Inphi Corp
Q2 2015 Earnings Call Transcript

Published: 29 Jul 2015

[Operator:]

Good day, ladies and gentlemen, and welcome to the Inphi Corporation Second Quarter 2015 Earnings Conference Call. At this time, all participants are in a listen-only mode. Later, we will conduct a question-and-answer session and instructions will follow at that time. [Operator Instructions] As a reminder, this conference is being recorded for replay purposes. I would now like to turn the call over to Ms. Deborah Stapleton, Investor Relations. Please begin.

[Deborah Stapleton:]

Thank you, Latoya. Good afternoon, everyone. Thank you for joining us today to discuss the financial results for the second quarter of 2015. With me today are John Edmunds, Chief Financial Officer; and Ford Tamer, our Chief Executive Officer. John will begin with the Safe Harbor, then Ford will give you an overview of our business; after that John will provide a financial summary of Q2 and the outlook for the third quarter of 2015. Then we’ll be happy to take your questions. John?

[John Edmunds:]

Thanks Deborah. Please note that during the course of this conference call, we may make projections or other forward-looking statements. These forward-looking statements and all other statements made on this call, which are not historical facts, are subject to a number of risks and uncertainties that may cause actual results to differ materially. These forward-looking statements speak only as of today’s call. We do not undertake any obligation to provide updates after this conference call. For further information regarding Risk Factors for our business, please refer to our registration statements on Form S-1, as well as our most recent annual and quarterly reports on Forms 10-K and 10-Q, all filed with the Securities and Exchange Commission, accessible at www.sec.gov. Please refer, in particular, to the sections entitled Risk Factors. We encourage you to read these documents. Also during the course of this conference call, we may make reference to non-GAAP financial information. A reconciliation of this information is included in the press release and on our website, which is available at www.inphi.com. This information is not a substitute for GAAP and should only be used to evaluate the company’s results in conjunction with corresponding GAAP measures. Now to begin our review of the quarter, let me turn the call over to, Ford Tamer. Ford?

[Ford Tamer:]

Thanks, John. Good afternoon, everyone, and thank you for joining us for our Q2 2015 earnings call. Once again, I’m pleased to report that we exceeded the midpoint of our guidance for Q2 and demonstrated excellent year-over-year growth. Specifically, we reported record revenue of $60.7 million for Q2 2015, and record non-GAAP EPS of $0.24. This is compared with revenue of $33.9 million and non-GAAP EPS of $0.09 in the second quarter of last year. Our growth remains robust, both from organic Inphi product and from the acquired Cortina product. Once again, I thank all of our employees for their hard work, enabling us to deliver these results to you. Regarding guidance, we are still cautious about the macro environment as discussed by our peers. Given we still get the majority of our revenue from the service provider market. Our business is being impacted by the pause in wireless infrastructure roll out in China. Therefore, we have adopted a conservative guidance for Q3 with a wide range of revenue outlook to account for the current industry slowdown. We’re guiding to a flat $0.24 EPS and a 1% quarter-over-quarter revenue increase at the midpoint. We believe that our second-half results will outpace our first-half results due to the strength in the server and data center segment. Now, let me offer some details around our continuing progress in 2015. Let’s start with our core communication business, where sales of our optical amplifiers, drivers and physical layer products were again up more than 50% on a year-over-year basis. We’re very pleased with this continued progress as it has been a re-growth engine for the company. While we saw consistent performance across all product lines, let me highlight the success of our legacy transport business, and specifically, the broad market endorsement of our high gain FEC, or Forward Error Correction IP. We are increasingly confident that our outstanding FEC design is well on its way to becoming the standard for top tier OEMs across the industry and carriers around the globe. As proof, earlier this year, Deutsche Telekom demonstrated the FEC effectiveness on its 100 gigabit per second long haul optical links. As a result, we subsequently notified their ecosystem partners to adopt it for both merchant and captive Tier 1 DSPs. Acceptance of this standard is clearly gaining momentum. Other carriers are also now considering a similar endorsement, further solidifying Inphi’s very efficient and elegant design as an industry standard. In Q2, we were pleased that IEEE 400-gigabit committee Endorsed Pulse Amplitude Four Level Modulation or PAM4. This endorsement by a 75% vote makes it clear that PAM4 is the only viable technology standard for data center interconnect for two-kilometer and above distances. It is also a compelling option for data centers wanting a unified solution to cover both two kilometer and above, as well as the lower end at 500 meter and below. The Inphi PAM4, two laser solution operating a 28 gigabaud looks increasingly compelling for a number of reasons. First, Tier 1 OEMs and data centers are already adopting the PAM4 architecture. We never build products on hope. We start our development projects based on the endorsement of large customers, who commit resources to taking the products into production. PAM4 is no different, with customers committed to taking to production in the spring of 2016. Second, the strong support for PAM4 by a strong majority vote in the recent IEEE committee meetings in March, May, and July, shows its attractiveness to data centre customers for 40, 50, 100 and 400 gigabit per second bandwidths. We have included a new slide in our updated Q2 corporate presentation on the Inphi website, which depicts these platforms along with our low-power TIA amplifier. The third reason that PAM4 is compelling is that is now out of design phase and currently sampling to several module and OEM customers. We forecast the list of adoptees to grow dramatically over the next few quarters. Finally, two laser PAM4 modulation is a better fit for data center interconnect, because of its superior performance, lower power consumption, lower cost, the ability to cover two kilometer reach and its immediate availability. All-in-all, with the IEEE endorsement we see our PAM4 solutions contributing meaningfully in 2016 and beyond. And we are on track to deliver to early adopters on the schedule. In fact, the buzz around 100-gigabit in cloud data center is intensifying. And some customers are even talking about skipping the 40 gigabit generation. We are convinced this can result in increased revenue for Inphi in the spring of 2016. I’m also delighted to announce that we just made our already strong board of directors even stronger with the addition of Elissa Murphy currently the CTO and EVP of cloud platforms at GoDaddy. Ms. Murphy previously served as Vice President of Engineering at Yahoo and General Manager at Microsoft, where she honed her expertise on massive scale computing necessary for today’s big data applications. I know that Inphi and our shareholders will benefit from the addition of Ms. Murphy’s expertise and we enthusiastically welcome her to our board. Finally, one more important product highlights for Q2 is from our memory business. The transition from DDR3 to second generation DDR4 is tracking to expectations for the year. We continue to increase our server OEM qualifications and market share as forecasted. We believe this will result in a stronger second-half 2015 for our memory products. Furthermore, we’re confident that the rollout of our latest generation memory register and buffer products will continue this momentum into 2016. In conclusion, we are convinced the demand for bandwidth will continue to rise. In the mid to long run, this will drive Inphi revenue and profitability growth that will see us through the choppy environment our industry is experiencing today. The tangible measure of progress in Q2 was our High-Gain FEC PAM4, and our optical networking and memory products and businesses, are all contributing to a steady march towards a bright future. In that future, the cloud is the network and the data center is the computer. Our current product positioning in these growing market is already translating into significant growth in our design-win funnel for revenue in 2016 and beyond. Now, let me turn the call over to John for a deeper discussion of the quarter’s result and our outlook.

[John Edmunds:]

Thanks Ford. Now, let me recap the financial results for Q2. As Ford mentioned, in the second quarter of 2015 Inphi reported revenues of $60.7 million, which were up 79% year-over-year. Well, this was due in part to the Q4 2014 acquisition of Cortina even when we add the Cortina 10-gig and 40-gig physical layer business into the comparable base for Q2 2014. The core communication business continues to show more than 50% organic year-over-year growth. As expected, the quarter was driven by sequential growth in core communication, which we define as amplifiers, drivers and 10-, 40- and 100-gig physical interface products. Our communications business represented approximately 80% of our total revenues in Q2. The core communications products represented nearly 70% of communications, while the legacy communications represented approximately 30%. Legacy products are also now showing some strength relative to expectations from the beginning of the year. As expected, Q2 sales of product into high speed memory applications were down sequentially as sales of DDR3 products began to decline, making way for the market transition to DDR4. Sales of our DDR4 products were flat in the quarter, but are now expected to roughly double in the second-half, relative to the levels booked in the first-half. This should result in annual revenue in high-speed memory showing some relative strength and the expectations, delivering in the range of 20% to 25% of overall revenue for the year. Gross margins on a non-GAAP basis came in at 68.6%, slightly ahead of where we expected based on mix and including approximately $0.7 million of IP licensing revenue for the quarter. Based on the current forecasted mix and the small ongoing stream of IP licensing revenue, we expect the gross margins in Q3 to be in a range of 67.9% to 68.5%. On a GAAP basis, we were able to break-even for the second quarter. The delta between GAAP and non-GAAP is approximately $9.8 million, made up of four basic adjustments. As discussed previously, the GAAP loss include stock compensation expense of $7.2 million, some acquisition transition expenses, mostly legal, accounting, and some cash retention bonus cost of $0.4 million, as well as purchase accounting related adjustments totaling $6.4 million, all offset by the related tax effects for these additional expenses, along with the valuation allowance adjustments and other tax differences totaling $4.2 million tax benefit, Again, all four items combine to a net $9.8 million of additional income on a non-GAAP basis. The Q2 GAAP-based result of break-even compares to net income of $2.6 million, or $0.08 per share in Q2 of 2014. The Q2 2014 GAAP net income was a positive, more as a result of inter-period tax allocation, which took place between the quarters throughout 2014. The more telling comparison was on a GAAP operating loss, which in Q2 of 2015 was $1.9 million, and included $6.8 million of expenses related to the acquisition of Cortina. Netting out the Cortina-related expenses would result in operating income of $4.9 million, which compares favorably to the year-ago GAAP operating loss of $1.7 million. Now to add further clarity to the analysis of operating results, let’s turn to some additional non-gap measures. On a non-GAAP basis, net income for the second quarter was approximately $9.9 million, or $0.24 per diluted share, which was slightly up compared with a non-GAAP net income in Q1 of 2015 of $9.3 million, or $0.23 per diluted share. Non-GAAP operating expenses for the quarter totaled $29.6 million. This was up $1 million from the previous quarter, a little less than expected due to less R&D co-funding from customers, which is offset by lower payroll taxes compared to Q1. Again, in Q3 and Q4, we anticipate reverting to a higher level of customer R&D co-funding, showing up in the form of additional NRE offset to R&D expense. Although, we will have a higher program spend with respect to certain R&D projects, we only anticipate operating expenses increasing by approximately $0.4 million, or 1.4% in Q3. In Q2, we were able to deliver an operating income margin of 20%. In Q3 with slightly less gross margins and slightly higher operating spending, we believe we can deliver operating income margin in the range of 18.5% to 20%. With regard to the non-GAAP tax provision in Q2 consistent with Q1, we are currently expecting approximately 18%, which does not include the potential benefit of a restoration of the federal R&D tax credit by the U.S. Congress. In fiscal 2014, when the R&D tax credit was restored in Q4, non-GAAP effective tax rate for the year dropped to approximately 13.2%. Other income in Q2 was down by about $200,000 from what we would normally expect, due primarily to some accounting related foreign currency exchange losses. They were the result of now transacting a larger amount of our business internationally, as a result of the Cortina acquisition. In Q3, we’ll institute a small hedging program to neutralize these accounting-based foreign currency exchange exposures. So we do not expect them to recur. Now, turning to the balance sheet, the overall cash was up $9.2 million in the quarter from $78.2 to $87.4 million. This was due to a strong cash flow from operations in Q2 of $13.5 million, up from a comparable cash flow from operations in Q1 of $12.2 million [ph]. The company also had capital expenditures of $3.2 million, which was down slightly from the $3.4 million in Q1, and somewhat less than we forecast. The remaining reconciling items were approximately $0.5 million used to pay withholding taxes on employee stock awards, net of proceeds received from option exercises. In addition, we had $400,000 used to tax benefits associated with stock compensation related to paid-in capital; as well as approximately $200,000 in non-cash accretion of premium in short-term investments, that is included as a source of funds in the operating cash flows. Overall, we generated $10.3 million of free cash flow in Q2, which is up from the $8.8 million in free cash flow in Q1. We do expect a higher level of cash expenditure in the back-half of 2015 for several masks that’s currently scheduled to be taped out at the various foundries. Accounts receivable came in at $35.5 million or 53 days in June, compared with $36.7 million, or 56 days sales outstanding at the end of March. Inventory on the balance sheet at the end of June was $24.4 million, which included $1.1 million of acquisition-related step-up in inventory value as compared to $23.3 million, which included $1.9 million in step-up in inventory value as of March 31, 2015. Excluding the relative step-up in inventories, inventory days were up modestly at 111 days or 3.3 turns at the end of June, compared to 97 days outstanding or 3.7 turns at March 31. Inventory balance is higher than normal due to a variety of reasons including the company having received a $3.4 million influx of inventory in March for a specific supplier-driven last-time buy of a widely used framer product. In addition, we recently purchased expensive long lead time packages associated with the next-generation amplifiers and drivers, as well as other changing rates of inventory absorption over time, which we currently believe will smooth out within a few quarters. Payables increased to $10 million or 48 days in June from $8.2 million or 37 days outstanding at the end of March. Now, let me recap the business outlook for Q3. I remind everyone again that, the following statements are based on current expectations as of today and include forward-looking statements. Actual results may differ materially. We do not plan to update nor do we take on any obligation to update this outlook in the future. Due to uncertainty in the revenue forecast associated with China, we have assumed a bit wider range of guidance for Q3. Our forecast of Q3 revenues is in a range of down approximately 3% to up 5% or $59 million to $63.6 million. We expect non-GAAP gross margins to be in a range of 67.9% to 68.5%. We expect non-GAAP operating expense to be in a range of $29.1 million to $30.9 million. We are currently estimating the non-GAAP effective tax rate to be 18% for the year. Again, this does not include the potential reduction if Congress renews the Federal R&D credit. We are confident these components should then result in non-GAAP operating margin of between 18.5% and 20%. This should then lead to non-GAAP net income of between $9.1 million and $10.7 million. On approximately 41.4 million estimated diluted shares, this would result in estimated non-GAAP earnings of between $0.22 and $0.26 per diluted share. We also estimate the stock-based compensation expense to be between $7.3 million and 7.5 million. In addition, we expect about $3.2 million from purchase accounting based adjustments. In Q2, we will also have another quarter of an additional flow through of stepped-up inventory estimated at $800,000, and we expect $900,000 in acquisition transition based expenses. This would imply a GAAP net loss in the range of $800,000 to $2.5 million. GAAP loss per share would then be in a range of $0.02 to $0.06 per share. A more complete reconciliation of the Q3 net income forecast is attached to the last page of the press release. We will not update this outlook during the quarter until the time of the next quarterly earnings release, unless Inphi publishes a notice stating otherwise. So, please ask any questions you may have today during the general Q&A period. And now, we’ll be happy to answer your questions.

[Operator:]

Thank you. [Operator Instructions] The first question is from Quinn Bolton of Needham & Company. Your line is open.

[Quinn Bolton:]

Hey, guys, congratulations on the nice growth in the optical business. Just wanted to ask first, Ford, on the PAM4 commitments that you mentioned, it seems like there is perhaps maybe even a pull in the PAM4 revenue, now looking for spring of 2016. Is there any more color you can give us as to what types of programs those are? Is that direct with some of the OEMs? Would that come in the form of more PAM4 modules? And then I’ve got a couple of follow-ups on the memory interface business.

[Ford Tamer:]

Thanks, Quinn. We are excited about PAM4 becoming – being pulled by customers for 40-gig, 50-gig, 100-gig, and 400-gig. So our excitement is going, because we have now more platform than we initially anticipated, and some of them would come to market sooner than the more traditional platform we had been discussing in the past.

[Quinn Bolton:]

It sounds like you expect now PAM4 revenue to be meaningful if I heard you properly in the script in 2016 and beyond. Is there any sort of sense you can give us for what the PAM4 revenue stream might look like next year?

[Ford Tamer:]

We expected to start in the spring of 2016 and ramp further in the second-half and ramp even further in the 2017 timeframe. So it’s going to be a continuous ramp, Quinn, not just a PAM ship, but also a whole platform along with our low power Transimpedance Amplifier that – and we’re depicting those platforms on the website. And in some cases it would pull some of the linear driver along with them as well.

[Quinn Bolton:]

Okay, great. And then just wanted to come back to the memory interface, I know the Broadwell qualification is sort of ongoing now. Can you give us some sense where you are on the second gen data buffer? Is that now sampling to customers, and how do you feel about Broadwell qualification looking into next year?

[Ford Tamer:]

Sure, the first comment, I would say is, we are also excited on a Haswell current platform that we are gaining share, and that will result into increased revenue in the second-half of this year compared to the first-half of the year. On Broadwell, we are in the middle of the qualification, we haven’t discussed specifics, we’re excited about our second generation data buffer, and believe that we will gain market share as a result. And that would translate into increased revenue starting in early 2016.

[Quinn Bolton:]

Great. And then just lastly, for John, can you give us some sense what you’re looking for in terms of IT revenue this year?

[John Edmunds:]

Quinn, so far it’s been bumping along at about a little more than $0.5 million a quarter. And so, I would expect that to continue into the back-half. We have a number of different parties that are interested, and we’re talking with a number of parties. So it could be better than that, but at this stage, I think, it’s better to just assume that we’ll continue at that pace until we hear otherwise.

[Quinn Bolton:]

Okay, great. Thank you.

[Operator:]

Thank you. The next question is from Ross Seymore of Deutsche Bank. Your line is open.

[Ross Seymore:]

Hi, guys. Thanks for letting me ask a question. Ford, you talked a lot about the technology roadmap, but it has a ton of things that are going to drive growth, looking a little further forward. Can you just provide a bridge from now until then? What are you seeing as far as bookings, I know, it’s not unique to Inphi as far as the weakness in service providers. But, generally, how are you viewing the current slowdown and when do you think we’re likely to come out of it?

[Ford Tamer:]

Right. So if you look at the latest Infonetics report, 400-gig coherent report, and that’s still probably a large part of our revenue. Couple of insights from that report. On our website, we’ve also updated the forecast from Infonetics for the past three report. And you can see the 100-gig forecast is slowly, but surely going up with every new report that they’re publishing. So that’s very positive. If you look at the share coming from Asia from that report, you’ve got about almost 60% coming from either China system OEMs or module makers selling into that geography. And so that’s a pretty large exposure to the Chinese market. We were expecting Phase 11 [ph] of the roll out – of the infrastructure roll out to be awarded for bids in September and translate into revenue. And now, this is being delayed to Q4, so we’re hopeful to see that Phase 11 bidding start, and that should translate into revenue in the late Q4, and then increase revenue into 2016. That’s probably the largest question mark for us in our forecast and hence the adoption of our more conservative outlook for us, Ross. Now, in the middle between now and then, if you look at the other vectors in our business, as we said in our prepared remarks, we are confident that our memory would be stronger in the second-half compared to the first-half, and it would be stronger in 2016 compared to 2015. So memory will end up being a bright spot. We are still seeing strength in the transport business and legacy, which is ahead of expectations. And then we are continuing to see deployment in our physical layer product in the data center market. And I just close with that final remark from the Infonetics report. The same Infonetics report, the latest one talks about 100-gig being the new 10-gig and how some customers would actually skip 40-gig, I’ll read from it. It says, 2015 would be the first-year where 10-gig WDM shipments decline. 40-gig technology experienced a brief moment of glory, but has soon been annihilated by a 100-gig. So we do believe that the growth vectors are still in place and hopefully after this momentary pause in China and resumption of that infrastructure buildup, we’ll be back on track.

[Ross Seymore:]

Thank you for all that color. I guess as one, hopefully not so long follow-up. But one follow-up is switching over in the memory side of things, John, just to make sure I got the math right. You talked about the second-half being almost double what the first-half was, and if I did get that right, can you talk about – are there any margin implications for DDR4 replacing DDR3, as a percentage of your mix?

[John Edmunds:]

So what I specifically said, Ross, was that our DDR4 revenue would double in the second-half relative to the first-half. And that will bring us back into, overall revenue for the year would – for our high-speed memory would be in the 20% to 25% range of – of total revenue, which if you use $250 million push you right around by $55 million sort of level at the midpoint. So we do believe there will be strength in the second-half, and I don’t think there’s really gross margin implications at those mix levels. DDR4 is new technology, and it does come in at a higher gross margin, it will amortize over time. But we’re excited to take share and volume-add spaces as quickly as our manufacturing partners can take that supply and bring it into the mix.

[Ross Seymore:]

Perfect. That’s all I had. Thank you very much.

[Operator:]

Thank you. The next question is from Jay Srivatsa of Chardan Capital Markets. Your line is open.

[Jay Srivatsa:]

Thanks for taking my question. Ford, I just want to ask a little bit more detail on the service provider comment you made in China. We’ve seen delays in china before and it’s highly unpredictable. I guess what gives you the confidence that you will see that being a near-term slowdown and not get prolonged into 2016? And then more importantly, I guess, are you sensing that there is a CapEx cut that’s happening within service providers, as the slowdown happens near-term, or you feel things will bounce back pretty easily in Q4?

[Ford Tamer:]

Yes, so our comments talked about the mid-to long-term coming back online. I mean, if you believe in bandwidth, if you believe and as one recent report pointed to slowdown in Morrisville, creating more bandwidths demand, which is good for Inphi, then that demand will come back. We cannot call the resumption of the China infrastructure brought out, so that that’s why we’re actually being cautious in our outlook. And so right now we’re all assuming it comes back in Q4, but it could be delayed to Q1. So it’s hard for us to call it and others right. But overall, I think that bandwidth demand is increasing because of the data center to data center and inside data center type of traffic, as well as video serving over the Internet. So we do believe the mid-to long-term demand is still very solid and is there, we can’t really call the short-term.

[Jay Srivatsa:]

Fair enough. And then on the memory control side, I think in the past call, you had talked about how you expect DDR3 to likely continue through the rest of the year and DDR4 might likely get delayed, but looks like that changed a little bit from your commentary in the transcript. So, can you help us understand what changed during the quarter that changed the mix going forward?

[John Edmunds:]

Jay, I don’t think anything has really changed. What we have said in the past is that we thought that in terms of unit volume the crossover between DDR3 and DDR4 would take place sometime in the second-half of the year. The revenue crossover may be earlier in the year due to high mix of our DIMM in the market. But what we were commenting on at the time is that the unit volume mix may be in the second – crossover may be in the second-half of the year. As you recall, our first quarter was strong for DDR3. And as we expected it began to fall off in the second quarter. Our DDR4 revenues have begun to ship, but they were relatively flat from Q1 to Q2, although we now expect that they will increase in the back-half of the year almost doubling the amount of DDR4 revenue that we expected or we experienced in the first-half of the year. And that will bring us back to a level, as I mentioned, at 20% to 25% of our total revenues.

[Jay Srivatsa:]

Okay. Fair enough. Maybe one last question on the step-ups, do you expect them to taper off by end of this quarter putting yourself on track to be GAAP profitable next quarter or do you believe that will continue for another quarter or two?

[John Edmunds:]

Well, we think they will taper off – the step-up in inventory value will taper off to be about $800,000 in Q3. And hopefully, the balance of that inventory will turn by the end of the fourth quarter, maybe another $800,000 in that quarter. As you can see, we were breakeven this quarter and a lot of this has to do with how the GAAP tax rate comes out. And again, that’s going to vary based on whether Congress passes the R&D tax credit or not, lot of variables overall. But we’re continuing to grow the business overall and we do believe we’ll be GAAP profitable within the next probably 12 months or so.

[Jay Srivatsa:]

Thanks John. Thanks, Ford.

[John Edmunds:]

Okay. Thanks, Jay

[Operator:]

Thank you. The next question is from Ruben Roy of Piper Jaffray, your line is open.

[Ruben Roy:]

Thank you. Ford, I just want to clarify something on the PAM4 commentary. It sounds like we’re looking ahead to two lane PAM4 being pulled in a bit. Any clarification on what that might mean if anything on when we might expect single lane PAM4, and then also, if you could touch on this production timing again on each of those technologies? Thank you.

[Ford Tamer:]

So, the commentary, Ruben, was to say that the two lane PAM4 has a 100 gigabit chip. It can also be used as a 40-gig and 50-gig offerings. And so, that’s what’s being pulled in. Regarding two lambda or two laser PAM4 compared to one laser PAM4, couple of points here. Number one, the two laser PAM4 is shipping available sampling today from us and our competition. So, there is no need to wait a year or two for one laser PAM4. We at Inphi, as well as our competition are all working on the one laser PAM4, but the IEEE made it clear that for two kilometer and above two laser PAM4 will be the preferable solution. So that would put the one laser PAM4 into a niche, which is the 500 meter and below. And so, for any data center that would like a unified solution for both 500 meter and two kilometer there is no need to wait two laser PAM4 is available today, shipping sampling from us and our competition. So, you could look at solutions for the two laser PAM4, not chip solutions shipping in the spring of 2016. And you could look probably for one laser chips available in the second-half of next year from those both Inphi and our competition. There will be a good year or year-and-a-half after the two laser solution have been shipping. So, there’s probably going to be a lag between two laser and one laser of about 18 months to potentially 24 months. The one laser solutions are going to be extremely challenging in designing PCB board that run at 56 gigabaud. They’re going to be extremely challenging in designing components to run at that speed. It’s going to be challenging to find test equipment that can run at 56 gigabaud. So, the point we’re trying to make is, look, there is no reason to wait until this bright future one laser comes around, which would be pretty challenging; use what’s available today and start building today, because your competitors are building 40, 50, 100 and 400-gigabit products using our two lambda PAM4 today.

[Ruben Roy:]

Thank you for that. That’s very helpful. I just had a quick follow-up on the macro commentary and sort of the caution that you’re approaching your guidance with near-term. I hate to nitpick on near-term when longer-term still sounds so good. But have you actually seen for Inphi order reductions or anything on the inventory side out there in the supply chain that is creating the caution or is this more kind of general market-driven caution? Thank you.

[John Edmunds:]

Yes, Ruben, it’s John. We have seen, I would say, some isolated pockets of the things being pushed or delayed. Nothing that looks like it’s the getting cut, as someone asked earlier. And I think the book-to-bill ratios did start to come down closer to 1. So we’re not seeing orders come in as rapidly as we as thought – as we had been experiencing in the – up through the first quarter I’d say.

[Ruben Roy:]

Okay. Got it. Thanks, John.

[Ford Tamer:]

Yes, Ruben, one more comment on this. The reason we are saying, but to long-term we’re okay, is starting 2016 we’ll have many more growth vector coming to the market and giving us diversification and increased revenue growth. Q3, Q4 is harder for us given these revenue growth vectors are not there. And so we’re still depending on the resumption in the China infrastructure.

[Ruben Roy:]

Got it. Okay. Thank you.

[Operator:]

Thank you. And the next question is from Srini Sundararajan of Hamrock [ph] Summit. Your line is now open.

[Srini Sundararajan:]

Hi, thanks for taking my call. My question is if you were to quantify the impact of China over this year, what would the quantified number be in terms of revenue lost?

[Ford Tamer:]

If you look at the guidance for Q3, we took it down by a couple of percent from where the prior consensus was. So, I mean, we are not guiding to increase 1% quarter-on-quarter from Q2 to Q3. And you could see what consensus was. That’s really the impact for us in Q3.

[Srini Sundararajan:]

Okay. And then as a follow-up, I just wanted to ask you, is there a difference between the product that you make for DDR3 versus DDR4, meaning in terms of manufacturing? Is it too much of a difference or it’s a relatively straightforward change in design?

[Ford Tamer:]

Yes, the difference is, we have a difference in both the product architecture. We also have a difference in the process node that we used for the DDR3 and DDR4 product. But our DDR4 products are now fully in production and we’re over any potential invention or process node issues we may have had in the past.

[Srini Sundararajan:]

Okay. Thank you very much. That’s all.

[Operator:]

Thank you. And the next question is from Brian Alger of ROTH Capital Partners. Your line is open.

[Brian Alger:]

Great, guys, nice quarter as well. Couple of clarifications; obviously, things aren’t really good on the PAM4 side of things. But just for my help here, as we look at 2016 and the total 100 gigabit opportunity there, we’re still not expecting PAM4 to be the dominant module, correct? That’s something that is going to phase into as we get into 2017?

[Ford Tamer:]

Yes, Brian, you’re right. I mean, we expect the first QSFP module into the data center to be based on a four laser solution that would use our Clock Data Recovery or CDR product along with our TIA amplifier product. So we have a set of product for that four laser solution, that is being taken to market by customers in the early 2016 timeframe, and that would be followed up by the PAM4 product in the spring of 2016. So, yes, you’re right, the dominant form factor would probably be still the four laser, but transitioning quickly to the PAM4. And for Inphi, the dominant revenue driver in 2016 is still going to be the long haul and metro coherent 100-gigabit solution that we’re selling for both the TIA driver and companion CDR product.

[Brian Alger:]

Right, right. Well, I understood. I was just looking at obviously the Infonetics data has been updated for the long haul side of things. But as I look at the data center forecast, it seems a bit stale, and I would anticipate that given the architecture the two laser PAM4, if anything that forecast for data center growth with 100-gigabit would have to be moving up into the right, correct?

[Ford Tamer:]

Yes, you’re right. The forecast actually has been moving up to the right. We need to work update and then we will [ph] over the next couple weeks on the website.

[Brian Alger:]

Okay, great. And, John, just a quick question with regards to the inventory, I know it’s been a bit elevated here for a couple of quarters, you’ve clearly defined why that is. I’m wondering how we should think about it coming, I guess, back in line, is that going to be through a reduction in the dollar value of the inventory, or is it just going to be from holding the inventory steady as the COGS line increases with revenue growth?

[John Edmunds:]

Well, I think, it would be a little of both, Brian. We expect that the inventory will move, it may take a quarter or two, but it should start to come down, as we move forward here. And we’re not really worried about anything, in particular, in the inventory, or in the balances that we’re optimistic that, that will come back into alignment and probably be within, I would rather see it around 80 days of inventory within the next 12 months or so.

[Brian Alger:]

Okay. So that’s a bit of a cash flow tailwind for us, great. And then just one last one with regards to the Cortina business, we didn’t talk much about it, but it sounds as though, the FEC acceptance here could be another leg of, in terms of a growth factor. How should we think about that in terms of scope and potential? Obviously, it’s a bit out there on the horizon, but what are we talking about in terms of dollar opportunities for Inphi?

[Ford Tamer:]

At this point that would be a IP revenue and that would be incremental to what, John, has mentioned as the baseline that we’re expecting.

[Brian Alger:]

Great. Thanks, guys.

[Operator:]

Thank you. The next question is from Tore Svanberg at Stifel. You line is now open.

[Tore Svanberg:]

Yes, thank you. Few questions, first of all, just coming back to the near-term guidance, just so I understand sort of the moving parts here, so you’re expecting the memory interfaces business to be up sequentially, but then you are expecting the core comm business to decline sequentially, right?

[Ford Tamer:]

I think, the core communications business will be relatively stable. I don’t know that it will be, if it’s down, it will only be down slightly as we expect and has potential to grow on the quarter. And as you might know, we gave a range of guidance, so we do think there is an opportunity for both sides of the business to be growing.

[Tore Svanberg:]

Sounds good. And as we look at the memory interface business for the second-half of the year, especially, with the DDR4 business doubling second-half versus first-half. Is there an acceleration happening in Q4, or is this going to be a fairly sort of even split in Q3, Q4?

[Ford Tamer:]

No, we’d expect it to increase in Q4 versus Q3. We’ll be – it won’t be instantaneous, because we’ll be getting opportunities and adding to the inventory flow, as we go here and as the module makers are able to work us into the mix. But we’re getting positive signals and a set of welcome arms that’s going to help us raise that inventory over the course of time here, not just in Q3, but in Q4 and then on into 2016.

[Tore Svanberg:]

Very good. And questions for Ford. Ford, so Infonetics is committing 250,000 ports, 100-gig next year. I know this is hard to answer, but what is sort of your guesstimate on PAM4 as a percentage of those ports?

[Ford Tamer:]

PAM4 as a percentage for 400-gig, sorry?

[Tore Svanberg:]

As a percentage of the 100-gig, the 250,000 100-gig ports expected in 2016.

[Ford Tamer:]

We are almost indifferent to the mix that we get similar content for the two laser and the two laser, Tore. So we are excited about the two laser PAM, because we feel it’s more differentiated and so we should increase our share as we go from a four laser to two laser, because the four laser market is a more crowded market with more competitors versus the two laser market being really just two competitors, us and one more company. So, but we’re almost indifferent as far as dollar content for one versus the other.

[Tore Svanberg:]

Sounds good. You also talked about sort of pulling in TIA, CDR with PAM4. I think between you and your competitor, you’re probably the only one that has that ability. So, how important is that? And as you’re now sampling PAM4, how much traction are you getting with the type of a pull-in activity?

[Ford Tamer:]

Actually, this is another big differentiation for PAM4. On the four laser NRZ solution that are shipping today, we have competitors that are able to field and sample both the CDR and TIA from the same company. When it comes to PAM then you got to go to two companies to compete with us. So, there is one company that makes up the PAM and couple of companies that are working on driver and TIA. But there isn’t one company that has PAM linear TIA and linear driver. So, we would be the only company in the PAM universe that would have the whole solution under one roof.

[Tore Svanberg:]

Very good. Thank you very much.

[Operator:]

Thank you. The next question is from Richard Shannon of Craig-Hallum, your line is open.

[Richard Shannon:]

Ford and John, hi, thanks for taking my questions. I guess, just a few from me. So, in your memory interface business you talked about your favorable trends here in the second-half of DDR4. Based on what you’re seeing from the qualifications for next year, what kind of share do you think you’d get in DDR4?

[Ford Tamer:]

Thanks, Richard. It’s too early to discuss the Broadwell share right now, as we’re still couple of quarters away from the main launch. We can just state that we’re actively engaged with the memory vendor and OEMs for that qualification. And that we feel confident our gen 2 product will enable us to capture a good position definitely better than what we did in our gen 1 product.

[Richard Shannon:]

Okay. Fair enough. I guess, last kind of quarter for now to see how that’s going. Second question related to PAM4, Ford, I think you mentioned that a number of data center operators have already started or committed to adopting your solution that uses PAM4. Can you characterize the set of guys who are looking at it and maybe can you declare whether all of them – all the major ones are using it and if not can you characterize the ones that would versus would not use it?

[Ford Tamer:]

Unfortunately, we are under NDA and really cannot breakdown how many and who are using or not using. I think it’s fair to say though that there is over 50% of the data centers, and I would say over – or a lot of the large Tier 1 system vendors are committed to it. So, I would say the majority of the data center and Tier 1 system OEMs are definitely committed to PAM4.

[Richard Shannon:]

Okay. Well, that’s fair. Another question on PAM4, the testing qualification times here look fairly lengthy. What do you see as the primary risks for holding to that timeframe? Are those things that are in your hands or outside of it? And what do you think the biggest risks are to that timeframe?

[Ford Tamer:]

It’s a good question, Richard. We do believe that the risk are somewhat low, because we have the chip in the lab, it’s working. It’s been tested. Its functionalities are green, being shipped to already tens of customers. So we are fairly confident that we’re on target to sort of enable customers to go production in the mid next year timeframe.

[Richard Shannon:]

Okay. And just to be to be clear, Ford, I was asking about the ecosystem, not just Inphi specific, with that would you characterize it the same way throughout the ecosystem as well?

[Ford Tamer:]

The risk on ecosystem is that, we have some of our traditional partner in ecosystem are busy fielding four laser, NRZ type of solutions. And so the risk there is that those traditional partners are not deploying the resources towards PAM4. And so we’re having to work with – it’s more a number of partners to take PAM4 to production and in some cases, new partners that are not part of the more traditional Inphi partners that we’re used to in the past, that’s probably the risk.

[Richard Shannon:]

Okay, that’s fair. And one last quick question from me, I’ll jump out of line. Can you characterize and quantify how much your – just your TIAs and drivers together grew in the second quarter, year-on-year, or quarter-on-quarter, however, you want to look at?

[Ford Tamer:]

I think, year-on-year, we characterize this as greater than 50%. And I believe the – you would find the amplifiers and drivers are probably above that average.

[Richard Shannon:]

Okay, great. That’s all the questions from me, guys. Thank you.

[Operator:]

Thank you. The next question is from Sundeep Bajikar of Jefferies. Your line is open.

[Sundeep Bajikar:]

Hi, guys. Thanks for taking the questions. First, if you could just give us an update on IP revenues in terms of composition, what types of technologies are being licensed and to whom?

[Ford Tamer:]

Sundeep, as John was saying, it’s a small number today. So I don’t think we need to really spend a lot of time on it. I think what we do, as if and when it becomes a larger percent of revenue, we’ll commit to do that. But right now, I think, given a small percent, maybe we’ll just skip, if that’s okay.

[Sundeep Bajikar:]

Okay. That’s fair. And then similar one on total revenues, if you could just give us an update on mix in terms of service provider versus cloud for total revenues?

[Ford Tamer:]

Yes, Sundeep, we talked about in the past, the issue for us is, when we shipped majority of our revenue into that coherent market, which goes into service providers, it could end up being used for either wireless infrastructure backhaul, it could be used for data center. Data center connection, it could be used for transport long haul. The difficultly for us is, we don’t really have a good visibility into the percent of each. And we’ve been asked the question in the past, we’ll try to do a better job trying to get you a better answer in the future.

[Sundeep Bajikar:]

Okay, great. And then last one from me, can you talk a bit about the potential contribution from non-volatile DRAM. Is this referring to the Intel, Micron, and new memory announcement today? And from an Inphi perspective, would this be an extension of DDR4 register and buffer technology, or would this be a new product?

[Ford Tamer:]

Yes. First, without sharing any NDA related information, today’s announcement is extremely exciting for Inphi and the whole server or storage industry. There is significant performance, capacity cost, and persistency gap between DRAM and then – that this technology could address. And, therefore, new storage architecture will emerge requiring high bandwidth and high-speed interconnect that Inphi can enable. So we believe today announcement is very positive for Inphi. As you can see, we’ve been reporting on our website in our corporate presentation for probably the past six months, PAM increased due to the non-volatile DRAM, and it’s still there in our corporate presentation. And we are working with different ecosystem partner to come up with solution in that space. We have been working on it for a while. And that non-volatile persistence memory segment will have different components, and will require companies such as Inphi to work with all the different ecosystem partner, the server and storage OEM, the CPU maker, the memory maker, the module maker, operating system provider and data center operators and user, and we’ll be doing so. It’s a bit early for us to say exactly what we’ve been working on. But we’d probably be in a position to do so end of this year, early next year.

[Sundeep Bajikar:]

Great. Thanks so much.

[Operator:]

Thank you. The next question is from Gus Richard of Northland. Your line is open.

[Gus Richard:]

Yes. Thanks for squeezing me in. Ford, I’m just wondering the weakness in China, is that primarily you mentioned in your script it was wireless? Is that also a metro and long haul?

[Ford Tamer:]

So…

[Gus Richard:]

Hello?

[Ford Tamer:]

Yes, Gus, the question is, are you saying the – is the question is, the weakness we’re seeing from wireless or from long haul and metro or – I’m sorry, what is the question, Gus?

[Gus Richard:]

Again, the question is, you mentioned in your script that you were seeing some weakness in your wireless business in China. And I was wondering if that weakness in China was also metro and long haul?

[Ford Tamer:]

We don’t believe so. We believe the metro, long haul, is still fairly solid. We ship the same components on 100-gig coherence into the same system OEMs, and in turn they end up using it for LTE infrastructure for metro, long haul, and for data center to data center connectivity. So, if you look at these three segments, the only weakness we are seeing is that into LTE infrastructure rollout related to that Phase 11 pause in China. We don’t currently see any weakness in the long haul metro or the data center to data center interconnectivity.

[Gus Richard:]

Okay. And then on the memory interface, John, I think you said that 20% to 25% of revs for the year would be the memory interfaces, as I got that correct for the whole year?

[Ford Tamer:]

Yes, that’s correct, for the whole year.

[Gus Richard:]

Perfect. And then, the last, when you mentioned the last time around for the OTN framers I believe. Has that started to – did that start the last time buy in the second quarter or do you expect that to sort of transpire in the third quarter? How is that revenue going to play out over the next – how is that playing out?

[Ford Tamer:]

So again, it was the last time buy driven by the supplier. In other words, they had a fab supplier, one of the offbeat fabs, I’ll say, in Japan that decided to not make that process available anymore. And so, they put us – really Cortina, even prior to the acquisition put them on notice and we were in a position where they had a very popular widely used framer product that – where they weren’t going to be able to buy anymore. And so they scheduled a last time buy which is about $3.4 million inventory receipt in this March quarter. And that will probably in our view continue to sell and ship out over the next four to six quarters or so.

[Gus Richard:]

Okay. Perfect. Thank you. Thanks for clarifying this. Thank you so much for taking my questions.

[Ford Tamer:]

Sure, no problem.

[Operator:]

Thank you. This concludes today’s Q&A session. I’d turn the call back over to John Edmunds for closing remarks.

[John Edmunds:]

Thank you, operator. Inphi plans on attending the Oppenheimer conference in Boston on Wednesday August 12. We’ll be at the Jefferies conference in Chicago on August 25 and 26. And we will attend the ROTH one-on-one conference in San Francisco on Wednesday September 9, as well as the Deutsche Bank conference in Nevada on September 16. Ford and Deborah and I would like to thank you for joining us today. And we look forward to speaking with you again in the future.

[Operator:]

Thank you. Ladies and gentlemen, this concludes today’s program. You may now disconnect. Good day.