Attached files
| file | filename |
|---|---|
| 8-K - 8-K - APPLIED MICRO CIRCUITS CORP | earningsreleaseq115.htm |
| EX-99.1 - EXHIBIT 99.1 - APPLIED MICRO CIRCUITS CORP | exhibit991pressreleaseq115.htm |
Exhibit 99.2
Applied Micro Circuits
Fiscal 1Q 2015 Conference Call Script
July 29, 2014
Traci Tsuchiguchi- Vice President of Investor Relations
Good afternoon everyone and thank you for joining today’s conference call. On the call with me are Dr. Paramesh Gopi, our President and CEO, and Doug Ahrens, our Vice President and 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™ and other products and our anticipated growth and profitability all constitute forward-looking statements for the purpose 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 facts are subject to a number of risks and uncertainties that may cause actual results to differ materially. We refer you to our most recent Form 10-K 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 want to point out that AppliedMicro has several analysts that cover our stock 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 analyst models and not necessarily the guidance that was given by the Company. With that I will turn the call over to Paramesh.
Paramesh Gopi - President & CEO
Thanks Traci.
As you’ve seen in the earnings release, our June quarter results came within our guidance range, albeit at the lower end of our revenue expectation. This was primarily due to a sharp decline in our legacy embedded PowerPC business - I’d like to point out that we had noted this trend on previous earnings calls. Connectivity sales, however, were very strong this past quarter and ensured that our base business would continue to support investment in our X-Gene Server on a ChipTM and X-Weave growth initiatives.
The June quarter was outstanding for X-Gene in many respects. It was the first quarter in which we recognized revenue for X-Gene. Importantly, I am pleased to announce that both X-Gene and X-Gene 2 are fully on track.
Purchase orders continue to come in and we are building backlog for X-Gene units. We are very pleased with the reports from our partners and customers of improved total cost of ownership, or TCO, and system performance attained using X-Gene.
Contribution from X-Gene is growing every quarter and we continue to expect meaningful revenue from X-Gene in the second half of our fiscal year - the December or March quarters. Consistent with what we’ve talked about on prior calls, our view of “meaningful” relates to market adoption. In our view the deployment of 10,000 X-Gene ARM servers represents a meaningful market shift.
I am also happy to report that our 28nm X-Gene 2 sampled this spring. X-Gene 2’s differentiated scale-out feature set including RDMA over Ethernet is performing as expected.
This past quarter, X-Gene was embraced by the supercomputing industry as a viable solution capable of addressing the needs of massive high performance computing, or HPC, systems.
Various customers and partners publicly validated X-Gene’s performance and TCO advantages for supercomputing environments. This leaves no doubt that X-Gene is in fact a high performance server class processor.
The work we started 12 months ago with our HPC partners NVIDIA and Mellanox, as well as the HPC scientific community, enabled us to unveil the world’s first ARM v8 HPC production platform at the International Supercomputing Conference in Leipzig last month. I would like to note that this is the first ARM platform that meets the performance requirements of the HPC community.
I highly encourage you to go to www.apm.com to see Sandia, Hewlett Packard and ARM talk about the benefits of and excitement about running HPC workloads on X-Gene. You can also view a panel discussion where partners like NVIDIA, Mellanox, ARM, as well as HPC end customers, discuss their positive experiences with X-Gene-based platforms.
HPC customers and OEMs have very stringent criteria to adopt new technology. Performance, production worthiness, and architectural superiority are key to their selection process. X-Gene’s architectural advantages are driving adoption in the supercomputing space. These include brawny cores, wide memory bandwidth, and high speed I/O, all of which are integrated onto a single piece of silicon.
As it turns out, the needs of the HPC market and the needs of more mainstream data centers are very similar. Like Big Data and other cloud and enterprise workloads, HPC workloads are highly parallel and are memory and I/O constrained. The integration of high performance networking and storage along with mixed-signal IO make X-Gene optimally suited for large, dense HPC farms.
Consequently, HPC has emerged as a tangible market expansion opportunity for X-Gene as evidenced by the number of customers that have announced X-Gene-based HPC platforms.
Earlier in the month, on the other side of the globe, we hosted an over-subscribed event in conjunction with COMPUTEX in Taiwan, where numerous Tier 1 ODMS demonstrated their X-Gene platforms that target various datacenter workloads. This proliferation of X-Gene platform starts in the Taiwan ODM community is significant because ODMs rarely invest in platform development unless there is end-customer demand. All of this, we believe, signals X-Gene’s momentum and reflects the size and growth of our opportunity funnel.
In addition to the progress we’re seeing on the ODM front in Asia, we are very excited about developments at some of the major Chinese Hyper-scale companies who have completed their benchmarking on X-Gene platforms and are currently in the process of porting their internal software and applications onto X-Gene.
We are nearing an inflection point in the process of commercializing what has been a long-term endeavor to build a revolutionary Server on a Chip solution and creating a brand new category. X-Gene is now synonymous with ARM in the server market, representing ARM’s entry into the data center as the only production solution available today. X-Gene was designed to address the needs of cloud and hyperscale data centers and remains differentiated by the following factors:
Number (1): Brawny, custom, general-purpose processor cores capable of seamlessly running all mainstream cloud workloads. X-Gene does not require partitioning or segmenting workloads into various appliances. This is critical for hyperscale customers that re-use servers for various applications over time to extend the useful life of hardware investments. It is also table stakes for those who do not know what requests will hit which servers at any given time.
Number (2): X-Gene is a highly integrated, monolithic Server on a Chip. In other words, it is a single piece of silicon, not a stacked die or a multi-chip package. It takes what are now four discrete components -- the processor, the chipset, the network interface card or NIC, and the board management controller or BMC --and combines them onto one die. This significantly benefits power consumption, physical density, and system level performance, which enable a dramatic 50% TCO reduction for cloud workloads.
Number (3): The ARM 64-bit ecosystem has been developed and optimized on X-Gene because of our differentiated time-to-market advantage. We have grown increasingly confident in our ability to preserve this lead. All subsequent offerings will have to pass the time-consuming and laborious testing and validation hurdles which X-Gene has already passed. This can take anywhere from a year to 18 months after the availability of functional silicon. We believe that by the time others are out with their first generation products, X-Gene will be nearing introduction of its third generation solution.
Now, with initial production units shipping and a broader funnel of production shipments in the fall, I’d like to remind you of the five steps to commercial deployment that we outlined back in our previous earnings calls. Steps one through four represent the major milestones towards commercialization that X-Gene and any other subsequent ARM 64-bit server silicon provider will have to complete in advance of commercial deployment.
The number and types of our customers who are now at Step 4, or pilot production, has continued to grow and now includes customers from all major classifications we’ve defined, including OEM, hyper-scale, networking and now high performance computing.
The two events that need to occur in order to move customers from Step 4 to Step 5, commercial deployment, are:
A): Receipt of production silicon. As mentioned previously, we have started to ship production X-Gene units. AND
B): Customer Insertion Point Decisions: These are based on our customers’ internal software schedules and their key ecosystem partner software customization deliverables.
We believe that as we continue to execute to our X-Gene roadmap, major Tier 1 OEMs and ecosystem partners will have less incentive to consider competitors’ ARM based product offerings, as they will have to repeat the entire platform development and software validation processes with no obvious performance or TCO benefits.
As you’ve seen, the ecosystem around ARM 64-bit and X-Gene continues to make significant progress towards commercial deployment. We are very excited to note that Canonical demonstrated Open-stack on a production X-Gene cluster managed by JUJU - their leading open source cloud management and provisioning framework. This signals X-Gene readiness for the growing SDN data center market.
We are also pleased to note that Oracle’s Java development Kit JDK has a managed beta program underway. This is key for broad enterprise adoption.
Finally, we anticipate Red Hat will launch a managed beta program in the very near future for Enterprise Linux - which will complete the readiness of the ecosystem for enterprise and private cloud deployments.
I would also like to share with you another important aspect of our ARM64b product leadership and it has to do with the embedded processor space. I am excited to share that we have secured multiple flagship, tier 1 design wins in the ARM v8 embedded space for high-end datacenter and service provider platforms. This is directly attributable to our first mover advantage with a production release ARM 64-bit implementation. We are particularly heartened by these design wins as they represent the fact that APM is the only provider of production worthy enterprise grade ARM v8 64b solutions today. Stay tuned for a formal launch of these exciting new products in the near future.
Consistent with our prior comments I’d like to summarize our computing business dynamics going forward:
While the legacy PowerPC business is in decline, our embedded ARM 64-bit family of products is poised to provide compensatory revenues and fuel future growth in our base business. This is a result of already secured Tier 1 flagship embedded design wins and an actively growing funnel.
Now to our connectivity business:
Our X-Weave family of products represents our newest offerings and will serve as the primary growth vehicle for this market in the years ahead. In fact, we have started to ship pilot production volumes of X-Weave 28 nm products and are on track to deliver production 28nm X-Weave units to Tier 1 customers in the December 2014 and March 2015 quarters.
Most importantly, cash flow from our base business will continue to support X-Gene and X-Weave development and production efforts.
With that, let me turn the call over to Doug.
Doug Ahrens - CFO
Thanks Paramesh.
June quarter consolidated net revenues were $50.3$51.8 million dollars, towards the lower end of our guidance range. Computing revenues were approximately $18.9$27.6 million dollars, which declined more than expected due to weaker legacy PowerPC sales. We are finding that revenue from the oldest products is declining rapidly. Many of these products are over 10 years old and were part of the legacy IBM embedded PowerPC acquisition that the Company made many years ago. The recent uncertainty regarding IBM’s foundry business and its possible sale to Global Foundries has impacted customer behavior, causing an acceleration of the transition away from the PowerPC architecture. This has resulted in the decline of this legacy business to be more front-end loaded in our fiscal year than originally expected.
Importantly, the June quarter was the first quarter in which we recognized revenue for X-Gene, which approached $1 million dollars. This is important in that it signifies the beginning of the ramp of our new, flagship product.
Because this is a new category, we felt it would be helpful to provide more granularity into the starting point of the X-Gene ramp. However, we do not intend to break out X-Gene revenue beyond this initial ramp as it is part of our Computing business.
Connectivity revenues were approximately $31.4$24.2 million dollars. In the June quarter within Connectivity, we experienced strength across the board, including 40 and 100 Gigabit Ethernet products. We continue to invest in Connectivity and leverage our IP to develop innovative, integrated solutions.
In terms of geographical splits, sales to North America accounted for approximately 38%42% of total revenues, sales to Europe contributed 10%10%, and sales to Asia contributed 52%48%. In the quarter, three customers accounted for more than 10% of our business: global logistics support vendor Wintec and distributors Avnet and Paltek. Distributor revenues for the June quarter were approximately $30.9$26.8 million dollars.
Non-GAAP gross margin in the June quarter was 59.8%61.9%, above our guidance range due to the more favorable revenue mix toward higher margin Connectivity products relative to our expectation. Our first quarter non-GAAP net income was $155$730 thousand dollars or break-even on an earnings per share basis. Our non-GAAP financials exclude certain items required by GAAP.
Our net loss on a GAAP basis was $13.1 million dollars or 17 cents per share, versus net income of $23.1 million dollars or 30 cents per diluted share for the prior quarter. A complete reconciliation between GAAP and non-GAAP
financials is available on our first quarter earnings release, which can be found in the Investor Relations section of our website. Please note, there is no reconciliation relating to forward looking statements.
Channel inventory, excluding certain non-cancelable, non-returnable orders, was 80 days, up from 47 days in the prior quarter. The increase in days of inventory was broad-based across our product lines.
Turning to the balance sheet, our cash and short-term investments totaled $98.6 million dollars, down from $106.6 million dollars in the prior quarter, in-line with our expectations.
Inventory levels were $20.7 million dollars, up from $18.9 million dollars in the prior quarter in preparation for the upcoming commercialization of X-Gene.
Now, turning to our payments related to the Veloce acquisition: During the June quarter, Veloce payments were comprised of $6.4 million dollars in cash and $300 thousand dollars in stock. To date, we’ve paid a total of $160.8 million dollars out of the maximum Veloce merger consideration of $178.5 million dollars.
Importantly, cash flow from operations was breakeven during the June quarter, excluding the merger consideration paid to Veloce.
With regard to guidance for the September quarter, we expect revenues to be in the range of $42 to $46 million dollars, and our backlog coverage as of the date of this call is over 70%. The book-to-bill ratio in the June quarter was below 1.
This comprehends our expectation that legacy PowerPC revenue will continue to decline rapidly. We believe the embedded computing business is approaching the bottom as the majority of the oldest products will have reached end of life. This guidance also accounts for a pause in Connectivity following a very robust June quarter. We view this as a recalibration of our base business upon which X-Gene can grow.
We expect gross margins in the range of 60 to 63%. Our gross margins in the September quarter are expected to benefit from the forecasted mix between the Connectivity and Computing businesses.
Non-GAAP operating expenses are expected to be in the range of $28 to $30 million dollars. We expect earnings per share to be a loss of $0.02 cents, plus or minus one cent on a non-GAAP basis. Furthermore, we expect the share count to be in the range of 78 to 81 million shares.
Importantly, the cash flow from our core business continues to enable us to invest in X-Gene and X-Weave development through commercialization.
Now, I’ll turn it back over to Paramesh for some closing remarks.
Paramesh Gopi - President & CEO
Thanks Doug.
In conclusion, we’d like to leave you with the following:
X-Gene is on track. We commenced initial shipments of production silicon, purchase orders continue to build, and we have backlog. Our 28-nanometer X-Gene 2 has started to sample and X-Gene 3 design work is well underway.
We will host a timely analyst event in conjunction with ARM TechCon in the fall - we hope you will be able to join us. We will also participate in the Roth Semiconductor Corporate Access Day as well as the Deutsche Bank Technology Conference in September.
With that, we’d like to open the call up for questions. Operator?
(Operator Instructions)
Cody Acree - Ascendiant Capital Markets, Analyst
Thanks, guys, for taking the questions and congrats on getting X-Gene initial revenue. I guess, Paramesh, to the extent you can, you said material revenue, about 10,000 units. In your guidance for the September quarter, can you give us any color on the contribution you expect for X-Gene?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
I think what I wanted to make sure to be very clear about is: We have already shipped about $1 million of revenue, which was not ever anticipated or disclosed to the Street, number one. Number two, we said that we have credible backlog. Number three, we clarified what meaningful would mean because it's not just us, but the entire category reaching 10,000 ARM which equals X-Gene servers in the market. So, suffice it to say that there is backlog today on the books for X-Gene, both in the September quarter and December quarter, as well as March quarter.
Cody Acree - Ascendiant Capital Markets, Analyst
And can you talk more about that step, that fifth step, to material revenue? So they now have production availability of silicon. What is the -- I know it is different for every customer, and maybe every software suite, but the time that it takes to get this done for customers?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
Let me go back and talk to you a little bit about the four markets. We have always had that too. The guys who own their own data centers, those are the guys that we have the least visibility into because they will do things per their internal cadence. HP and Dell, we publicly said where they are. They have actually publicly gone on record -- HP has, at least, last quarter after ISC, talking about how excited they are about X-Gene, the platforms, and the types of workload that they are talking about.
On the networking side, we have always maintained that was a very important part of our business going forward and there has been huge progress there. All the software hurdles have been removed.
And the fourth piece, which is really neat, is the HPC piece.
So I want to clarify one thing on the HPC side. NVIDIA and, I will call it, the scientific computing community have actually benchmarked and have equated -- based on the HPCG, which is the high-performance computing benchmarks -- us to Xeon E5.
So there is no doubt that there is no performance hurdle that we haven't met that the industry requires for us to see the type of traction that we are seeing.
So from an ecosystem point of view, we've always talked about it; if you go back and look at the last three calls, we have talked about compilers, we have talked about Java, we have talked about Red Hat, we have talked about Canonical. And all of -- if you were to look at the time sequencing of all of this, we are here today because of all that work.
Cody Acree - Ascendiant Capital Markets, Analyst
And then, lastly, on X-Weave, thanks for the update there, but I think in previous calls, you had said that you expected that to ramp to material revenue at about the same pace as X-Gene. Is that still the case? It sounds like maybe that is just a little bit behind where X-Gene is tracking.
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
I think we have always maintained that we would have X-Gene revenue this fiscal year. And as far as X-Weave, I think we told you on the last call, if I recall correctly, they have already shipped samples -- engineering samples -- of
28-nanometer X-Weave to our Tier 1 customers. What we are telling you is that we will start to ship pilot production and production units towards the second half of our fiscal year, right on track. No change.
Cody Acree - Ascendiant Capital Markets, Analyst
Okay, great. Thank you, guys.
Matt Ramsey - Canaccord Genuity, Analyst
Thanks for taking my questions and, again, congratulations on the first revenue from X-Gene. Paramesh, a couple of ecosystem questions. You laid out the compelling case there of why X-Gene would be appropriate for HPC workloads. I wonder if you could give a little color on what in that ecosystem of HPC for the server space has changed over the last three, six months, because at least from my point of view, it didn't seem part of the initial X-Gene TAM you guys were targeting, and now you're very bullish with some new partners in that space. Thanks.
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
Really good question. What has changed in the last, probably, six months?
So a year ago, as you probably know, HPC is primarily dominated by E5 and E7 type Xeon processors. That's fact, right?
About a year ago, we started discussions with two of our partners, NVIDIA and Mellanox, because obviously the HPC community is looking for bigger, denser, lower energy clusters that they want to deploy. And one of the big things that the HPC environment offered us, either through HP's Moonshot initial kind of testing that they did with our cartridges almost a year ago now; it was a very interesting point of view into the fact that the software is all open source. It is all the same LAMP stack, so it's Linux, Apache, MySQL, and PHP.
In fact, I was reminded by some folks at CERN that I met about three weeks ago that they invented the Web. So, it was interesting coming back to see data centers from where they started.
So the same software stack and the software readiness that we saw for the mainstream data center ecosystem is fully applicable to the HPC space. Having said that, the biggest challenge is how do we make sure that we can play nicely with the two key incumbent connectivity and, I will call it, DSP or very, very high mathematical compute engines that exist in that space.
So, the biggest work that happened there was we had to actually get drivers for the NVIDIA graphics engines ported onto our X-Gene platform, as well as really work with Mellanox to get to where we can see latencies that are comparable to Xeon-based architectures that are shipping today. And if you were to go and look at what happened at ISC, the entire scientific community was taken by storm because they couldn't believe that an ARM could perform like an E5 and they had us look at HPCG benchmarks. They had us look at a whole bunch of other benchmarks, and data speaks the most loudly, right?
So, it was very clear that it is equivalent in performance to a Xeon. So, I think all of the claims that we could not really talk about in terms of the data center access that we were getting with our key Tier 1 OEMs got wrung out by the scientific community and put up for everybody to see.
Matt Ramsey - Canaccord Genuity, Analyst
Thanks, that's really helpful. I guess for Doug, maybe a couple longer-term questions. I guess I appreciate the PowerPC declines and the guidance you have given for the next quarter, but you had laid out some longer-term guidance profile in the not-too-distant past about doubling the Company's revenue from an X-Gene ramp with a certain timetable and also the legacy combined businesses before X-Gene remaining flat as a foundation. Maybe could you give some updates if those two guidance points still stand and the timeframes on those, and any color at all around a ramp trajectory. I know it's tough to give, but anything like that would be super helpful from our perspective. Thanks.
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
Sure, so just looking at the September quarter, first of all, we talked about the structural decline in the oldest PowerPC products pretty much running its course here. And so, we also see that pause in the connectivity side and that leads to our pause in the -- or our guidance for the September quarter.
And then, you start to see X-Gene meaningful revenues coming in in the second half of our fiscal year, and then the long-range objective we have put out there is to double the size of the Company within three years and that is very tenable, given the bottom's up -- the sockets we see out there and the design win opportunities. Customer by customer, we can see a clear path to that goal in the next three years.
So the early stages, of course, are hard to call, given this is a brand-new category, and we haven't put a slope on that line. We have just given you a starting point and an objective that we see a clear path to, three years out.
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
So to add to Doug's comments, I want to make one thing very clear. We are not changing the view that I had about nine months ago, saying that we are on track to doubling the Company's revenue. And when I talked about it, the average run rate I think over the last four quarters was $200 million or so.
So we are not changing -- at least from a view point of view, we are not changing that target for ourselves, number one. Number two, I want to remind you that we are now in possession of securing design wins in the embedded space that are now firm and set. Because of our leadership in the server space, we have been able to essentially really drive ARM into the embedded networking space, so that should provide you a real substrate to understand the fact that our base business, after this PowerPC inflection point, will have a very, very tenable path to growing because we will drive that embedded business with the secure design wins that we have, and that's just the tip of the iceberg on the embedded stuff.
Matt Ramsey - Canaccord Genuity, Analyst
Thanks very much, guys. I will get back in the queue. Appreciate it.
Ambrish Srivastava - BMO Capital Partners, Analyst
I had a quick couple ones for Doug, and then for you, Paramesh. For Doug, why is inventory in the channel jumping around so much, Doug? It was down to 47 and then back up again. And then on the guide for September and looking at the PowerPC-related legacy business, are you calling a bottom in September and you expect that to -- that piece, that bucket, to start to move back up again?
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
Sure, so I will answer the questions in the order you gave them. So regarding the channel inventory, we did reach a very low level in the prior quarters. So 47 days is historically very low for us and it is now up to 80, and it was pretty broad based, but I can give you a couple of examples to give some color.
So for example, with the older PowerPC products, the sell-through slowed down quite a bit. So the inventory did not move through as expected. That's one example.
Another one is there has been high demand for our connectivity products in Asia. The sell-through has been pretty good, but what you see is a lot of demand for those products. We have filled those orders and the June quarter was very robust, and now we see a pause heading into this quarter.
So, those are two examples contributing to that. But it was broad based, as we said.
Regarding the PowerPC trend and whether this is the bottom, we are approaching the bottom here because of the fact that the oldest PowerPC products have essentially reached end of life by the September quarter. There is still some in there, but it is getting quite small. So now you are getting down to some PowerPC products that are TSMC based, rather than IBM based, that still have several years of life in them, and so those create the new base for the
PowerPC business. And then, we get to the embedded ARM products down the road that Paramesh talked about, which resumes the growth in computing with just the embedded space alone. And then, you add X-Gene on top of that for the computing segment.
Ambrish Srivastava - BMO Capital Partners, Analyst
Okay, and then for you, Paramesh, clearly on the ARM-based server front, you have been ahead of the competition and we have maintained that. But now that you have actual products out there, one question is -- first question is, what are you seeing in terms of response from Intel?
And the second question, I am just scratching my head because the ARM-based servers are such a big cornerstone of what you guys have been doing for the last not just one year, two, three years. Why are you saying that after this quarter you will not break out the revenues for ARM-based servers?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
Maybe I will answer your last question first. Because between our embedded business on the ARM front starting to gain traction in design win funnels and X-Gene, they are all part of our computing businesses. One of the things that we wanted to make sure that we categorically take away from all investors' minds and perception issues is that there is no more hurdle in terms of our product delivery or factors under our control to tip the market. I want to be categorical here.
We are shipping production parts today. These are all 2.5 gigahertz type parts. They are all fully production worthy. They are going to Tier 1 guys, so there is no notion of is this real or is this not real. That is the only reason we wanted to make sure that we are categorical about breaking it out and giving you granularity in terms of what has already happened a quarter earlier than we ever said anything would.
As far as the second piece and Intel is concerned, from our perspective we have a very unique view in terms of what makes the key pieces of what we are putting together really shine. And part of that is, again, I come back to the fundamentals: Most of the cloud workloads, as has now been evidenced outside any data center operator, but you can go check the data, it's all there, look at the HPCG benchmarks. Look at how much I/O, look at the latency that we get even with InfiniBand and Mellanox in Nvidia today. If you look at the leading platforms that are shipping today with E5s, the latency bounds in terms of fractions of a microsecond for communication across CPU and memory and I/O, along with HPCG benchmarks that talk to you about the performance of memory streaming subsystems, are all visible to all parties.
Therefore, from our perspective, that whole notion of we have now -- we can talk about competing. Let me try this on for size. We have parts that are non-FinFET that are competing with leading-edge FinFET parts. That was the whole premise of what we're trying to build.
Ambrish Srivastava - BMO Capital Partners, Analyst
Okay, thank you.
Rick Schafer - Oppenheimer, Analyst
I guess I will start off with a follow-up to one of the last questions, and that's just, Doug, I don't know, or Paramesh, if you guys would care to put a dollar figure or a timeframe around where you see the PowerPC business stabilizing or bottoming out or however you want to phrase it. Is it two quarters out or can you just -- some kind of frame it up some way?
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
So we, of course, don't give guidance past one quarter, but trendwise, I can tell you that we see that we are approaching the bottom or very close to it, if not already there, by the September quarter.
Rick Schafer - Oppenheimer, Analyst
Okay, okay. And maybe just a high-level question for you, Paramesh. I have been asking myself this question for a while, but why do you think the overall data center Cloud Server market has been slower to take off, just generally speaking?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
I think that's a question of relative view. (laughter). I think there was a lot of expectation, Rick, set very early on, two years ago, if you look at what was said out there by people like Calxeda, 32-bit stuff. I can go back and I can look at our earnings transcript calls and questions that were asked to me about how do you compete with people like Calxeda in saying that they will have revenue end of two years ago.
So, I think it takes a lot of time to build a mature product and I think we have learned a lot, that you cannot live on PowerPoint and you cannot live on false claims. I will give you an idea. Nvidia, Mellanox, and the scientific community, you are talking about DoD types here. They don't run anything frivolously. So mission-critical applications, scientific computing applications are very data driven, and data driven means production-worthy parts at speed, at test.
That's why I am saying if people have not experienced what it takes to bring a server-class market part to production and if you have never built an x86 before, I invite you to go through the -- I will call it the pain -- to go get something done. Which is why I want to emphasize the amount of confidence we feel in our lead from getting -- to get this market started.
Rick Schafer - Oppenheimer, Analyst
Got it. And just one more and I will try to ask this a different way. But is there any way you can give us a sense -- maybe it's a rough idea of the number of POs that you have out for X-Gene right now or an idea of the size of the average PO that you guys have. Just some way to get a sense or gauge what that design win backlog looks like for you guys right now?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
So I can give you some color. The POs come from all four spheres of the markets that we talked about. And the POs, in terms of the people that we have been engaged with for many, many months now and many quarters, they are repeatable, increasing POs from those people. There are POs from the HPC community. And I think I can tell you for a fact that the PO color has broadened; the funnel has broadened. I can tell you that.
Rick Schafer - Oppenheimer, Analyst
Okay, thanks, Paramesh.
Chris Rolland - FBR Capital Markets, Analyst
Thanks for the question. Nice bounce here for connectivity in Q2. I'm imagining a decent drawdown for the third quarter here. So maybe you guys can talk about the puts and takes here, why the upside and then why the downside in 3Q? Thanks.
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
Sure. There was very high demand in the June quarter for us, for our connectivity parts, particularly in Asia. There are a lot of Tier 1 vendors competing for these build-outs in China with -- our 40-gig parts and our 100-gig parts were very -- in very high demand for those deployments. So you've got the usual names all moving very aggressively last quarter into final deployments.
And so, there is a lot of demand for our parts, so we see that being a growth rate, though, that can't be continued into the next quarter. We grew 30% in our connectivity area quarter on quarter, so we don't see that type of growth rate being sustainable into the following quarters. So we think we are going to see a pause going forward.
Chris Rolland - FBR Capital Markets, Analyst
Okay, great. And then two quick ones, I guess, one for Paramesh and one for you, Doug. On the visibility into the 10,000 ARM servers, is that in a single quarter or was that over the back half of the fiscal year? And then, Doug,
more of a housekeeping. Can you tell us about some non-GAAP but cash charges that you might expect for the rest of the year? Thanks.
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
Sure. So first of all, I always talked about meaningful in that 10,000-server range. I wanted to actually publicly state it in a prepared call statement to say that we are fully confident that is what we would view it as something that is an industry category adoption metric.
That does not in any way hamper us from doing more or less or any of that stuff. But more importantly, I want to say that we are still fully on track, fully committed to saying that we will have meaningful revenue towards the second half of our fiscal year.
One thing I can tell you is if we didn't have our $1 million already shipped or close to $1 million already shipped, that would not be a possibility. It is sequentially linear. One causes the other.
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
Chris, to your second question, the primary driver between GAAP and non-GAAP is stock-based comp, so it's a non-cash item. So the other items are in our non-GAAP P&L. That's the main item to point to.
Chris Rolland - FBR Capital Markets, Analyst
Sorry, no more additional restructuring charges? Or anything like that?
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
I'm sorry. So, there is none that we have planned right now in terms of restructuring charges, no.
Chris Rolland - FBR Capital Markets, Analyst
Thanks, guys.
Christopher Longiaru - Sidoti & Company, Analyst
Congratulations on the X-Gene progression. My first question has to do with just relative pricing. You have made some comments about your major competitor on the X-Gene front in the past as being around half the price from a total BOM perspective. Is that still a correct assessment or can you make any relative comments on total cost?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
I think architecturally we are talking about -- there is nothing changed from an architectural point of view. I will point out to you that X-Gene 2 actually enhances the notion of a lower TCO because it does integrate very high-performance RoCE or RDMA -- an RDMA NIC over Ethernet.
So fundamentally, I think there is no change to our initial sizing of the market, and BOM savings, TCO savings are the two keys that drive adoption.
Christopher Longiaru - Sidoti & Company, Analyst
So those relative statements that you had made in the early stages of X-Gene still hold true in terms of around half the cost?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
Yes, yes, absolutely.
Christopher Longiaru - Sidoti & Company, Analyst
Okay, okay. And then, just, so you had mentioned to Mark -- this was maybe three or four quarters ago -- an anticipated total market opportunity for X-Weave to be around $1 billion a couple of years out. Is that still the anticipation, or as this progresses, do you see -- you sound a little more bullish. Do you see more of an opportunity? Can you update us on your expectation for that a ways out?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
No, I think from my perspective you should -- the TAM has not changed appreciably. In fact, I would almost tell you that I don't want to set too high of an expectation relative to the TAM expanding, but I can tell you for a fact going forward that X-Weave will have a lot more applications even in that TAM to get more share of that market as it starts to wind its way into things like backplane cables, as well as next-generation digital backplanes and optical backplanes.
So I am very, very excited about the fact that we are already going to be shipping production and pilot production units of X-Weave at 28 to multiple customers this fall.
Christopher Longiaru - Sidoti & Company, Analyst
So in terms of that progression, does that -- I know that this was something that you jumped in with a little bit later and you always talked about X-Gene leading the way and X-Weave maybe being a couple of quarters behind it. Now it seems like you have even moved that up slightly, but the idea here is - are you going to be selling these two things in tandem? Are they essentially going to be sold to a lot of the same customer base at the same time, and so the progression will be in lockstep, or do you see X-Weave following by a couple quarters versus X-Gene?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
I would tell you that there is -- the initial ramps of them are not connected. Let me start there, number one.
Number two, I would tell you that X-Weave can stand and will grow on its own, independent of X-Gene, and once they get lumped together in the 24-month timeframe that we've talked about before, it has a very dramatic affect on building systems for data center guys. So initially, we are not counting on any interplay between the two parts.
Christopher Longiaru - Sidoti & Company, Analyst
Okay, that's helpful. And then the easiest question you're going to get, I had a couple different calls so I missed the revenue guidance. Can you give me that again?
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
Yes, the revenue guidance was $42 million to $46 million for the September quarter.
Christopher Longiaru - Sidoti & Company, Analyst
All right, I will jump out with that. Thanks, guys.
Krishna Shankar - ROTH Capital Partners, Analyst
Paramesh and Doug, congratulations on the X-Gene production milestones. Would you anticipate, given this trough in the PowerPC embedded space here in the September quarter, would you expect the connectivity business to resume growth in the December quarter?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
I can tell you for a fact, Krishna, that we don't guide that far out, but I can tell you that from a service provider point of view and a data center point of view, the reason Doug talked about the pause earlier was primarily because multiple people competed for two major parts of a carrier RFQ in China for 100 gig, and I think everybody bought a lot.
So, I think it represented a very interesting, but slightly deviant, demand cycle than we usually see. We're used to seeing a lot more smooth demand from a data center to data center or carrier buildout for very high performance 40 gig, not necessarily 100 gig. So, I think what Doug meant by the pause is that there was a deviant demand going, driven by two or three RFQs and service by seven guys. So all of them buy from us for the 100-gig module market, so we saw that.
I think going forward, I can tell you for a fact that none of the macroeconomic factors that we know of today would suggest that you would have any of those types of deviancies in the future. And I can also tell you that we are very --
being driven by our customers to get the X-Weave products even earlier, if we can. So there is no -- from a macro market perspective -- we don't see any major factors that could affect that.
Krishna Shankar - ROTH Capital Partners, Analyst
Okay, and then turning to X-Gene, do you anticipate a steady ramp over the next few quarters or will there be a pause? I know you are shipping some production shipments now to customers, but will there be a pause after the initial -- after they verify the initial production shipments and put their data centers or networks into deployment, will there be a pause before you ship the next larger volume of X-Gene production shipments?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
I can tell you for a fact that at this point in time, as I said, the reason we broke out the big -- the chunk that we said earlier -- is that is what I call the real causal point that gives us confidence in the backlog we have today, it's going into meaningful deployments, number one.
A lot of that testing, just to address your question directly, has already happened. I think we have said that we have shipped many, many, many hundreds of parts over the last three quarters essentially to multiple customers to do a lot of this pilot testing and bring up, so the step from now to basically when they start to deploy stuff has to do with two major pieces. One is obviously ordering production silicon, which is taken out -- is already there. The other is things that they may want to do to insert certain applications or certain pieces of their own infrastructure ported over to X-Gene, and that's what I call insertion point in their data centers.
In some cases, we have visibility; in some cases, we don't. As with any new category, Krishna, I want to be very open. The fact that we are building backlog today, or seeing things happening, gives me great confidence that we're on a very steady kind of path to achieving growth in this category. Having said that, though, we are not privy to all of the data center deployment dynamics within some of our larger customers.
Krishna Shankar - ROTH Capital Partners, Analyst
Okay, and then, so of the four buckets, you would expect the web scale -- how would you rank order the four revenue opportunities -- server OEMs, the networking converged OEMs, web scale, and HPC? Could you rank order the revenue opportunities for those four buckets over the next year?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
I would probably say that we have equal funnels. If you look at the funnel and if we were to divide it, I would say that the web scale part of the funnel is probably higher than the HPC part of the funnel, and the HPC part of the funnel over the last, probably, year has grown to be equivalent to the funnel that would combine the networking and the storage pieces together.
So I think -- I would say that it is strength from all pieces, from all four legs. The HPC leg was something, as I told you before, we never disclosed or didn't -- when we started this, we didn't know, frankly, whether we would have the chops to go and really compete in the supercomputer space, but I think the data proves that we are there.
Krishna Shankar - ROTH Capital Partners, Analyst
Great, thank you. And then, Doug, can you talk a little bit about the cash burn for the September quarter and how you see working capital needs as you do the X-Gene ramp?
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
Sure. So when we look forward to next quarter, we still expect cash to be in the upper $90 million, so we don't see a significant cash burn. There are just the puts and the takes in the working capital. Some inventory build will continue to unfold as we build X-Gene, but other than that, there are no particular puts and takes that would cause the cash balance to change significantly. So we have close to $100 million on the balance sheet in cash and short-term investments.
Mike Lucarelli - Evercore, Analyst
Thanks for taking the questions here. Most of my questions have been answered at this point, but I thought I might make another stab here at your guidance, maybe. I know you talked here and there about where it is going to go, but is the magnitude of the decline in your legacy business going to be similar this quarter as it was last quarter?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
Mike, you are a little feeble. Can you repeat the question again? Sorry.
Mike Lucarelli - Evercore, Analyst
Sure. Your legacy business, that was down $10 million last quarter. Should we think about the magnitude of that decline as similar this quarter as it was last quarter?
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
Mike, we don't give the mix in the guidance. We never have. But the driver, the primary driver, of this decline is a continued decline in the PowerPC business, the oldest products. So you can infer that is where the revenue guide is coming from.
Mike Lucarelli - Evercore, Analyst
Okay, then I should take it as a pause being flattish to the connectivity business?
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
We don't know exactly how that's going to play out, but that's a fair assumption for now.
Mike Lucarelli - Evercore, Analyst
Okay, fair enough. And then, you gave the number for X-Gene, which is great, for last quarter. Was that included in your number for processor revenues?
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
Yes, it's in the computing numbers we gave.
Mike Lucarelli - Evercore, Analyst
Got you. Okay. And since you had this big decline in the PowerPC business, does this give you any more line of sight to when your embedded ARM design wins ramp? Will they ramp sooner now than you originally thought?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
I think what we have always said is that they will start to ramp end of 2015 and most of 2016. At this point in time, I think that's what we are maintaining, but I can tell you that the design win funnel for that embedded ARM business continues to grow at a very, very nice pace. So, the reason we are confident in saying that it will be compensatory in terms of making sure that the base is back up to where we would have been, given all the secular kind of PowerPC going away, is primarily because we have confidence in that design win funnel. These are already granted design wins from our current -- some of them being our current PowerPC customers, too.
Mike Lucarelli - Evercore, Analyst
Got you, okay. Then last question here, you have some good confidence in meaningful revenues in this fiscal second half. Do you think you're going to have another 10% customer before fiscal year-end?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
Wow, that's a really good question. At this point in time, I don't -- we haven't -- I actually haven't contemplated that, but clearly, given the success of where we are going, right, maybe we need to go and do some math to go figure that out. But I don't think so. I don't know whether we can break that out at this point in time. No.
Mike Lucarelli - Evercore, Analyst
Okay, fair enough, and congrats on the first revenues, guys.
Hans Mosesmann - Raymond James, Analyst
Congrats on the -- . Paramesh, the benchmarking that you talked about with the X-Gene versus Xeon E5, that's a 40-nanometer versus 22-nanometer comparison. Can you tell us just at a high level what would X-Gene 2 be comparable to in the Intel lineup? Is it fair to say that's now reaching into the E7?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
No, no. I want to be very clear. A really good question, Hans, and I think the one thing that we have always talked about is when we started out X-Gene, remember that Sandy Bridge was the prevalent shipping processor. And if you really look at it, when we announced X-Gene, Sandy Bridge was starting to hit its mainstream status and FinFET was already announced two years out. This was two years ago.
So, fundamentally at that time, we always said that we would be in the range of two-thirds of the performance of an E5, high-end E5, and basically at roughly about 40% the TCO. And we maintain that. Now having said that, what we wanted to demonstrate, without any marketing, PowerPoint, any of that stuff, is to say a completely third-party set of benchmarks, which essentially exercise the key parts of X-Gene, whether it is high-performance CPU, high-performance memory subsystem, and the networking subsystem, is stuff that the high-performance computing community uses and it is called HPCG.
So they run that both on a current shipping E5, both on the Ivy Bridge and the Sandy Bridge type platforms, as well as on X-Gene 1, which is the 40-nanometer product, and they have seen equivalent performance for their HPC applications. And when you actually run it, combine I/O, combine the graphics performance that is required to sustain one HPC node in a cluster, you pretty much can do exactly what an E5 does with X-Gene one at a fraction of the TCO.
With X-Gene 2, you get even more advantages because now you have lower power. Obviously, that means that you can squeeze more DSPs, more InfiniBand, more processors into one rack, but more importantly, you have the ability to scale out, so you can have two X-Gene ones and you can build essentially a 16 core high-performance MEMS cache configuration, along with X-Gene 2.
So, if you really think about it, we have now broken one more big piece that has only been native to high-performance computing, which is how do we really scale out workloads?
Hans Mosesmann - Raymond James, Analyst
Okay, that's very helpful. Two other questions. The inventory increase in the June quarter, what part of that increase was X-Gene related?
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
Very, very little. We are talking about the channel inventory, I believe, is what you are referring?
Hans Mosesmann - Raymond James, Analyst
No, on the balance sheet.
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
I am sorry. On the balance sheet, yes, it was a significant part of that increase.
Hans Mosesmann - Raymond James, Analyst
Okay, and then the last question is the availability of 40-nanometer at your foundries, is that an issue or could that be an issue over the next several quarters?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
Answer is no. No issue.
Hans Mosesmann - Raymond James, Analyst
Okay, thank you, and congrats.
Matt Ramsey - Canaccord Genuity, Analyst
Thanks, guys, for letting me hop back in. Paramesh, I just wanted to ask a couple of longer-term and maybe change dynamics in the embedded or the networking space where it sounds like you are more excited about enterprise networking design wins. What really changed there competitively over the last six months or so? Are these just partners or potential customers in that space wanting to get off of legacy architectures into ARMs sooner, and you guys are the first to market? Is it new sustainable partnerships for you? Is this business defensible that you may win as other ARM designs come in?
Just anything around the dynamics in that market, because it seems like, again, similar to HPC, but it sounds like those are some new design wins that you might not have expected maybe a year ago or weren't as confident about. Thanks.
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
Really good question. Let me give you some color. About a year ago, I think, and I may be off by a quarter or so, we announced on one of our earnings calls that our key embedded customers and the embedded market is transitioning away from both PowerPC and MIPS architectures. I think we said a lot of our key Tier 1 customers, the likes of the Huaweis and the Ciscos and the ZTEs of the world, have made a commitment to only really focus on two key architectures in the future, which is the ARM architecture and the x86 architecture.
So, what we said was that the industry will go through a massive transition over the next three years, and the ARM architecture and people who possess the fundamental ingredients for the ARM architecture will make big headway into the embedded space. And the x86 architecture will down-flow their server parts into the embedded space. That has already come to be.
So, what changed? A lot of the embedded system requirements for high-end networking, whether it is SDN boxes, SDN controllers, or things like control plane and/or data plane processors and switches in high-performance data networking routers are all moving and looking for an alternative that will give them the ability to increase their effective networking density, which is more pipes in, more pipes out, and have the ability to take one code base that runs, surprising enough like data center guys, on Linux and move it between x86 and ARM to get better TCO.
The big hurdle for all these guys was twofold. One is some of them were using Wind River Linux, and Wind River Linux happens -- Wind River happens to be owned by Intel, right? So, we had to go through and prove beyond doubt that we can match the performance of what they are already using on x86 toe to toe. That happens only when you have production silicon that can be evaluated by people who write code for these networking applications.
The last year, we spent proving beyond doubt that when you move mission-critical networking code or embedded code, you now have a real robust platform that you can use. So that's really tipped the scales on that part of the business. I will tell you that the number of design wins that we have put on the list over the last six months have primarily been driven by the reality we have brought to the "ARM in networking" construct. So, that's a big change.
The second piece of it is if you're going to go and do an embedded ARM design for a captive socket that is either MIPS or PowerPC, the proof point is what you run today. And it is really neat when you can go and say, "Here is actually a really high-performance processor and, yeah, we have built PowerPC, small PowerPC stuff, but it is really easy for us and it is not untenable for us to take production-grade working v8 and obviously build an embedded core from it." So it's a different product, so that's what changed.
Matt Ramsey - Canaccord Genuity, Analyst
That's really helpful. Thank you very much for taking the last question. I appreciate it.
Chris Rolland - FBR Capital Markets, Analyst
Just a quick follow-up here. Doug, back to the non-GAAP cash charges, can you remind us of remaining payments to Veloce? Is it about $17 million left and what is the timeframe there and minimum cash payments? And then, also, the $7 million this quarter, why did you guys decide on all cash versus stock there?
Doug Ahrens - Applied Micro Circuts Corporation, VP & CFO
Sure. So, we have about $18 million in total merger consideration left to go for Veloce, and the mix of cash and stock is at our discretion. And we factor in a variety of things, retention, cash versus stock, we factor in taxes, liquidity, lots of things. So, it's a moving thing.
This particular one last quarter was a singular choice to do all cash and has no reflection on what to expect going forward. It remains at our discretion on how to pay that out.
Chris Rolland - FBR Capital Markets, Analyst
Okay, great. Maybe one last one, if I could squeeze it in. So, lead times for X-Gene, are they going to be similar to your manufacturer stated lead times? And then, also, will they ever be available through distis for sale or is it all going to be direct?
Paramesh Gopi - Applied Micro Circuits Corporation, President & CEO
Lead times, similar. The latter, it depends on customers and who is buying.
Chris Rolland - FBR Capital Markets, Analyst
Thanks, guys.
Operator
Ladies and gentlemen, that concludes our question-and-answer session. I will now turn the call back to Traci Tsuchiguchi for closing remarks.
Traci Tsuchiguchi- Vice President of Investor Relations
Thank you all for your time this afternoon and we look forward to seeing you all at ARM TechCon this fall.
Operator
Ladies and gentlemen, that concludes today's conference. Thank you for your participation. You may now disconnect and have a great day.