Mellanox Technologies, Ltd. (NASDAQ:MLNX)
UBS Global Technology Conference Call
November 14, 2012 10:30 am ET
Jacob Shulman – Chief Financial Officer
Steven M. Milunovich – UBS Securities LLC
Steven M. Milunovich – UBS Securities LLC
Good morning. Why don’t we get started? I’m Steve Milunovich, UBS IT hardware analyst, and I’m joined by my colleague, John Roy, who’s our storage specialist. And we’re very excited to have the management of Mellanox with us. Jacob Shulman will be doing the presentation. He served as Mellanox’s CFO since November, so very recently. Previously he was the VP of Finance from March 2012 to October 2012, and also Mellanox’s Corporate Controller from June of 2007 to March of 2012. Prior to joining Mellanox, he served as Corporate Controller of Atrica, a carrier Ethernet company which was acquired by Nokia Siemens. Also with us today is Michael Gray, the former CFO.
So Jacob’s going to give us a 10- or 15-minute presentation on the company, and then we’ll have a little discussion. And please think up some questions as well. Jacob?
Good morning, guys. It’s been 10 days since I officially became CFO, so I’m still enjoying it. Before we start, please examine our Safe Harbor Statement. I think it’s pretty straight forward, and I’m sure you’re all familiar with it.
So what is Mellanox? Mellanox is a fabulous semiconductor company. We manufacture, design, and sell high performance interconnect products that help to facilitate data transmission between servers, storage systems, and embedded solutions. We target today various markets: HPC high performance computing, Web 2.0, storage, enterprise data center, cloud, and financial services.
We’re dually headquartered. We were founded back in 1999 in Israel. And we’re dually headquartered, all of our R&D and operations guys located in Israel, and all of our business, sales, and marketing guys all in Sunnyvale, California.
We have approximately 1,100 employees. We’re dually listed. We traded on the NASDAQ and the Tel Aviv Stock Exchange. And Q3 was record quarter for us. Our revenue year-over-year grew 130%. Our operating income was approximately 39% of the revenue.
So what do we do? Each compute and storage system consists of three main blocks: CPU, memory, and interconnect. Over time CPU vendors and memory vendors introduced more and more powerful products. And if the pipe between memory and CPU is not wide enough and the interconnect cannot feed data to CPU, then the CPU would be idle and the system would be inefficient. So what we enhance efficiency of data centers and clusters, and we’re doing that using two main protocols: InfiniBand and Ethernet.
So when we first introduced our product, we offered 10-gigabit-per-second product. Then the market progressed to 20, 40. And last year we introduced our latest generation InfiniBand product which offer 56-gigabit-per-second bandwidth. The name of the game is bandwidth and latency. We are the only game in town for 56-gigabit InfiniBand products. We also sell Ethernet products, and we provide bandwidth of 10 and 40-gigabits-per-second in Ethernet.
HPCs have historically been our predominant market and historically Mellanox has been viewed as a HPC player. But things have certainly changed over the past few years. Today we target additional markets as Web 2.0, database, enterprise, cloud, and storage. There are multiple examples of how our products are used in those markets.
In HPC, there is a list of top 500 computers published every six months. The latest list was published yesterday, and out of top 500 supercomputers in the world, 224 computers connected via InfiniBand. InfiniBand is the most used interconnect in HPC market today, with 45% attach rate.
Purdue University is a great example of how our systems are used in HPC. They were able to build a system between 187-teraflops with only 648 servers. That’s a double performance; half of the server come from prior cluster.
In Web 2.0 we provide unmatched ROI. Great example is Microsoft Bing Maps. When Microsoft was deciding on their solution for the Bing Maps, they had two alternatives, 10GigE solution, 10-Gig Ethernet, or Mellanox’s 40-gigabit InfiniBand. Eventually they went and decided on Mellanox’s solution. So they were able to increase by a factor of 10 performance from prior cluster, and they were able to achieve that with a cost cut, comparing to 10GigE solution.
In the enterprise database market, Oracle has put us on the map. Oracle uses our InfiniBand in their Exa family products, Exadata, Exalogic, Exalytics, all around our 40-gigabit InfiniBand as a backbone. So Oracle customers able to get 10x performance and sometime achieve a reduction of their CapEx by a factor of five.
Storage is a big market for us. It’s a greenfield for us. We just recently announced several design wins in the storage market. With EMC, we have two public design wins. It’s a green plan and standard. We also have design wins with IBM XIV. Oracle says we truly believe that InfiniBand has the potential to replace Fibre Channel in the storage market. If you look at the trendsetters today, HPC, Web 2.0, the cloud guys, none of them use Fibre Channel in their storage applications. Also additional reason: Fibre Channel highest bandwidth today is 16-gigabit-per-second. InfiniBand provides today 56 and going to 100 in the future.
We play in a big market and we expect this market to grow. The way we assess or calculate our total available market is by the number of endpoints sold annually. So we believe that 13.2 million endpoints sold on annual basis with 8.6 million being in servers, 4 million endpoints is in storage, and 600K in embedded.
We apply our average selling price per endpoint of $400. That’s how we derive $5.3 billion available market today. If we take our midpoint for our Q4 guidance, we expect to achieve revenue of $526 million in 2012. So our penetration in the market is high single digits, maybe 10%. And this market is growing. We believe that in 2014 there will be 15.1 million endpoints sold, and the market would grow to $6 billion.
Who do we sell to? We primarily sell to big OEM customers, big server, storage, and embedded OEMs. HP has historically been our number one customer, followed by IBM, Dell. Oracle has an ownership position in the company. They’re slightly below 10%, and our revenue from Oracle is approximately 5% on a quarterly basis. 10% of our revenue comes from distribution, and we also work with multiple software partners to enable compatibility with different operating systems.
We provide full end-to-end solution. We sell ICs, silicon. We sell adapter cards. We sell switches and gateways. We sell cables. And our gross margin difference between these product categories, we garnish highest margins from ICs, followed by boards, followed by switches and cables. So for us, it’s our gross margin on a quarterly basis is a matter of mix of the products. The more ICs we sell, the higher margins; the more cables we sell, the lower margins.
In terms of the roadmap, today we’re the only game in town for 56-gigabit-per-second bandwidth. Our closest competition on InfiniBand side is Intel. They have 40-gigabit-per second. And we currently working on the new generation of 100-gigabit-per second, and we plan to introduce that in the first half of 2014.
Few words on the financials, Q3 was exceptional quarter for us, was a record quarter in all parameters. Our revenue in Q3 was $156.5 million. That’s 130% growth year-over-year. Our gross margins on the non-GAAP basis was 70.5%. And when we measure our non-GAAP result, we exclude stock-based compensation as well as amortization of intangible assets acquired. Our operating margins were close to 39%. We had $1.37 EPS and generated $74.4 million in cash flow.
People today look for growth stories, and Mellanox is one of them. Our CAGR over the first five years is 44%. Our 2011 annual revenue was $259 million. And assuming midpoint of the Q4 guidance, we expect to achieve $526 million this year.
On a quarterly basis, 2012 was impacted by significant large unique deals that came our way. In Q1, we said that our large and unique deals were $10 million. In Q2, there was a pent-up demand of $30 million associated with Romley introduction. In Q3, large and unique deals were $25 million.
So if you look at the baseline of the business excluding those large deals, our Q1 baseline was $80 million, growing to $100 million in Q2, to $130 million in Q3. And when we guided Q4, we said we don’t expect any large deals and unique deals in the Q4. So the baseline in Q4 growing from $138 million to the level of $145 million, $150 million.
Today we’re very diversified company. Back in 2009, 2008, most of our revenues came from sales of silicon and boards. In 2011, we acquired a customer and a competitor, Voltaire. And since then we provide full end-to-end solution. Switch system is a significant portion of our business today, being 33% to 35% of our revenues. Boards account for 31% of our revenues, silicon approximately 18%, and the rest is cables and accessories.
On the data rate, breakdown by data rate, 50% of our revenue come from newest generation FDR products, 56-gigabit products. People many time ask us how long of a run we could have in 56 because Romley production is slowing. So if we compare it to our previous generation product, QDR 40-gigabit-per-second, there were quarters where QDR accounted for 70-plus percent of our quarterly revenue. FDR products, latest generation products ramp up faster than QDR, and we think that there is still room to grow to 70% of our revenue.
We generate healthy amounts of cash. We don’t have any debts on our balance sheet. Cash in investments today represent 55% of our total assets. We exited Q3 with $405 million in cash and investments.
On the long-term model, today we outperform our long-term model. But we think that long term for us is like two to three years. We will face competition on InfiniBand side from Intel. Today we compete on Ethernet side with the Broadcoms and Intels of the world. And as a result of competitive landscape as well as higher volumes that we expect to ship in the long term, we expect our margins to go down from 71% to the level of 65% to 69%.
This year we had a huge leverage from the model. Our revenues increased 100%. Our operating income increased 270%. Frankly, we were unable to hire enough people to keep up with the growth, and we want to catch up in the future. We continue to invest in R&D. That’s why we expect that our R&D expenses would grow from current levels to the range of 26% to 30%.
We will invest more in sales and business development. We’re expanding into additional geographies like China, Australia, Japan, and Europe. And we expect our SG&A expenses to grow to the level of 15% to 18%. That would still leave enough room for the profits. We expect our net income to be in the mid-20s, in the range of 21% to 27%.
So just to summarize, interconnect is very strategic to each computer storage system. People today realize the importance of interconnect. We play in the various markets in HPC, Web 2.0, storage, database, cloud. We believe that Web 2.0 and storage would be primarily growth drivers for us in 2013.
We’re uniquely positioned in the market. We’re today the only game in town with 56-gigabit-per-second InfiniBand products. And we’re a growth story. We target in big markets, $5.3 billion today, penetrating only 10% of the total available market.
With that, I will turn to you to the questions.
Steven Milunovich – UBS Securities LLC
Great, Jacob. Thank you so much. Maybe you could talk a little bit more about what you expect out of Intel. They look like they’re going to be getting back into InfiniBand. How do you expect them to enter the market, at what time? And if they do, how will you compete with them?
So Intel was one of the founding members of InfiniBand Trade Association back in 2000. Then they decided to abandon this protocol. Recently they’re coming back. They acquired InfiniBand assets of our closest competitor, QLogic, back in February of 2012 this year.
So what we hear is that Intel will skip their current generation. Currently, they’re one generation behind us, so their fastest product of 40-gigabit-per-second. It seems that they decided to skip 56-gigabit generation, and then will try to intersect us at 100-gigabit. What we hear is that they planning to introduce their 100-gigabit product in 2015. We’re targeting to first half of 2014. Our game plan is to stay ahead of Intel. Intel is a big company. They could put resources into development. But we have learned a lot from selling 56-gigabit-per-second generation, and we would like to execute better than Intel.
(Inaudible) think up some questions, but I’ll go ahead and go with one real quick here on Fibre Channel and storage. Certainly Fibre Channel’s been a predominant interconnect in storage for quite a while. I know you’re making some progress on the back-end for the storage guys with EMC. How do you see that playing out on the front-end? And how do you see the storage market playing out for InfiniBand in general?
So we think that InfiniBand has a great opportunity to replace Fibre Channel. As I mentioned, trendsetters today don’t use Fibre Channel in their storage applications. You don’t see Fibre Channel used in HPC, most of it over cloud. Storage market is very slow market. It takes time to get a design win in storage market. It takes years to get approved by EMC. But then it tends to be very sticky. So the replacement of Fibre Channel will not happen overnight. But we think Fibre Channel market is $2-billion market, and we see great opportunity to have penetration there.
Clearly you have a speed advantage at this point, but latency is actually, at least from what I can understand, even more important. When you look at the 13 million or so some odd endpoints that you discussed, are those latency challenged endpoints in your opinion? Or how do you think of the market if you were to kind of segment out speed and just really look at latency? Thank you.
So the name of the game is bandwidth and latency. Bandwidth is the amount of data you can use. Latency is how fast you can do that. Today we compete with competition on Ethernet side, with Broadcom and Intel. And we’ll win because of the latency. We have RDMA technology which is unique technology that offloads some of the CPU tasks on the interconnect, and therefore that allows CPU to focus more on heavy lifting. And therefore our products have low latency, and that’s how we win. We think latency is very important, and we’re very well positioned in that regard.
(Inaudible) You discussed types of applications that 13 million represents. Are these latency-challenged applications? What subset of the overall server market is your TAM, your $5.6 billion template?
Okay, so 13 million endpoints, if that’s the entire market, there are always applications that not that latency-sensitive. But we believe that, with the data growth in the future, the growth in unstructured data, the need for analysis of that data, there will be more and more latency-driven applications.
Jacob, on the balance between CPUs and how fast they’re getting and they’re continuing to get faster obviously on one side, and you got flash on the other boosting storage, I mean, how do you track what you think people need? Which, to kind of get to the point in your question, are we I/O-dependent on most of the applications you see out there? And do you really feel that flash is going to push you guys to another level here in terms of demand?
So there are multiple trends on the market today that require faster interconnect, for example, virtualization, solid state drive, etc. So using our products, for example, we could double number of virtualized servers provided by VMware. The more virtualized servers, the kind of more bottleneck for connection between those servers. Our products can resolve those. So the trends that we see today like virtualization, like solid state drive, like faster and faster CPUs, they all like software-defined networks. They all play in our hand because their need for interconnect is greater.
I think you said that the kind of quarterly baseline run rate is around $150 million. But you do have lumpy business and large deals that occasionally occur, and we saw the effect of some of that in the third quarter. Should we assume that the run rate revenue should be near $600 million on that $150 million run rate? Or does that not take into account potentially lumpy business on top of that?
So we all have to keep in mind that we have very low visibility. We only guide to one quarter because we don’t know what, the majority of our revenues come from turns. Turns mean that we have to receive bill from a customer and fulfill it in the same quarter. Q2 and Q3 were exceptional quarters for us, where turns accounted for more than 50% of our business. In Q4 we’re back to our historical trend where Q4 is more than 50% of our business should be turns.
So sometime we have visibility toward big programs. It could be six months ahead of time, but we don’t know. There are so many moving pieces. We don’t know exactly when these opportunity will come our way, in what quarter. We just own part of the data center, and when the customer decide to build in new data centers, there are so many other considerations in addition to interconnect. So it’s not always depend upon us.
Also keep in mind we sell through OEMs, and how well OEM products will be adopted in the market is also unknown to us, right? Therefore we said that $150 million is a baseline. That’s purely based on our gut feeling. We have huge number of design wins and potential in the market, but again we don’t know exactly when those design wins will convert into revenues.
Steven M. Milunovich – UBS Securities LLC
Okay. I think we better stop there. But, Jacob and Michael, thank you so much for joining us today. We appreciate it.
Copyright policy: All transcripts on this site are the copyright of Seeking Alpha. However, we view them as an important resource for bloggers and journalists, and are excited to contribute to the democratization of financial information on the Internet. (Until now investors have had to pay thousands of dollars in subscription fees for transcripts.) So our reproduction policy is as follows: You may quote up to 400 words of any transcript on the condition that you attribute the transcript to Seeking Alpha and either link to the original transcript or to www.SeekingAlpha.com. All other use is prohibited.
THE INFORMATION CONTAINED HERE IS A TEXTUAL REPRESENTATION OF THE APPLICABLE COMPANY'S CONFERENCE CALL, CONFERENCE PRESENTATION OR OTHER AUDIO PRESENTATION, AND WHILE EFFORTS ARE MADE TO PROVIDE AN ACCURATE TRANSCRIPTION, THERE MAY BE MATERIAL ERRORS, OMISSIONS, OR INACCURACIES IN THE REPORTING OF THE SUBSTANCE OF THE AUDIO PRESENTATIONS. IN NO WAY DOES SEEKING ALPHA ASSUME ANY RESPONSIBILITY FOR ANY INVESTMENT OR OTHER DECISIONS MADE BASED UPON THE INFORMATION PROVIDED ON THIS WEB SITE OR IN ANY TRANSCRIPT. USERS ARE ADVISED TO REVIEW THE APPLICABLE COMPANY'S AUDIO PRESENTATION ITSELF AND THE APPLICABLE COMPANY'S SEC FILINGS BEFORE MAKING ANY INVESTMENT OR OTHER DECISIONS.
If you have any additional questions about our online transcripts, please contact us at: firstname.lastname@example.org. Thank you!