Good day, ladies and gentlemen, and thank you for your patience. Welcome to the Q3 2016 Applied Micro Circuits Corporation earnings conference call. (Operator Instructions) As a reminder, this conference may be recorded.
I would now like to turn the call over to your host, Ms. Suzanne Schmidt with Investor Relations. Ma'am, you may begin.
Thank you, operator. Good afternoon, everyone, and thank you for joining today's conference call. On the call with me are Dr. Paramesh Gopi, President and CEO; and Marty McDermut, CFO.
Before we begin, I would like to remind you that various remarks that we make on this call, including those about future financial results, including revenues; gross margins; operating expenses; design wins; product plans; our competitive situation; market trends; statements about future development, production, and adoption of X-Gene, X-Weave, HeliX, and X-Tend, and other products and technologies; and our anticipated growth and profitability all constitute forward-looking statements for the purposes of the Safe Harbor Provisions under the Private Securities Litigation Reform Act.
These forward-looking statements and all other statements that may be made on this call that are not historical fact are subject to a number of risks and uncertainties that may cause actual results to differ materially. We refer you to the most recent Forms 10-K and 10-Q filed with the SEC, in particular to the section entitled risk factors, and to other reports that we may file from time to time with the SEC for additional information on factors that could cause actual results to differ materially from our current expectations. These forward-looking statements speak only as of the date hereof, and we disclaim any obligation to update them.
I would like to also point out that a number of the securities analysts that cover our stock use various financial measures, and this creates a range of variability relative to the Street financial models. When we say Street estimates, we mean the consensus of the major financial analysts' models, and they are not necessarily the guidance that was given by the Company.
With that, I would like to turn the call over to Paramesh.
Thank you, Suzanne. I am very pleased to report that we delivered our fourth consecutive quarter of solid operating results. Building upon the increasing momentum in our X-Gene and HeliX families of computing products, we remain well positioned to continue to drive the proliferation of ARM-based servers into the data center market. Additionally, we are also well poised to benefit from our strong heritage in optical networking for the emerging high-growth data center connectivity market as our X-Weave family of products ramps into production.
Key business highlights from this past quarter include: first, on the computing side of our business, I'm excited to report that as announced last Monday, X-Gene was chosen as the sole ARM server solution for the European Union Supercomputing Hyperscale 2020 Initiative. The European market is demonstrating that it wants an alternative to x86, and APM was chosen amongst all of the other ARM server providers.
X-Gene is clearly the best alternative to x86 and superior to the competition in HPC and hyperscale environments with our enterprise-hardened, industry-leading products. This decisive win is testimony to our superior product and technology offering versus other ARM competitors. Additionally, as evidenced at the recent ARM Technology Conference, customers such as EMC, Hewlett Packard Enterprise, and Morgan Stanley have all validated our leadership, and we've demonstrated solid traction.
Second, we announced X-Gene 3, the industry's leading high-performance FinFET ARM server solution, which extends our technology leadership while rounding out our ARM server product portfolio. We also announced our X-Tend technology, which provides ultra-high density scale-out and virtual scale-up servers by seamlessly connecting multiple X-Gene nodes to each other while presenting multiple terabytes of shared DRAM memory as a single cluster. It also positions us to compete effectively against incumbent and future generations of x86 solutions.
Third, on the connectivity side of our business, I'm excited to report that two of the world's largest networking system OEMs have committed to fund the development of 100 and 400 gigabit per second custom products based on our 16-nanometer PAM4 and mixed-signal technologies. This underscores our technology and product leadership amongst other offerings. This success is largely due to our early commitment to invest in FinFET for PAM4. I will remind you that PAM4 is expected to be the key technology that enables the high-growth inter- and intra-data center connectivity market in the coming years.
Let me now update you on each of our main businesses. Turning first to X-Gene: as I just mentioned, our third-generation X-Gene, which was announced at ARM Tech Con last November, uses 16-nanometer FinFET and has 32 brawny cores running up to 3 gigahertz, with a nonblocking, high-performance, low-latency fabric interconnect, featuring eight DDR4 memory channels addressing up to 1 terabyte per CPU socket. X-Gene 3 has the right balance of compute and memory performance and will be a game-changer in data center and hyperscale computing, with the highest performance per watt provided by any single-socket platform. Our results to date lead us to expect X-Gene 3 performance will be comparable to Intel's E5 and E7 class of products.
In China, home to three of the Super 7 global cloud data centers, we are seeing excellent traction for X-Gene. We are excited to report that initial pilot deployments with mega data center players will result in early production deployments in the second half of this calendar year.
Also in November at Supercomputing 2015, we announced a breakthrough interconnect technology -- X-Tend, which enables ultra-high density scale-out and virtual scale-up servers. X-Tend evaluation platforms are currently shipping to AppliedMicro customers, including E4 Computer Engineering and Eurotech.
Eurotech's Aurora Hive HPC system is also based on X-Gene and was also demonstrated at SC15. The high supercomputer leverages a new, original high-performance computing system architecture, combining extreme density and best-in-class energy efficiency. This new architecture is based on an innovative, highly modular, and scalable way of packaging electronics' water cooling, boasting record-breaking computational density.
We are also excited to report that X-Gene-based storage server platforms are entering production ramps, and we have secured backlog for these. These platforms span OEM, mega data center operators, and service provider customers.
Turning now to the embedded side of our business and HeliX. We are very excited to report that as we projected a year ago, HeliX shifted to one of the world's leading data center switches this quarter. We expect to see HeliX to do very well in calendar year 2016, as we expect design wins with Tier 1 OEMs in both the enterprise networking and video surveillance market segments to ramp into production.
Regarding the development of the ecosystem for ARM-based server processors, I would like to note that Canonical released the next production version of Ubuntu 15.10 for ARM64 and X-Gene. This release has built-in support for OpenStack, with the Canonical OpenStack Autopilot, which provides a fast and easy way of building an Ubuntu OpenStack cloud based on X-Gene. This has been extremely well received by customers.
Onto our connectivity business, and specifically our X-Weave family of products. We continue our successful penetration into the high-growth data center networking market. I will remind you that 100 and 400 gigabit per second PAM4 technologies are key to driving this market.
Building on the success of our PAM4 products over the past two quarters, we continue to see strong customer validation in the form of design wins and partnerships. During the December quarter, we received our first 16-nanometer FinFET PAM4 silicon back, and it meets our customers' and their end-customer expectations in terms of power, performance, and density, which has then translated to multiple design win awards, the most important of which I elaborated upon earlier.
To reiterate the point I made earlier, this has resulted in a groundbreaking partnership with one of China's largest networking OEMs to develop a leading-edge, custom silicon solution that will power their next-generation optical networking equipment. We envision that this and other such partnerships will be the first step towards a significant market opportunity whereby we capture the value created by our commitment to investing in leading-edge technologies.
Based on committed design wins, we expect our PAM4 and mixed-signal products to contribute significant revenue in the coming fiscal year. Our base connectivity business remains very stable and continues to track well, with solid backlog coverage, a lean channel, and book-to-bill consistent with our internal forecasts.
Let me conclude by reiterating that we remain very optimistic about the continued adoption of ARM-based server processors, and this is based on orders we have in hand today. We have a clear line of sight to meaningful volume. We are building backlog as we start to see the first storage server platforms ramp into production.
With that, I'm very pleased to introduce Marty McDermut, our new CFO, for a review of the December quarter financials and our March quarter guidance. Marty?
Thank you, Paramesh; and good afternoon, everyone. Since joining AppliedMicro, I have found the Company is in an excellent position, given the significant product cycles that are just starting to ramp and the ample design wins that are in place, with many of these having larger revenue opportunities than the Company has had historically. I've been dealt a good hand.
I'll start with a review of the financial results for the fiscal third quarter, follow with comments on our balance sheet, and finish with our fourth-quarter fiscal 2016 guidance. Fiscal third-quarter revenues, gross margins, and earnings were in line with our guidance. Total revenues for the quarter increased 2% sequentially to $40.6 million. This marks the fourth consecutive quarter of increasing sequential revenue.
Regarding product revenue by market, computing revenues increased 21% sequentially to $19.8 million. We saw strong sales of our PowerPC products, along with strong sequential growth in our X-Gene revenue, albeit off a smaller base. Connectivity revenues declined to $20.8 million. Datacenter demand was robust, as expected; while telecom service provider demand was softer than we thought. Service provider business can be lumpy, but over the long term it's been a good business for us.
Given the increasing number of design wins in both the carrier and data center markets and Tier 1 customer-sponsored engagements for our 16-nanometer technologies, we expect to see robust long-term growth in our high-margin connectivity business. Total gross margin for the quarter was 55.4% and in line with the prior quarter and slightly above the midpoint of our guidance. Non-GAAP operating expenses totaled $24.6 million, a slight decline from the prior quarter and within our guidance.
I've inherited an organization that's very focused on controlling costs, and I'll do everything I can to continue to drive incremental improvements. The tax expense for the December quarter was $200,000, which is consistent with the guidance provided on last quarter's call.
The third-quarter non-GAAP net loss narrowed to $2.1 million or $0.03 per share. On a GAAP basis the net loss for the quarter was $8.2 million or $0.10 per share. Our non-GAAP financials exclude certain items required by GAAP. A complete reconciliation between GAAP and non-GAAP financials is available in our December quarter earnings release, which can be found in the investor relations section of our website.
Now, turning to the balance sheet, our cash and short-term investments were $76.4 million, up slightly from $75.5 million in the prior quarter. Cash flow from operations totaled $3.1 million, more than offsetting capital spending and other uses of funds.
And finally, turning to our outlook for the fourth quarter of fiscal 2016, these are estimates based on our current knowledge and are subject to change; as such, they're covered under our Safe Harbor statement. We estimate revenue for the March quarter will range from $40 million to $42 million. We have good visibility into our March quarter revenue, as we have backlog coverage of more than 90%, and channel inventories are lean.
We expect non-GAAP gross margins will be 57% plus or minus 1.5 percentage points. Non-GAAP operating expenses are expected to be in the range of $26 million to $27 million, which is up slightly compared to the prior quarter, primarily due to the timing of mask tapeouts. In the March quarter we expect our tax expense will be approximately $200,000. And then, finally, we expect our non-GAAP earnings per share for the quarter to be a loss of $0.03, plus or minus $0.02.
In summary, we continue to demonstrate financial and operational discipline as we execute on our operating plan. We will continue to invest in the Company's future growth and remain focused on maximizing shareholder value.
Thanks, Marty. As we close out the call, I'd like to highlight how proud I am of our team's accomplishments with respect to X-Gene 3. Building on the robust X-Gene 1 and 2 product lines, we set out to deliver an unprecedented level of performance, and the team has executed.
Now to summarize the key points of the call: number one, our underlying financials remain very healthy with the lean channel and healthy backlog coverage, despite some softness in the telecom end markets. Number two, we've established leadership pole position in the emerging PAM4 market, with leading packet optical networking OEMs collaborating with us and underwriting custom silicon development for a new generation of X-Weave products. Number three, we continue to be the market leader in ARM-based server processors, as demonstrated by X-Gene 3, the industry's highest performing SoC (technical difficulty)
Computing revenue was stepped up quite a bit in the quarter. Maybe you guys could walk us through the contributions there of what really drove that.
And I guess the guidance for next quarter is maybe a little bit below the growth rate that at least I had projected. So do you see connectivity snapping back up? Or there's some one-time lumpy orders in the current quarter, and connectivity comes back down? Maybe it's some of the -- walk us through some of the moving parts as we go from December to March quarters on the revenue side. Thanks.
Sure. Maybe I'll take a first shot; and, Marty, feel free to add. Number one, I think, as we said, the connectivity revenues have primarily been affected by a certain amount of lumpiness in the near-term service provider spending. From our perspective, that business is really good for us long-term. In fact, if you look at the number of design wins we closed, we had a record number of design win closures for our new next-generation, both X-Weave -- X-Weave PAM4 and X-Weave security, network security products. So from my perspective, the lumpiness in terms of service provider stuff is a fairly, I'll call it, a short-window phenomenon, and really doesn't affect the long-term trajectory of that business, number one.
Number two: as far as you asked about the computing revenue shooting up, there were two factors. One was, obviously, as we our product managing PowerPC and our old IBM architecture products to a lot more tighter envelope, that's really helping, because we can really drive things appropriately to make sure that we can extract the most margin and revenue out of the business over the foreseeable future -- while we are replacing it with both HeliX and X-Gene. And I think we had a really, really good, strong sequential growth relative to X-Gene, even though it was off of a small base. So I think both of those are key contributing factors for this quarter's results.
Got you. That's helpful. I guess in follow-up to that, the gross margin was guided a good bit higher for the next quarter, but it sounds like from your commentary there that connectivity revenue may be a little bit weaker in the near term. So how do we balance that? I always had thought about connectivity as well above the corporate average margin.
So are there new products in the PowerPC business or in the ARM business that are driving margins higher? Or maybe the connectivity snaps back in the March quarter?
So I think, Matt, to make it very clear, the connectivity component has two pieces. One is a service provider component; the other is a data center component. The third is what I'll call a new component, which is more in the data center bucket, which is all of the new next-generation 100 and 400 gig PAM4-type products. I think if you refer to my prepared comments, I said that we have multiple customers engaged with us to actually fund development of custom silicon products. So as you go through the year, we would expect to see some of that actually change some of the effective margins that we put out, number one.
Number two, it kind of shows that we've taken this big leap forward with FinFET and PAM4. And it's really going to pay off, even in 2016, which is a huge difference than where we were last quarter, right? On a second and related note, I think it's fair to say that the one big thing that we are really excited about is -- if you look at the blend between service provider products and data center products, I'm happy to say that we are managing those in a very tight way to make sure that in the future you'll start to see lesser and less modulation on the margins from a service provider softness-type scenario, right? So both those should give people a lot of comfort in the fact that our base business and the vectors associated with the growth are actually coming to bear in a much more accelerated fashion.
Thanks for that, Paramesh. And then, Marty, I guess first of all, welcome. Second of all, the OpEx jumps up, as you mentioned -- you called out tapeout costs. Should we expect that to be sort of a one-time phenomenon, and you get back toward the $25 million quarterly run rate in the June quarter? Or how should we think about it progressing off this one-time jump? Thanks.
Well, the base is what you saw in the past quarter; but that will vary as we do any additional tapeouts. So that can push it up or down. But generally, you know, our costs are very predictable. It's just these -- any project costs that come in or these tapeout costs come in. And we'll give guidance on that as they occur.
Hey, guys, nice navigating in a tough environment. And I'll add my welcome to Marty as well.
My first question is: you guys have been managing costs pretty well. You've lowered costs considerably, but you have all this design activity going on. So how should we think about what incremental R&D costs you are going to have to support this activity? Can you give us a little bit of a long-term idea of how you're looking at that?
I think what we've said so far is that there were three major factors. In fact, if you look back a few quarters ago to now, we've basically managed to keep this cost envelope that we said we were going to keep to in the wake of all of the design win activity that we have generated, and in the wake the design win closures that we had for both the X-Gene and HeliX family of products, as well as our new X-Weave products.
So from our perspective, I'll go back to some comments I made. The set of customers is not that big. You talk about the Big 7 data center guys; you talk about the half-dozen OEMs that really make the space tick, right? So the set of customers is actually -- as we move into transforming our Company, the set of customers becomes a lot more heavy and a lot more narrow. We're doing less and less distribution business.
So support-wise, I think the biggest costs, as what Marty identified earlier, as we go forward would be investments in things like R&D; and investments things like platform-level support for all of these new products. But I think they're all contemplated. And so far I think we have managed our cash to the point where -- I believe that -- I think we've topped all estimates coming out of this quarter. So we are very, very happy with the way things have gone. And we don't anticipate any massive forklift upgrades to expenses going forward.
Okay, great. And then just to kind of get a better idea, or if you have any more clarity into timing -- so I'm sure you were working on this European Union project with them for a while before you could mention it. But now that it's out there as of a week ago or so, how do you look at revenue ramping for project like this? What should we think about when we look out, you know, a couple of quarters, a year, two years -- how this filters through?
So first, let me answer the question right off. One is that the EU project is for the H2020 supercomputer initiative. That stuff has been -- they were looking at a number of technology vendors for the ARM server part of that big private project, and it's sponsored by a bunch of the European labs, the key universities. And what that was is basically a way to make sure that they have a bat for a 2020 supercomputer cluster that has the right server technology.
So what it did fundamentally was over the last -- I think it was almost a year that this project was used, that the bake-offs happened, and all of the pieces happened -- and we won hands-down versus any other ARM server vendor, primarily because of three things: shipping production parts today, very robust. Extremely -- it's actually the only part that the entire ecosystem kind of has used as the key part to develop software on.
And number four, if you go recall, look at all the data points over the last two years. The performance per watt is unparalleled to any other ARM vendor. Plus, when we announced X-Gene 3 in our future roadmap, it was a clear view that, okay, we actually have thought through the requirements to round out the portfolio. So that's what made the EU piece happen.
I do want to take you back -- when you asked about ramps and revenue, if you were to look at what I said last quarter, I said we have good visibility. We had really good visibility to 50,000 to 100,000 units over the next 18 to 24 months. Today we have actually backlog and POs supporting that, which is awesome, right? So I think that's the really, really important piece of showing the entire evolution and the momentum behind all of the work that we are doing.
And then, just to follow up on that: as we move forward, you're sending out a bunch of these chips so your customers can start to build out these data centers and get comfortable with it, which was part of the revenue that you saw, although you said growing off a small base. But that -- with each of these announcements, that number -- in terms of your shipping to customers for them to get comfortable -- should continue to go up in the short-term at that high growth rate off a low base. Is that fair to say?
I think what we've done -- so I think there's -- you asked two questions in one; let me try to answer it. One, we talked about all of the announcements we've made. All of the announcements we've made clearly illustrate that the shipping production platforms, whether it's from folks like HP; whether it's from folks in the HPC space, like people like Cirrascale, and E4, and Eurotech; or whether it's with guys who are building merchant storage platforms, whether it's ODMs in Taiwan, like Wistron and MiTAC. All of those have been used.
For instance, last quarter we talked about China telecom, and I gave folks an idea of the fact that the China telecom guys have decided to partner with us on specific storage server initiatives. So what is happening is with each quarter, we are talking about an end customer, not just an OEM. We're talking about specific things that they use to drive a solution, and why we were differentiated and undefeated in terms of driving ARM into these markets.
And the third piece of it was: we said, okay, each one of these will have a particular ramp rate. So in the case of China Unicom, they are buying -- I think last time I said their annual server buy is about 100,000 -- so 100,000 to 120,000 servers; and their storage tier is about 10% of that. And they'll migrate a percentage of that tier over the next 12 to 18 months to ARM. And so that's -- you're consistently following that trajectory.
So we may have seen an interesting change for you guys this quarter, perhaps, in your business strategy. It seems like the custom or semi-custom PAM work you're doing in X-Gene stuff was pretty interesting. Would you describe that as maybe a slight change in your business plan or go-to-market strategy -- these combinations? And then, secondly, has anyone else approached you regarding custom or semi-custom work, particularly for HeliX or X-Gene?
So I think the PAM4 part of our strategy -- if you were to go back two calls ago, we actually were very clear that all of this IP was residing in our wheelhouse for the longest time. I'll take you back to the days of AMCC producing the first 2.5 and 10 gig SerDes in their own fabs. A lot of the transmission IP already existed.
The one thing that we did, and we worked over the last 2.5 years to do, was migrate all that IP to 16-nanometer FinFET. And why 16 nanometer? Because we decided that's the right thing to do for X-Gene -- big market. And, by the way, when we did that, we accomplished two things. One is: we had to -- we had an unprecedented power, density -- a low-power, high-density solution that could really enable and jumpstart 100 and 400 gig -- single-laser 100 gig, multi-laser 400 gig type markets.
And that really -- its proof is in the pudding, right? We got our first silicon back. People loved it. And I can tell you that the waterfall of design wins has just been staggering; it's awesome. So that's just been really, really good for us.
Now, the customer-sponsored stuff becomes really interesting, because if you go to FinFET and do this type of stuff for the connectivity business, you need -- you know, we've always maintained that anchoring these types of programs is the key to driving revenue; it's the key to making customers become sticky, sole-sourced, and more importantly, key to building a long-term business on this piece. And that's happened, and that's huge.
As far as the X-Gene piece is concerned, you can clearly tell that because we'll have technology in the same node, it could make sense for people in the future to approach us to do that stuff. But we can't comment on whether they have or have not approached us at this time.
Okay, great. And then, also, I don't know if you can maybe put some dollar numbers towards a lifetime value for the PAM win or for the X-Weave win, and maybe even take a second stab at the European win as well.
And then, also, do you have any kind of harder metrics that you can give us for either X-Gene or HeliX? Your biggest competitor is now giving their ARM-based server stuff as a percentage of revenue. Might you be able to do that? Or might you be able to do that in the near future?
So I think -- let me first start with the PAM4 and take you back to my prepared comments. We said we will have extremely -- we'll have significant revenue this coming calendar year. And that'll be -- I believe, I think what we said was high single-digit percentage points is what I said in my prepared remarks, if I'm not mistaken. So there is an important piece of data for you.
The other piece that I'd like to kind of refer you to is it takes a long time to go off and build a real business that has real legs and has displaced Intel. So the one thing I want to be clear here is that every single X-Gene design win has displaced an Intel E3, an Intel E5, or equivalent Intel family of processors. So in other words, when we talk about building a revenue funnel or building -- you know, talking about revenue, that's what we talk about.
So if you look at it today, I think I'll take you back to my comments last quarter. I'll also take you back to the comments I made today. We have secure backlog. And if I were to go extrapolate that, from our perspective, for that kind of backlog and the orders that we're seeing and firm POs that we're getting on the books today, it wouldn't -- you know, I'd fully expect it to approach 10% of our revenue exiting the fourth quarter of our fiscal 2017.
So that's kind of giving you the color that I see today. And that's just between, you know, I would say, three to four big guys. That's it. And that's not even the extent of the total universe that is actually looking at X-Gene. And it does not include any of the UniServer stuff.
I'll just ask one more quick one, then. On the 90% backlog coverage, this seems to have trended up over time. Is there some -- you know, is that changing? Are order patterns changing? And then also, why not higher guidance, considering we still more than two months left?
Well, I think this is -- we gave you the guidance based on our visibility right now and where we think we're going to end up for the quarter. So it's based on what we actually think we can deliver. And we are pretty -- very confident about that, as I said in my comments.
So the other thing, Chris, is that given the fact that -- I think we talked about -- despite the lumpiness, we wanted to do two things. One is clearly illustrate that our business is really, from a margin perspective, is going to be starting to become invariant with the lumpiness as we -- more invariant as we go forward.
The other piece of it is that we want to make sure that we are executing to a plan. And so far we have sequentially executed really well. We've controlled expenses really well. Products are coming out, and I think we're in really good shape.
Paramesh, just on the telecom side of the business, where you noted some softness -- is that relating to customers in China? Can you talk a little more about the weakness that you're seeing in the telecom service provider market? And would you expect that to recover by the June time frame?
Yes, so I think -- you know, we look at these types of things; we've not -- it's not like we've not seen this in the past. Generally speaking, when you have this kind of telecom softness, a lot of it is to some extent geography-specific. Believe it or not, I don't think it's -- in this case it's not really geography-specific.
We don't see it that way. We just see it's almost a small inflection in this. We don't at all view it as a long-term issue for us in any way. And from our perspective, you know, it's something that is part and parcel of being in this telecom hybrid service provider and data center business. So to answer your question, don't see it as a huge trend; don't see it as some kind of big flu. It's just a small infection.
Okay. And then in the computing part of the business, good momentum there. Is that driven by the PowerPC business, which you did through foundry but not IBM? And do you see continuing legs for growth to that, or is that kind of some near-term strength that you're seeing in terms of one-time purchases? Can you talk about the PowerPC piece of the business and also when HeliX will start contributing to revenues?
Sure. So, one, we said that we had a good, strong quarter on the PowerPC stuff. Yes, that does include a lot of the non-IBM -- I'll call it our own TSMC moved PowerPC as well, which is really good. So we control that a lot more.
Second of all, on the HeliX side, I want you to note that we shipped -- as we said we would, our first HeliX parts have already started shipping in December, actually. And very happy to say that we expect that to do really well in this calendar year.
And third of all, I think the one thing that I can start to tell you is that given the fact that now we have a common software base between the server pieces -- think like OpenStack, and Autopilot, and the networking pieces -- you'll see a lot more HeliX getting into more smart networking gear going forward.
Okay. And then, finally, on X-Gene, can you talk about the progress with existing customers like PayPal, China Unicom, TSMC? Maybe share some of the customer experiences and when you might receive follow-on orders from some of those customers you've already announced?
Sure. So I think what we've said is we've shipped revenue every quarter. We had a sequential increase in revenue. Those all come from customers who are not -- as you know, Krishna, this is an enterprise cycle. People take things; they buy things in chunks. They continue to build out their infrastructure.
But I want to know two things. One is that from last quarter to this quarter, the biggest change is that we actually have firm POs for a lot of the end customer deployments, which is extremely exciting for us, number one. Number two, we also see -- we have also seen some of the biggest mega data center guys in China start to build large pilot clusters based on X-Gene, which is huge, because it's massive for guys like us.
Third, and most important, for the first time I can see how a small handful of customers, big customers, can really start to build a nice stepping stone to a large revenue contribution towards the second half of our fiscal year.
Okay. And then how's -- are you shipping X-Gene 2 in terms of production silicon for revenues now? And then based on the prototype 16-nanometer X-Gene 3 FinFET that you have, what changes do you need to make before you sort of turn that over to production silicon?
So just to be clear, right, yes, we are shipping X-Gene 1 and X-Gene 2 today. All of the POs that we've talked about, firm orders, are all for both those parts. There are no POs for X-Gene 3, because what we showed is a test chip, and we said that X-Gene 3 will only have meaningful contribution 1.5 fiscal years out.
But my point is the measurements, the FinFET test chips that we are back with, and the entire execution of a FinFET part is fully under -- you know, we're going really well, making extremely good progress. So I would tell you that we are really up and up the leader in this space, because we not only have the existing customers going to production, but most importantly we're basically hooking them up for three generations at once.
Yes, so we've always said it'll be a 2017 part. In fact, what I said on the last part is that don't count -- there's no -- given the fact that we are developed -- it's in development during fiscal 2016, it's all 2017 and 2018, right? So.
Good afternoon. It's actually David on for Cody. Thanks for taking the question, and congrats, Marty, on your new role there. Paramesh, if you could, maybe give us your thoughts on the recent QUALCOMM JV in China and how this might impact any opportunities you envision for that market?
Okay. So I think, you know, it's been going on for a while now that QUALCOMM had an investment going on in China relative to some of the early settlements that they had to do for some of the mobile patent transactions that they were dealing with. I think they used a -- this is one of the most impoverished provinces in China. I think there's some investment that was tied to it, and I think there's some development that will be tied to it local to that geography.
Really, from our perspective, it has really has zero impact on our business in China to date, because most of this stuff -- if it starts today, it's is probably going to show up three years from now. So from our perspective, that's kind of where it stands. That's the long and short of it.
Great. And with X-Gene, you talked a lot about the potential of the China market. Can you kind of tell us -- what is it that you believe you bring to China that gives you an advantage over the larger players, like QUALCOMM and Cavium?
For starters, we've had the only shipping parts, production parts for the last year. That's a really big thing, number one.
Number two, we have built on two generations of parts. We've actually got our first FinFET test silicon. So that really proves what we are doing for our next-generation devices.
And most importantly, if you really think about what it takes for an IT user to convert what he has on an existing infrastructure to a new infrastructure, it takes him to be comfortable with something that's been battle-hardened, robust, and really -- you have to have a value proposition that is performance -- that is really good performance with a significant reduction in cost and total cost, thinking power and price. So I think all of that is key.
I will also say one more thing, which is: the amount of software ecosystem work and proprietary software work done by a lot of our mega data center customers -- whether it's porting databases, whether it's porting applications for the mobile space -- has all happened only after they have found that platforms of production where they have shipped to a certain minimum threshold, either within the US or within other geographies. So that's a huge, huge deal. So anybody who has to go into the enterprise will have to go through that tunnel.
Thanks, guys -- and Marty, great to be working with you again. In terms of the X-Gene ramp, the China Unicom, is it going to be more driven by projects that are lumpy? Or will it be replacement as they put in new ones? And as you get to that 2017 sort of bogey of 10%, is it going to be concentrated on a few customers, or do you envision it will be spread out -- the X-Gene revenues -- across a broader base of customers?
No. So from my perspective, Suji, it's basically going to be a few customers. And, remember, this is just the beginning of the larger revenue ramp, right? So we have firm POs today. Really excited, because these -- their end customers identified, their entire application stuff ported. And by the way, the visibility has increased in the form of monetizable purchase orders for all of the three segments, the storage server segment being very key; the in-memory database segment being very key. And behind that is the Web segment. So all of these three segments are in line.
We clearly see how people are going to build things out and have really good -- we've translated all of that visibility now into some solid orders and bookings, number one, right? Number two, we've really not lost to any -- if you really look at where we are winning, the only guy that we are displacing is Intel. So there's really no ARM competitor that can even compete with us. And we actually -- I would say that on the majority of these designs we've not lost to a single ARM competitor.
Okay, great. And my other question is -- I know you interoperate with FPGA CPs, but does it make sense for your approach to partner more closely with an FPGA vendor going forward? Is that something that would be strategic?
I would think that it's going to seem -- you're going to see more X-Gene and HeliX in the same platform than X-Gene and FPGA in the same platform.
I'll give you one interesting example. If you will look at Mobile World Congress -- and I think a lot of the buzz around there is: how do you do more software-defined network virtualization, right? So if you really think about it, our HeliX parts that are going into things like switches for placket controlled planes, that are going into things like video surveillance servers, where they are being -- they are almost an add-in card, or they're almost an add-on processor or an accelerator processor to a bigger processor, like a big Xeon.
I see the ecosystem evolving -- where you see HeliX with x86 will essentially become where you see HeliX with X-Gene in the next two years. I really don't see an FPGA in the mix.
Thank you, ladies and gentlemen. That does conclude the Q3 2016 Applied Micro Circuits Corporation earnings conference call. Thank you for your participation and have a wonderful day. You may disconnect your lines at this time.
"Chris Rolland, FBR & Co. - Analyst [21] -------------------------------------------------------------------------------- Okay, great. And then, also, I don't know if you can maybe put some dollar numbers towards a lifetime value for the PAM win or for the X-Weave win, and maybe even take a second stab at the European win as well. And then, also, do you have any kind of harder metrics that you can give us for either X-Gene or HeliX? Your biggest competitor is now giving their ARM-based server stuff as a percentage of revenue. Might you be able to do that? Or might you be able to do that in the near future? -------------------------------------------------------------------------------- Paramesh Gopi, Applied Micro Circuits Corporation - President and CEO [22] -------------------------------------------------------------------------------- So I think -- let me first start with the PAM4 and take you back to my prepared comments. We said we will have extremely -- we'll have significant revenue this coming calendar year. And that'll be -- I believe, I think what we said was high single-digit percentage points is what I said in my prepared remarks, if I'm not mistaken. So there is an important piece of data for you. The other piece that I'd like to kind of refer you to is it takes a long time to go off and build a real business that has real legs and has displaced Intel. So the one thing I want to be clear here is that every single X-Gene design win has displaced an Intel E3, an Intel E5, or equivalent Intel family of processors. So in other words, when we talk about building a revenue funnel or building -- you know, talking about revenue, that's what we talk about. So if you look at it today, I think I'll take you back to my comments last quarter. I'll also take you back to the comments I made today. We have secure backlog. And if I were to go extrapolate that, from our perspective, for that kind of backlog and the orders that we're seeing and firm POs that we're getting on the books today, it wouldn't -- you know, I'd fully expect it to approach 10% of our revenue exiting the fourth quarter of our fiscal 2017.
So that's kind of giving you the color that I see today. And that's just between, you know, I would say, three to four big guys. That's it. And that's not even the extent of the total universe that is actually looking at X-Gene. And it does not include any of the UniServer stuff."
The backlog and PO's that Paramesh is talking about, the visible pipeline he keeps raving about is...10% of revenue exiting the 4th quarter of fiscal 2017...so are we talking about $5mln bucks? Fiscal 17 ends March 2018...is that anywhere close to some 20% of server market share?
You know, Gopi always sounds very upbeat and enthusiastic, but there is little in the way of actual substance to what he says. If I recall, I thought all this was supposed to take-off in the second half of 2016. Well, it's the second half of 2016. Now what? Did he mean calendar 2016? They're going to miss guidance for the year by $1M if they fall in the middle of current guidance for the fourth quarter. That means that next year would have to grow at about 8% quarter on quarter to reach analyst estimates of $200M for 2017. $11/share is consistent with a quarterly revenue of $76M at current multiples (!) If the $200M analyst estimate turns out to be true, then that would be consistent with a $8 valuation by the end of 2017, which I would take at this point. What they're telling the shareholders does not seem consistent with performance.
AMCC needs a better case besides smoke and mirrors.
It's not a disaster, but it is definitely NOT the Mosesmann $11/share any time soon, unless and until we can get some strong Q on Q revenue growth and P/S valuation increase consistent with a company that has cornered the market in a new type of very popular product. Until then, this is starting to sound like Chris Gardener's perpetual "softness in China" pronouncements ("lumpy revenue").
Like I said, it's not bad, but it's definitely not great. It's certainly not in line with ARM's predictions of garnering 20% of the server market.
I made the above estimates using the revenue estimates (and actual) on Yahoo, no additional dilution, and a P/S of 3. If they hit the exact center of guidance for Q4 '16, they should come in about $1M light for the entire year ('16), which is not a disaster.
They are still growing their numbers, but it would be difficult to support organic growth with less than the prognosticated stellar projections. Like I said, this is no disaster, but it is not as rosy a picture as first painted. At this rate, we will not see the analyst-projected $11/share until Q2 '19. I hope Gopi has some prospective buyers lined up, so as to maximize shareholder value in a timely manner.
Post by christopher on Jan 31, 2016 18:50:58 GMT -5
I wonder what type of new product announcements are expected in the near future?
We may be better off to put on our over sized rubber boots so we are able to wallow around in what obviously will be another large pile of steaming shit, that will soon be fed to us. yet again!
"The story at Applied Micro today is pretty much the same as it was a few years ago - and that's not a good thing.
HP seems to have stopped playing up its Moonshot server, and investor excitement around AMCC has faded somewhat - yet the stock still trades at a robust multiple.
A revenue ramp based on X-Gene 3 is promised for late calendar 2016, but previous expectations have not materialized meaningfully.
While bulls will undoubtedly say that there's a massive opportunity here if X-Gene can steal even a few percent of Intel's share, that hope seems to be priced in.
If the company does not demonstrate the viability of its technology via significant revenue from production silicon, shares could have much further to fall."
"I will assume Christopher Zepf of Kingdom Ridge Capital, will most definitely have an interest in shareholder value."
Right you are Chris. A sale might even be the reason that Marty was hired. I don't know what leverage Zepf has on the board. I know that BM actually placed people on the VTSS board.
?