fair amount of new players that we've been able to partner with and because of our technology been able to gain their market.

So that's the major drive.

The 2018, and we had stated it back then, really was the trough of what we would have seen with the mobile segment because of cognitive decisions we have made.

We had stated very strongly that we weren't getting out of that segment and that some of it that didn't make sense to be done at that time in either Newport Beach Fab 3 or in Migdal Haemek Fab 2 was still a good margin and made sense to be doing, for example, in San Antonio.

So the ability now with new products or having had time to qualify existing products and run them in other factories that were not utilized with products that were running higher margin to substantially higher margin, that's what we're seeing the fruits of now.

And then additionally, you have the 300-millimeter platform.

That itself is a big enabler in many ways.

You have the benefit of a 65-nanometer digital with the SOI being combined, so you have then big benefits for a combined LNA, for example.

And then we continue with a very aggressive road map at 200-millimeter with new tape-outs at 200-millimeter, as we talked about, because of extremely good figures of merit.

I mean our onset, offsetter and linearities are excellent.

So it's just a question of having regained products or requalified products from some customers and factories that we now wanted to run them in and having built up additional business and increased market share with others.

Okay.

Actually, I have one more.

As far as the 5G rollout goes, would you benefit more from that, from the handset side or from the infrastructure side?

Or what areas will you see the 5G rollout in?

I think in both.

And I don't know exactly how I would quantify one versus the other.

Certainly, you'll need better figures of merits, so we have very good platforms for that.

We have a lot more content, I mean whether we're dealing with at least 4 antennas per phone.

So with the MIMO requirements, the -- really the requirements off of the LNAs will be very, very strong.

So initially, just the increase of content will have a benefit for us no matter what.

Although it's not in the initial rollout, but probably 1.5 to 2.5 years from now will be the millimeter wave addition to the 5G.

And within the millimeter wave, that has big benefits for us both for infrastructure as well as for the handset itself.

So I mean that's the sweet spot of everything we do at -- well, not the sweet spot but one of them.

So I think we'll have big benefit there.

As far as 5G rollout in and of itself, I mean, certainly, 5G enables higher data rates.

Higher data rates really drive then more information going all the time and the ability then to be able to transfer data, the need then to store more data, it all ties into what is the core of our RF business.

So I think in all of the handset as well as the infrastructure with base stations as well as the increased demand on data center storage, on cloud computing, it all ties in together with the 5G rollout, doesn't it?

Yes.

This ends the question-and-answer session.

Mr. Ellwanger, would you like to make your concluding statement?

Yes, certainly.

So really, we're very encouraged to guide a Q2, Q1 organic growth of about 10%.

Especially in the present market condition, if you look at year-over-year, to be able to have a 1%.

As we go forward throughout the year, we expect in the second half start rates to increase.

In the SiGe, we expect Q4 revenues to be back up in the SiGe, the rest of our business to continue.

I think we're extremely positive and encouraged as we go forward.

We know that we're doing the right things, and we have amazing customers that are partnering with us, multiple that show appreciation for what we're doing for them.

The ability to have them depend on us, especially for themselves within a difficult cycle when they have upsides, for them to come and ask can you do this special for us, to come through and do it, it just continues to solidify very, very strong relationships.

So although, again, short-term cautious, we're doing the right things with the right people in the right spaces.

We're very confident of our future and where we're going.

As we stated, the financial position of the company is now our balance sheet.

We're looking at many opportunities that we couldn't have looked at before and seriously evaluating them.

And hopefully, we'll go forward with multiple areas of very cost-conducive capacity expansion as well as entrants into new markets.

And we'll update on that as times are appropriate and as things really flesh out.

I look forward to meet as many of you as we can face-to-face to show you really heart-to-heart, face-to-face what we're excited about, why we're excited.

Within the next months, we have 4 events that we'll be presenting at.

On May 29, I'll be at the Craig-Hallum Investor Conference in Minneapolis.

The whole day is set up with one-on-one meetings that typically turn out to be much more than one-on-ones.

But I would love to meet as many of you as we'd like to be there and to discuss anything in depth that you might have.

On June 4, Dr. Racanelli will be attending the Needham Automotive Tech Day with a presentation and also one-on-one meetings.

That will be in New York.

So anyone that would have opportunity to be there, please, we'll love to see you and to have as much interaction as possible.

On June 4, we have the Jefferies Israel High-Tech Conference in Herzliya.

I'll be presenting at that.

That's a presentation of one-on-one meetings/and on June 6, we have the Baird's Global Consumer Technology Conference, also in New York, and Dr. Racanelli will be presenting there as well as one-on-one meetings.

So any of you that would like, we would certainly love to see you face-to-face.

As always, we're very open for phone calls, so if you need more clarifications, please contact Ms. Noit Levi, and we'll follow up as quickly and as best as possible.

Thank you very, very much.

Thank you.

This concludes the TowerJazz First Quarter 2019 Results Conference Call.

Thank you for your participation.

You may go ahead and disconnect.