Altera Corporation's CEO Presents at Citi Global Technology Conference (Transcript)

| About: Altera Corporation (ALTR)

Altera Corporation (NASDAQ:ALTR)

Citi Global Technology Conference

September 04, 2013 09:45 AM ET


John P. Daane - President, Chief Executive Officer and Chairman


Glen Yeung - Citigroup

Unidentified Speaker

(Call starts abruptly)

This presentation contains forward-looking statements which are made pursuant to the Safe Harbor provisions of the Private Securities Litigation Reform Act of 1995. Forward-looking information generally refers to any information relevant to a future time period. Investors are cautioned that actual results may differ materially from these forward-looking statements and that these statements must be considered in conjunction with the cautionary warnings that appear in our SEC filings, available from the company without charge.

Further this discussion and presentation may repeat elements of prior guidance and then in so doing the company is neither reaffirming nor modifying their prior guidance. Thank you.

John P. Daane

Good morning. We typically get from investors, our questions really boil down to our central question, just a couple of themes which include our expense structure going into 2014, cash usage or our capital structure, our growth opportunities in some of the key vertical markets, market share of 20 nanometer and then why are you moving to Intel for foundry business for ’14.

And I am sure we are going to have plenty of time to go through on those questions and others but in particular if you talk about 28 nanometer and why we are doing Intel with '14 hard to talk about, a little bit easier to show on some (inaudible) with some graphics. So what I am going to do is talk about those two questions and then what we’ll do is also then we’ll go through questions that Glen and the audience has in just a second.

This breaks down, this chart is the first five years of revenue per node for both us and the competition. And what we’ve done in the top half of the chart have shown 130 nanometer and 65 nanometer. The high end is in blue the low end or low cost is in green. You can see what's fairly consistent in it is the high end was about three quarters of the revenue of the overall nodes.

When we get into 40, 45 this is really where we introduce as an industry the mid-range family and as you can see the mid-range family started to take a slice predominantly a little bit from the high end a little bit from the low end.

As you move into 28 nanometer this is what Xilinx announced about a year ago in terms of what they thought the splits would be, what we also we also announced a year ago in terms of what we thought the family splits would be for 28 nanometer, again over the first five years and what you see is the high end still is over 50% of the revenue, of a node, mid-range is little bit higher, taking again a little bit more market share from the high end, also the low end may be at about 30% market share or a percentage of the overall node and then lower cost at about 20%.

If you look at the revenue difference in 28 nanometer between the two companies first of all what you see is that Xilinx has really captured the red. They probably have over 80% market share in the mid-range in 28 nanometer. If you look at the blue in terms of the company’s public announcements we have about 80% market share of the revenue in high end which is in blue and right now the low cost area is roughly a split between the two companies.

Now Xilinx does have coming out the last quarter obviously a higher revenue amount within 28 nanometer and that difference is in that silver slice which is the prototyping business and it’s important that everybody understand what this is. We as an industry do sell our chips to other semiconductor companies who use them in order emulate a prototype design before they go take that out and manufacture the particular design.

We sell those very high end devices for a lot of money usually between $25,000 per unit to 50,000 per unit. As such they are again very expensive, nobody is going to carry it in to production but early in a technology's life it actually can accumulate a lot of revenue. They usually last for about two years and then when you introduce your next generation technology all of that business moves to the next generation, because they’re really interested in the highest density parts.

So in the first two years at $25,000 to $50,000 per unit you can actually get a lot of revenue very early on from the prototyping business but when we look at five years of cumulative revenue that business is obviously swamped by the production that you have in the other families. So if you look at the delta between the two companies right now it’s predominantly in that silver band which again will transition over to 20 and 14 as we’ll talk about in just a moment as we get into next year.

And ultimately if you look at the production that's left we have 80% of the blue, got about 20% currently of the red, we split the green. We obviously have over 50% market share for 28 overall, again looking at cumulative revenue at this node, our overall market share within FPGAs is only about 42%. So we feel very comfortable like we had in the last generation well over 50% market share we do again in this node we’ll continue to take market share as we go forward.

Now as we go into 20 nanometer and 14 nanometer the reason I wanted to show again that pie chart is also explain the strategy. We will be updating our mid-range family that will address our competitiveness in the red as we move into 20 nanometer and then we’re going to be going to 14 nanometer for the high end.

And really what that’s going to do is position ourselves as I’ll talk about as the only company that will participate in high end FPGAs as we move into the next generation node, which means Altera is really set up to capture well north of 50% as we move forward. And let me sort of address this through the next couple of foils.

First of all, in terms of the products, Arria is our mid-range family. We’re moving to 20 nanometer planer with TSMC, a very successful product. We will be prototyping our first chip or phasing out our first chip, next month in October. We already have over a 1,000 customer engagements. So this is going extremely well for us in terms of traction with customers.

Stratix is our next generation high end. It will be with Intel. And again as I’ll talk about in terms of the next die prototype, this will really position ourselves as the only company, competitively in the high end, which is again over 50% of the revenue of the industry.

In 20 nanometer, with TSMC both Xilinx and ourselves are developing devices. And our device will be a mid-range, I would bet Xilinx will also introduce a mid-range off of that platform. We’ll be very competitive because there are a number of things that we’re doing differently within 20 nanometer, I think that are going to give us good performance, good cost structure, good power structure to compete.

Xilinx will be moving to what TSMC calls, the 16FF or a FinFET vertical transistor for its high-end generation. Two things I would say about this and this really points out why we moved to Intel. Number one, what TSMC has said is 16FF does not scale from 20 SoC, which means that die size will effectively be exactly the same.

Number two, what Xilinx has said is that the wafer cost for 16FF is higher than 20 SoC. So on your right hand side of the foil, what you end up with is a more expensive die than on the left hand side of the foil, if you move to 16FF. And you cannot sell more expensive product into the mid-range, because again, mid-range is price sensitive. So that really then determines it can only be high-end product.

We are moving to Intel at 14 is a true 14, which means it scales as Intel said [inaudible] continues for them, you get a much smaller die size, which means your costs go down versus our competition which actually sees our cost structures go up. Because we’re on a more advanced transistor we get higher performance, and lower power. Obviously the cost advantage that I mentioned, and then we also get a significant density advantage.

And because of that as we move forward we can scale the performance of the Stratix series, double the performance, more than double the density, more than halve the power over the previous generation. And there is really nothing that our competition can do because if you look at the roadmap of the semiconductor industry from a foundry perspective, there is nothing that catches up with Intel’s 14 nanometer for the next four to five years.

So we think we’re fine at 28 nanometer in terms of market share and then as we move forward, our selection of Intel as well as our selection of TSMC will put us really in a lead in terms of programmable logic to come.

So with that, I’ll go back to questions from Glen.

Question-and-Answer Session

Glen Yeung – Citigroup

Thanks John. I want to ask a question after that. But let me do one thing just to get out of the way, we’re asking three questions of all the companies. It can be relatively short answer because we’re really just trying to compile the answers across a group of companies.

First one is, would you characterize your overall visibility in your business as better or it's the same versus the quarter or a year ago?

John P. Daane

Visibility is better, our book-to-bill is over one last quarter. With that people are still very cautious though, and so they’re ordering only really when they need to. So certainly visibility is better.

Glen Yeung – Citigroup

Okay. Would you characterize Altera’s opportunity for productivity gain better versus, let’s thing about that kind of factor.

John P. Daane

I think that gets into the discussion that we had in terms of expenses for next year. We have gone through a three year period of time where we've expanded our R&D spending to implement microprocessors to open up the embedded market for some IP in order to replace ASSPs. We’re also doing three FPGAs in parallel today, we’re doing Intel, high end TSMC mid-range and then we’re doing another with TSMC for the low end. That’s more FPGAs than we’ve done.

Now going into 2014, what we’re doing is flattening out expenses. So slight increase in base expenses, two acquisitions where you get -- two small acquisitions where you get 12 months of expenses versus six months, slight increase in OpEx initially in our business model because our gross margins are high. With revenue growth we should be able to significantly expand our earnings at a much faster rate than revenues moving forward.

So in general, I would say from a productivity perspective, going into a time where it’s much better for our company because a lot of our base investments are done.

Glen Yeung – Citigroup

Okay, and then the third question is use of cash, can you just rank them between buyback, M&A, dividend or other?

John P. Daane

So, we’ve announced a couple of things. Number one is, we’ve just recently increased our dividend by 50%. Number two we’ve announced that we are going to over the next two years purchase about 41 million shares of our stock, which is about 13% of our outstanding shares and so a very aggressive moves to return that cash to shareholders. We’ve done a few minor acquisitions over the last 20 years, got a tremendous number of opportunities in that space. So mostly buybacks and share repurchases.

Glen Yeung - Citigroup

Okay. Just because of your size I want to understand a little bit just talking about the Intel relationships. And I think Intel's been a bit more open about this than Altera has. But it’s exclusive at the moment and may be if you could just articulate the sort of nature of that exclusivity particularly as you move beyond your initial work with them at 14 as you go further ahead.

John P. Daane

So as people may know Intel was already working with a couple of start-ups within the FPGA field. Obviously they can’t then exclude those individuals. So what we did in the agreement as we said this was exclusive to mainstream FPGA companies for FPGA or products that are mostly FPGA or have FPGAs in them.

Who does that include? Well obviously the mainstream companies would be ourselves, Xilinx, Lattice, Microsemi. Again anybody outside of the two start-ups that they already are working with, it’s exclusive at 14 and then it's exclusive moving forward beyond that.

We think that advantage is key because again if you look at the foundry industry while people are calling their products different names like 60 nanometer or 40 nanometer or 10 nanometer if you actually look at what they are implementing it’s something more closer to 20 and then moving forward may be you get to the 16. But nobody is getting close to where Intel is on process technology.

Probably I'd say for the next five years it’s probably much longer than that at this point.

Glen Yeung - Citigroup

Great, your history with TSMC is a long one and so when we think about this shift to Intel do we notionally think the same way this is going to probably be a pretty long term relationships?

John P. Daane

Yeah we have very few relationships that we do. We have two packet assembly companies that we work with, we have two substrate manufacturers we have one foundry. So we’ve been a company that’s been very loyal to few companies for a very long period of time.

Note that TSMC, we had a 20 year exclusive relationships, still most of our business in-fact all our business today is still centered with TSMC and that will be the case for a while and even as we ramp up Intel TMSC is always going to be very significant to us.

So we are going to have two foundries. We are not going to replicate products. We’ll have some products that will be with TSMC and then we’ll have some products that will be with Intel, will depend on what you are doing with the product. But right now the high end is certainly being centered with Intel to take advantage of their leadership and process.

Glen Yeung - Citigroup

One of the things TMSC has consistently done in our view is that in order to progress to leading edge of the foundry they would take the customers most likely to go there. Altera would be an example of that. And you would share some of the cost, initially development cost initially to get yourself on the leading edge at an appropriate time. Is that something you will do with Intel or because they don’t actually need to do that per se?

John P. Daane

No we don’t. So we do not actually with TMSC either, we don’t pay for except the technology we don’t pay for development of technology. We do as an industry and most customers that are significant do this, there are some requirements for our types of products which are slight variants off of the mainstream process technology. Intel would do that as well.

But those generally come free. What you pay for is Mass cost and for your design and then also what’s you spend on the wafers and all that you pay for.

Glen Yeung - Citigroup

Okay, and then sort of a last thought around Intel is your ability to get the capacity that you may or may not need looking forward. Is there some guarantee from them that you have the appropriate capacity or how is that issue be resolved if there is one?

John P. Daane

So we had the question of why not use Intel for about probably three years before we signed the agreement. And my answer was always you know they have great process technology but what about service. And so what we did is we wrote a contract with Intel and the contract really covers a lot of the service elements of you would not necessarily have it in any other contracts. We don’t have those sorts of things with TSMC for instance, that's covered end of life because they tend to retrofit their facilities very quickly, discover.

So they’ve agreed to manufacture the products for us for at least ten years. For our customer base this included things like capacity, service lead times all of this is taken care of so we have no concerns with that. Then also really if you think of Intel right now they are operating I think from what I’ve heard with the utilization level that's well below 100%. So we believe that they are investing and have the capacity that’s adequate for us to work with them.

Ultimately I think they have grander ambitions within the foundry industry. They’d like to obviously take on some other big customers as well, naturally because of that I think they are going to try to make us very successful to hold up to others as an example what they can do.

Glen Yeung - Citigroup

Okay. Let’s switch topics a little bit and talk about why we upgraded you which is really among other things looking ahead at this opportunity in China that we see in terms of what you can capture, maybe just from your perspective what’s your understanding of what exactly the opportunity in China is going to be?

John P. Daane

I think there are two things that are going on. One is China recently announced that they have a broadband initiative. So they want to significantly increase the broadband access to people both rural and within cities, very similar to what Japan's done, for several generations which will require a very significant spend in the telecom and networking area.

And then secondly we have the 4G build outs ahead of us for the three major wireless companies. If you think about the wireless side what you’re really seeing is today only one of the three started to deploy and that’s China Mobile.

They currently have and are going through an RFQ process for about 200,000 new base stations, plus an undisclosed number of radios and there is some technical reasons I can explain if anybody wants to know where they are going through a bidding process but it also involves the government to decide how many radios are deployed.

And that’s to do an initial trial around TD LTE. That spend is predominantly going to be this year and probably a little bit early next year as their first round and there will be many. For those of you that sat through the TD-SCDMA which was their third generation they did actually six bids or six rounds of build out of the infrastructure which lasted probably four or five years.

And the sixth round was actually two, so there were in fact seven rounds where they built out over a million base stations and many millions of radios and again they are going to go through that with TDLT as well.

Followed by that you are going to have China Telecom and then China Unicom also do wireless build outs. They are going to be, I think China Telecom may start next year. I would bet China Unicom probably starts the year after. The best thing is you are going to see the spread out then over many carriers through many years in terms of the spend, all for 4G build out. So we think China is actually going to be very strong within communications, both telecom and wireless for many years to come.

Glen Yeung - Citigroup

As we think about the opportunity for a PLD within a given base station and I had two questions kind of my first one is I know there are lots of variants of what might -- but can you help and sort of try to pinpoint content per box but at least can you help think through the range of content possibilities for PLD.

John P. Daane

Yeah we typically have not broken down the actual dollar content per card or per customer. I guess the way I'd kind of frame this is if you move forward from 2G which mainly was GSM to 3G Y band CDMA or CDMA 2000 our content doubled going from that to 4G the content from 2G to 4G triples.

So you do get a substantial increase in PLD content as you move forward within generations of equipment, the complexity is higher. Typically within the base stations you will see high end FPGAs, the Stratix series where we’re very strong and the radios you will see the mid-range or predominantly the low end which we’ve also been very strong. Radios ship generally in the millions of units per year, base stations are in the hundreds of thousands of units per year in terms of the rough range of the equipment.

We’ve been very strong in China. We’ve always been the number one and continue to be the number one PLD vendor in China. Huawei is our largest customer and we’re number one at Huawei and we'll continue to be so into next year, certainly in the year after. We’ve also added to that we’re now I believe the number one at Ericsson as well.

So if you look at the two largest wireless equipment manufacturers, we are number one and I think that really sets us up both for China as well as deployments throughout the rest of the world.

Glen Yeung - Citigroup

And just to delve into the question a little bit more, when we think about LTE relative to I guess the China Mobile TD-SCDMA, what we call 3G generically we’re talking about spectrum which is a little bit higher and that base station coverage is not as good as it was at 3G hence more base stations. Do we think about LTE both in terms of how much enhanced when we think of, when we talk about radios but also more base station just to cover the…

John P. Daane

So you end with more radios due to that higher frequency which is good for us because obviously radios are those [oblong] usually they are kind of base boxes which you will see all around on tops of buildings, those have up to several PLDs per box and so they are substantial portion of revenue.

You have more that you are doing in particularly as a government both in the U.S. just talking about auctioning off more spectrum but as part of 4G in China, the government will also give more spectrum to the carriers. That means they’re more radios that are going to be deployed that's beneficial.

Question that you’re asking really gets into the base station itself, there are macro base stations which provide the overlay. Several radios feed into each base station. And so there was a discussion a few years ago, will small base stations come and if small base stations have less PLD content, will you actually lose in the marketplace? So we were on the defensive for a couple of years explaining no.

In reality everything that China Mobile is deploying, and for that matter, the other carriers will deploy is macro base stations, because that's what they need to do is provide the broad coverage. The area that the small cell or small base stations comes into play is with higher frequency, higher frequencies have difficulty penetrating buildings.

And so, where you have high density housing or high density offices like New York, what you will do is then deploy small cells within that building with their own low power radios in order to provide cell coverage. Those will augment the macros, they will not replace the macros.

So ultimately if you think of it, since our business predominantly has been a macro, nothing replaces our business and then because in some of the micros and picos we do have FPGA content, those new base stations that would be put in buildings do provide some additive business for us. And that’s really very similar to what AT&T announced in terms of their strategy.

Now over time, reasons the macros will go away or excuse me, the micros and picos will go away and will just be macros is because as an example, with AT&T, they plan to get rid of GSM in the United States to recover all of that spectrum, which is low frequency spectrum. And when they do that, they won’t need to put, low frequency again goes into buildings very well and then they won’t need the small cells for buildings anymore and so that whole deployment goes away, very similar to the strategy within China.

So what you’ll see is some initial deployment of small cells for high density areas for within buildings, does not replace the macros, really augments it for a short period of time and then ultimately the micros and picos go away.

Glen Yeung – Citigroup

So, China is an opportunity that’s relatively visible today as you see the firm information about it. But recently Taiwan's talked about assuming LTE, India is talking about expanding their cell networks in general. U.S. has been relatively good, Vodafone yesterday, how do we think about rest of the world?

John P. Daane

Yeah, rest of world has actually been outside the U.S. and Japan and South Korea, has been very quiet in terms of telecom spend. We will eventually have Europe start to pick up. Europe has been doing very little. I think that’s probably a couple of years away but will be additive at some period of time.

India has really done very little on either 2G, 3G or 4G. And so there is a real pent up demand. To some extent the reason India has been slow is the government sort of getting in the way of auctions and whether the auctions were legal for the spectrum. As those things resolve, you’re going to see a very rapid deployment.

You’ve got areas like Brazil which has the World Cup and the Olympics coming up, and yes their coverage for cell network is extremely poor. So they’re going to have to go through an upgrade cycle. So we really look at it on a worldwide basis.

Communication spend is in very early innings. And as smartphones really start to take off around the world and people are using data more with the plans as PCs, tablets or phones, that requires an upgrade of the networks, both in the wireless side as well as in the telecom side for backhaul, all of which is great.

So if we look at communications over the next five years, we can really see not just one geography driving the demand but multiple geographies. And again all phase differently, so we should get some pretty good demand over several solid years.

Glen Yeung – Citigroup

And if you were to sort of put that into relative picture within Altera, does that mean that comps become a -- they are already a big portions of your business?

John P. Daane


Glen Yeung – Citigroup

Does it actually get bigger. And what’s the implication of that in terms of gross margin?

John P. Daane

Well, I think within any particular vertical that we have, the margins really not dictated by the vertical, it’s dictated by the volume.

Glen Yeung – Citigroup


John P. Daane

So the higher the volume, the lower the price for the same device; the lower the volume, the higher the price. So naturally, if you think of radios, very high volume, the margins you will get in radios are lower than base stations and then if you go into industrial as an example or defense military, those prices are higher as margins are higher. And we really run a portfolio of business.

In general we’ve been able to manage ourselves over the last five years, seven years on gross margins that are well above our corporate model, no reason to think that any development within wireless necessarily drives that differently.

I do think there is some great growth opportunities beyond communications for us. We talked about telecom, we talked about wireless. Computer is another area. FPGAs are now being used as co-processors to CPU’s in order to accelerate algorithms. CPUs are not and what you have in your PCs and servers were really not designed to run mathematic algorithms very well.

So GPUs were initially adopted as a way to accelerate math over applied search like compression which are all used by companies that build out their own data centers and are some of the largest consumers or developers that make the servers and computers in the world. The FPGA will run those algorithms as fast as a GPU, at actually an order of magnitude lower power consumption. And power is their largest expense.

So what we are finding now is FPGAs are being adopted into computer manufacturers and people are building their own computer for data centers for search and for compression which are huge if you think of really the companies that are buying a hardware today.

It’s a new market for us. They buy the largest FPGA, the most expensive they can get, they replace it every two three years, it’s grown now to about 10% of our overall revenue. As a company it was up 20% year-over-year and we think can continue to grow at that clip. So we think we’ve got some other markets beyond communications side that are adopting FPGA’s at an accelerated pace which also provide great growth for us.

Glen Yeung – Citigroup

Is there an ecosystem in the sever market or the data center market that needed to have like [software] for example, I know when graphics came on there was some software within these two…

John P. Daane

So if you’ve got search there are couple of companies that develop their own search algorithms And so therefore they can hire RTL engineers, hardware engineers in order to program or reprogram some algorithms on their own and that’s fine for those guys.

If you want to take it more broadly to for instance oil exploration or high frequency trading or some of the other markets that would like to take advantage of running math algorithm fast but don’t want to hire hardware people they just want to program it and then see, what Altera did is we developed to open CL. And actually IBM was the company that pushed us into this originally as a partnership with them and we are working with the other computer manufacturers as well.

And by having this, which is unique to Altera we are the only PLD manufacturer, we’ve got a way to program our devices through C, don’t have to touch any of our tools and now you can program the FPGA in exactly the same way that you program your CPU. And it’s opened up the FPGA opportunities suddenly to the software developers.

And the importance of that is out of every ten engineers nine of them are software engineers only one is a hardware engineer. So we’ve literally opened up programmable logics to the other nine-tenths of the engineering world, to server, people who program for servers across many different applications of market. So we think that’s going to provide and really accelerate our computer growth story.

Glen Yeung – Citigroup

It sounds like you are typically oriented towards the high end server.

John P. Daane

High end server won’t be in the low end Blade servers that people are using for generic capability. It’s really that you have some scientific application that you need to accelerate. And of you think of math algorithms generally that is where we will be adopted and there will be an exploration, that will be scientific, high performance computing for agencies within the military environment for encryption decryption, for search for compression of photographs, high frequency trading, just general trading in general.

But actually it proves to be a very large portion of the total servers that are being shipped today. And if you look at Altera we are now in four of the five largest server manufacturers and users which is great progress in a very short period of time which is why that business is going so fast.

Glen Yeung – Citigroup

Do you have a sense of what your share in this opportunity?

John P. Daane

I think probably we are number one and I think we’ll definitely because we have the open CL, C like programming which is unique to us, because we’ve have so much momentum with the computer manufacturers I think that’s a market from which we will certainly get the most benefit from.

Glen Yeung – Citigroup

Interesting, thinking about industrial and what do you think characterize as an industrial automotive? Every chip company that says this is really what we want to be. Right so and you get to know on time so that’s the good news. Where is the opportunity, what does the landscape looks like there for you?

John P. Daane

So industrial did bottom for us in calendar quarter Q1, grew in Q2. There are a number of applications from which we can grow and we have the embedded CPUs as an industry both ourselves and Xilinx have done this where we put ARM microprocessors inside of our FPGA that really it allows us to displace all those embedded microprocessors that many people developed in the market place.

We are much more flexible, we can make -- you have the CPU you can use the FPGA in order to put it in your own blocks to make your own microprocessor if you will which is very exciting for a lot of industrial companies. So we think we can grow very well there. Automotive has been in cabin electronics , camera systems, car navigation systems today adding a driver-assist radar system is where FPGAs are being adopted.

I would say out of the FPGA industry today which is roughly a $5 billion industry, automotive is about 3% of that. But for us as an example, automotive last quarter was up 25% year-on-year. So it’s growing at a very good clip as our products are being integrated.

Why FPGAs? Well, if you look actually look at automotive, the numbers of models is actually fragmented. And so there is a lot of low and medium volume applications within these auto manufacturers. Additionally, for guys like Bosch and Denso, who make the equipment for the end auto company themselves, what they like to do is leverage their R&D, create one card and then reprogram that through software in order to add new features or make it display differently for each auto manufacturer without having to redo cards from scratch. And the FPGA is perfect for that because of it allows that re-programmability.

And so we’re seeing a lot of adoption within the auto industry and certainly growing very quickly for ourselves and I would bet our competition is growing just as quickly there.

Glen Yeung – Citigroup

And just to clarify the portion of business that’s associated with auto and industrial and mobile…

John P. Daane

Yes, industrial for us, and again, industrial is different for everybody. We break out tests separately. We broke out medical separately. We break out military separately for industrials, just industrials. So this is the equipment that will go into your automated bread slicers and everything that goes through manufacturing. And there is probably -- industrial's about 10%ish, a little bit higher as a percentage of our overall business, automotive is about a 3% as a breakdown.

Glen Yeung – Citigroup

Defense is an area where Xilinx has had some success recently.

John P. Daane


Glen Yeung – Citigroup

Can you talk about, one, what’s going on there, is that -- how do you reinterpret that situation and then see what's the opportunity for you?

John P. Daane

So Xilinx certainly has higher market share within military. And fortunately Altera in the mid-90s decided to exit the military business and really decided to re-enter when I came to the company in the early 2000s. And it takes some time to build up the business, and Xilinx certainly has higher market share within that field. And obviously it’s going to enjoy that to a greater degree than we would.

Second thing, this is quite obvious is they are obsoleting a lot of their product portfolio at this point. I would estimate based on the products they’ve announced, it's probably 20% of their revenue. Obsolescence I think is probably not something that you do within your industry. But what we do is we say to our customers, we cannot manufacture this device anymore. So, you have to take the rest of your lifetime requirement and buy that from us today.

So it’s really accelerating all of the future life of your product portfolio revenue in today’s period of time. And they’re doing that for a good portion of the revenue.

If you think about the products that are older, generally that’s going to be with your telecom, customers, your military customers, and your industrial customers. And if you look at Xilinx last quarter, obviously half of growth came from military and half of growth came from their mature old products. Generally your old products decline every quarter, they don’t grow. And that shows you that they are getting some benefit certainly on a revenue line and substantially on gross profit or gross margin line from that obsolescence that they’re doing. And that’s probably military is contributing a lot to that right now.

Glen Yeung – Citigroup

One last question is thinking about, you’ve come through this issue with Huawei where it was PLD to ASIC transition for them. And let’s clarify that that’s now behind us. And then two, remind us of the opposite, right, are we still comfortable with the idea that it’s in general ASIC to PLD world?

John P. Daane

Yes. In general, what happens is that there are some conversions that are done in our industry and really you hit a volume point for which it’s more cost effective to be in that ASIC than the FPGA. You have to pay an upfront price in order to implement the ASIC. And so it takes a very high volume in order to get that cost differential between the ASIC and the FPGA to make up for that high upfront cost, because that cost will run you $20 million to $50 million today that you’ll pay the ASIC company in order to implement your chip.

So when I joined Altera about 12-13 years ago, the cutover point was about 10,000 units a year and then somebody would move from an FPGA to ASIC. Today, it’s about 1 million to 2 million units that they will cutover. And if you think of the radios, radios are in those millions of units per year, there is an opportunity to cutover to the ASIC from the FPGA.

So, when I started, hundreds of conversions will happen a year. Now you’re into just a few conversions will happen per year. And ultimately what Huawei has said is they’re not going to do any more FPGA conversions, where they may have made sense to start four years ago. If they look at it today, they do not economically make sense any longer and they don’t anticipate doing it.

Overall, our industry grows because for every one that you lose, you have hundreds of new ones that come in from people that can’t afford the ASIC anymore that are now new to using FPGAs or used FPGAs but are now not using that ASIC and are start going to use our FPGA product and that will continue to be the trend I would say going forward.

What happens is obviously with Huawei they can revert some business to an ASIC, that lowers our revenue. Our competition's revenue at Huawei is actually not growing year-on-year for that matter, probably next year. So we took a little bit of a market share decrease because our revenue went down but overall we are still number one at that account and expect to continue to expand and I would say in general what you are going to see is those conversions will be rare.

Now what happens with Huawei is a lot of people heard of HiSilicon as a subsidiary of theirs. HiSilicon is an ASIC company therefore they are going to replace every PLD and they really lost sight of HiSilicon is. HiSilicon is not meant to replace PLDs. HiSilicon has 5,000 engineers. The entire ASIC industry does not have 5,000 engineers put together.

What's HiSilicon's intention is to develop their own IPs so that they no longer need ASSPs. And so we are not their target, as much as it is the ASSP companies that are their target as they develop their own MPU's, their own switch engines, their own base band chipsets for cell phones, that's really what they are targeting.

Glen Yeung - Citigroup

John, we out of time but I appreciate the chat today.

John P. Daane

Thank you very much, Glen. It's a pleasure, thanks.

Copyright policy: All transcripts on this site are the copyright of Seeking Alpha. However, we view them as an important resource for bloggers and journalists, and are excited to contribute to the democratization of financial information on the Internet. (Until now investors have had to pay thousands of dollars in subscription fees for transcripts.) So our reproduction policy is as follows: You may quote up to 400 words of any transcript on the condition that you attribute the transcript to Seeking Alpha and either link to the original transcript or to All other use is prohibited.


If you have any additional questions about our online transcripts, please contact us at: Thank you!