Intel Corporation (INTC) Q1 2013 Earnings Report, Transcript and Summary

Good day, everyone, and welcome to the Altera Q4 2012 Business Update and 2013 Financial Guidance Conference Call. As a reminder, this call is being recorded. At this time, I would like to turn the call over to Mr. Scott Wylie, Vice President, Investor Relations. Please go ahead, sir.

Good afternoon. Thank you for joining this conference call, which will be available for replay telephonically and on Altera's website shortly after we conclude this afternoon. To listen to the webcast replay, please visit Altera's Investor Relations web page, where you'll find complete instructions. The telephone replay will be available at (719) 457-0820, use code 258712. During today's prepared remarks, we will be making some forward-looking statements. In addition, management may make additional forward-looking statements in response to questions. In light of the Private Securities Litigation Reform Act, I would like to remind you that these statements must be considered in conjunction with the cautionary warnings that appear on our SEC filings. Investors are cautioned that all forward-looking statements in this call involve risks and uncertainty and that future events may differ from the statements made. For additional information, please refer to the company's Securities and Exchange Commission filings, which are posted on our website or available from the company without charge. With me today are John Daane, our CEO; and Ron Pasek, Chief Financial Officer. Ron will open the call with a few brief remarks before turning the call over to John. After John concludes his remarks, we will take your questions. Prior to the Q&A session, the operator will be giving instructions on how you can access the conference call with your questions. I would now like to turn the call over to Ron.

Thank you, Scott. Earlier this afternoon, we released our mid-quarter update for the fourth quarter and our 2013 guidance. As a reminder, we are discontinuing the practice of giving a mid-quarter update. So from here forward, we will give quarterly guidance only on our earnings call. A few highlights about our plan for 2013. You should expect gross margin to be in the 69% to 70% range for the year. This is essentially flat to 2012. Gross margin will be above our long-term model for the fourth year in a row and is the result of good cost control as well as prudent pricing practices. R&D is targeted for $404 million, an increase of 12% from 2012. Keep in mind, nearly all of this increase is the result of 3 factors: increased variable compensation expense, as well stock-based compensation; annualizing partial-year salaries for people we hire in the course of 2012; and to a far lesser extent, a small amount of incremental hiring in 2013. While these factors explained the change year-to-year, what amiss is, of course, are the strategic priorities, which John will elaborate on in a few minutes. We're planning SG&A at $315 million, an increase of slightly less than 9%. The majority of the SG&A increase is the variable compensation for the bonus, sales commission and stock-based compensation. Headcount growth and SG&A will also be very nominal. Keep -- please keep in mind that in both the FY '12 and FY '13 SG&A, there's an $8 million to $10 million cost relating to our IRS litigation, as well as patent litigation brought on us by several non-practicing entities. Our core tax rate for 2013 will be 13% to 14%, which is slightly higher than 2012 and relates to an increase in U.S. income as a percent of total. The rate assumes current tax policy, including the absence of an R&D tax credit in the U.S. Not counting any share repurchases which we may be do in 2013, fully diluted share count will be in the low 320 million shares. Finally, CapEx will be in the $60 million to $70 million range. Now, let me turn the call over to John.

Thank you, Ron. I would like to spend the next few minutes outlining Altera's strategy and the opportunity and in turn, some of the details behind the 2013 numbers that Ron just presented. The rising cost of chip design with each new process node has provided an increasing advantage for programmable logic. As we opened a 3-process node lead over ASICs, we created the tipping point where our FPGAs provided a lower-cost total solution for many markets. With the tipping point, we were able to accelerate FPGA adoption and our revenue growth. With a few of the top 10 ASIC suppliers now exiting the market, we have a $12 billion ASIC replacement opportunity in front of us. Our goal is to continue to grow at least twice as fast as the semiconductor industry on average, and ASIC replacement is one avenue to achieve this. We have an equally exciting opportunity to replace ASSPs, a servable market of $36 billion; and microprocessors in the embedded market, a servable market of $9 billion. Our thought process here is the same as when we embarked on a development of a true low-cost FPGA family with Cyclone or integrated DSP blocks and transceivers into our FPGAs. These investments build upon our programmable heritage, open new markets, increase our ASP through IP integration and, in total, enable faster growth through an expanded servable market. Success in the embedded and ASSP markets requires incremental investment to our base FPGA spend as we, in effect, invest several years ahead of revenues. We embarked on this a few years ago with our silicon convergence strategy, and we are now just starting to see the success of our investments in design wins and early revenue. I would like to take a moment to provide some additional insight into some…

Before we get into the Q&A, with respect to this quarter's guidance as is typically the case, we'll not be providing today the type of granularity that accompanies our end-of-quarter earnings releases. We're happy to clarify, if necessary, but in terms of any questions, we'll not be able to spell out more of the quarter's puts and takes or other fourth quarter details today. Best, therefore, to aim your questions at our 2013 guidance or other topics that we spoke to during our prepared remarks. We would now like to take questions. Please limit your questions to one at a time. So that we give, as many callers as possible, the opportunity to ask questions during the call. Operator, would you please provide instructions and poll for questions?

[Operator Instructions] And we'll go first to Vivek Arya with Bank of America Merrill Lynch.

John, I'm wondering what -- how should we think about sales growth expectations for 2013? Because when I compare your OpEx growth, they trail the current 10%-or-so sales growth expectations that a lot of investors are assuming for next year. And if that is the case, 2013 will be the third year in a row when OpEx growth is outpacing sales growth. And I understand that you are investing for the long term. But when should investors expect to see more operational efficiency in the model?

Vivek, this is John. I -- at a high level, we are not going to provide guidance for 2013. It's been consistent with our practice for many, many years. We'll provide one quarter guidance at a time, and we'll continue to stick to that. So we're not here prepared or willing to provide prediction for revenue growth next year. I think the opportunity is quite clear. I mean there's a very large market to continue to replace ASICs. There are a number of ASIC players who are leaving the industry, which tells you a lot. There is equally a large opportunity to get in and replace microprocessors in the embedded space mainly, again, because these applications do not have the volume for embedded microprocessor companies to develop solutions. And for the same reason, we're seeing a very large opportunity in the ASSP industry. So we think the growth opportunity is there. It's required some additional investment above and beyond our base FPGA products to develop IP solutions for ASSPs; to integrate microprocessors; to develop OpenCL as a compiler, specifically for the server space; 3.5D packaging, many of these areas were incremental investments. But each is proving to be a very wise technology investment, either because it's opened up a new space for us, or we're getting current traction with products. And so we think all of that will show itself in the revenue growth an opportunity. And fundamentally, we still believe that we will grow at least twice as fast as the semiconductor industry on average going forward. We have the opportunity. We have the products. I think we have the design win momentum. Now long term, we still absolutely believe in our financial model, and that is to achieve a high 30% operating margin. Some years maybe higher. Some years maybe lower. Because this, after all, is a cyclic industry and revenues do fluctuate. But overall we still believe in that, which fundamentally says going forward, we would expect revenues on average over a series of years to grow faster than our OpEx. And we believe that's absolutely attainable.

And just as a follow-up, as we look at 2013, I understand you're not providing numerical guidance for sales. But are certain end markets where you have better visibility than others?

I would say, Vivek, in general, the visibility from any of the markets is very low right now, simply because customers are unsure about the macroeconomic environment. In general, I would say where we are currently this year is in a lull in a couple of our markets. Military has decreased quite substantially. Really, as we had some programs that bought in 2011 and then did not repeat in 2012. Even though we expect the spending from governments in the military space to decrease in the future, we do not see that having much of an impact really on Altera. We think the areas that we've concentrated will continue to get budgeting. And we would expect military, as an example, to grow from here. Communications has been slower this year. We've seen a slowdown in China and North America in terms of deployments. I think looking forward, since 4G is still not well built out in the U.S., we still have that spend in front of us. And obviously AT&T, as an example, announced that they're also going to spend more money on wireline. We have China 4G and continued 3G build-out in front of us. We have 4G build-out in South Korea and Japan currently going on. Eventually, we'll get India doing 2G and 3G deployments and then Europe, eventually doing 4G. So we think we're probably at a lower space in terms of communication spend than what we may see in the future -- coming years in terms of communications. And then some of our other markets are actually newer for us and have grown well this year, and we would expect that to continue in the future. Automotive is a newer market. It's probably a few percent now of the PLD industry but growing very well. We would expect that to go. And also, as I mentioned, the computer industry is doing very well, particularly for Altera. We're still on the early innings of that. We would expect very strong growth there. So hard to predict exactly what's going to happen in 2010 -- or excuse me, 2013. But overall, we think that probably looking forward, the markets that we participate in, some of them are at a low point, some of them we're achieving very solid design win progress. And we think we've -- we're really set up for some good growth in the future.

We'll go next to the James Schneider with Goldman Sachs.

First of all, on the OpEx side, can you maybe give us a little bit of color in terms of, first of all, what percent of your R&D you plan on spending on 20-nanometer programs? And then, how many tapeout do you expect on 20-nanometer versus the ones you did on 28 in 2012?

Jim, fortunately, I'm not going to provide that sort of color in terms of our product or product roadmap. As I mentioned, what is different about this is we are spending on both the 20-nanometer planar and the 20-nanometer FinFET simultaneously. So there is a little bit more development, plus we're also, as I mentioned, doing a new, low-end product. So we certainly have a lot going on in the FPGA side. But within that, obviously, some of the additional investments we're doing, in terms of ASSP functionality, microprocessors, OpenCL, 3D, are additional spend. We've been doing some of that for the last series of years. But they certainly play into next year as well. So while I can't break it down, what I don't want to do is leave you with the impression that all of the additional spend is necessarily because of 20 or 16. We have added a new low-end product. We haven't done in quite a few years and, as well, are doing or continuing a number of strategic initiatives for revenue expansion.

That's helpful. And then as a follow-up, on the gross margin guidance, flattish to this year's gross margin. But can you give us some color around your thinking both around mix of small and large customers versus [indiscernible] what your assumption in terms of the balance of those markets? Is it just what you're seeing [indiscernible] in Q4 going forward? Or is that more [indiscernible] that you saw traditionally?

So Jim, that's a good question. It's -- again, with not great visibility, it's kind of hard to answer that. But at the same time, there's nothing that I see that would really fundamentally change that mix of large customers and small customers. So I don't think you should assume that's going to drive a whole lot of margin variation.

We'll go next to Ross Seymore with Deutsche Bank.

Last year, you guys had a little bit of a different linearity to your OpEx, specifically, on the R&D side. Is there anything you can tell us about how that will trend through the year? Will there be a first-half or second-half bias in tapeouts, et cetera?

So Ross, on the R&D side, it's very, very slightly front-end loaded for the first half versus second half but not very pronounced at all. Certainly, not to the extent we thought this year would be front-end loaded. And SG&A is almost perfectly linear.

Great. And then as my follow-up question, on the gross margin side of things, you guys have kept at the higher end of your range, I think, even a little longer than you had originally guided. But does the 2015 target is still hitting down the 67% still exists? Or has that also been pushed out, given your current success?

Yes. So remember, absent this year, what we tried to guide was what could happen potentially with gross margin. It could possibly go down to as low as 67%. And the caveat to that was, remember what John said earlier on the call and we've said many times, we want to --- we believe we can continue to grow twice as fast as the semi market in general. So part of the variability in pricing would be to attract current business, that's ASIC or ASSP, thereby maybe having a slight degradation in margin to get that. As of -- until now, we haven't really needed to do that. And again, you have to kind of excuse us here. Because I don't think any amount of pricing would have really helped this year. So as long as we're continuing to penetrate the large opportunity John described in both ASSP, ASICs and embedded, if we can continue to do that and get above-average gross margin, then we'll do it. But we're not going to sacrifice growth to get that margin.

We'll go next to Ambrish Srivastava with Bank of Montreal.

John, question #1 is, could you at least help us understand the delta in the next node that you're spending from an R&D perspective versus the last node and just with respect to where you are in that spend? So that's the first question. And the second is the newer opportunities that you're seeing -- that you are opening up with the PLD space. We saw this ASIC conversion that was bigger than normal. Can you help us understand? In the newer markets, is the conversion going to be similar kind of profile, less or more?

So I would say in general, first of all, from an ASIC conversion standpoint, we do not see any potential ASIC conversions within any of the base business that we have or any of the customers that we have. We typically get anywhere between, as we talked about before, 0 to perhaps 2 conversions per year where our customer finally reaches a volume point where it makes sense to transition. With every generation of new products, that transition point, of course, moves into higher and higher volume. So in other words, 10 to 12 years ago, the conversions might happen in 10,000 units a year. Now we're seeing that conversions happen in the low millions of units per year. So over time, there is fewer and fewer conversions possible, simply because the costs continue to rise to implement that ASIC NRE. And as you amortize that into the component run, what you find is you have to have a very, very high volume application in order to really get the cost reduction to make it make sense. So we don't see many going forward. The markets that were talking about going into automotive and server, we don't see any more of an ASIC conversion opportunity. In fact in servers, what they're counting on is the fact that the device is reprogrammable so they can download new algorithms to be able to accelerate those. So -- and ASIC would make no sense in those applications whatsoever. So I'm not concerned with any additional business converting either in the existing business space or really going forward. And I think even a few more years, probably, there will be almost no ASIC conversions available just because of the rising costs. And again, I go back and highlight the point, which is that several…

And what return are you targeting, John?

So I'm -- our overall return is to -- I guess, the 2 simple metrics is to grow twice as fast the semiconductor industry and to achieve an operating margin that's in the high 30% range, roughly about 38%. And that's been our target for many years. On average, we've, over the last several years, done pretty well to that. And we expect that we can continue doing that going forward.

We'll go next to Romit Shah with Nomura.

This is Sanjay Chaurasia for Romit Shah. John, my first question is about SoC trends. You're seeing some better SoC trends. Some of the companies, like Cavium, and I was just wondering if this has any impact to your design wins in wireless base station space.

So the answer is no. I mean fundamentally, if you look at the base stations, they boil down to roughly 4 base components. Generally, one of them is a multi-core processor, which would be a Cavium-like device or a number of other suppliers that are in that space. FPGAs, in prior generations, were not servicing that function. So there's been no competition there. There generally is a series of DSPs or an SoC or an ASIC that combines multiple DSPs. That's something that we've not competed with, and it's not replacing us. And then there's a series of FPGAs, which usually, a couple 3 to 4 that provide different functions. And those are remaining within the system. So honestly, I think it was irresponsible for some of the SoC or many core guys to say they were competing with ASICs or programmable logic because they're not. We haven't seen any displacement and don't expect that we will in the future based on existing or new architectures that are being developed within the wireless space.

Okay. And as a follow-up, some of the conversation we are hearing by talking to industry participants is that they see FPGA moving to control plane from data plane. Could that change the mix between the low-end and the high-end FPGAs for you guys next year?

Well actually, it started the other way around. FPGAs were used for glue logic, and then they moved into the control plane. And then when we got higher-performance and higher-density products, they moved into the data plane. And so we continue to be really in all of those functions. Overall, obviously, the adoption of the higher-end FPGAs is more data path. The lower-end functions might be microprocessor combined with -- or control might be microprocessor combined with FPGA. And then of course, you will always get CPLD for board-level functions. So generally, within this system, you've got a combination of different devices that fill different needs.

We'll go next to Shawn Webster with Macquarie.

A couple of quick ones. And I understand you're not going to provide a lot of this on Q4. But I was wondering in terms of the slightly lower expectations for Q4, if that was related to your ASIC conversion problem in Q4 or if that was elsewhere in your business. And then I have a quick follow-up.

Shawn, the lowering of the guidance to the bottom end of the range for Q4 had nothing to do with the ASIC conversions. That was the easiest thing to call, because we knew exactly what that had been in Q3 and prior quarters. It was roughly, if you do the math, about $20 million of revenue in the third calendar quarter. If you annualize it over 4 quarters, it was less than $80 million as an impact to our business. And ultimately, obviously that was well known. Really, what we're seeing is just softness broadly. Nothing really is standing out, in particular, from any particular customer or program within the fourth calendar quarter.

Okay. And then you had mentioned in some of the new initiatives you're investing in, from R&D, you specifically called out applications that one of your competitors, Lattice, is in. Are there any specific end markets or product types that you're looking out there?

Yes. I think those types of products are very popular in a broad range of applications: industrial, automotive, computer server. And there's some consumer applications that we see those devices integrated into. They're also in communications equipment. Generally, those low-end devices are very, very popular across a broad spectrum of platforms. And again, it's a space where we've seen Lattice for a number of years. Lattice's management comments that they really only see us in the space. But it's a space that we've not invested in for a few years, and we think that it makes sense for us to go back and add some new features and capabilities to those product lines, cost reduce the capability. And ultimately, I think our advantage is we have the customer reach to be able to really reinvigorate that product line for Altera and start to grow some significant market share against our competitor in that space.

We'll go next to Christopher Danely with JPMorgan. Christopher B. Danely - JP Morgan Chase & Co, Research Division: With the tax rate and the U.S. exposure going up a little bit next year, can you just tell us what percentage of cash will be in the U.S.? And does this leave open the possibility of the dividend increasing?

So Shawn, yes, our U.S. business is going at least slightly. It has a small effect on the tax rate, obviously, because the higher percent of business is in the U.S. And [indiscernible] that means there's more cash here in the U.S. And I can't comment on any plans for the dividend this time. But as I said before -- I've said many times that we were committed to increasing the dividend over time, making it a regular and meaningful return to shareholders. Christopher B. Danely - JP Morgan Chase & Co, Research Division: Sure. And for my follow-up on -- I guess just a question on the near-term environment. Can you just remind us what normal seasonality is for Q1 and how your book-to-bill has tracked so far? And John, do you think that the fiscal cliff issues are having any impact to your customers? And are you guys changing anything this quarter in reaction to that?

So I'll do part of that. Maybe Ron, if I could ask you to do the other part. I think, in general, from a corporate standpoint, are we doing anything different with the fiscal cliff in front of us? No. The reason is, if you look at that, it's a very gradual spending reduction. So we're -- I think the initial impact is more through taxes and consumer consumption in the United States rather than it is anything that's impacting any spending of the U.S. government. And ultimately, I haven't found anybody yet who said that they're doing anything in preparation to that. I think, really, what people are reacting to is the macro environment. They're seeing slower growth. And many of the -- and economists and they themselves are slowing down their business, trying to minimize the amount of inventory that they hold. They're waiting until they get orders and then ordering products from us, which means visibility is very low. And it's typical during this cycles. What happens is people place orders and want product immediately and then tends to try to expedite you for that. And I think we're in that classic, low-visibility, high-expedite sort of a business cycle. And I think until we get some clarity really on what's happening with economic growth in Europe, China and the United States, overall, I think people are going to remain cautious.

So the first part of your question, Chris, I think was about seasonality. From where I sit, I don't really see any predictable pattern to seasonality for quarter bias. So it's difficult to say anything about Q1 at this point. And we will give guidance on Q1 on earnings call at the end of January. Christopher B. Danely - JP Morgan Chase & Co, Research Division: And book-to-bill so far?

Don't want to talk about that at this point.

We'll go next to Steven Eliscu with UBS.

First of all with respect to your SG&A for this year, it looks like it will be about 5 points above your target model and remains such in 2013. So without any actual cuts to SG&A, is it realistic that you can get to your target model in the next -- even in the next few years?

Yes. Steve, so this is Ron. I think if you look -- what we said historically, when I gave the long-term model up to 2015 a couple of years ago, it was relying on revenue growth to get there. We didn't feel that we needed to cut SG&A. But we expected to grow SG&A slower than revenue. As you've noticed this year, our revenue doesn't unfortunately happen year-to-year in a completely up into the right fashion. So this year, actually, obviously, it's down. Typically, you've seen fairly large variances in revenue growth or decline. So I don't think you can model in a way that says next year is just a regular year, and you assume a linear progression out to 2015. And as John said earlier, I guess the point on that was, and that's our kind of a long-term guidance, the 38%. There may be puts and takes by the time we get out to 2015. You may see gross margins, as we said earlier, higher than the 67%. You may see SG&A slightly higher. But on average, we still return that high 30% operating income.

I think if you sort of take a step back, Steve, 5 to 7 years ago, I don't know the time but roughly in that time frame, we were operating with a SG&A structure in the mid-20s, about 25%. And we said our goal was to get it down to about 16% -- 15% to 16% with revenue growth. We didn't need to necessarily invest a lot in SG&A as we move forward. We could get leverage with the revenue growth, and you've seen that. Obviously, given the last 5 to 7 years, it's dropped down to about 15%. And we expect going forward that SG&A can certainly grow at a much slower pace than revenue growth and that we can get that number down. And if you were to look at other semiconductor companies that are roughly within our size, range or larger, they're all operating with an SG&A around 10% to 12%. And so for those reasons, we feel very comfortable that it's, first of all, a good model to have. And then secondly, based on history, we believe that this is something that we can get. But it is going to take some revenue growth to achieve.

Okay, that's helpful. And as a follow-up, regarding longer-term growth and some of the end markets you talked about having lulls, specifically, if we look at communications. Is there a reason to believe that perhaps the end market growth may not be what you think it is, as you have dynamics where carriers are trying to maintain certain CapEx to sales ratios or even lower them and be able to drive that through higher-priced data plans that limit demand for FPGAs?

Well, I think in reality, Steve, what we're seeing is quite the opposite. I think some of the leading carriers are actually now increasing their spend. You're seeing AT&T and Verizon purchase a lot of spectrum in the United States, as an example, from other companies. You're seeing AT&T recently announced that they were increasing their OpEx spend for the next 3 years. You've seen China Mobile is aggressively advertising 4G in China even though they do not have a license. And they're also pushing the government really hard to get a license to be able to deploy equipment within that space. So I would say, generally, what they've done in order to make the business more profitable, and again, wireless is generally with many of these carriers far more profitable than their wireline business, they have done caps. And they made those caps stick in terms of data rates, and they more than made up for that by the fact that more individuals have adopted cellphones as a way to get information. And so they've increased the number of subscribers who are using data. And it's far, far more made up than the fact that they might turn off some people that were unhappy that they got a cap. So I actually see the communication spend, based on the carriers' desires, probably doing very well over a series of years. And again, I'd point out the fact that FPGAs in general have an increasing dollar content as we move forward into newer generations of equipment. So even if CapEx is roughly flat, as it's been for a series of years, we still grow. And you see that in our base numbers. We've significantly outgrown the CapEx figures. And it's, again, all because our dollar content rises as we introduce each new generation of telecommunications and wireless equipment.

We'll go next to Srini Pajjuri with CLSA Securities. Srini Pajjuri - Credit Agricole Securities (USA) Inc., Research Division: John, as you look to expand into some of these newer markets, to what extent do you need to have some of the IP required in-house? For example, you talked about microprocessors, and also, I think there's some analogue content in here. I'm just wondering what your strategy is in terms of the IP that you need for this market.

So some of the IP like the ARM microprocessors we've licensed. At the same time though, one has to develop the organization in order to implement both the hardware and software, build the systems, build the bridges between the FPGAs, do the software in some of these ASSP cases. We, for instance, bought a company that was working on OTN. We've licensed some other IP from other companies. In some cases, we've developed the IP ourselves. So it's a combination thereof. But again, in order to do it, it's either a combination of licensing and/or individuals in order to implement the technology. And that's something that we've been doing. I think the good news is, if you think of silicon convergence, which is the idea of bringing together IP blocks with FPGAs, DSPs and microprocessors, one can get all of the technology very readily with the exception of the FPGA technology. Obviously, there's only a few companies that have that. So it's very easy for us to branch out and pull in microprocessors into our chips as an example. By pulling in that device into our chips, we increase our ASP because we replace what was a separate microprocessors or are selling price goes up by $10 to $30 just by integrating that processor. Our margins are still the same. So it's a great uplift for us. It's been similar to our strategy in the past years where we did this first with transceivers and DSPs as an example. And It's sort of a stair-step function as we go in -- go along. And it allows us to grow faster than our customers. So those are the combinations, hiring people and licensing. And the path I think is -- forward is quite simple. Srini Pajjuri - Credit Agricole Securities (USA) Inc., Research Division: Great. And then as a follow-up, there's lot of talk about the design cost increase at 20 and beyond. My understanding is that 10-nanometer and FinFET also require more fab time and more steps involved. And I'm wondering, what sort of wafer cost increase should we kind of think about going forward? Is it going to be similar to what you saw in the previous nodes? Is it going to be different? And if it's different, to what extent do you have the ability to pass on some of those incremental costs to your customers?

Well, so if you go back in time for our industry, the wafer cost for every new generation goes up. And the reason it goes up is because we've always added new materials in order to help with performance and density and also minimize the power consumption. So you've seen us add low-k dielectrics, copper, barrier metals, now high-k. You're doing double patterning, which means you're moving through the steppers a couple of times in order to create a critical layer. We're moving the FinFETs. Each of these have more processing steps on more expensive equipment. And fundamentally, the wafer cost goes up. In the past, what happened is by moving forward, of course, you were able to shrink your die and that more than made up for the increase in the wafer costs when you look at the die costs. Because again, you were shrinking the die enough that it more than made up for the fact that the wafer was more expensive. Now and generally, you're finding that, that formula does not work with a few exceptions. Cost is an exponential curve based on die size. As you get to the larger die sizes, that's where that you really start to get on the sort of exponential end of the cost curve. PLDs are large die. So we still have the ability, even with the significant increase in wafer price going from 40 to 28 to now 20 to FinFET technology to achieve a very significant die-size reduction, and because we're on that exponential part of the curve with an increase in wafer price, we more than make up for it with our die-size reduction. And as such, we still have the ability to push forward the new generations and achieve a total cost reduction of our silicon. Most ASSPs, I think, are going to be challenged, which is why you hear some of the discussion of -- out of other semiconductor companies saying that, that avenue of moving to the next generation to achieve the cost reduction is no longer there for us. It's still absolutely intact for our industry. And because it's intact, we are aggressively moving forward with 20, and we're aggressively pushing into FinFETs. And by moving to the new next node and reducing our costs and increasing our complexity and performance, we're able to displace more ASICs, ASSPs and microprocessors. So it continues to fly well. So overall, you're right that the costs are going up. They have been for a long period of time. I think that inflection is starting to change many semiconductor businesses. But the base PLD business, pieces is still absolutely intact. We can move forward. And by doing so, again, we open up a lot more market and can grow faster.

We'll go next to David Wong with Wells Fargo.

This is Parker Paulin for David Wong. Just wanted to see if we should ever expect something, the sort of on par with Xilinx's zinc chips to -- for you guys?

That's called our SoC products. And as I mentioned, we're sampling those this quarter and expect to be in volume production early next year.

But nothing -- we shouldn't expect anything than outside of the Arria and Cyclone lines?

That's the 2 product lines today that we're doing the development in 28-nanometer. We haven't announced what we're doing in 20, but I did mention that we are doing SoC products and 20-nanometer as well currently.

We'll go next to William Stein with SunTrust.

I'm hoping you can dig a little bit into the low end of the market that's mostly served by Lattice. You highlighted this is a new area of R&D focus. Is there a particular market opportunity you see there that is growing faster than you'd seen in the past? What's the motivation for the new investment there, please?

This is an area actually that we've been in for many, many years and Lattice has been in for many, many years. So we compete head to head. That's the CPLD market and the low end of the FPGA market. And this is a space that we have not invested in for some time. As I mentioned, it's a very broad application space. In other words, customers really across almost every market that we service use these types of products. By going back and really developing or refreshing our product line and adding some new features and cost reducing it, we think our competitive side will go up significantly. It’s a market that we are already very successful in, have a brand recognition there with our MAX series and obviously, much stronger customer attraction than our competition. And so it's something we think by refreshing that we can, not only grow from a market share perspective but there are a number of new applications that we can actually grab and displace over other types of semiconductors that we plan in that space. And again, I'm a little bit difficult about going into too much detail, because we have not announced this product yet. But I just did want to highlight it because it is an incremental spend to what we've been doing over the last several years.

Sounds like an interesting new market. One more question on automotive. I saw your products demoed at the trade show recently for lane detection application. Can you talk about other applications and what you see as the volumes? And whether those would potentially grow enough and -- what you see is the impact on your model from that?

Yes. What you find is, actually, automotive is far more fragmented than I think people appreciate. And that -- what you -- if you look at the in-cabin electronics for most cars, by model they're different even within one manufacturer. And so the volumes are not necessarily high for a lot of the content that we're doing, which is the -- first of all, the in-cabin electronics, which should be the car navigation systems, rear camera displays and then also now the ADAS or driver assist, which is things, as you mentioned, like lane departure or radar systems for collision detect or automatic parking. And those are applications which are perfect for programmable logic based on the volume. The other benefit that they have is for many of the manufacturers of automotive equipment such as, if you look at DENSO or Conti or Visteon, by developing one platform, you can reprogram our device and software and make it look slightly different for different models, and therefore, reuse a lot of their base R&D investment. And as you know, many of those companies are challenged by their overall business model. So they find, from an R&D perspective, using programmable logic allows them to create leverage and reduce their overall expenses so that they can compete with more manufacturers. And so we do really have a value added there as far as car manufacturers are concerned. It's an early market. 5 years ago, I'd say probably our revenue as an industry was 0. Today, it's probably 2% to 3%. So it's growing quite quickly. And we think it's a great market for us and continue to grow very well over the next several years.

We'll go next to Brendan Furlong with Miller Tabak. Brendan Oliver Furlong - Miller Tabak + Co., LLC, Research Division: Quick question. The last quarter conference call, you mentioned that you didn't expect anything significant in the wireless CapEx front to happen until the back half of 2013. Since then, AT&T has announced the big CapEx spend for next year. Just wondering if that has altered your opinion or that was already factored into your opinion over the second half of next year.

We haven't really provided, I don't think, any guidance out into next year what we thought would happen with communications or wireless or any of the verticals. I think, generally, what we did say is that we felt that starting in the second half of 2011, really through this year, we have seen a slowdown in communications spending. Second half of last year was really North America-specific, started to pick up in the second half of this year. This year, we saw, probably starting Q2, Q3, really a slowdown in China spend. So overall, the 2 largest geographies from a spend perspective had been slow. We did expect them to pick up over time. I think AT&T's announcement was very positive. Obviously, China Mobile's push to get 4G license is also positive. But we'll have to see how that develops. It's very hard to predict exactly what will happen in any given year. Brendan Oliver Furlong - Miller Tabak + Co., LLC, Research Division: Okay, great. And then a follow-up question I have is on the R&D front. Looking out into 2014, I know it's a little bit away, but if we see the higher R&D this year for variable comp and new hires that happened this year, the assumption would be then that you have -- R&D remained high in 2014 because of tapeout costs and actual production of 20-nanometer or early production of 20-nanometer, is that correct?

Yes. You're asking a question we don't have all the answers to you. But there's a slight increase in 2014 for those tapeouts, but it's not significant. It's not huge.

And the other thing is there's not as large of a headcount increase overall for the corporation this year -- or excuse me, in 2013, as we had in 2012. But given that, very hard to predict exactly what will happen. Obviously, the variable side is people have looked at our proxies, see if it's driven by revenue and operating margin. And without knowing what those 2 numbers will be, it's very difficult to say exactly what the variables side will be 2 years from now.

We'll go next to David Wu with Indaba Global Research.

Yes, just one clarification. If I look at the difference in the R&D spending of roughly $44 million between next year and this year, would you say roughly half of that was because of going to the FinFET or taking on additional FinFET investments and the other half would be related to the development of incremental low end and any SoC ARM chip investment. Would that be roughly the breakdown between that -- those 2 buckets for 2013 over 2012?

David, no. Actually, most of the increase for R&D from '12 to '13 is variable comp, mostly bonus and stock-based comp, as well as some focal. And really, that's probably 2/3 the increase. To a lesser extent annualizing the headcount, some of which, John described working on new initiatives, is the remainder. But fully 2/3 of that is just variable comp that gets reset going into 2013.

So basically 2012, you benefited from the missing of your operating expense -- operating income number, and that's reset back up next year.

That's correct. It's the operating margins and the revenues both.

We'll go next to Ian Ing with Lazard Capital Markets.

First of all for John, in your workings with TSMC, do you have sense of how much volume and demand they're expecting for 20-nanometer planar and FinFETs? There were headlines that TSMC is breaking ground on a 16-nanometer FinFET fab extension. And also their highest-volume customer, which typically is [indiscernible] mobility, haven't really revealed 20-nanometer plans yet.

Yes. So I think the way to think of this is, 16 is after 20. And so what you find is all of the people that can afford to do development in a next-generation node are doing development in 20. And based on what I know of their customer base and the demand, they're going to have a lot of demand in 20. So yes, I guess, basically, what I'm saying is people are not waiting for 16. The people that can afford and were in 28 are in 20 and doing development. Some of those will move to 16 for products from that's available. But remember, these products are not -- or technologies are not coming out exactly at the same time. And the capacity ramp for 16 will be well after 20.

Okay, great. And my follow-up is you talked about some die-size shrinks at the latest nodes. My understanding is these are big SSEs, which are typically patent-limited designs, a lot of analogue mixed signal blocks that don't scale as well with Moore's Law and that was more of a silicon integration play with more IP here so...

Yes. So I think that's absolutely right. If you got a design which is predominantly analogue or only has a small digital section to it, you're not going to achieve much of a die-size shrink moving to the next generation. And therefore, if the wafer cost is higher, your unit cost is higher. PLDs, on the other hand if you look at it, is mostly digital logic, whether it is the FPGA, the DSP, the memory blocks or the microprocessors. And that all scales. And these are big die. And so therefore, we do get cost reductions and can move forward for the next several generations, whereas many of the others stop. I think by having other companies say that they can't move forward in process technology is just a tipping point accelerated. Tipping point was where we were moving forward a new process technology while the ASIC industry stopped and tried to use what they viewed as cost-effective older technology for as long as possible. As we opened up a 3-node generation lead, we really became far more cost effective and accelerated ASIC replacement and our revenue. By having other ASSP companies, microprocessor companies freeze and not move forward, that just opens up more of an opportunity space for us to replace their products as we move forward with new designs and new processes.

We'll go next to Gus Richard with Piper Jaffray.

John, the CoWos technology looks rather disruptive. A couple of questions, how -- when do you think the intersect is? When do you think it will -- you'll start to use it? And when do you think the higher-volume markets will start to adopt it? And does it contain any potential threat to the PLD market?

So Gus, I think if you sort of take a look at it, I said probably a year ago that we thought that this was more within the 20-nanometer time frame and really because of 2 reasons. One is the technology had to mature to work on reliability and materials, which is doing better. And then as well, the costs are still very high for the substrates. And so really, we need to see the overall costs to come down before volume adoption will happen. So I think initially, what you're going to see is markets that are new, you create a niche where the customer will be willing to pay for the added cost start first. And then maybe broad volume adoption comes a little bit later. But we think probably within the next 3 years, you're going to see higher and higher adoption of this technology. And so we prepared our 20-nanometer products to automatically work with the CoWos technology. So we're setting ourselves up well. In terms of what it does, it obviously allows integration of devices and the interconnective devices in a different way than has been possible in the past, which can significantly increase performance and lower power, which are 2 great things from a customer perspective. It is an advantage for us. Because if you think of the things that you want to integrate, it's mostly things from analogue to memory to microprocessors to FPGAs. Really, we're the -- we and a couple of others are the only sources of the FPGA technology. So we're really in a cupboard seat. We can integrate other devices and other technologies either by creating them ourselves or by working with some other semiconductor companies. But it's very hard for them to integrate our devices. And so we see ourselves in a really good position to be an integrator of components to make system solutions. And so, therefore, we've been investing in this technology very aggressively.

And then as a follow-on, I noticed an increasing number of companies going vertically integrated and doing their own design in-house. Apple is the premier example. I understand that that's all very high volume. But I'm wondering if, increasingly, you're going to see more and more silicon developed at OEMs as opposed to -- as Apple is a SoC company. Is this a trend you're seeing? And is this something that's going to drive PLD designs into more designers and cloud companies, for example? And are there things you need to do to your tools to accommodate that?

Well, I think that the way to move into the cloud was through OpenCL. And again, Altera has been the only one to adopt the technology and work with the server manufacturers. So we're in a unique space really there. And it opens up opportunities for us really to replace CPUs and GPUs with very expensive FPGAs. They don't care about the price. We can accelerate the algorithms that they run at a much faster rate. And we've created a way to program these with C-like code, which was the barrier for us getting our products adopted. So we've opened that market up with a different type of technology. Overall, I think the fact that you're seeing some systems companies design their own solutions means that probably, over time, there'll be fewer ASSPs available. Because if some of these larger companies, like Apple or Cisco or Huawei or Ericsson or other types, develop their own IP, that means they'll need FPGAs in order to implement their technology. The ASSP companies find themselves with less of a market to sell to because some of the market leaders are not interested in ASSP solutions. They're doing the IP development themselves, and they may exit the market. And with one of the examples that I used earlier, we actually -- with our ASSP solution, we actually moved in with one of the market makers, and we saw the ASSP companies that had dominated that space announced that they were exiting and were no longer going to develop chips for that. So we become the only supplier of technology now within that space. So I think that trend probably will continue over time.

And we'll take our final question from John Pitzer with Credit Suisse. John W. Pitzer - Crédit Suisse AG, Research Division: A couple of questions. John, when you look at the gross margin guidance for next year, to what extent is the revenue stream still dominated by 40-nanometer, where you guys have had a very good lead over the competition? And how do we think about, as 28 kind of leaks into the top line, the margin implication? As you know, there's a lot of concern in the investment community about whether or not that relative lead holds.

So John, this is Ron. The last 2 quarters, I've mentioned specifically the 40-nanometer was the single, largest revenue node. You should continue to see that throughout the next year. December, this month, marks the fourth year that we started shipping 40-nanometer. And as you know, it takes about 5 years to ramp to peak revenue. So you should continue to see 40-nanometer dominate.

I guess at a higher level, if you look at the programmable logic industry, no matter who's winning and who's losing, the margins did not really fluctuate a lot with new product introductions. It's more really dependent on the market segments that you're serving at any particular time. So in other words, your market segment mix. And so -- and we just kind of -- if you go back and look over time, what you find is the gross margins for the industry have actually, over the last 10 years, gone up. And we've taken consistently market share within nodes. And so I don't see a change between one vendor or the other as having an impact. As you've seen, we've taken a lot of market share. And our margins have gone up, and our competitions have as well. So we think as we continue to take market share, not only in 40 but also 28, that, that will just continue. John W. Pitzer - Crédit Suisse AG, Research Division: Great. And then, John, given the escalating cost and complexity around Moore's law, is there a point in time where it makes sense to look at a multi-foundry strategy, either from a capability standpoint or a cost standpoint? Or does the complexity of adding an additional foundry partner just worth any of the benefits?

It would be more expensive to add an additional foundry partner, but it may make sense based on technology or commercial purposes. And so as we've also said, we've tried to, as a company, maintain a few large strategic suppliers. We think we get better service, quality, reliability and pricing by doing that. But there may be a time where that doesn't work anymore, and we have to look at others. So we're always out in the marketplace, talking and understanding what's going on. And right now, we're happy with our situation. But I can't say that, that's something that will always be that way. John W. Pitzer - Crédit Suisse AG, Research Division: Is any of that embedded in the next year's OpEx guide?

Yes. I hate to even get in to discussion about what are strategy is. But ultimately, I mean, again, we're doing a few products. And it's not -- nothing is really dramatically changing there.

Thanks, John. And as a reminder, we will announce fourth quarter results on January 23. This concludes Altera's conference call. Thanks for your interest and participation.