Nova Ltd (NVMI) 2017 Q4 法說會逐字稿

Good day, and welcome to the Nova Measuring Instruments Ltd. Fourth Quarter 2017 Results Conference Call. Today's conference is being recorded.

At this time, I would like to turn the conference over to Miri Segal, MS/IR. Please go ahead.

Thank you, operator, and good day to everybody. I would like to welcome all of you to Nova's Fourth Quarter and Full Year 2017 Financial Results Conference Call. With us on the line today are Mr. Eitan Oppenhaim, President and CEO; and Mr. Dror David, CFO. Before we begin, may I remind our listeners that certain information provided on this call may contain forward-looking statements, and the safe harbor statement outlined in today's earnings release also pertains to this call. If you have not received a copy of the release, please view it in the Investor Relations of the company's website. Eitan will begin the call with a business update, followed by Dror with an overview of the financials. We will then open the call for the question-and-answer session.

I'll now hand over the call to Mr. Eitan Oppenhaim, Nova's President and CEO. Eitan, please go ahead.

Thank you, Miri. Let me add my welcome to everyone, and thank you all for joining our fourth quarter and full year 2017 Financial Results Conference Call.

I will start the call today by speaking briefly to our December quarter results. I will then spend some time summarizing 2017 performance highlights and their relevance to our growth trajectory in the years to come. I will conclude my part with the guidance for the first quarter of 2018. Following my commentary, Dror will review the quarterly and full year financial results and details.

Our fourth quarter performance was an excellent conclusion to a fifth consecutive record year, demonstrating the strength of our diverse portfolio across all technologies and device segments. Revenue and (inaudible) profitability for the fourth quarter both exceeded our quarterly guidance. The new record levels in the fourth quarter contributed to yearly revenue growth of 35%, where products alone grew at a pace of 42%, outperforming the market annual growth rate. This significant milestone is driven by growth in all our product lines, platforms and technologies. Our consistent performance through 2017, which reflects compound annual growth of 18% in the last 5 years, has resonated with our customers as we have been able to meet their growing demand for advanced metrology in a very challenging wafer fabrication environment. All of that is also reflected in the market share we gained this year with strength in the most complex applications, which are crucial to enable our customers' fast and stable ramp-ups.

During the quarter, we continued to effectively execute against our strategic targets to diversify our markets, customers and products, with multiple success in all key performance metrics. Our initiative to expand our presence in memory segment continued to bear fruit this quarter as well, with strong contribution from leading memory customers.

As reflected in our quarterly results, the memory segment accounted for 40% of the overall product revenue, a trend that will continue during 2018, while we make more significant inroads into the space, driven by growing demand for vertical NAND and advanced DRAM. This successful milestone was also supported by our recent press release about 2 major memory customers that selected our dimensional solutions following on-site competitive evaluation. As a result, our efforts to create a balanced customer mix are progressing well with 3 large customers exceeding each more than 10% of our quarterly product revenue, including 2 leading memory customers. Investing in disruptive technology continued to be our main focus as we believe that our ability to create an innovative and differentiated portfolio will expand our available markets and contribute to our growth while we meet our customers' most difficult process challenges. This initiative has been executed successfully with 3 new recently launched metrology hardware models to enhance our dimensional and materials metrology portfolio in both the X-ray and optical product lines. All of our newly launched models, including new all CD standalone, also all CD-integrated metrology and XPS tools have been already installed in various customers and are generating revenues.

On top of our new hardware models, we also recently launched a breakthrough machine learning software switch to complement the traditional physical modeling software engines that have been used in the industry for many years. The new Nova Fleet software significantly enhances our metrology capabilities and accelerates time-to-solution in the most advanced 3D and high-aspect ratio devices. Together with our Big Data Fleet Management infrastructure that has been well received by all leading customers, Nova Fleet utilizes fleet-wide information to provide adaptive metrology solution based on continuous trainings. The platform is also open for various other technologies and various other hybrid metrology information to improve training and validation.

Moving now to the full year. 2017 turned out to be an outstanding year for Nova with record financial performance, demonstrating a fifth consecutive year of growth and reflecting the transformation we went through to establish Nova as a multi-technologies end-products company. Our attractive offering is generating a lot of traction among our customers, and as a result, we posted annual record revenues in both our dimensional optical portfolio and our materials X-ray portfolio. Notably, as part of this achievement, we could broadly leverage our new software capability to generate record annual software revenues, which more than doubled year-over-year. This achievement supports our long-term consistent vision of tighter coupling between hardware and advanced software engines to address complex challenges that are no longer related only to hardware performance. With an outcome of our commitment to support our customers' most challenging transitions, we currently have more engagements and opportunities across all industry segments and devices, which we believe will strengthen our competitive position, expand our available markets and allow us to differentiate our offerings. Because of these successful engagements, our customer and segment mix in 2017 were more diversified and balanced than ever before. A key part of our strategy for the last few years was to broaden our customer base by expanding our position in memory. In 2017, we made significant progress in achieving that by gaining memory market share with both our opticals and X-ray solutions. This remarkable progress is reflected in the yearly revenue mix with more than 35% of our product revenues generated by memory customers. Moreover, we can -- we could expand our memory exposure even further this year, by making significant inroads to all major memory providers. The results are strongly evidenced with sequential growth in memory revenue to record levels, embedding multiple expanded position in existing customers, as well as adding new customers. These efforts are also evidenced by the yearly balance customer mix that included 2 large memory customers and 2 large foundry customers, which contributed more than 10% each for the overall product revenue. Moreover, we are very encouraged by the sizable growth in both segments, where foundry contribution grew at 35%, and memory grew at 62% from 1 year before.

As it relates to our financial model, the strong yearly profit demonstrates once again, the value our offering brings to the customers, the synergies we've built in our portfolio and the operational efficiency we embedded into our financial model as we continue to grow. As a result, we reported this year, yearly profit with 76% growth in non-GAAP EPS.

Despite the challenges we faced with the fast trending of the company, we continued to maintain our [fleet] model, which can accommodate elevated investment in R&D, while still meeting our long-term profitability targets.

As part of our evolution into a bigger company, we are focused on changing our R&D spending to be divided by investment in sustaining innovation through our traditional portfolio and disruptive innovation to develop new products for new emerging process control challenges, which are not addressed currently. This strategy is already embedded in our product rollout, where we have expedited our development cycle, and could launch multiple new solutions through the year. We should expect a similar pace in the coming year while we differentiate our offering and expand our available markets beyond the traditional metrology space.

The solid performance of our product sales is also complemented by ongoing strength in our Service business with a new yearly record high of approximately $48 million, which reflects 50% growth. Our continuous investment in enhancing our install base is bearing fruit, with various solutions that enable enhanced productivity, better utilization and improved metrological capabilities than in previous models.

As part of our growth, we are entering 2018 with a more balanced geographical mix as well. While Korea continues its growth in revenue contribution to around 27% from the overall revenue in 2017, in light of the increased investment in DRM Flash, we are also experiencing a tremendous growth in spending in China, which now accounts for approximately 17% from the overall revenue in 2017. The investment in China, both by domestic and global players, is driving significant growth in semiconductor CapEx, and Nova is well positioned to benefit from this growth in 2018 as well.

As for our product portfolio, our growth as high-end technology company is driven by successful execution of a differentiated product strategy, that in our opinion, is becoming more attractive as challenges are growing in the most complex next-generation memory and logic devices.

Our strategy to invest a large portion of our income back into R&D to create new organic engine is bearing fruit with a holistic approach with coupled hardware sensitivity with advanced software capabilities, which intersects customers' needs in the most complex step. Entering 2018, we will continue investing in this direction to rollout new solutions and products to address an even wider range of applications, and in return, accelerate our growth in the new metrology segment. The ability to create synergies between our technologies is expected to drive another growth year in 2018 for both the X-ray and optical product line. Encouraged by our intensive growth in recent years, we've deduced at the beginning of 2018, new multi-year strategic plan to meet $300 million in revenues and further growth in profit. While we continue to look for new acquisitions, we believe that meeting our new target can be achieved largely by investing in our own technology to drive faster organic growth. In order to meet this aggressive milestone, we are going to focus on leveraging our technology innovation to create a meaningful competitive edge, investing in customer partnership in early development stages, and continuing to maintain our solid operational model.

In summary, our 2017 results can be attributed to the transition we made from being a CMP-integrated metrology company, which was highly exposed to foundry customers, to a multi-product, multi-technology company, with a broader customer base across all device segments.

Based on our 2017 achievements and yearly record order backlog, we expect another growth year in 2018, and see a clear path to meeting our long-term strategic model. Based on our current estimations for the overall industry demand, we believe that the industry momentum will continue in 2018, while the markets benefit from solid catalysts, mainly in the world of data management and high-power computing. In this environment, where our customers are going to continuous inflection points, Nova is well positioned to benefit from their needs to fabricate better devices, which also [bank] to the market.

As for the first quarter of 2018 guidance, we expect the following: revenues in the range of $54 million to $60 million; diluted EPS on a GAAP basis, in the range of $0.29 to $0.40 per share; and non-GAAP basis diluted EPS in the range of $0.34 to $0.45 per share.

Now let me hand over the call to Dror to review our financial results in detail. Dror?

Thanks, Eitan. Good day, everyone. In my following prepared remarks, I will refer to both GAAP and non-GAAP results. You can find a detailed reconciliation between GAAP and non-GAAP results per item at the end of the earnings press release. I will start with overviewing 2017 fourth quarter, and afterwards, will review 2017 results on an annual basis. I will then review the expected financial model of the company as part of the Nova 300 strategic plan, and will conclude my prepared remarks by providing more details regarding the guidance for the first quarter of 2018.

Total quarterly revenues in the fourth quarter of 2017 were $57.4 million, higher than the company guidance for the fourth quarter, reflecting 6% increase quarter-over-quarter, and 14% increase year-over-year. Product revenues in the quarter were $44.8 million, of which 60% came from the foundry segment, and 40% from the memory segment.

During the quarter, the company had 3 customers that exceeded 10% of product revenue. GLOBALFOUNDRIES accounted for 26% of product revenues; Samsung accounted for 22% of product revenues; and Hynix accounted for 11% of product revenues. As expected, and as communicated in the previous quarterly results conference call, blended gross margin decreased in the fourth quarter and came in at 56%. This reduction was attributed to lower service and software revenues, and to the impact of the recently announced new product introduction, which bear higher costs during the initial penetration phase.

Operating expenses came in higher than planned at approximately $19.7 million on a GAAP basis, and $18.4 million on a non-GAAP basis. Most of the increase was due to higher R&D expenses as the company accelerated introduction and development of new products, and the rest of the increase was in G&A expenses, which included higher consulting expenses, some of which are onetime expenditures.

Tax expenses in the fourth quarter were higher than expected and were impacted by several elements. The first element is the tax legislation change in the U.S., which reduced the company's U.S.-based profit tax rate to approximately 20% from approximately 35%. The implementation of this change during the fourth quarter of the year affected the deferred tax asset balances at year-end.

Additionally, tax legislation change in Israel reduces the Israeli-based profit tax rate to approximately 13% from approximately 16%. Following a final analysis of these new legislations, the company implemented the relevant changes during the fourth quarter of the year. The third and final element is a onetime tax provision of $3.5 million for years prior to 2017 and 2016, which was required following tax assessment discussion with the tax authorities regarding these prior years. As a result of these changes and updates, the effective tax rate of the company on a GAAP basis came in at 37%, and the effective tax rate on a non-GAAP basis, which excludes changes in deferred tax assets and dimensions tax provisions for prior years, came in at 11%.

GAAP net income in the quarter was $8.2 million, or $0.29 per diluted share. Non-GAAP net income in the quarter was $13 million, or $0.45 per diluted share, higher than the company guidance for the fourth quarter.

During the fourth quarter of the year, the company generated positive cash flow of $9.3 million from operating activities. In parallel, the company essentially completed most of its $3 million investment in expanding its manufacturing capacity for Optical CD in Israel.

I will now move to review the annual results of the company for 2017. Total revenues in 2017 increased by 35% over 2016. This growth is approximately double compared to our peer group process control companies, and reflect the market share gains of the company in 2017.

Product revenue distribution within 2017 were 65% from the foundry segment, and 35% from the memory segment. Customers that accounted for more than 10% of product revenues on an annual basis were as follows: Samsung was the company's largest customer in 2017, accounting for 27% of 2017 product revenues, relative to 12% in 2016; TSMC accounted for 23% of 2017 product revenues, relative to 37% in 2016; GLOBALFOUNDRIES accounted for 14% of 2017 product revenues, relative to less than 10% in 2016; and Hynix accounted for 10% of 2017 product revenues, unchanged from 2016.

Then in gross margin for the year was 59% on both GAAP and non-GAAP basis, reflecting 410 basis points improvement year-over-year on a non-GAAP basis. This increase in the annual blended gross margin was attributed mainly to a significant increase in software revenue, which more than doubled in 2017 over 2016, and accounted for 10% of the company's product revenues.

Operating expenses in 2017 increased by approximately 13%. A combination of a 35% increase in revenues, significant expansion in gross margin and well-controlled operating expenses, resulted in more than 50% of the incremental revenues in 2017 flowing into the operating profit.

On a percentage basis, operating margins expanded in 2017 to 26% on a GAAP basis, and 28% on a non-GAAP basis. The annual effective tax rate in 2017 came in at 23% on a GAAP basis, and 15% on a non-GAAP basis. GAAP net income for the year was $46.5 million, or $1.63 per diluted share. Non-GAAP net income for the year was at record level of $55.4 million, or $1.94 per diluted share.

During the year, the company generated positive cash flow of $62 million from operating activities and concluded the year with $150 million in cash reserves. In parallel, the company's days sales outstanding in 2017 were 69 days within the company target of 70 days, and inventory turns were 2.8x a year, higher than the company target of 2.5x inventory turns per year.

As communicated during recent investor conferences, the company revised its target financial model as part of updating its strategic target to reach $300 million in revenues organically. As a $300 million revenue level, we expect gross margin to be between 56% and 59%, and operating margins to be between 26% and 29%. We have monitored closely industry trends and benchmarks, and we believe that this model represents the right balance of competitive pricing of products, proper investment levels in R&D and infrastructure to support continuous growth and a healthy profitability level, which is higher than the benchmark for semi-capped companies in our scale.

On the tax front, following the recent tax legislation changes, we expect the long-term effective tax rate of the company to be approximately 18%.

Regarding the company guidance for coming quarters -- for the coming quarter, as Eitan mentioned, revenues in the first quarter of 2018 are expected to be between $54 million and $60 million. At these revenue levels, we expect the following: blended gross margin is expected to be approximately 56%, which still reflects lower-than-average software revenue; operating expenses are expected to be similar to the previous quarter; effective tax rate is expected to be approximately 19%.

With that, I will move the call back to Eitan.

Thanks, Dror. With that, we will be pleased to take your questions.

(Operator Instructions) We should now take our first question from Edwin Mok of Needham & Company.

The mix, go in the first quarter and for this year, if I remember, you had reached strong growth in 2017. Do you expect that mix to continue to increase in the coming year? Or do you expect it to be more balanced?

I think, Edwin, we didn't hear the first part of your question, but I think that you're asking about the mix between memory and foundry, right?

Yes, that's correct. Yes.

So as I said, we think that the strength in memory will continue in 2018 as part of the inroads and valuation that we are doing will materialize. We think that it will continue in several customers as well as in both segments in the DRAM and the vertical NAND or the NAND, and we believe it can even grow. Our long-term target, as we see that, should be at around 40%, 45% memory and around 60%, 65% foundry. This is -- sorry, 55% to 60% foundry. We think that in foundry, the intensity is higher, and therefore, we believe that we need to be exposed largely to foundry.

Okay, great. That's helpful. And then on the software side, I remember middle of the year, there was -- I remember in 2017, you guys had 3 sizable upgrades, software upgrades for your customers. Because of the new software you guys rolled out and that roll just kind of basically more than doubling of your sales in 2017. And you just guided for 1Q software to remain light. Do you think your software level is -- do you think that your software sales can grow this year? And regarding the new machine learning software, can you help explain what you guys are doing there? And what is the, call it, a tax rate or customer response after you launch your software?

So I would like to divide the question into few levels. The first one is that we are encouraged from the results that we had in 2017, and we continue selling and continue delivering products according to our model. That's saying that conservatively, we would like to be in 10% out of the revenues coming from software. I think with the potential that we have, although we don't see the end of the year, there is a potential to overachieve what we did in 2016. So this is regarding the revenue. Secondly, regarding the tax rate, the thing that we need to divide between the regular software upgrades that we are doing to our system that are not counted in this software revenue that we are talking about. Most of the software revenue that we are talking about is the new modeling capabilities that are copying -- coming on top of our fleet. And their main target is to make sure that we are solving faster in the application mainly in the optical side. As you know, physically, in order to solve -- in order to measure CDs in the semiconductor, you can't actually see what you measure, you need to model it, then interpreting the spectrums that you extract from the measurement. This takes time. It always becomes a limitation in very fast cycle. And therefore, we looked for ways to shorten the time. And the shorten -- shorten the time coming from looking on a different angles on the way that we are doing modeling. So if previously, you need to do a full model, a full physical model, it could take even 3 months. Today, we can use some automatical algorithm engine that fits on top of the Big Data bases that we have, take information from various tools, can be metrology tools, can be other process control tools and can be even process tools. And actually, by getting large amount of information, we can tighten the measurement, and sometimes, even almost predict the measurement. So therefore, the combination of physical model and mathematical model bring the model for us to sometimes to semi-models, or even sometimes without a model at all. And this can be reflected in the time-to-solution that we bring to the table, which coming very close to other process control tools that we have in the field, and actually removing the limitation of time-to-solution even in R&D, or in a [very] cycle changes in development of the structure. The third element is what kind of software we are selling. So we sell this machine learning. And of course, this machine learning has a client-base -- sorry, several client bases on our install base and it's been sold according to the licenses that we sell there too. And secondly, there's all the Big Data base to collect all the information and manage the fleet, which is coming on either sub-wides, or node-wide, or it can be even done on a tool basis. So this is the 3 elements that cover the software side.

Okay, great. That's extremely helpful. And then finally on the target model, as you guys raise your prior model, even though on the revenue side as well as your margin. I want to understand beyond, probably just the steps in memory what's driving this increase? Is it just increased metrology intensity? Is it you are more confident about your ability to grow a successful new product. And then I think about at conferences you guys talk about beyond 2020, you think you can eventually get to 450,000 number with acquisition. Any color you can provide around those, even longer term outlook? 5-years' outlook?

Yes. So, I'll start from the fact that we didn't neglect the idea of doing M&A and continue to do further acquisitions. I think that following 2017 and the very successful integration that we went with ReVera in the last 2.5 years, we definitely are encouraged and we'd like to do other acquisitions as well. Majorly this, the market condition and the strength of the economy have been creating some borders and limitations to the time that you can make it, but still, we are looking for target lists and we are -- aggressively would like to expand our portfolio with other product and technology as well. However, when we're looking at the successful growth in the last 2 to 3 years, we're also encouraged from the traction that our products are getting in the market. So if you're looking right now on the memory side, the growth actually came from introducing new models, introducing their hardware and software coupled solution, introducing some disruptive innovation that could not be sold before by our competitors. And if you look right now on the market share -- now our growth this year, on the 40% growth in our products, it's definitely more than the overall process control growth. Overall, it's more than the overall optical metrology growth. So it's actually more than even the wafer fab equipment growth rate. So we believe that most of it came from poor market share. When we're looking on the future, we don't count to be $300 million company organically by just trusting the intensity. I think that the intensity is -- will grow, obviously. But the major element for us being in $300 million in couple of years is 2 things: one is continuously taking market share and establishing our leadership in foundry; and second, what I discussed about the technology. We truly believe that through our product, we could differentiate our offering in the last 2 years. It will continue in 2018, where we continue to find the regular metrology side, but we still bring the new innovation to the field to new application which are currently not sold. So the combination of new products, new innovation, new disruptive innovation, together with the market share that we will take with our technology in existing markets, give us the confidence that we can reach $300 million.

We should now take our next question from Patrick Ho of Stifel, Nicolaus

Eitan, maybe first off in terms of your efforts in expanding your memory customer base. I think, obviously, that's impressive and it brings you into the market segment, which is showing a high level of growth. Can you discuss some of the applications and some of the manufacturing challenges your solutions are addressing? And I guess how that could continue to grow particularly, as 3D NAND has -- grows more layers and DRAM continues to shrink. What are some of the additional growth opportunities for Nova in these industry transitions?

So Patrick, I'll look at that from the 2 technology product lines that we have. I'll start first from the dimensional part and the optical technology. I think that as more -- as the xenon structure is becoming more weighted into the high-aspect-ratio structures, and as more as they are adding layers and pairs of memory, the parameters and the CDs that we need to solve through these towers of memory are becoming more and more challenging. You can imagine that if the customer is asking a couple of CDs along the profile of the high-aspect-ratio structure, it's becoming very, very complicated to do it with optical. And we believe that we have some mechanism to use the optical system that we have, which represent different technology than our competitors, we'll be able to solve part of the challenges that we see in the vertical NAND. Actually, most of the market share that we took in vertical NAND came from this capability. Secondly, in the vertical NAND, it is built on -- from batches of memory. So if we can solve the ability to measure CDs on nonstructural device, the [non-equivocal] device, which we can, actually give us the benefit to solve more applications. So this is regarding the vertical NAND and the optical. Secondly, regarding the DRAM. DRAM is a dimensional structure that keeps on scaling and shrinking. And as in other devices, once it shrinks and scales and goes to 3D dimensional structure, the need for metrology is increasing and you need to measure more parameters as the design rules increase. So this is regarding the DRAM -- and sorry, in the VNAND. Now regarding the X-ray part or the XPS parts, when you're looking right now specifically on VNAND. And the combination of ONON or OPOP, the demand for measuring very ultra-thin layers as well as the composition and the other characteristics of the materials are becoming very crucial in stabilizing the overall structure. This is why we see in the memory customers, we see in 2017, a change from taking tools to just [in-fab] tools, to actually in inline tools, which doesn't measure only (inaudible) wafers or monitor wafers, but measure actual production wafers. So this is increasing the attach rates and also increasing the market share in the memory. So overall, the combination of dimensional and materials in the memory beside intensity itself is increasing.

Great, that's helpful. Maybe as a -- oh, yes. Yes, as a follow-up question. Hello? Hello?

Yes, Patrick, we...

Okay. As a follow-up question, I was just -- in terms of your R&D investments that you've talked about, are they to introduce new hardware solutions, software? And are they going to be within your current scope of dimensional and X-ray? Or do you see yourselves expanding into potentially new other metrology markets that are peripheral to what you already serve?

So I think, Patrick, in the next 2 years, '18 and '19, we're definitely going to use our current technology, the dimensional and the materials, the X-ray and the optical, to come with new tools to the market. But the new tools that we are going to come to the market are not necessarily solving the traditional metrology application that's used to be sold in old CD before. Saying that, it means that the process control itself, starting from testing, inspection and metrology, has different application that we think that we can solve by our own optical and X-ray technology. On top of it, nobody says that optical can be used only for dimensions and X-ray can be used only for materials. So the combination of the 2 technologies can join together for some unique technology that actually can solve some of the application that today customers are required to break the wafer in order to find out the ultimate profile, we might find some ideas and ways to do it differently. Operator?

Yes. I think that the line is not open. And because I see that -- I think that there is a question.

Operator, can you hear us?

Yes sir, I can. I do apologize. We should now take our next question from Mark Miller.

In projecting growth again in 2018, I'm just wondering in terms of what are going to be the major drivers? Is it penetration and more memory? Is it the new products? I'm just -- and also the growth of the China opportunity. I'm just wondering what do you -- if you can rank those in terms of what are the greatest growth drivers you expect in 2018?

Yes. So thanks, Mark. So we have 4 elements, the 4 growth engines when you're looking on 2018. First is the -- we are going to keep the momentum in growing our market share. This is utmost crucial target for us in the next coming year, and definitely, memory is one of the main targets for us to increase the market share. Secondly, I think that the territory mix where China will continue growth, it's the second element. We see continuous growth in the last 3 years, and we think that it will continue growing substantially in 2018 as well. The third one is the new growth engine that we bring to the market. So if we could bring to the market last year 4 new type of products, which 3 out of them is hardware and one of them is software, we're going to stay on the same pace in 2018 as well. And there's a lot of traction in the market before these tools are going outside to the market and we believe that we can generate revenues from those tools as well in 2018. And the fourth growth engine, all the market is talking about memory and the growth of VNAND and DRAM. But we also look on this year as the growth year for us in foundry as well. Although the foundry is a bit muted towards the second half, we still believe that the continuous investment in variety of technology now starting from 28, 22, 16, 10, both in China as well as in Taiwan and in other places, will continue and the [5-for-10] and 7-nanometer will continue as well in the advanced node in the second half. So we believe with the fourth growth engine for us should be the leadership or establishment of the leadership in the foundry. So we believe that we can grow in foundry as well. So that's the fourth -- those are the 4 growth engines as I can mention for the fourth -- for 2018.

Okay. I'm just wondering if -- are you trying to see any replacement of some of your legacy tools? People upgrading to new models whether it's for better resolution on H3D devices through into light source or detectors? Or are there upgrade opportunities on your existing tools, your legacy tools?

So as we've discussed in our service revenue, which a record high this year, we definitely see a lot of opportunities in upgrading the install base. Actually, more than 1/3 out of the revenues are coming from utilizing our install base in a better way, either by upgrading them to the most updated hardware or upgrade them with the new software or changing panels [of parts], in order to utilize those tools that are -- so like definitely there are a lot of opportunities for us in the install base.

Great. Just last question, a little more color, what's going on with the X-ray business last quarter?

Are you asking about specifically, about the X-ray in the fourth quarter?

Yes.

So Mark, we don't break out the revenue from X-ray to the optical or from dimensional to material. I can just refer to the overall year saying that both product lines grew significantly and both of them presented record all-time yearly revenue, for both the X-ray and optical.

Do you think X-ray will grow in 2018?

I said X-ray will grow when we build them and it's proven to be a very nice growth rate. We truly believe that X-ray can keep on growing in 2018. We actually -- we mentioned also the backlog that we have for 2018, actually, the backlog that we came exiting 2017, is the highest that we have ever had for our tools going forward to a new year. So we're definitely confident that those product lines are growing well and will continue to grow in 2018 as well.

So that was a record [turtle] backlog you ended 2017, is that correct?

Yes.

Mark, from '17 to '18, this is the record backlog that we ever had.

We should make now take our first question from David Wu with Indaba Global Research.

I've got a question on your customer list in memory. You've done a very good job in penetrating both Samsung and Hynix, and I was wondering what is keeping you from achieving similar success at Toshiba and Micron?

So we didn't say that we have an issue, with those customers, we just detailed the biggest customer that contributed more than 10% of our product revenue. Looking right now, on all the 6 memory customers, we have presence in all of them. I think that looking on the amount of spending this year, 2017, for both Hynix and Samsung, they were actually higher than any other memory provider. And, therefore, I think that their results are better. But, definitely, Nova has acquisition and a growing position in all the memory customers.

Okay. The other question I have to follow-up is TSMC. They've been digesting their spending from first half of calendar '17, and I noticed that they pulled in their 60-nanometer fab in Nanjing into the second quarter. I was wondering whether there's going to be a pickup in TSMC orders, starting midyear onwards from just where you're sitting.

The way that we look at TSMC this year is going to be affected from 3 type of spendings, okay? One type is the continuous expansion of the 7 nanometers and the 7-plus which has multiple customers that are waiting for capacity and I think that they will increase the capacity along the year. The second one is the launch of a new 60-nanometer line in Nanjing, which will continue throughout 2018. And the third one is opening the 5-nanometer pilot line somewhere in Q3 or Q4 2018. We definitely -- when you're looking on those potential opportunities, they are more weighted toward the second half than the first one.

This concludes all the questions for today. I'd now like to hand the call back to Mr. Eitan Oppenhaim for any closing remarks. Thank you, sir.

Thank you, operator, and thank you all for joining our call today. By that, we conclude our fourth quarter and full year 2017 Earnings Conference Call. Thanks and goodbye.