高塔半導體 (TSEM) 2017 Q3 法說會逐字稿

Ladies and gentlemen, thank you for standing by. Welcome to the TowerJazz Third Quarter 2017 Results Conference Call. (Operator Instructions) As a reminder, this conference is being recorded, November 7, 2017. 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 third quarter of 2017. 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 Forms 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. Now I'd like to turn the call to our CEO, Mr. Russell Ellwanger. Russell, please go ahead.

Thank you, Noit. Welcome all of you to our conference call, and thank you for joining us. We reported another strong quarter with record revenues of $355 million or $1.42 billion annualized run rate with incremental net profit margins strongly outpacing the revenue growth, demonstrating added value being provided to our customers and to our shareholders. We reported record EBITDA for the quarter at $109 million, up 12% over last year, GAAP net profit for the third quarter was $55 million, representing an increase of 8% year-over-year or 24% on an adjusted basis.

For the fourth quarter of 2017, our guidance is $358 million, plus or minus 5%, resulting in annual year-over-year growth of 11%, of which 20% of that is organic business unit-based growth. Oren will cover our financials in much more detail in a few minutes.

Excluding the Panasonic and Maxim long-term contracts, which are committed and stable, or in other words, considering only our business unit-driven growth, we recorded year-to-date organic growth of above 20% year-over-year. I would now like to review the main activities in these various business units that will ensure mid- to long-term growth and further cement our leading position in the relevant markets.

In RF, and high precision analogue, we're investing in technology to address the next generation of connected and smart devices. For wireless, we're focused to bring to market technology for upcoming 5G standards that will substantially increase mobile data transmission rates. This includes continuing our aggressive roadmap and RF SOI, but as well augmenting this with RF MEMS, which promises to leapfrog RF SOI performance with switching speeds down to 50 femtosecond or less to address the most demanding switching in tuning applications and smartphones and IoT terminals. 5G also includes a much higher frequency band, 28 gigahertz versus less than 6 gigahertz, which is used in today's wireless communication systems, and we are working with partners to demonstrate full chipsets with our high performance silicon germanium technology that can achieve high data rates in this band.

As we recently announced with UCSD, we're reached data rates of up to 12 GB. In infrastructure, we're working to support the strong growth in global data traffic with continued roadmap in silicate germanium technology using fiber optic communications networks and data sensors. We continue to see strong strength in this SiGe market, and based on our customers' forecast, we expect continued strong growth. This is driven primarily by data centers and cloud connectivity requirements. We expect this well through 2018 and beyond.

Last quarter, we announced initial developments of silicon photonic platform that can augment our silicon germanium offerings, and increase our TAM in this growth area. We're making good progress. This quarter, we demonstrated all components of the system, including photodetectors, wave guides and modulators and have initial customer prototype products taped into our factory well ahead of initial plans.

In RF sensors, we announced a very important partnership with DENSO Corporation in Japan, which is delivering RF radar systems, built with our silicon germanium technology for the Toyota Camry, which was released earlier this summer in North American market.

We see automotive radar as a good additional growth factor for advanced silicon germanium technology over the next several years as more vehicles are adopting this safety technology and the number of sensors per vehicle will be increasing from the 3 to 7 today to perhaps as many as 12, which is predicted for a fully autonomous vehicle of the future. We are emphasizing all such partnerships.

As we stated coming into this year, our RF focus is to engage in deals on those platforms where we supply high value to our customers and hence, we'll reduce the quantity the lower value products where price rather than technology becomes a differentiator in favor of high-end, high-value mix.

Moving to power management, and beginning with automotive segment. Most new cars include a wide variety of new technologies, such as ADAS, 48-volt battery architecture, headlight LED, advanced safety features, power savings and charging capabilities.

Also the aggressive CO2 emission reduction targets in many countries is driving the need to develop more hybrid and EV vehicles. Examples can be the stop/start systems, electric power steering and more, which are using 48-volt operating voltages, enabling a potential 10% to 15% gain in fuel economy.

Our 0.18-micron 200-volt SOI power management platforms serves the automotive power management needs with a very high breakdown voltage, up to 200 volts, combined with very high immunity to noise. This platform is qualified for consumer grade during the past quarter and automotive qualification is ongoing with a very high overall market interest.

To increase our footprint in the power management SOI market, we continue to invest in development of next generations to include extended voltage ranges, reduced RDS(on) for better efficiency an die size and a variety of different isolation options.

A second platform that we've invested in for automotive power management market is our [RESIR] technology that offers the best-in-class RDS(on) for voltages up to 80 volts. This technology will include also advanced isolation schemes allowing high current capabilities and negative voltages, while still maintaining smaller die size. The platform will be released by the end of this current year with customer designs in Q1 '18.

The low-voltage 5-volt market, mainly needed for consumer applications, covers about 50% of the overall power management IC market and amounted to about $22 billion in product revenue in 2016 according to MarketsandMarkets reports. To address this highly competitive segment, we've recently introduced our 5 volt 65-nanometer technology at a 300-millimeter wafer size, enabling a significantly better value proposition for our customers than our competitors' 110 nanometers to 180 nanometers 200 millimeter platforms and is already in mass production.

We're expanding the 65-manometer 300-millimeter offering to support also higher voltages up to 20-volt operations and will benefit from best-in-class RDS(on) for improved efficiency as well as the suite of low-mass count embedded solutions at an overall low-mass count flow, also supporting high-frequency switching speed.

In parallel, we continue to maintain our leadership position with best-in-class RDS(on) across wide voltage ranges to support increased consumers and industrial markets needed for more efficient solutions.

In the CMOS image sensor market, we're investing today in technology for 3 main directions: next generation global shutter technology for the industrial sensor market, backside illumination and stack wafers for the high-end photography market and special pixel technology for the automotive market.

In the industrial sensor market, we're working with several leading customers on the development of a 2.5-micron state-of-the-art global shutter pixel on a 65-nanometer 300-millimeter wafer size platform. As you may recall, we previously announced the availability of our 2.8-micron global shutter pixel on a 110-nanometer platform, the smallest global shutter pixels in the world.

This is still the case and our 2.5-micron pixel on the 65-nanometer 300-millimeter platform will continue our leadership in this area, providing an even smaller pixel, being then, again, lowest pixel size in the world, and thus allowing higher sensor resolution for any given sensor size.

The first tape out in this new platform is expected to happen by the end of this year. One of the most important parameters of figures of merit of global shutter pixels is the shutter efficiency. Our shutter efficiency gets to 100 DB, which is several orders of magnitude higher than the competing 3-micron pixel in the market. This, combined with our stitching technology, allows us to take a large portion of the high-end industry market as well as even use this technology for high-end photography, especially for video applications.

We believe that our current generation of global shutter technology on both 180-nanometer and 110-nanometer, will continually grow, but will gradually be augmented and eventually replaced by this next generation 300-millimeter technology, enabling us to maintain our market edge for many years to come.

In the digital SLR market, we have engaged with one of the leaders in the world in the development of their next generation sensors and in parallel, are on track with our 300-millimeter backside illumination stack wafer technology development with outstanding pixel performance.

Our first silicon runs of full stack wafer flow where the circus are on the bottom, CMOS wafer, and the pixels are on the top, CIS wafer, show improvements of more than 2x in quantum efficiency with only slight increase of dark current in the 2.4-micron pixels.

We expect to bring the dark current to the same outstanding level we already show in front sight illumination, about 10 electrons per second at 60 degrees Celsius, in next silicon runs. This technology will be ready for customer tape out in Q1 next year, so customer designs are going to start very soon on this platform.

Lastly, in the automotive area, we've developed both SPAD, single-photon avalanche diodes state-of-the-art technology, as well as ultrafast global shutter pixels for automotive solid-state LIDAR, based on time of light principles.

Our SPAD show very large -- very low dark signals at the same level as discrete SPAD devices, but with the advantage of full integration of CMOS circuits, allowing low-cost, low-power consumption and system-on-chip capabilities. We've engaged with several companies, among them, one of the most promising in this area, in develop end of automotive LIDAR, and expect to be a major player in this market in the future.

An autonomous vehicle will require multiple sensors or, if you will, sensor fusion, combining the capabilities of radar, LIDAR, and standard, albeit advanced, CMOS image sensors, to accurately identify and enable real-time classification of all images within the field of view of interest. Our expertise in these 3 areas puts us in a strong position to take advantage of this anticipated, high-growth market.

With regard to our TOPS business unit, the discrete power semiconductor industry is seeing the highest growth rate since 2012, and most likely will continue to see a pronounced growth, led by automotive applications, in particular, inverter powertrains used in electric and hybrid vehicles as well as the need of power devices for 5G, IoT base stations. Currently, there are shortages of 8-inch diameter silicon wafers with discrete power devices.

Our partners in TOPS projects, Infineon, previously International Rectifier, ON Semi, previous Fairchild, and Vishay-Siliconix are the leaders in the market and continuously widened their device portfolio with the state-of-the-art products, many of which are transferred from 6-inch to 8-inch wafers and have long-term high demand.

Our success in discrete power device transfers to high-volume production is due to our high level of engineering capability, enabling fast transfer and first-time flow success within an environment of process flow IP protection. This also attracts additional potential customers, interested in the fabrication of state-of-the-arts diodes and transistors, and joint development of novel device flavors with enhanced performance, including switching times, voltages, low dissipation and high-voltage.

These products include best-in-the-world deeply scaled vertical, high power MOSFETs including integration and trench aspect ratios well beyond 10:1, advanced super junction diodes and transistors and also includes the most advanced protection devices in the world.

Our strategy in discrete power devices has proven itself, namely to codevelop with leading strategic customers, their next generation platforms with intrinsic flexibility, enabling numerous designs, focused on the most challenging applications.

In the sensor field, TOPS activities are focused on the development of diverse sensor platforms, also targeting automotive and mobile application. This includes a variety of the best-in-class magnetic field sensors, based upon magnetic tunnel junction technology, and using a novel principle of sensing, enabling best-in-class sensitivity, developed with Crocus Technologies.

We're currently in the production ramp of these sensors, including engagement with a Tier 1 automotive customer. High sensitivity radiation sensors, gamma rays, x-rays, are being productized in radiation measurement systems by our partners, and are employed in prestigious academic circles from medical and environmental studies.

The obtained experience in sensor science is applied to other sensor fields, in particular to low power sensors and sensor systems for environmental control, for example, gas sensing, temperature sensing,

(technical difficulty)

including devices suitable for operation in harsh environments. The developed production platforms employ novel device configuration and principles protected by TowerJazz intellectual property.

As part of our sensor hub sensitive, a leading Israeli academy hosted for us a sensor incubation seminar where over 40 companies entered dialogue and showed strong interest in our sensor offering and capabilities.

Regarding current utilizations, as previously stated, our operational model is to run it around 85% utilization, which provides us with the best balance between line flow and wafer outs. Utilization rates for the third quarter in Fab 1 in Migdal Haemek, Israel, our 6-inch factory, we were at 91% utilization. Once again, this is a very strong testimony to the long-term viability of our analog business and operational model, as this Fab was built in 1983, and is running at even above the designed utilization model.

Fab 2 in Migdal Haemek, Israel, our 8-inch factory, was at 86% utilization. Fab 3 Newport Beach, California, another 8-inch factory, was at 81% utilization. This is a bit lower than the usual utilization level of the Fab due to maintenance activities that were held during the quarter.

The 3 TPSCo factories, as a percent of total photo layers, was about 58% utilization, 8% higher than the Q2 utilization, with year-to-date doubling of third-party revenue in 2017 over 2016. Fab 9, our San Antonio factory, was at 60% utilization.

Turning to China. For our standard foundry business, China is our highest growth region with over 80% year-over-year revenue growth, against the base of multiple tens of millions, and a continued strong growth forecast. With over 20 strong active customers, being served with activities from all of our business units, and in all of our Fabs excluding our 6-inch Fab 1, and as well many more Chinese customers in the sales funnel, China growth has both tactical and strategic importance to the company.

Considering the government-backed semiconductor initiatives, and the Chinese stated goals to develop a complete semiconductor infrastructure to serve their growing needs of IC manufacturing and IC integration, we view it as strategically important to take part in this infrastructure build out as we had previously presented. Therefore, we have been, and continue to be active, in exploring multiple opportunities to expand our presence there.

In August, we were pleased to announce the framework agreement with Tacoma, for a new 8-inch semiconductor fabrication facility in Nanjing, China. As part of this project, we'll provide technology expertise together with operational integration consultation. To date, we've been paid $18 million, additional payments will be fleshed out in the definitive agreement. With this opportunity, we can gain a strategic footprint in China, while extending our offerings and manufacturing capabilities.

In addition, we recently announced a partnership with Yuanchen Microelectronics, a manufacture for advanced CMOS image sensor backside illumination processes for manufacturing in Changchun, China. This partnership allows us to provide our customers with advanced BSI technology mass production for high-end CMOS image sensors starting in mid-2018. This is the first time BSI will be offered by a foundry to the high-end photography market, including large formats requiring stitching.

A new BSI technology will be utilized for high-end photography automotive, augmented and virtual reality sensors as well as other growing CIS markets. We're excited with the potential from this partnership, which further enhances our leading CIS offerings.

To summarize, we continue to focus on full circle value creation for our customers, partners, employees and shareholders. As we continue our lead in the analog semiconductor space, we are taking significant strides to increase our activities and capabilities and look to further our market potential and competitive advantages by investing and focusing on additional high-growth, high-margin markets as well as several disruptive technologies, partnered with lead customers that can take a major share of new and emerging needs with these new breakthrough technologies.

We look to provide our customers with the right platforms to support their needs to capitalize on the trends that are prevalent now as well as those that are emerging, which are driving and will continue to drive the world for years to come. With that, I'd like to turn the call over to our CFO, Mr. Oren Shirazi. Oren?

Thank you, Russell, and welcome, everyone. I will start my review by providing the highlights of our P&L results and then discuss our cash flow report and our balance sheet. For the third quarter of 2017, we reported today a year-over-year increase of $28 million in revenues to a record $355 million as compared to $326 million in the third quarter of 2016, resulting in a record EBITDA of $109 million, net profit of $55 million, record cash generation from operations of $104 million and record free cash flow of $62 million.

Both gross and operating profits for the quarter increased year-over-year and were at $89 million and $55 million, respectively, as compared to $81 million and $49 million in the third quarter of 2016, respectively. With EBITDA for the third quarter of 2017 of $109 million, which is $12 million higher than the third quarter of 2016.

Net profit for the third quarter of 2017 also increased and was $55 million or $0.56 in basic earnings per share and $0.54 diluted earnings per share as compared to $51 million or $0.58 basic per share and $0.52 diluted earnings per share in the third quarter of 2016.

As our TPSCo foundry business and revenue continues to grow, we reached a profitability level entailing royalties to TowerJazz and Panasonic. This results in a higher cost of revenue, proportionally yielding a lower gross margin, creating a lower tax expense and lower noncontrolling interest, which in turn generates greater net profit, greater cash and greater free cash flow for the company. This is seen in the 55% incremental net profit margin increase as compared to the second quarter of 2017. Our adjusted net profit for the third quarter of 2017 was at a record of $61 million, 24% increase as compared to $49 million in the third quarter of 2016.

I would like now to review the results the first 9 months of 2017. Revenues for the first 9 months were a record of $1.03 billion as compared to $909 million for the same period of 2016, reflecting 13% year-over-year growth. Gross and operating profit for the first 9 months of 2017 increased year-over-year and were both at a record of $265 million and $165 million, respectively, as compared to $215 million and $120 million, respectively, for the same period of 2016.

EBITDA for the first 9 months of 2017 totaled a record of $318 million, representing 22% increase as compared with $261 million in the first 9 months of 2016.

Net profit for the first 9 months of 2017 was $151 million or $1.57 in basic earnings per share and $1.49 diluted earnings per share. Net profit for the first 9 months of 2016 was $156 million or $1.81 basic earnings per share, and $1.61 diluted earnings per share, with net profit included $51 million net gain from the acquisition off San Antonio Fab recorded in 2016 being that.

On the balance sheet and cash flow report. So during the third quarter of 2017, we achieved record free cash flow of $62 million, with a record $104 million cash from operations and we invested in fixed assets $42 million net. The cash flow for the first quarter -- for the third quarter of 2017 included $18 million net cash received from Tacoma as announced by us in August 2017, with respect to the new China Fab. The other main cash activities during the third quarter of 2017 were $16 million loan repayment and $50 million invested in marketable securities.

Debt. As of September 30, 2017, our total gross debt was $335 million comprised of outstanding principal amount of $155 million of bank loans and $180 million in debentures.

Cash, including marketable securities and net of gross debt as of September 30, 2017, totaled a record of $195 million as compared to net cash of $37 million as of December 31, 2016.

The strong growth in our cash position was mainly due to the positive $147 million free cash flow that we generated during the first 9 months of 2017.

Our current -- our net current assets -- or current assets less current liabilities increased to $599 million as of September 30, 2017, from $451 million as of December 31, 2016.

Current ratio as of September 30, 2017, increased to a record 3.4x as compared to 2.8x as of December 31, 2016.

Shareholder's equity as of September 30, 2017, was also at a record of $874 million as compared to $683 million as of December 31, 2016.

Share count as of September 30, 2017, included 98 million ordinary shares and the fully diluted share count is 180 -- 108 million. Our fully diluted share count as of September 30, 2017, included 10 million maximum potential shares to be issued comprised of 5.8 million shares underlying convertible bonds, 3 million warrants is related options and our receivables, and 1 million shares underlying capital notes. This ends my summary and I would like now to turn the call to Noit Levi. Noit?

Thank you, Oren. Before we open up the call to the Q&A session, I would like now to add the general and legal statements made and to be made with regards to our results. Please note that the third quarter of 2017 financial results have been prepared in accordance with the U.S. GAAP and the financial tables in today's earnings release include finance information that may be considered adjusted financial measures and non-GAAP financial measures under Regulation G and related reporting requirements as established with the Securities and Exchange Commission as they apply to our company.

Namely, this release also presented financial data, which is reconciled, as indicated in the tables or in the call, on an adjusted basis after deducting: one, amortization of acquired intangible assets; two, compensation expenses in respect of equity grants to directors, officers and employees; three, gain from acquisition, net; four, noncash financing expenses related to bank loans early repayment; and five, other nonrecurring items, such as acquisition-related costs and Nishiwaki Fab restructuring cost and impairments.

Adjusted financial measures and non-GAAP financial measures should be evaluated in conjunction with and are not substitute for GAAP financial measure. The tables and the earnings release also contain the comparable GAAP financial measures to the adjusted financial measures as well as the reconciliation between the adjusted financial measures and the most comparable GAAP financial measures.

EBITDA is reconciled in the tables from GAAP operating profit. EBITDA is not a required GAAP financial measure and may not be comparable to a similarly entitled measure employed by other companies.

EBITDA and adjusted financial measures and the non-GAAP financial information presented herein should not be considered in isolation or as a substitute for operating income, net income or loss, cash flow provided by operating, investing and financing activities, per-share data or other income or cash flow statements that are prepared in accordance with GAAP and is not necessarily calculated or presented on a basis consistent with the same or similar data presented in previous communications.

And now we will open the call for Q&A. Operator?

(Operator Instructions) The first question is from Cody Acree of Drexel Hamilton.

Russell, maybe if we can start with you, I got a couple for you and a couple of Oren. Russell, we'll get back to your statements about China. I think you said multiple initiatives or multiple opportunities you're exploring. Could you just maybe provide any further color, as to maybe the kinds of projects or the kinds of engagements that you're looking? I know there only so much detail, but if you can at least maybe provide color on the types of things we're looking at?

Certainly, Cody, to the degree that I can. So we've announced the partnership on a technology capability where we're doing the backside illumination. We had announced the engagement in a certain amount of our IP as an in-kind investment for which we would also get paid and gain capacity with Tacoma and Nanjing. We're pursuing other such opportunities, not necessarily identical to that of Tacoma, but where you would have other municipalities that are interested in creating semiconductor manufacturing capability and aligned maybe with a different funding structure from the central government or the municipality, one such we're in very strong discussions with, but nothing to a point that there is a definitive agreement or even an MOU or a binding MOU, so nothing at the level of presenting. But the overall drive and incentive that we have is to continue our growth within China from the pure foundry capability, operational capability that we have in our other sites and there, we've been doing extremely well, as I mentioned, about -- of a bit over 80% year-over-year growth against the base of multiples tens of millions of dollars to begin with, of an extremely good application support team, very, very good sales and country manager. So those activities continue. We believe to have more local presence there with manufacturing capability would only accelerate our growth rate within China and the very fact that there's so much government initiative at some point, we believe that certain technologies, for example, those that would have embedded memories with specific encryptions might need to be manufactured in China and that if we didn't have the capability there, we might not have that manufacturing, we might lose the business that we already have, for example. So our desire is to continue to grow, and if we can take advantage of helping municipalities with their stated goals under a model that's very advantageous to our shareholders and supports our customers with a regional benefit, it certainly falls into where we're going and what we want to do.

And then just -- congratulations on pushing above the $1.4 billion annualized revenue rate, but if you look at even just a 10% growth annually from here, you start to get very quickly into that maximum capacity that you've talked about, available capacity of about $1.6 billion. What do you see as your opportunities for either organic capacity growth or for inorganic capacity expansion from here within a time frame that would be able to address customer growth within the next 2-plus, 3 years?

So a 3-prong question and I'll give as strong an answer as I can for each of the 3 prongs. The organic growth, for the most part, is within our hands. We talked about the fact that San Antonio, we can build out another 50%, 60% with a reasonable amount of infrastructure cost, and then on top of that whatever the equipment cost would be. At this time, candidly, the 200-millimeter used tool market is not so readily available, though a lot of that at this point is dealing with OEMs that are using some used content in manufacturing tools with a little bit longer lead times. But we have within our hands the ability to trigger the facilities work in San Antonio at any time. In addition, we have a substantial amount of buildout that we can do at the Uozu 300-millimeter factory from Japan and I just, personally, with our Chief Operations Officer, visited there, worked with the facilities group and looked at the opportunities of growth there. So from the present capacity, we can probably increase that by a factor of 3. Now that, again, will not be necessarily inexpensive to do, to -- the facility work wouldn't be prohibitive but the tools are expensive. The beauty of organic growth is that you already have a certain amount of fixed cost that's covered, so the incremental fix is very, very small against the full wafer revenue that's coming in. Although you do have some increase in fixed costs, you'll have some additional depreciation, the incremental margin off of organic growth, I think is a very, very good model for us to have. And I had stated, I believe it was in Q2 in the conference call, certainly stated in Q1, that at our Q4 release, we'll be very clear on our organic our growth road map and that's what we still plan on doing is announcing to the market in the beginning of 2018, where and how we will grow organic capabilities, which makes sense, in general, to do that, because the incremental margin that comes in off of it very, very good, albeit, there is some investment. Now the other side comes into deals that we've done, such as Maxim or such as we've done with Panasonic. There are multiple opportunities that are on the table right now that maybe will happen, maybe won't happen, which is one of the reasons for not triggering the organic growth until a point that you must trigger it. Because the ability to take on a factory that's fully paid for under a model that you are getting very, very high capacity at very small money. If you look at the Maxim deal and you say an analogous $12 million to $14 million per thousand wafer capacity, you're looking at an organic buildout of somewhere about $350 million for what we were able to get there at, if all in, pretty much net 0 investment. So the model that we have had in the past is a very good model for continuous growth. And that's a model that we're pursuing and we're looking at with, again, multiple opportunities at the moment. But difficulty with that is it's, obviously, not 100% in our control, does such a deal happen or not, but certainly, we're active on it, and that's why we've stated that for our fourth quarter release, we'll give a very concrete plan about organic growth because the opportunities that we're looking at on this inorganic capacity acquisition would take preference and priority for us over the organic until a time that we really need to trigger. Now even if we do trigger organic growth and announce that, the first portion of it will be the facility buildout that at sometime make sense because we'll fill it with equipment and once the facility is built out, you have then a little bit more freedom to trigger the equipment purchase. So hopefully, that gives you a very complete answer there, Cody. I believe it's a complete answer.

It does. And then just for Oren. Oren, if you could, maybe, help us with the Panasonic royalty payment. I think this is the first time that we've seen this addressed in the cost of goods sold line. If you could just kind of walk through -- is this the structure of how you're going to be paying Panasonic going forward? And you had a credit in your noncontrolling interest line and so, I guess, I'm just asking, is this the new baseline for kind of gross margin and royalty payments for costs of goods sold? And then what are your future expectations for that noncontrolling interest and any -- just help you can give us there?

Yes. So like stated before, basically, there is a trigger point that -- or code, which is a very nice improvement in the TPSCo business, revenues and net profit. This triggered or entailed, like I mentioned, a royalty that TPSCo is starting to pay to Tower and to Panasonic. The impact of that is that on the one hand, increase in the -- from the consolidated P&L point of view, increase in the expenses. These expenses are, of course, tax deductible. So this enables, to the consolidated financial statement, a benefit in the tax line and also a benefit in the noncontrolling interest. And all those 3 components result in a net benefit to the net profit, which you can see that in the bottom line. I mean you know very well that, just for example, if we compare Q3 2017 to Q2 2017, we have almost a $10 million improvement -- betterment in the revenue line, which, by model, or by our 15%, 20% net profit margin incremental should have resulted maybe $2 million better net profit, but it resulted $5 million better net profit. So the net impact -- the net positive impact is this $2 million to $3 million. And to your question, yes, you should expect it going forward, the same continued benefit. And in order for modeling or for focusing point of view, I suggest to actually look at the 9 months year-to-date P&L, which shows, is in our reports, $1.4 million income tax expense and $2.2 million noncontrolling interest. And this is a run rate that is -- pretty much should sustain. And that's very good news to our investors, I believe, that actually income tax expenses are at the order of maybe $0.5 million per quarter instead of previously much higher in the same for the noncontrolling interest.

That's a definite improvement. And so just to be clear, structurally, our taxes are likely to be in this $0.5 million a quarter, $2 million-ish run rate, and that non-controlling interest is going to be significantly lessened going forward?

Yes.

The next question is from Quang Le of Crédit Suisse.

If I could come back to that TPS revenue and how you got a profit out of that if you could explain to me. Because I believe you grew revenues in TPSCo, and then it resulted in higher cost of sales, right?

Yes.

And for some reason, I believe, your margins -- gross margins are lower, but you still ended up in profits. Could you explain it in more detail? I am still confused of that, sorry about that.

Okay. So from a consolidated point of view, payments by TPSCo of royalties are part of COGS expenses, right -- in the cost of revenues. So obviously, an expense in the cost of revenue line is reducing the gross profit, reducing the operating profit, on the one hand. On the other hand, this is a tax-deductible expense for TPSCo in Japan. Now the amount, that is an expense increase in the comp line, let's call it 1x, okay? Which is the amount payable to Panasonic, is actually half of total amount the TPSCo is paying, right? Because TPSCo is paying, let's say, 1x to Panasonic and 1x also to Tower. So those 2x are now an expense in TPSCo P&L. Now assuming that the tax rate in Japan is 30%, so this gives a benefit to the tax line of the group of 2x times 30%, which is 0.6x benefit on the tax line. And now you have 2x less 0.6. So after the tax, it's 1.4x times the minority share, which is 49%, you have a 0.7x benefit to the noncontrolling interest. So all in all, you have 0.7x benefit to the non-controlling, 0.6x benefit to the tax line. So together it's 1.3x less the 1x of the expense in TPSCo books to Panasonic. So you have a net positive impact of 0.3x. So the benefiting tax line and in noncontrolling interest line is higher than the downside of the COG (sic) [COGS] and this brings, what I indicated to Cody's question, $2 million to $3 million net impact this quarter. That's the 0.3x -- the 0.3x is the $2 million, $3 million.

So 0.3x is $2 million basically?

Yes.

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

Just a follow-up question on the capacity, the internal versus external. I know you will go into detail in Q4. But I'm kind of wondering in terms of how you think about, at least qualitatively, return on investment versus, say, the incremental gross margin -- or the incremental margin that you would get from expanding capacity, organically? Can you balance the 2 in terms of how much CapEx is needed to expand capacity internally, versus say, buying capacity through one of these innovative joint ventures that you've done in the past? How do we think about that? How does the board think about that as well?

Yes. So basically, like Russell mentioned, for example, purchasing of the Panasonic's 3 Fabs and the Maxim San Antonio Fab, it's actually cost us a very minimal amount 1,000 wafer, because even if you consider Panasonic was $8 million and Maxim $40 million, so divided by 25,000 wafers a month of San Antonio or higher, of course, the amount in Panasonic, it comes out to be a very minimal amount of less than $1 million to 1,000 or so, if you also consider what we got. But when you need to now buy new -- not new, used equipment in the markets from the OEMs or even from other dealers, you will usually, in the history, if you check our financials, it costs us between $3 million to $5 million per 1,000 wafers. And created an ROI over 12 to 18 years. So obviously, it's good to buy CapEx for $3 million to $5 million for 1,000 and have ROI of 12 to 18 month, but it's even better to buy it at almost 0, which is the deal that we did for the capacity expansion. Now as Russell now mentioned about 2 possibilities that -- to buy not -- that we need not only the capacity tools, we also need to invest a little bit in facility, which is for San Antonio expansion and for Uozu. And then it will cost a little bit more than -- or not a little bit, but more than $3 million to $5 million per 1,000. It could cost like, he mentioned, between $5 million to $10 million. Still it has a very good ROI of 18 to 36 months. But, of course, the best is to buy it through an acquisition of an existing Fab with a long-term take-or-pay commitment from the seller. So this is the first priority of what we're trying to do and like Russell mentioned, if we see that it does not happen, we may consider this internal capacity extension, which is higher, of course.

And just a follow-up...

Raji, if I could just answer something, Raji?

Yes, go ahead.

So what Oren referred to with our history, we've been able to, in the past, do bulk deals of buying tools. So we bought almost an entire factory of tools at one point, which greatly reduces the cost. Now that really is very, very conducive to organic growth if you can buy tools, that you're going to 30, 40, 50 tools at one time, you can buy them at a very, very good cost and then the organic growth makes very, very big sense. If we're buying everything through OEMs, then it's is a bit higher cost. So it really depends a little bit on tool availability and what the deals are. And what I stated was that, in all instances, the triggering would be a 2-prong triggering that first, we would go ahead and build out facilities. Nominally it would be done in a modular way. And then, you would upon need, bring in the equipment, while you still have the ability to, possibly, take advantage of a big sale of a defunct Fab. I am sorry, to add that. I just wanted to put a little more color in there.

No, I think that's very good. I'm just trying to put some numbers around it. I think in the past, you had mentioned for San Antonio that it would be capped at, maybe, $10 million per 1,000 wafer. So if you were to increase capacity 20,000 wafers, that would be $200 million. And if, say, you sell that, those addition 20,000 wafers at $700 a wafer, you might get a $200 million revenue opportunity, but at 20% free cash flow margin, it would take 5, 6 years to recoup the additional CapEx that's required.

Yes, we will. Yes, everything you said was true and correct until the 20% because you have to know that this will be an incremental free cash flow and the incremental -- our incremental free cash flow is the same like our incremental EBITDA or gross profit margin, which is about 50%, 55%, maybe.

50%. Okay. So then I would cut it, maybe, in half?

So it's in 2 years, so the $200 million additional revenue that you calculated will result in $100 million additional free cash flow, so...

Free cash flow so to recoup the CapEx might be 2.5 to 3 years?

Yes, even less.

(Operator Instructions) The next question is from Richard Shannon of Craig-Hallum.

I apologize. I'm just boarding a plane now so background noise could be high. I'm going to keep my questions short here and get back in queue. But I guess my first question, Russell, you gave us some good detail on all your power management initiatives going on. I'm wondering if you could help us understand the cycles here. How long the designing cycles are by markets and by voltage levels? And whether we're going to see an acceleration in growth in that power management segment that seems to be doing so well right now? Any detail you can provide, even a few quarters out, that would be great.

It really depends a bit on, if the customer's is an existing customer, that is designing in something in addition or if it's a brand-new customer. Our platforms been very modular platforms and the fact that they are

modular allows person that's familiar with our designs to, without a lot of work, add certain features to a PDK that they're very familiar with. So for an existing customer, it's basically, for their first tape-out, however long to order mass set, it's a number of weeks; to run it through the factory, you're dealing with 3 months. So you're, basically, from basically from the start of the design, until they see wafers for prototypes and can start seeding the market, they are dealing probably on a 5, 6-month type of a time frame. And that's with an existing customer that's familiar with you that has the market that they want to be serving. If you're dealing with someone that doesn't know you at all, then the typical from a design win until you really have something out on a prototype, you're adding anywhere from 3 to 6 months to it in the area of power management. But power management flows are, for the most part, not very heavy layer flows so they move through the Fab fairly quickly. Now you're dealing somewhere between the 300-millimeter, it's 17, it's a very, very low mass count especially with some of the mix signal power now, up through, at the most, 30 layers. So you're not dealing with the very, very long flow. So the cycle time in the Fabs are fairly quick.

Power management has been a very, very big growth market for us. And I believe it will continue that way. We had stated at the beginning of this year that all of the business units -- half would be seeing about 25% year-over-year growth and with the RF, we'd started by saying it would be mid-single-digit, which we changed to say that (inaudible) structure optical markets, although the power management did growth at that type of rate and we haven't given any guidance for (inaudible) spoken about in '18 (inaudible) I would invite is, and will talk about it in a little bit, have an investor day coming at the end of this month and we'll get into some strong specifics on all of the offerings, and maybe then you'll share a great amount of the excitement that we see in our offering.

I'm getting on a plane, the reception is getting bad. I'm going to jump out the line, and maybe follow-up with you later.

The next question is from David Duley of Steelhead Securities.

First question is, you mentioned that you received an $18 million payment from Tacoma. Could you give us an idea about what you think the -- how may more payments that you might receive? Or what the total payment might be? Or when you might receive further payments?

So what I stated in the -- as I talked about the quarter, hasn't changed. The next payments will be defined in the definitive agreement. How much would we expect, that's -- I don't know I really want to put that out in to the market, but you could say that we're transferring licensing certain technologies and then we'll have a certain capacity in the factory. The $18 million is the first payment. You would think that other payments would be on that same order of -- and that there would be multiple.

Okay. And as far as your revenue guidance goes for the fourth quarter, I think it was up $3 million or 1% sequentially, roughly. Last year, I think you had 4% sequential growth, and perhaps, $12 million or $14 million worth of revenue growth. I was wondering what you were seeing in the marketplace as far as why this year's growth in the fourth quarter is a bit less than last year's?

I'm not seeing much difference in the marketplace. We have customers that are very excited and things that are moving. On any given quarter, you have some certain inventory corrections. We have still very, very big demand within the -- especially within the optical markets. We're changing mix, the Newport Beach to be able to have more silicon germanium capacity. So I don't think there is anything specific about Q4. It's very difficult to look at a 4% versus 1%, and think that there's a difference year-over-year, especially against a different base of revenue. But I don't see anything specifically different in the atmosphere, in the environment. If anything, I see tremendous excitement going forward because of a lot of new initiatives, maybe more than we had coming into this year. So I don't see anything specifically different other than, as I stated, we are, by design, moving from what would be price pressure, lower value products within some of our RF portfolio to a higher-margin value-add mix and that's the only difference that's a cognitive decision from ourselves.

Could you give us an idea roughly what the size of your silicon germanium business is now as far as the base goes? And what would you expect the growth rate to be over the next year or so?

We have a demand right now that's probably, if I look at the -- at our forecast, the demand is presently sitting at least 30% to 40% higher than our capacity. So that's why we're increasing that. We've said before that our optical business was about -- what we said, we announced 30% plus minus was RF and of that, circa 10% was the infrastructure. So you could -- now there is some mobile that is silicon germanium as well, but if you were to look at that and you would say that the SiGe is sitting then somewhere about $140 million, just from previous numbers.

$140 million annually?

That's from previous numbers. Again, I haven't broke it down, just from what we had said in the past. We talked about 30% was RF and about 10% of it was infrastructure.

There are no further questions at this time. Mr. Ellwanger, would you like to make your concluding statement?

Certainly. So again, thank everyone for your interest in the company. We sit extremely exciting -- excited about where we are, where we're going. We have, on November 29, an Investor Day at the NASDAQ market site in Times Squares at New York. We really invite everybody to come there. We'll have more time to share in what we're excited about. You'll have presentations there from the President of the company, responsible for the BUs, from Marco Racanelli, responsible specifically for the RF and HPA, but he'll talk about a variety of different platforms, operations. Oren will speak. Obviously, I'll be speaking there. But you'll also have opportunity to hear from and interact with our Chairman, Amir Elstein, that is semiconductor veteran, having been an executive at Intel Corporation. So I think it will be a very good day, and we'll have much more time to interact. A good portion of time for Q&A, and we'll be able to get into details there that nominally we'll be able to share our excitement for our future, which we really have. The activities we're doing and I mentioned, I believe in the PR about really certain activities that we're doing that are very, very strong, disruptive technologies with leading partners. These technologies, again, when you work on something like that, you don't have 100% surety that it's going to pan out, but the fact of having leading customers partnering with us on disruptive technologies, it does work and it'll take off like crazy. And if you have a standard road map, that is customer partnered for advancement, multiple disruptive technologies and new initiatives to increase your served market, it's a very, very exciting future forecast. So again, in closing, thank you very, very much and look forward to see everybody that can possibly be there in New York at the NASDAQ on November 29. Thank you.

This concludes the TowerJazz third quarter 2017 results conference call. Thank you for your participation. You may go ahead and disconnect.