Executives: Rick Muscha - Senior Director, Investor Relations Moshe Gavrielov - President and Chief Executive Officer Lorenzo Flores - Senior Vice President and Chief Financial

Officer
Analysts: Joe Moore - Morgan Stanley Ambrish Srivastava - BMO Capital Markets C.J. Muse - Evercore Tristan Gerra - Baird Steve Smigie - Raymond James Ross Seymore - Deutsche Bank Philip Lee - Citi John Pitzer - Credit Suisse William Stein - SunTrust Chris Rolland - Susquehanna International

Group
Operator: Good afternoon. My name is Kelly and I will be your conference operator today. I would like to welcome everyone to the Xilinx Third Quarter Fiscal Year 2017 Earnings Release Conference Call. [Operator Instructions] I would now like to turn the call over to Rick Muscha.

Thank you. Mr. Muscha, you may begin your conference.

Rick Muscha: Thank you and good afternoon. With me are Moshe Gavrielov, CEO and Lorenzo Flores, CFO.

We will provide a financial and business review of the December quarter and then we will open the call for questions. Let me remind everyone that during our conference call today, we may make projections or other forward-looking statements regarding future events or the future financial performance of the company. We wish to caution you that such statements are predictions based on information that is currently available and actual results may differ materially. We refer you to documents the company files with the SEC, including our 10-Ks, 10-Qs and 8-Ks. These documents contain and identify important risk factors that could cause the actual results to differ materially from those contained in our projections or forward-looking statements.

This conference call is open to all and is being webcast live. It can be accessed from our Xilinx Investor Relations website. Let me now turn the call over to Lorenzo.

Lorenzo Flores: Thank you, Rick and good afternoon everyone. In the December quarter, Xilinx’s sales were $586 million, up 1% sequentially and up 3% on a year-over-year basis.

Our sales for the first three quarters of the fiscal year are up 6% compared to the same period of the prior year. Growth was driven by our advanced products, which increased 2% sequentially and reached a new record. For the first three quarters of the fiscal year, advanced products have grown nearly 50% compared to the same period of the prior year. In the December quarter, sales from our 28-nanometer Zynq family, our 20-nanometer family and our 16-nanometer family also passed our targets and reached new records with Zynq approaching 10% of total sales. Total 28-nanometer was approximately flat overall.

From an end market perspective, the largest sequential gains during the quarter were driven by communications with Industrial and A&D up slightly and Broadcast, Consumer & Automotive down. Profitability remained strong during the quarter as both gross margin and expenses were slightly better than guided. Gross margin in Q3 was 69.6%, slightly better than guided as we continue to benefit from our focus on operational efficiencies. Operating expense was $244 million, $1 million lower than guided. Recall that the primary driver of the increase in our R&D investment from the prior quarter was for 16-nanometer tape-outs as we continue to accelerate our technology leadership and brought in our product portfolio.

Moshe will discuss how this R&D investment is beginning to pay off. Operating income for the quarter was $163 million or 27.8%. Other income and expense was an expense of $392,000, lower than guided due primarily to an investment distribution. Our tax rate was 13% for the quarter, better than guided due to several small discrete items. Our net income for Q3 was $142 million or $0.52 per share.

Some key points on the balance sheet and cash flows. We ended the quarter with $3.4 billion in gross cash and $1.8 billion net cash after our debt. Accounts receivable ended the quarter up $113 million at $341 million. This relatively high level of receivables was caused by the pattern of shipments and collections in the quarter and we see no collectibility issues. Inventory was $206 million, up $9 million from the prior quarter and up $10 million from the same quarter a year ago.

Operating cash flow was $106 million and was impacted by accounts receivable, as I just discussed. We expect operating cash flow to improve significantly next quarter as we normalize accounts receivable. In the quarter, we paid $83 million in dividends and we repurchased 3.9 million shares for $214 million, an average price of $54.72. For the first three quarters of the fiscal year, we have returned $665 million to shareholders through a combination of dividends and share repurchases. This is more than 100% of the operating cash we have generated over the same period.

We ended the quarter with diluted shares of 271 million, which included 16.9 million shares from the convertible. A further comment on the convertible. In December, we received conversion notices for $140 million of the $600 million convertible notes consistent with the term of the notes. These conversions will be settled this quarter. Our expectations based on the current range of share prices, is that this will reduce diluted shares by 2 million to 2.5 million.

As we have discussed, capital allocation remains a top priority for the company. We continue to invest in our business to capitalize on our leadership position and we will return capital to our shareholders with a commitment to an increasing dividend over time and through share repurchases. We continue to execute the more deliberate approach to share repurchase I discussed last quarter with the intention of exhausting our $1 billion authorization over the next several quarters. As I turn to guidance, I would like to once again refresh the key points of our annual guidance from our Analyst Day in May. At the Analyst Day, we said we expected revenue growth to be between 4% and 8%, our gross margin to be between 68% and 70%, and our operating expense to be between $930 million and $950 million.

Our performance through the first three quarters of the year and our guidance for Q4 indicate that we are on track to the guidance provided at our Analyst Meeting. In the March quarter, we are expecting sales to increase to $590 million to $620 million, which would put at the center of the revenue guidance provided at our Analyst Meeting. Our backlog is up heading into the quarter and advanced products will continue to grow, delivering new records for sales of our 28-nanometer, 20-nanometer and 16-nanometer products. On end markets, we expect communications to be down slightly with declines from wireless offsetting growth from wired. Industrial and A&D is expected to increase with all three secondary markets, flat to up.

Lastly, Broadcast, Consumer and Automotive is expected to increase driven by a significant increase in advanced driver assistance systems. Our gross margin will be approximately 68% to 70%. For operating expense, we continue to invest in accelerating our leadership position and we expect operating expense to be approximately flat at $244 million. Our operating expense outlook includes $1 million of amortization. Other income and expense will be an expense of $4 million.

Finally, our tax rate is expected to be 14%. Let me now turn the call over to Moshe.

Moshe Gavrielov: Thank you, Lorenzo. I am very pleased with the financial results of the third quarter. We delivered our fifth consecutive revenue growth quarter.

Revenue reached its highest level in nearly 2 years while delivering better-than-expected profitability. We continue to benefit from our diversified, multi-market strategy and technology leadership portfolio. In the December quarter, wireless rebounded significantly, largely driven by 4G business in China and India as well as pre-5G deployment. Sales also increased in A&D and ISM, more than offsetting declines in other end markets. Overall, revenue growth continues to be driven by our advanced products.

For the first 9 months of fiscal year ‘17, advanced products were up nearly 50% versus the prior year period. We anticipate our advanced products will again grow significantly in the March quarter and set a new sales record. This momentum in our advanced products is fueling our confidence in meeting the midpoint – the revenue growth forecast 4% to 8% we provided at last year’s Analyst Meeting. Sales from our 28-nanometer Zynq product family, the industry’s first All Programmable SoC, increased by nearly 20% led by growth in wireless communications and ADAS applications. Our overall 28-nanometer revenue, including Zynq, is expected to increase in the March quarter, set a new record driven by broad end market deployment.

20-nanometer revenue again reached a record level, significantly exceeding our $50 million target. We expect to ship over $60 million of 20-nanometer products in the March quarter, setting a new record. 60-nanometer sales grew significantly in the December quarter to a new record, exceeding our forecast while shipping to multiple end markets, including automotive, data center and communications. We expect 60-nanometer sales to increase again in the March quarter, set a record as broad based customer adoption continues to accelerate. Our decision to increase our fiscal year 2017 R&D investment to both capitalize on our technology leadership and expand our product portfolio continues to pay off.

After reaching critical production milestone that extended our 60-nanometer competitive lead to well over a year, we have accelerated our production tape-outs of this node and are currently shipping 12 unique products to over 300 active customers. We announced the expansion of our 60-nanometer portfolio with devices that integrate high bandwidth memory and CCIX technology. This product family is architected to support performance intensive applications, such as compute acceleration in the data center, high data rate, Ethernet connectivity, 8K video and radar systems. Last quarter, these expanded R&D investments helped us to further develop the market potential for data center acceleration. Xilinx introduced the reconfigurable acceleration stack with a goal to enable mainstream adoption of our FPGAs for accelerated machine learning, video transcoding and big data analytics.

Amazon Web Services recently announced they are deploying our 60-nanometer UltraScale+ FPGAs in the cloud. This has the potential to create a new disruptive business model for our technology and expand the reach to a very broad set of enterprise customers. We continue to invest in data center market development, anticipating the longer-term horizon for revenue returns. The $590 million to $620 million sequential revenue guidance driven from significant growth in our advanced products will manifest itself in many of our end markets, including wired communications, data center, automotive, ISM, A&D, consumer and broadcast. This broad based growth underscores the strength of our diversified, multi-market portfolio.

We now turn the call back to the operator for Q&A.

Operator: The floor is now open for questions. [Operator Instructions] The first question comes from Joe Moore from Morgan Stanley. Your line is open.

Joe Moore: Great.

Thank you so much. I had a question on the broadcast, consumer and auto being down sequentially quite a bit and I guess the ADAS business within that is obviously doing very well, can you just kind of put in context kind of what are the puts and takes in that business and is there something anomalous about the decline that you saw in the December quarter?

Lorenzo Flores: So I don’t think there is anything anomalous in the decline, Joe. I think the issues we had, some inventory positioning on the part of some of our key customers in ADAS. We still see strength in all of the design wins that we have talked about and all the – on all of the end markets – I am sorry, end manufacturer models that we have talked about before, so we don’t have any concerns over that. And as Moshe described, we expect to see a very strong rebound in automotive driven by ADAS in the fourth quarter.

A large part of the decline, as we have been talking about before, has been in the infotainment business as well. So we are not concerned.

Joe Moore: Okay, great. Thank you. And I guess maybe following-up just on that business, can you talk about – there is infotainment headwinds, are there other headwinds to understand and when you have talked about pretty strong growth in ADAS, should we see that result in much higher revenue growth in this segment in the next 12 months or are there offsets to that?

Moshe Gavrielov: Well, if you go back 1 year or 2 years, infotainment was by far the largest portion of the business.

And over the last 2 years, ADAS has grown and infotainment has shrunk. Now, ADAS is by far the largest portion of the business and it generally speaking, continues to grow at a faster rate. So we expect our automotive business to be one of the growth drivers over the next year for sure and see overall growth in automotive in fiscal year 2018 versus fiscal year 2017.

Lorenzo Flores: I think we have discussed before Joe that the breadth of our design wins and we are at a point where it’s the take rate of the option packages that will help determine the growth of our business there.

Joe Moore: Okay.

Thanks so much.

Lorenzo Flores: Thank you. Next question?

Operator: Your next question comes from Ambrish Srivastava with BMO Capital Markets. Your line is open.

Ambrish Srivastava: Thank you.

Moshe, I had a question on OpEx and really not looking for the guide, but really the framework of how should we think about what are you and Lorenzo looking at as we look at the next fiscal year, because last year you negatively surprised everybody, but then you articulated why you are doing it and then since then you have kind of shown to us why and you gave some good metrics today. So how should we think about how you are thinking about what you and Lorenzo are looking at over the next quarter or two quarters. And then my quick follow-up for you Lorenzo, accounts receivable are up a lot and you said it should normalize back in the quarter, did I get that correct that there will be a big delta on a positive side for CFO?

Lorenzo Flores: Thanks. So the quick answer is yes. And I will let Moshe answer the bigger picture framework question and I will probably add in at some point.

Moshe Gavrielov: Okay. So we will be giving guidance for next year, some initial guidance at the next earnings release and obviously full guidance in May when we have our Investor/Analyst Day in New York City. But generally speaking, if you look at what we had committed to do, we had said that we are investing. We are investing to exploit the technology lead we have and it would be foolish not to do that both in the short-term and definitely in the medium-term and long-term. Having said that, we are committed to getting back to the 30%, right.

So what you will hear in the next earnings release and in May, you will hear more details with regards to that, but we should be moving back in that direction generally speaking. And how quickly we will get there, it’s too early for me to give you any clear guidance on that, but that direction we are committed to regardless.

Lorenzo Flores: Yes. So actually, I don’t have much to add to that Ambrish, but I think I will expand a little bit on accounts receivable because I am sure other people may have the same question. If you look back, we have had an occasional spike of accounts receivable and you probably generally followed up with the quarter where they come down and we obviously collect the cash.

Obviously, we are trying to manage this, but most importantly, when we service our customers, when they want the product. And so that impacts when we can collect for the revenue that we have. So and looked it very closely, we don’t see any issues, as I have said earlier, on collecting the accounts receivable. So we expect to turn that back down to a more normal level in this quarter.

Ambrish Srivastava: Okay.

Thanks for the clarity.

Moshe Gavrielov: Next question please.

Operator: Your next question comes from C.J. Muse from Evercore. Your line is open.

C.J. Muse: Yes. Good afternoon and thank you for taking my question. I guess, one question, two parts, so as you think about 60-nanometer, your share gains there and the acceleration that you are seeing there with your customer base, I am curious on two fronts, one, when do you think we could get upwards of $100 million or so incremental revenues either on a run rate basis or overall across ADAS and data center. And then secondly, now that you have put all the spending on the OpEx side in fiscal ‘16 – fiscal ‘17, sorry, how should we think about OpEx trajectory into fiscal ‘18? Thank you.

Moshe Gavrielov: Okay. Let me do – I will try to answer your first part of your question probably a little differently than the way you had asked it. Just to remind you, we – typically takes a very long period of time from when we introduce the technology to when it actually provides huge revenue. And an example would be 28- nanometer, which is still not at its peak and we actually think that it’s a minimum of 2 years, maybe even further away from its peak. Now, if you look at 20 and 16 together, we tend to look at them as one node just due to the way they were introduced.

They came relatively close together. And the combination of those – these two nodes is doing very well across all markets. 16, it’s still early days. We have these 300 customers and 12 unique devices that are shipping, but none of the customers are yet in volume production. You are right in highlighting that 16, two of the markets that will unquestionably drive that are automotive and data center.

They are not the only two markets. They actually had much broader applicability. And of the two – those two data center tends to happen a little faster than automotive, but data center for us is a nascent market. It’s very early in its development and that’s why we are very careful in highlighting that the investments we are making in data center now will likely pay off a few years down the road, and we sort of highlighted 2020 is when it will be substantial in the potentially hundreds of millions of dollars. So to answer on 16-nanometer, overall, I would be disappointed in 2 years if it isn’t well over $100 million worth of business.

Which markets will constitute that $100 million of sales, it’s still early to tell. And for this year, it will be significant, but it won’t be at that $100 million for this fiscal year for sure.

Lorenzo Flores: And the other part with the OpEx trajectory?

Moshe Gavrielov: Sure.

Lorenzo Flores: Okay. So C.J., I think I’ll just echo a little bit of what Moshe said in response to Ambrish’s question, which is we will obviously provide more details coming out of our fiscal year and at our Analyst Day.

But I think that what we said in the past is what we are thinking now, which is we don’t expect the slope of the R&D investment to be at the same angle as this past year and we are managing the company to get back to the 30% up and greater operating margin level in the intermediate term, if you will. C.J. Muse: Very helpful. Thanks so much.

Moshe Gavrielov: Thank you.

Next question, please.

Operator: Your next question comes from Tristan Gerra from Baird. Your line is open.

Tristan Gerra: Hi, good afternoon. Could you provide an update, if you could in terms of the timing for the next geometry node initial ramp? And also, what type of product should we expect, will it be all very high end or stacked architecture or will you be targeting mid-range as well with that new node?

Moshe Gavrielov: Okay.

Well, it’s the risk of quoting our President it’s going to be phenomenal. But back on the serious note, we are very excited about the 7-nanometer. We are delighted that TSMC continue to provide leadership product. We definitely benefit from the fact that they now have caught up and providing leadership technology at the earliest point in time. We expect our 2018 to be the year that we tape-out and provide first silicon.

It will be – and this is where the phenomenal ends it stripes, it will be a very broad product offering, it won’t be just the high-end product offering and we believe we have – we will have the engineering innovation that enables us to cover most of the market, definitely not the low end with the 7- nanometer technology. So, it will be a very broad technology, applicability and we will have very significant impact on numerous of the markets we service. It will likely not at least initially have a low end footprint just because when you get to those new advanced nodes, they tend to be a lot more expensive, both in terms of mask sets and in terms of wafer costs and it takes time to get that to the point where you can actually address a low end portfolio. So, what we have done to address the low end portfolio in the interim is expanding 28-nanometer, which is a much more cost-sensitive and mature node to address that market, so we can protect and actually expand our position at the low end, because we feel that the low end has been relatively underserved, and it’s an opportunity for us. Hence, the additional investment we are making at 28-nanometer at the low end.

Tristan Gerra: Great, thank you.

Moshe Gavrielov: Sure, thank you. Next question, please.

Operator: Your next question comes from Steve Smigie from Raymond James. Your line is open.

Steve Smigie: Great. Thanks a lot, guys. Appreciate my questions. My first question was a little bit on the competitive front, you saw Intel/Altera product there, 14-nanometer part. And I am just curious now it’s a little bit into that what we have actually seen from customers, does it matter or are you guys already so far ahead since you are a year out that it will take a while to even really get manage lion’s share of customers?

Moshe Gavrielov: Well, we never underestimate our competition, least of all, Intel in that regard.

But having said that, they are doing extremely well at 16-nanometer. The product is very mature. The yields are exceptional. The functionality of the product is great. That’s why we are investing in getting the production mask sets with very fine tweaks to solve the teething issues that the product had at the very beginning.

And it’s doing really, really well. So our competitive position, we feel is incredibly strong at this node just due to the maturity of the product and the breadth of the family and the quality of the design and the operations at this point. So, it’s going really well. And we obviously are intent on continuing to exploit this lead and we assume there will be competition, but we are moving forward as quickly as possible to enhance and consolidate our position and it’s working quite well for us.

Steve Smigie: Great, thanks.

And just as my follow-up, you have had quite a bit of success with Zynq with good growth rates there. Now, it’s been out there for a little while. Can you talk about what the growth trajectory over the next year or two might look like at this point?

Moshe Gavrielov: Well, it’s – Zynq was the groundbreaking new product for us and it was introduced relatively late in the 28-nanometer cycled product introduction, it was about a year after the initial device is. It was targeted at three markets, wireless, automotive and industrial. What has happened is because it is a really a system-level solution, because it has a very powerful CPU system embedded inside it, and that’s the first time something of this sort was available.

It took a little longer for those designs to move into production. But what has happened now is that they are moving into production. Wireless and automotive are further ahead. The big ramp in industrial is still ahead of us. We have a mountain of design wins, which are now moving into production.

And if you look at 16-nanometer, we are already seeing very fast adoption of the follow-on MPSoC device, which is going to market much faster than the initial generation was. So this is a significant portion of our business, because it started from zero and was competing with over $2 billion of other markets. It’s of other technologies. It took time, but it’s already at 10%, and we expected of our overall revenue and we expect it to be a very significant portion and a growing portion of the revenue with each subsequent node that we come up with. And it fundamentally moves us to a different role in our customer systems.

And we are now competing against ASSP quite successfully in a broad range of markets. And if you look at our automotive business, our position in ADAS is very different from our position in infotainment. In infotainment, we were a companion device. In ADAS, we are the central device in the system. And that transition that we have made into the hearts of our customer systems is continuing at a fast rate.

And Zynq is – in the 60-nanometer version and definitely the 7-nanometer version, which will be even more powerful and capable will enable us to do that in more and more of the markets we serve.

Steve Smigie: Okay. Thank you very much.

Moshe Gavrielov: Thank you. Next question, please?

Operator: Your next question comes from Ross Seymore from Deutsche Bank.

Your line is open.

Ross Seymore: Hi guys. Thanks for letting me asking you question. I just want to go back a little bit to the December quarter and could you explain a bit more about what broadcast consumer did, you mentioned that auto was down from an inventory digest and that’s understandable, but it seems like between what you delivered in December quarter in that entire segment and what you are guiding for March, it – there must have been a pretty steep decline in auto and then a pretty big snapback implied in your guidance, is there any color you can give on that, please?

Lorenzo Flores: Yes. So Ross, the other end markets in that group were also down a little.

And you are right, our guidance for next quarter assumes – we really are expecting a significant snapback from automotive.

Ross Seymore: Got it. Thanks for that color. And then I guess following up on the prior question about Zynq, Moshe. Conceptually, how do you look at that as far as being an incremental adder of revenue to the company versus something where there is some resemblance of a substitution effect and I guess the reason why I ask it that way is the 20% growth sequentially, the 25% going from very little to now 10% of your revenues, those are very impressive numbers, but if I look segment-by-segment where that Zynq product resides, those are not growing nearly as fast as what I would think if Zynq is growing in there, so is there are some offset to that we are not appreciating or any description on that dynamic would be helpful?

Moshe Gavrielov: Okay.

Well, generally speaking, if you look at our business, we tend to serve a lot of markets and those markets move at difference rates. And as a result, we typically have numerous – revenue from numerous technology nodes at the same time. So first order effect and this is before I talk about Zynq against vis-à-vis other non-Zynq technologies. So first order effect, you have older technologies that are shrinking and you have newer technologies that when we do well in those nodes and now we have three nodes in a row where we have done well, they tend to be growing. And what has happened over the past few years is it took time for the advanced nodes to get to the point where they could, even though we had well over our natural market share, but it took time for these advanced nodes to be able to compensate for the old nodes, which were shrinking.

Now, if you sort of do the arithmetic, we are at 47% on the advanced nodes and 53% on what we call core technology. And so the growth – the accelerated growth in the core technology is generally compensating for and then some for the shrinkage in the old nodes. This trend of new node cannibalizing old nodes is just parts and parcel of the semiconductor industry. We are not very different than that. So every time we introduced a new node, it – to some extent, if all the customers need is the exact same device, but a little cheaper, they can achieve that by going to the new node and that cannibalizes.

What sort has happened is over the years, we have provided a very significant integration play and this is where I am getting to answering your question, just in case you are wondering. This is where the integration enables us to capture a larger part of the customer’s bill of material. And Zynq is part of that. And so if you look at the bill of materials now, we actually have the opportunity to replace ASSPs or standalone processors or standalone microcontrollers, at least in some of the markets, not all that we read. So yes, in that regard, Zynq is a market expansion play, but it very similar to the core business, it does tend to have some element of cannibalization.

And that’s why when you sort of look at the numbers, which we have projected, we are sort of on course to deliver the first of what our hopefully several 6% growth years, which we said was our goal last year. And we are pleased to have delivered on that. And part of it is through market expansion. And Zynq is a market expansion play. But it provides some market expansion, but you are right, it does cannibalize some of the revenue from the previous generation.

And if you look at wireless as an example, which is a market, which we are now serving with Zynq, and before, we used to serve with what I would call advanced, but cleaner FPGAs, then yes, that’s a market where we are cannibalizing the old revenue. But truthfully, if we didn’t have the Zynq, we would not be playing in that market with these advanced nodes. So in some cases, it’s an expansion. In some cases, you just need to move forward in order to standstill. And that’s the name of the game is the semiconductors.

Ross Seymore: Thank you very much for that detail. It’s very helpful.

Moshe Gavrielov: Thank you.

Operator: Your next question comes from Chris Danely from Citi. Your line is open.

Philip Lee: Hi guys. This is Philip Lee on behalf of Chris. Thanks for letting me ask the question. I just want to dive a little deeper into the data center, the [indiscernible] opportunity there, how much revenue do you think you guys can do this year and how much of that is in new applications, such as hyper-scale or compute?

Moshe Gavrielov: So the numbers we have projected are potentially hundreds of millions of dollars in and around calendar 2020. And we are still at the early stages, right.

So and that’s why our choice of words has been consistently very careful in terms of setting expectations. Now what has happened is we are seeing tremendous pickup and tremendous interest in public design wins with Baidu and AWS position us to – as this market evolves and as it emerges to service it in a unique way. But it would be premature to assume that there is going to be a huge growth in terms of revenue in fiscal year ‘18, right. It’s still too early. These deployments are new deployments.

We are committed to supporting them extremely well to enable them over time to develop into significant business, but this is not fiscal year 2018 where it will make a significant dent in whatever number we commit to for this upcoming year. It’s still a little further out than that.

Philip Lee: Thanks, Moshe. Very helpful. And as a follow-on can you talk about your end markets the relative growth you expect this year, do you see any of the segments going faster or slower relative to the other ones?

Moshe Gavrielov: We will provide more insight, significantly more insight into that in May.

It’s still premature, but no, we feel that the technology leadership and the broad market participation are playing in our favor. And sometimes we predict that market A will grow and actually, it’s a market B that grows. And that’s one of the benefits of the broad market participation that we have. So, we will give you more data in May.

Philip Lee: Okay, thank you.

Moshe Gavrielov: Thank you.

Operator: Your next question comes from John Pitzer from Credit Suisse. Your line is open.

John Pitzer: Yes, good afternoon guys. Thanks for letting me ask the question and congratulations on the good quarter.

Lorenzo, maybe I can ask the OpEx question a different way. If you look at sort of the ramp of OpEx throughout the fiscal year, you’ve got a much bigger spend in the second half of the year such that if you just kind of straight line the March quarter guide, OpEx would be up 5% year-on-year in fiscal ‘18. I am just kind of curious given sort of the pace of tape-outs, especially in the back half of the year, this fiscal year, is there actually opportunity to take OpEx absolute dollars down going forward or is the opportunity set such that it’s really about sort of the slope of the ramp and the rate of growth?

Lorenzo Flores: Yes. So John, I certainly understand both why you asked the question and how you kind of framed it, but I will go back to something Moshe has said and we are going to try very hard to stick to over the next 3 months. We want to provide you a much clearer view of what our 2018, or FY 2018 looks like as we come out of this fiscal year and into our Analyst Day.

But I will address a couple of your points may be to help you out in the interim. We do have, as you noted, significant increase in the second half of the year from tape-outs that we are taking advantage of our leadership position and we think it was the right thing to do. We told you we are going to do it at our last Analyst Day and as we are doing it. We are also growing our overall R&D capability. So, those are factors we are going to have to balance as well as what we are doing on the next nodes in the next fiscal year.

And I think it’s probably more beneficial to your understanding of our operating model next year to wait rather than try to guess at it right now.

John Pitzer: That’s helpful. And then Moshe maybe as my follow-up, notwithstanding your comments earlier about putting expectations around acceleration, just given what we are seeing going on in the GPU market right now and the uptake of GPU to the data center, given that it would seem that the FPGA architecture in a lot of instances for acceleration applications might actually be superior, what is it that that’s allowing GPU uptick to be faster than FPGA uptake for these applications in the data center in the near term? Is it really a software stack issue or can you help me understand that?

Moshe Gavrielov: You are absolutely right. This was identified as a growth market by the GPU companies primarily NVIDIA, a long, long time ago and I give them credits for identifying that in their desire to diversify beyond their core PC and gaming focus, they invested very heavily on the software front. What we are doing is we are now taking a page out of their book.

It’s not necessarily the same page. It’s a little bit of a different page. And we have to do that in order to enable broader deployment fundamentally, because the customer base – the target customer base is not FPGA savvy and it’s very unlikely that they will become FPGA savvy. Hence, you need this higher level of abstraction. And so the core benefits in the silicon architecture are there and we are continuing to drive those in all of the current generations.

Some of the customers have the capability to deliver on that by themselves. Some of them need quite a bit of help and we are providing that through things like this rest architecture solution, which should make it easier for them to access the technology. Now as that is done and then you can look at some unique deployments like the Amazon one is unique and that it’s FPGA as a service, as we support that effort to enable them to expand the markets that could have a significant leverage with regards to the breadth of people who have access to it. That’s why we are so excited. That’s why we are getting some benefit from probably the biggest customer in the world and their business model and their approach to doing this.

So that could, as it gets deployed and it’s brought to the market successfully that could accelerate this entire process. So between that and the rest architecture, I think those are the things that we believe will enable us to get to market much faster. And it won’t take us the same period of time, but it took other competitors in the past to develop this market.

John Pitzer: That’s helpful. Thanks, guys.

Appreciate it.

Moshe Gavrielov: Sure. Thank you.

Operator: Your next question comes from William Stein from SunTrust. Your line is open.

William Stein: Great, thanks. Actually, I have a similar question to the last one, but maybe from a more generic perspective, you are highlighting this deployment at AWS. And as you just said software has been a big differentiator that’s perhaps given NVIDIA some early traction there. But can you talk a little bit more generically about the architecture of FPGAs versus GPUs and then your architecture versus what Intel/Altera has, and why you expect to capture more share of this market relative to both of those solutions?

Moshe Gavrielov: Okay. So, let’s step back.

The market really is very, very nascent and the biggest player today is Intel through their CPU offering. And now a fast emerging player who has pretty significant deployments is NVIDIA with some of these hyperscale players. What has become very, very clear is as the market is evolving, it’s also segmenting. And there are – it’s becoming clearer what elements are served really well by NVIDIA, for example and what elements which sometimes overlap, but sometimes are totally different are better within all programmable solution. So in these early days, its first-mover advantage is significant.

It does give a lead, but there are inherent advantages, which we believe we have. And Intel clearly recognized those and that was one of the major drivers as they publicly have stated for their acquiring Altera was to be able to address that market. So, there is a whole set of applications, which are learning-oriented where NVIDIA and GPUs have probably a better architectural fit, but most other applications, which are potentially larger, are better serviced by us and programmable logic. Now with regards to the competition against Intel, it’s again very early 30 days. We believe that, a), anything which is non-Intel related were going to be the obvious solution, right? And so there is not insignificant business, which potentially could be serviced by AMD on x86 by IBM with their power solution and by the host of ARM guys and that will fall very naturally into our lap for obvious reasons.

Having said that, that’s not the only market we can play in, because we think that even – and we are hearing this from the major players even where they are currently using Intel x86 processors when they are looking for an acceleration, they are finding that our leadership in terms of silicon technology and the approach with the pooled architecture where basically you could have a pool of Xilinx’ FPGAs, which provide acceleration, is more attractive to them than the approach, which Intel pursues which is more likely to strengthen their CPU hegemony with the one CPU, one FPGA solution, right. And that’s where we think that for a lot of these players, we – even if they end up using Intel x86, they are likely to want to pursue an acceleration solution, which is not dictated by Intel. And that’s where we have the opportunity. All-in-all, we are very careful in terms of setting expectations. And hence, it’s still early days.

We expect to see hundreds of millions of dollars in the 2,000 – calendar 2000 [ph] timeframe. And if we do well and all of these other things work our way, potentially it could happen maybe faster or maybe larger, but it’s too early to commit and we are very careful in terms of highlighting that.

William Stein: Great. If I can get one follow-up, please, in the ADAS market, the big sort of category that I am aware of is this sort of hovering view stitching together for a camera corner view in some orders, but I am suspecting maybe it’s much more diverse and at the minimum, you might have some sensor fusion applications, could you comment on the diversity of the design wins and the pace of ramps in those ADAS application? Thank you.

Moshe Gavrielov: Okay.

So you are right that this is emerging and it’s emerging very, very quickly. And the needs are quite fluid and they are growing and there are different requirements. And so there is a broad set of design wins, which address different approaches. And sensor fusion is sort of a buzzword for highlighting that there is a whole family of different sensors that are required, some of which point forward, some of which point inward, some of which point sideways, some of which point backward. And this market is evolving at a very, very fast rate.

What we are seeing is due to be uncertainty in the market and the – and due to the fact that for automotive, it’s moving, it’s an unprecedented upgrade [ph]. There is a lot of uncertainty, and that’s where we believe we have tremendous benefit because our solutions tend to enable a lot of flexibility as the markets change. And we are seeing that the major benefit with regards to competition with other companies that have a more ASSP oriented rigid solution. And we are seeing just a broad range of these different approaches and the design wins actually fall into numerous categories for us that – and enable us to address a lot of these. The early mover advantage we had here is helping us, because we have the relationships and it’s already approaching 10% of our business, which actually makes this one of the biggest players in this market.

Just don’t have the marketing budget to sort of spray that over the Super Bowl, but we are doing really well.

William Stein: Great. Thanks very much.

Moshe Gavrielov: Thank you.

Operator: Your next question comes from Toshiya Hari from Goldman Sachs.

Your line is open.

Unidentified Analyst: Hi, this is Charles on for Toshiya. Thanks for taking my question. Just going to head on the data center side again, such tremendous opportunity, but I was wondering, as you kind of size your opportunity in hundreds of millions of dollars out to 2018, 2020, what are kind of the share – what kind of share does that imply for you guys in the acceleration market and I was just wondering what your TAM assumptions might be. And then I had another one, in terms of displacing NVIDIA, just wondering if you have any sense of the hurdles, maybe from a performance standpoint or a power standpoint that you have to overcome given kind of your weight in the game on the software side?

Moshe Gavrielov: So the TAM is huge.

The TAM today is – it has been growing maybe not as quite as quickly as it was before, but it has been growing and it’s approaching $20 billion. And that’s if you look mostly at the service side that primarily with x86 and the GPU side and that’s sort of a market, which is already at that level now. That market is changing and it’s changing to be serviced by more cost effective and power effective solutions and that’s where we actually, for a broad range of applications can provide high performance, which we have demonstrated at a lower power point than competitive solutions, be they processors where the gap tends to be quite big and GPUs where the gap is somewhat smaller, but it’s still substantial. And so if you look at the big market, you can say well, $20 billion market and we do not want to leave anyone with the impression that, that is all serviceable by Xilinx. However, I do believe that there is billions of dollars available to be addressed by alternate solution.

And of that, we believe we can easily capture 10%. And if we do well, we can capture more than 10% of that over time. And the benefit will be higher performance to lower power point. And again for us, we have to accept the fact that we recognized this leg. And we now have to – we have the relationships with the sever hyper-scale guys.

We have design wins with several of them in significant programs, but it’s still going to take time for that to translate into massive deployments. And that’s why we are providing these numbers, right in the hundreds and millions of dollars, which sort of correlate to the market share – potential market share of the serviceable market that I shared with you.

Unidentified Analyst: That’s helpful. Thank you. And then just as a follow-up, you had mentioned some revenue from pre-5G builds, I was wondering if your outlook kind of on the timing of 5G has changed at all and if so, what’s kind of driving that? Thanks.

Moshe Gavrielov: So 5G originally because monetization of 4G took a little longer, there was an expectation that 5G would get pushed out about 2 years ago, maybe a little longer than that. It appeared that the tide had turned and there was a mad rush to provide 5G solutions. And we are benefiting from that because technology leadership has given us the full position with regards to this pre-5G deployment. Having said that, we think that the prime time for 5G is still likely in the 2020 timeframe and it’s not going to be massive before that. If it is, we will be in a good position to address it.

But we are looking at 2020 for the big wave of 5G deployments. And we think between now and then, there will be a lot of interim steps, which we are well positioned to service, but they are less likely to be true 5G. In terms of marketing statements, I am sure there will be a lot of ultra stating marketing statements, which will claim 5G and all of that, but we won’t be – the real thing there will be a step in the direction.

Lorenzo Flores: Operator, we can take one more question.

Operator: Your next question comes from Chris Rolland from Susquehanna International Group.

Your line is open.

Chris Rolland: Hey guys. Thanks for squeezing me in. So I am just thinking about India, perhaps that’s the next area of infrastructure of growth and I know won’t be as large as China, but also, we can understand that the equipment may not to be as heavily dependent on ASICs, so thinking in terms of actual FPGA content, what might India look like in revenue terms for you guys compared to China?

Lorenzo Flores: So we are actually pretty pleased with our position in India. We are benefiting right now – the strength in our wireless business, a significant portion of it comes from India-based deployments.

So as that expands over time, we think we will continue to take advantage of it. But the outlook – specific outlook for the deployment is a little bit more murky, but as I said just a second ago, we are really well positioned with respect to our customers that have a position in India. We don’t think it’s going to be of this scale in the near-term that the China-based deployments were and we also don’t think it’s going to be as rapid deployment has happened in China. We still think it’s a good amount of business for us.

Chris Rolland: Great.

Also, on the auto side of things, at CES I think there was at least one announcement from a large German Tier 1. I think they currently use Zynq that they will be switching to NVIDIA solutions. I think ADAS is a great market for you guys, has been Zynq it will continue to be, but what gives you guys confidence that we won’t see ASIC replacement in ADAS as kind of market standards start to mature here?

Moshe Gavrielov: So I am not quite sure who you are referring to because we have a plethora of design wins and we also tend to have follow-on design wins with all of the current players and new ones. The market is evolving and it’s changing quite rapidly. And there are people who have the strong position in infotainment who are trying and planning on playing in the market.

It’s – they are not coming at it from an ADAS-like approach, which tends to have a lot of distributed processing. They tend to have a approach, which is very, very super CPU or super GPU oriented with the central control. It’s still too early to know which of these approaches will work. And what is happening is as the market is very fluid, the needs are fluid and the competition is very fluid and the automotive market is going through a huge transition in this regard probably in unprecedented transition. We actually believe that having been there early on ADAS and being involved in all of these next steps is very attractive for us.

And that’s why we believe that this will continue to drive our growth going forward. 10 years from now, it’s quite possible that the market will be commoditized and there won’t be the room for the changes and the uncertainty as to how the market will evolve. And at that point in time, it’s quite possible that ASIC/ASSPs will gain market share, but I think there is several years ahead of us of fluid market where the flexibility of our capabilities and the strength which a lot of the strength, which is inherent that we just talked about in data center is actually very germane to automotive, right. And that’s where you will find a higher performance at a lower PowerPoint solution with the programmable logic element. So we are quite confident that we are – for the next several years, we are in a very good spot with regards to monetizing this market.

Chris Rolland: Yes, I agree. Great answer. Thanks, Moshe.

Moshe Gavrielov: Sure. Thank you.

Rick Muscha: Thanks for joining us today. We have a playback of this call beginning at 5 p.m. Pacific Time, 8 p.m. Eastern Time today. For a copy of our earnings release, please visit our Investor Relations website.

Our next earnings release date for the fourth quarter fiscal year 2017 will be Wednesday, April 26, after the market close. We will be attending the Morgan Stanley Technology, Media and Telecom Conference in San Francisco on March 1. Also, we are hosting an Analyst Meeting in New York City on May 22. Please save the date and we do look forward to seeing you there. More details will follow.

This completes our call. Thank you very much for your participation.

Operator: This concludes today’s conference call. You may now disconnect.