Feb 1, 2017
Executives
Angel Atondo - Senior Mark-Comm Manager Syed Ali - CEO Art Chadwick - CFO
Analysts
Blayne Curtis - Barclays Anil Doradla - William Blair Harlan Sur - JPMorgan Josh MacArthur - Oppenheimer & Co Steve Smigie - Raymond James John Donnelly - Stifel Gary Mobley - Benchmark Chris Rolland - Susquehanna Joseph Moore - Morgan Stanley Brian Alger - ROTH Capital Partners Matt Ramsay - Canaccord Genuity
Operator
Good day, and welcome to the Cavium, Inc. Q4, 2016 Earnings Conference Call.
Today's conference is being recorded. At this time, I would like to turn the conference over to Angel Atondo, Senior Mark-Comm Manager.
Please go ahead.
Angel Atondo
Thank you. Good afternoon, everyone, and welcome to Cavium's Fourth Quarter 2016 Financial Results Conference Call.
Leading the call today are Mr. Syed Ali, President and CEO of the company; and Art Chadwick, Vice President and Chief Financial Officer.
Before we begin, I would like to remind you that various remarks that we will make on this call will constitute forward-looking statements for the purpose of the Safe Harbor provision under the Private Securities Litigation Reform Act and will be based on information currently available to us. We disclaim any obligation to update these forward-looking statements.
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 and Form 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 these risks and uncertainties that could cause actual results to differ materially from our current expectations. In addition, during this call we will discuss non-GAAP financial measures.
Reconciliation of the GAAP to non-GAAP financial data can be found in our earnings release issued earlier today, and we ask that you review it in conjunction with this call. I will now turn the call over to Syed.
Syed?
Syed Ali
Thanks, Angel, and thanks to everyone for joining us today. In brief, Cavium's fourth-quarter revenue was $226.2 million, up 56% sequentially.
Non-GAAP gross margins were 65% and non-GAAP net income was $39.8 million, or $0.56 per share. Art will discuss our Q4 financial results, along with Q1 guidance, in more detail shortly.
In Q4, sales were up across all end markets. Sales in the service provider markets were up, driven by strong wireless infrastructure sales.
Sales in the enterprise and data center markets were also up, driven by broad strength across our product lines. Sales into our broadband markets were also up sequentially.
Now, as usual, I will go into the highlights of our product traction this quarter. Overall, design win momentum remained strong for both our MIPS-based OCTEON and ARM-based OCTEON TX processor product lines.
During Q4, we began sampling our mid-range 8- to 24-core CN83XX and CN82XX families of products, which are now addressing a new range of performance points for both control and data plane applications. We have also successfully demonstrated very high packet throughput achieved on these SoCs, by leveraging our proven OCTEON packet and scaling architecture.
Q4 also saw robust shipments of the OCTEON Fusion M to our lead OEM customer, who continues to deploy significant numbers of macro and micro BTS systems in large-scale LTE rollouts in India and other geographies. In addition, we are also beginning to see initial prototype volumes for a few smaller OEM customers who will start field trials this year, with production deployments planned for 2018.
In addition, we are seeing good demand trends for our run rate OCTEON-based wireless infrastructure business from our other large wireless customer. ThunderX deployments continue to make steady progress.
Two large mega cell data centers in Asia continue to deploy production workloads on ThunderX servers, while continuing to evaluate ThunderX for additional workloads. In addition, data centers in Europe and U.S.
also continue to steadily expand the installed base of ThunderX servers, while exploring use of ThunderX for additional applications through large-scale proof-of-concept trials. In Q4 2016, Packet.net, a leading provider of bare metal cloud services, added a new ARM V8 bare metal cloud service called Type 2A, based upon a dual-socket, 96-core ThunderX server platform.
The cloud service has been launched in the EU, US, and Europe and more recently in Japan in conjunction with Softbank. We also continue to see increased interest from service providers in our workload-optimized ThunderX products in conjunction with OCTEON Fusion and platform for NFC and cloud RAN applications.
In Q4, we announced a long-term collaboration for 5G with China Unicom and entered into several other trials with major service providers in the U.S. and other geographies worldwide.
Now moving on to the XPliant switch product line, as we discussed last quarter, we have achieved full production status on the XP80 family of products. During the fourth quarter, customers announced products based upon XPliant switches and highlighted the strong differentiation they achieved due to the highly programmable architecture.
Arista Networks announced a full product line, including a 32-by-100 gig switch, a 48-by-25 gig switch with 100-gigabit uplink, as well as a 48-port, 10-gig-based D switch with 100-gig uplinks. Arista has programmed a switch to develop custom profiles based upon their customer market needs.
Most notably, Arista has invented AlgoMatch technology by taking advantage of the programmable capabilities of the XPliant silicon. In addition, another customer, Brocade, has announced a family of switching products as well, including a 32-by-100 gig switch and a 45-port 25-gig switch with 100-gigabit uplinks.
We continue to develop additional silicon and software products in the XPliant family and we are now opening the programmable technology to a wider group of customers. We expect additional customers to announce products throughout 2017.
We also have won a few designs in our traditional embedded networking and communications market for switch fabric applications in products such as transport routers, telemetry, and other applications. Moving on to LiquidIO II, where product revenues continue to ramp very nicely.
Last quarter, we closed additional design wins at another major hyperscale cloud company and also at another OEM. We also have a strong pipeline of engagements at several other customers in the hyperscale and telco cloud provider markets.
I would now like to provide an update on the QLogic acquisition. The December quarter was our first full quarter of operation with the acquisition, and the integration is gone exceptionally well.
I would like to take this opportunity to congratulate the Cavium and QLogic teams that did a great job driving the integration. We are at full productivity from the acquired businesses, which are now fully integrated into Cavium.
These product teams are continuing their focus on market and technology leadership, serving the connectivity needs of top enterprise and data center customers worldwide with both ethernet and fiber channel technologies. Since the close of the acquisition in August 2016, we have seen an acceleration in design win rates across the product lines, as customers have become significantly more confident about the Company and its ability to drive strong competitive product roadmaps at very competitive prices.
We are increasingly being seen as a key strategic partner for connectivity solutions by our customers. In the December quarter, the QLogic storage products, which include both market-leading fiber channel and ethernet target connectivity solutions, have seven new design wins and completed qualification on 11 programs that are starting to go to production.
The new design wins are expected to start contributing in the second half of 2017 and beyond. We continue to see strong new design activities for connectivity solutions for storage systems in general and all Flash area storage in particular.
Our QLogic Ethernet FastLinQ products have continued to gain market traction in Q4. We won several new designs at major server OEMs, including our first large U.S.
cloud customer, and have engagements ongoing at multiple other leading OEMs. We have also significantly increased our engagements in the hyperscale, cloud, telco cloud, private and hybrid cloud, and networking OEMs, targeting the entire application spectrum from LANs to L2 basic NICs up to the feature-rich converged NICs at 10-gig, 25-gig, and 100-gig speeds.
Our recent HBE and gigabyte announcements are symbols of the traction that we are experiencing in the market. Moving on to our security products, we have shipped LiquidSecurity production units to our lead customer and are in the last stages for completing quals at two other cloud customers.
In addition, we have a strong pipeline of customers doing evals. We expect revenues for this product line to start ramping this quarter and over 2017.
Our NITROX products continue to win designs in our traditional application controller and security gateway segments and, in addition, at cloud customers. Overall, we continue to see strong need for encryption across all our cloud data centers and OEM customers.
I would now like to update you on another recent development. In Q3 2016, a developer of an ARM service CPU decided to exit the market, as it did not fit their product portfolio.
The IP portfolio includes core IP for high-end and high-performance compute applications. We went ahead and have acquired the IP assets.
The acquisition was not financially material and will be EPS neutral in 2017. We see this as a strong addition to our current extensive portfolio of IP.
Over the coming months, we will update the market on our plans for this IP asset. Now I would like to move on and give a brief outlook on the market environment that we are seeing for Q1.
In Q1, we see a market environment similar to Q4, with stable demand trends across our product lines. We see flattish trends in the enterprise, along with incrementally higher demand in the data center markets, with some incrementally lower demand in the service provider markets, which is actually much better than seasonal trends that we have seen in the past.
On that note, I would now like to turn the call over to Art Chadwick, who will provide a detailed discussion of Q4 financial results and guidance for Q1 2017. Art?
Art Chadwick
Great. Thanks, Syed, and thanks to all of you for joining us today.
I'll first go through Q4 financial highlights and then provide guidance for the first quarter of 2017. First of all, Q4 was a truly excellent quarter on many levels.
It was a record quarter in terms of bookings, sales, non-GAAP EPS, cash flow from operations, and other metrics. Revenue in the fourth quarter was $226.2 million, up 35% sequentially and up 124% over the same quarter last year.
Revenue strength was due to continued growth in Cavium-branded products, as well as a full quarter of QLogic. Non-GAAP gross margins were 65.0%, up 100 basis points from Q3.
Non-GAAP operating expenses were $98.1 million, up from $75.3 million in Q3, due primarily to a full quarter of QLogic expenses. Non-GAAP operating income was $48.9 million, up 51% sequentially from $32.4 million in Q3.
Operating margins were 21.6%, up 240 basis points from Q3. Net non-GAAP interest expense was $7.8 million, up from $3.7 million in Q3, due to a full quarter of debt-related interest expense.
Non-GAAP income tax expense was $1.4 million. The GAAP net loss was $121.6 million, or $1.82 per share, impacted by acquisition-related charges.
But non-GAAP net income was $39.8 million, or $0.56 per share, up 42% sequentially from $28.0 million, or $0.43 per share, in Q3. For the detailed reconciliation between our GAAP and non-GAAP results, I would like to refer you to the financial press release we issued today.
But, in summary, our non-GAAP results exclude $161.4 million in non-GAAP adjustments, comprised of an $83.2 million one-time, non-cash, acquisition-related tax adjustment; $30.1 million in amortization of acquisition-related intangible assets; $21.0 million in acquisition-related purchase accounting inventory expense, $19.3 million of stock-based compensation expense; and $7.8 million in other acquisition- and integration-related charges. The quarter-end AR balance was $125.7 million, down from $144.9 million in Q3.
Resulting DSOs were 51 days, which is down from recent quarters. But I'd also like to point out that our long-term DSO model of 60 days, plus or minus, remains unchanged.
Inventory was $119.7 million, essentially flat with $118.9 million in Q3. In Q4, we generated $76.0 million in positive cash flow from operations.
During the quarter, we paid off the full $50 million acquisition-related bridge loan. We then ended the quarter with $221.4 million in cash, up from $192.4 million at the end of Q3.
As Syed mentioned in his commentary, integration of QLogic has gone exceptionally well in regards to integrating the organizations, eliminating redundant expenses, developing new product roadmaps, and new customer design wins. In addition, we continue to monetize acquired assets.
And in December, we completed the sale of the QLogic headquarters for $36 million. As a result of exercising manufacturing rights to certain QLogic chips, we are now purchasing wafers directly from the fab, which will have a positive impact on gross margins beginning in Q1.
I would now like to provide more specific guidance for the first quarter of 2017. Beginning with the balance sheet, I'd like to report that in January, using cash generated from operations, as well as proceeds from the sale of the QLogic headquarters, we paid down an additional $86 million in debt.
Our debt balance at the end of January was $612 million, down from $750 million at the time of the QLogic acquisition. In regards to the income statement, we expect sales in Q1 will be between $225 million and $229 million.
We expect non-GAAP gross margins will expand by between 50 and 150 basis points during the quarter, which at the midpoint would be approximately 66%. We expect non-GAAP operating expenses will increase between 2% and 3% sequentially, due to beginning-of-the-year employee raises and employment taxes.
This would put Q1 expenses somewhere between $100 million and $101 million for the quarter. Interest expense is expected to be approximately $7.5 million, down from Q4, due to the lower debt balance.
Income taxes in Q1 are expected to be approximately $1.5 million, increasing to between 7% and 9% of non-GAAP income for the remaining of the year. The Q1 non-GAAP share count is expected to be approximately 72 million shares.
And based on those assumptions, we expect Q1 non-GAAP EPS will be between $0.55 and $0.57 per share. And on that note, I'd like to hand the call back to the operator for our Q&A session.
Operator
[Operator Instructions] And we'll take our first question from Blayne Curtis with Barclays. Please go ahead.
Blayne Curtis
Art, you talked about working through the QLogic inventory. When you look at the guidance of 66 gross margin, how much do you have left?
Is that now reflective of a full run rate or do you still see a tailwind of gross margins for the rest of the year?
Art Chadwick
Oh, no, we absolutely see a tailwind on gross margins for the balance of the year, both from improving gross margins on the QLogic products, as we've talked about, but also continued improvement in Cavium-branded products. So we do expect gross margins to get incrementally better throughout the course of 2017.
Syed Ali
If you remember, in our guidance we had said that -- in our prior quarter's guidance -- that QLogic roughly had about six months of inventory and that it would take us to roughly Q2 for it to start affecting it in a much more significant way. We did see some benefits in Q1, but we expect a tailwind of benefits there in the next couple of quarters, too.
Art Chadwick
Along with improved margins with Cavium-branded products.
Blayne Curtis
And then maybe, Syed, on the service provider, you had obviously the big ramp of Fusion-M. Maybe you can talk about what stage that is at and just some general perspective on the overall service provider market.
You got it down to normal seasonal. What is your view as an overall market for service provider for the rest of the year?
And do you see any tailwinds from Fusion-M above that?
Syed Ali
Yes. I think, if you take a look at the Fusion-M, it has been really two quarters, right, Q3 and Q4.
So it's in the early stages still. Additionally, our lead customer is winning other geographies, other than the larger India rollout.
So we expect to see healthy growth from that customer through the year, and also we have some other customers that are going to start coming online later part of this year to early part of next year. Overall, on the service provider market, when you look at our run rate wireless infrastructure business and combine that with the OCTEON Fusion-M, we see healthy growth growing from 2016 to 2017.
So we feel pretty good about that market segment this year with both of our large lead customers.
Operator
We'll take our next question from Anil Doradla with William Blair.
Anil Doradla
Clearly, Syed and art, there are so many moving parts from the QLogic, many of your products. And I'm sure all of us will be interested in getting more color.
But a big-picture question, as we enter to 2017, the demand environment seems to be reasonable. You've got all these new growth vectors; XPliant is clearly moving in the right direction; Thunder taking a little bit more time, but moving in the right direction.
So how should we be looking at fall of 2017 from a moving parts point of view? Is there any way you could qualitatively give some color as to how should we be looking at it from a growth point of view, new products point of view, and the core business?
Syed Ali
Overall, when you take a look at 2017, and we are right at the beginning of the year, we feel like we are in a pretty good place. When you take a look at our base business of OCTEON and NITROX, with the ramp of the OCTEON 3 and the NITROX V happening in 2017, we should see good growth there.
Also across our product lines, whether it is ThunderX, whether it's XPliant, whether it's LiquidIO II, whether it's LiquidSecurity, all of them should grow nicely year over year. And on top of that, when you throw in the QLogic business, we expect growth in that business, too.
So we feel, at this point -- always 12 months is difficult to predict. But where we are right now, that we are in a good place with many growth vectors pretty much across our product line.
Anil Doradla
Very good. And as a follow-up, you talked about the IP acquisition on the ARM front.
Clearly, the ARM server from a competition point of view, it clearly has proven that Thunder is the only game in town. So why do you think this particular vendor exited the market?
And why are we not seeing many peers within the ARM server market on the high end?
Syed Ali
At the end of the day, we are the only shipping ARM server vendor in production today. When you take a look at it four years ago, I think there were like six or seven vendors, and we are down to just one shipping right now.
So overall, each company makes their decision if it fits their product portfolio or not, and evidently this didn't. So they didn't want to do it.
But from our side, they have some good IP, especially some very good core IP, which really would add very nicely to our extensive collection of IP, and we expect that this will be a great addition to our IP portfolio for our ongoing products for this particular market.
Operator
We will take our next question from Harlan Sur with JPMorgan.
Harlan Sur
Good to see the adoption of XPliant by Arista, Brocade. And I know you've got other few white-box switch vendors as well.
Outside of Arista and Brocade, I know you guys have a number of other tier-one switchbox opportunities. I'm just curious when do these start to ramp?
And maybe more importantly as it relates to keeping the growth engine going here on XPliant, your biggest competitor is out there sampling Tomahawk 2 50-gig NRZ. Cisco is developing their 50-gig ASIC as well.
I'm just curious as to when is Cavium going to be out with XPliant 2 and ready for customer evals.
Syed Ali
So as I said in my prepared comments, we are developing products both for the higher-end spectrum beyond XPliant 1 and for the lower-end spectrum. So, you will see announcements from us in 2017.
The competitive product, by the way, is not 56 gig; it is 28-gig SerDes only. It's 256 ports of 28 gig.
So -- from our lead competitor. So, overall, as we have said before, we will have other OEMs start -- who are in development and will start rolling out products through 2017.
And as I also mentioned in my prepared calls, we have also started getting designs in the embedded market. These are in places like routers and other chassis-based equipment where there's a need for a switch fabric.
So that's a nice, new, interesting market that we are seeing penetration in. But overall, again, I think from an overall viewpoint, since we are shipping production units at 3.2 terabit, we clearly are only the second vendor beyond the market leader to actually have a product in production that is being shipped by customers.
So overall, we feel that we are in a good position here in this market.
Harlan Sur
I would agree with that. Thanks for the insights on that, Syed.
Wanted to talk about your storage business. If I look at industry data last year, I think 16-gig fiber channel ports crossed over 8 gig, so obviously they've still got some runway for upgrades in that cycle.
32-gig fiber channel is just now starting to fire. So the performance migrations still seem to be a potential growth driver here.
I think the other dynamic for you guys is end gain [ph] over fabric. Maybe if you guys could just talk about how all these dynamics is potentially going to drive growth for your HBA business in 2017.
Syed Ali
Right. So when you take a look at the storage market, there are two basic pieces.
One is obviously based upon fiber channel and the other is based upon Ethernet. And on both sides of the equation, whether it's the fiber channel products or the ethernet products, which are going into storage, we believe that we have the most comprehensive feature set on both the fiber channel and the ethernet side in our converged NICs for that application.
Regarding the overall 8 gig to 16 gig, yes, we are right at the midpoint right now, so there's plenty of room to grow there. But also in 2017, we have started to ship our first 32-gig fiber channel products also, which obviously will have a ramp in 2017, with much stronger ramps in 2018 and 2019.
So, obviously, the more number of ports that switch to a higher-speed fiber channel, obviously the ASP per port is higher. So that should be a good driver for us.
On the other side of the equation, the all-Flash array, we have a very good position in that target storage controllers for all-Flash arrays and we are seeing good revenue traction in that area. We believe that we are a strong leader in the all-Flash array market with our storage controller products.
So that's a summary of the overall market. This is our first quarter running this particular business.
And I've got to tell you this is a really nice business, great visibility; nice, long lead times; very predictable. And it's a really good business.
Operator
Okay, we got a next question from Rick Schafer with Oppenheimer and Co. Please go ahead.
Josh MacArthur
This is Josh MacArthur on behalf of Rick. Thanks for taking my question and congrats on the results.
I was hoping if you could maybe broadly explain why XPliant -- in the Arista press release, they pointed to it seems like your functionality is a big driver of the value add of their products. I was wondering could you maybe talk about how the programmability helps the total cost of ownership for their end customers and why they went with your piece?
Syed Ali
Right. So overall, when you take a look at programmability, what it gives you is it gives you several key benefits.
So, first of all, when you take a look at a switch, it can be deployed in three, four different locations inside the network. In each of these, they have different requirements for the number of lookups, for table sizes, for line rates, right?
So the way many of the customers have been handling it is having two or three different pieces of silicon essentially addressing these multiple market segments. But with the programmable architecture, you can have one piece of hardware and then, because of the programmable pipeline, the programmable table sizes, and all the programmable features we have, you can have one piece of hardware and with the software load have it such that it can go into a top of the rack, it can go into the spine, it can go into different configurations.
So that is definitely one key benefit. The other benefit is new features rollout.
So when a new feature needs to be rolled out, typically customers have to wait for a new piece or a new silicon spin from the existing vendors to add that feature. So, for example, Geneva is a brand-new feature right now and we are able to support it on our current silicon and not have to wait for a next generation to be able to support it.
Actually, when VX LAN was introduced a few years ago, this was a very major problem for customers where they couldn't support VX LAN features because the silicon couldn't support it. So having that upgradability, infield upgradability, and being able to continuously upgrade features on almost a quarter-over-quarter basis is also one of the big benefits of the programmable architecture.
Josh MacArthur
Thank you, that's very helpful. And then, I was hoping you could maybe provide an update.
I believe you had a LiquidIO customer that was evaluating your 25-gig offering sometime in the first half. Is there any update on the timing of that and whether it's won in the 25 gig?
Syed Ali
Yes. So essentially when you take a look at it, we do have multiple customers for LiquidIO -- for the LiquidIO II, excuse me, at both the 10-gig and the 25-gig level now.
And all of them have product to evaluate.
Operator
We'll take our next question from Steve Smigie with Raymond James.
Steve Smigie
I was just hoping, as a follow-up to a previous question, would you be able to rank order which of the new products from classic Cavium you'd expect to see the greatest growth this year, if you just sort of put them in order of the opportunity for this year?
Syed Ali
Yes. I think, for example, if you had asked me OCTEON Fusion-M the middle of -- in the first half of the year to rank order it, I would have miserably failed.
But I think the key takeaway here is that every one of the new products is going to grow and grow meaningfully to add revenue for us in 2017. It's difficult to rank order it, but we see very strong growth pretty much across all our new product lines.
And on top of that, the OCTEON III ramp is also adding some nice growth in 2017.
Steve Smigie
Okay, great. Thanks.
And just as a follow-up on XPliant, you've already given some decent color there. But now that we are further in, you've got a couple wins publicly announced, how has that changed the dynamic of the conversations you are having with other potential customers?
Has that accelerated it or were things already so far in the process that it doesn't really change much?
Syed Ali
That's a very good question, actually. Until a customer announces that he's shipping product, people always assume the worst.
Oh, the chip doesn't work, it's not competitive, it doesn't do this, it doesn't do that. There's a lot of doubts, right?
So once a major customer releases products and starts shipping into the market, that is a big stamp of approval that, A, the technology works; number two, the technology has a value proposition; and, number three, that there is an end market for this functionality. So we think -- and we have, frankly, heard back from some of the other guys developing this.
And obviously, this was a big sign of approval for them that they made the right choice in developing systems with us.
Operator
We'll take our next question from Kevin Cassidy with Stifel.
John Donnelly
This is John Donnelly on for Kevin. Thanks for taking my question.
I was just wanting to know, on the NIC side, are you guys seeing momentum in terms of the move to the 25-gig market, and any changes you may see in the competitive environment there?
Syed Ali
In the ethernet market, in the NIC market, still the single biggest volume is 10 gig. And we expect it to actually probably peak next year, not even this year.
And the 25 gig is obviously the fastest-growing element in the entire NIC spectrum, whether it is for the two larger products or whether it is for the LiquidIO products. But the percentages of the overall market still will be small in 2017, but significantly accelerate in 2018 and 2019.
So we have a very good position on the 10 gig. We have a very good position on the 25 gig, especially with us taking over the manufacturing rights and becoming significantly more cost competitive.
And also, we are starting to see the first shipments into the 100-gig ethernet market. So we feel very good about this market and our market position in there.
As you know, at the 25-gig level the main incumbent doesn't really have a product, still, which is Intel. We expect them to come into the market at some point in the future.
But at this point, there is a very limited number of companies, two, maybe three, from where you have to pick the choices. And as I said, the functionality that we have, the performance that we have, and the feature set and software that we have is absolutely world class.
So we feel that this will be a very nice growth market for us over the next few years.
John Donnelly
Great, that's very helpful. And congrats on the results.
Operator
We'll take our next question from Gary Mobley with Benchmark.
Gary Mobley
Art, if I'm not mistaken, the fourth-quarter revenue guide that you laid out assumed about 8% quarter-over-quarter growth for the Cavium-branded products, and I think that translates to 25% year-over-year growth. Did we track to that in the quarter?
And as I look out into 2017, I realize that you integrated the QLogic business and the lines are blurred and whatnot. But as I look at consensus, isn't the organic growth assumption about 20%?
And could you give us a sense of how that growth can be divided between new products and, say, OCTEON III?
Art Chadwick
So let's start with the 2017 question. We don't get that granular as to what portion of people's expected growth comes from the various products.
But I will say this. On the Cavium-branded products, we think probably two-thirds, give or take, of the growth year over year are based on our new products, right, our Thunder, XPliant, LiquidIO II, LiquidSecurity, and Fusion-M.
And about one-third of the growth comes from the OCTEON and NITROX business as OCTEON III continues to ramp very nicely. And then, we are also, as I had mentioned earlier, expecting growth in the QLogic products.
We are doing a lot of things there, both with our cost reductions that allow us to be a little more aggressive on pricing and drive some growth there, as we've talked about in the past. So that's what we are looking for in 2017.
And you are right. We didn't break out the mix this quarter.
But things were pretty much where we expected them to be.
Syed Ali
Yes. When we acquired QLogic, I think we gave pretty good guidance on what we expect the growth of QLogic to be for 2017.
And we are pretty much coming at or maybe even above that particular rate. So we feel pretty good about that business.
Gary Mobley
Okay. I know in the past you've broken out the percentage of revenue contribution from some of the new products.
Can you share with us what it was in the fourth quarter and perhaps what it might be in the first quarter?
Syed Ali
Yes. We are not going to get into that kind of granular detail.
We used to do that primarily because the OCTEON ramp was a little bit behind our schedule, so we wanted to give an idea of new products. Now, each one of these new products is adding a reasonable amount of material revenue to it.
So, overall, it's a good mix. And as Art pointed out, in terms of 2017 growth, what the breakout expectations of that, and you can backfill what the numbers look like.
Gary Mobley
Alright, thank you guys.
Operator
We'll take our next question from Chris Rolland with Susquehanna.
Chris Rolland
Congrats on a nice quarter and excuse me if any of my questions have been asked. My first question is an update on LiquidSecurity.
You guys used to mention the strong value prop for that product versus the hardware-only solution that was offered by competitors today. Just wondering how that market is materializing for you guys.
Syed Ali
Yes. So as we said, we shipped our first production units to our lead customer, which is one of the large U.S.
mega-scale data centers. And we are currently in the very late stages for two others.
So it's a pretty good stamp of approval from, obviously, the biggest guy and from the other two guys who are getting close to finishing up their evaluation and going to production. The value proposition at all these customers has been verified.
But one of the things is that this particular product, obviously it's a complete product. It has all the software.
And for some of the customers, we do some customized features for their end application. So it takes maybe a quarter longer than we would have liked, but the overall direction is right on track.
The overall value proposition is being established pretty nicely. Also, we are seeing some good traction in the financial market segment for this particular product.
So recently, we added some sales efforts into the large financials because this product also should have a very good fit in that segment of the market.
Chris Rolland
Great, it sounds like good progress there. So you guys can become a beneficiary from NFV.
But it can also potentially be a net negative for you guys. And Intel has been a little bit more loud about x86 NFV and their opportunity there.
I was wondering if you could talk about maybe the puts and the takes in that market and whether there's a situation where maybe x86 wins in the early innings of NFV and then ARM takes over after that. And then, I knew you guys have talked about NFV in the past at Mobile World Congress.
Maybe you guys can tease Mobile World Congress a little bit, if you guys have any updates on NFV there.
Syed Ali
Yes. So regarding the NFV market, again, in terms of having competitive solutions, there is just two.
There is Intel, obviously, with the standard x86. And our ThunderX is getting some good traction.
I think you will hear more about it at Mobile World Congress, so that's a preview for you. So, overall, when you take a look at the market, there is a small portion of the market that is moving from appliances to the cloud.
And for that, we have ThunderX, which is able to address that in a very, very definitive way. Actually, in certain applications for NFV and cloud RAN, because of all our packet processing hardware acceleration and all the bells and whistles we had from the OCTEON days, our solution is actually a very, very competitive solution.
So, essentially, we are on both sides of the table. And if there's any movement from one side to the other, we have a solution on that other side, too.
The other thing that we are starting to see a lot of traction is LiquidIO. So when you take a look at LiquidIO II today sitting inside a server, NFV has a lot of networking, security, and packet processing applications.
So, actually, the most competitive solution is actually having -- even within x86 is to have a LiquidIO II to offload all of that stuff. So that combination gives us also a very significant value proposition, and, of course, we have the Thunder aspect of the equation, which has all of these features built in.
So the way to look at LiquidIO II as a NIC inside the NFV server, it's almost like having an appliance -- you normally have appliances that will sit in front of, say, 50 servers, 100 servers. Now by putting LiquidIO inside each server, you have an OCTEON-based, hardware-based appliance at every node.
And that is what is gaining a lot of traction with customers.
Chris Rolland
I like that last insight. Thanks, Syed.
And congrats again on a good quarter.
Operator
Will take our next question from Joseph Moore with Morgan Stanley.
Joseph Moore
I wanted to ask a bigger-picture question about the China ARM server market. You guys have obviously done well in that region in the early stage.
There are Chinese-developed and JV-developed products that are there that'll -- it looks like to me they are trying to replace more of a traditional server infrastructure with local content versus you guys having a more differentiated approach. But how are you thinking about competition from Chinese incumbents?
And is there opportunity for you? And are there threats from that development?
Syed Ali
When you take a look at servers, the majority of the market, I would say probably 70% to 80% of the market, is dual-socket servers. So in terms of x86, obviously they have dual socket.
But in terms of ARM, we are the only vendor who has dual socket. So China will continue to be a strong market for us.
Obviously, we will have competition from the other two vendors. One is a local Chinese vendor and obviously Qualcomm has announced a product.
But both are single-socket products only. So, essentially, we still will have a very nice market position.
And we see that as probably one of the largest growth markets for ARM servers looking through in 2017 and 2018.
Joseph Moore
Okay, great. That's helpful.
And then, just a shorter-term question. LiquidIO I, it seemed like it bounced back a little bit after declining the early part of last year.
Is LiquidIO I going to be stable or up or down from here as LiquidIO II ramps?
Syed Ali
Actually, Joe, it has actually bounced back more than a little more. So it's doing better than what we even thought a few months back.
But overall I think from a modeling viewpoint, I think the way to look at it is from our lead customer, a relatively stable business in 2017. Obviously on a quarter-to-quarter basis, things can move up and down.
But the other important aspect is, irrespective of that, the second customer that we have in this particular segment for LiquidIO I is ramping pretty nicely, also. Obviously, they don't have a hockey stick ramp.
It is a much more kind of linear ramp there. But when you take a look at the combination, we feel pretty good about LiquidIO I business for 2017, but on top of that, LiquidIO II seems to be ramping a bit faster than our expectations in terms of revenue.
So, our combined LiquidIO business should be one of the nice growth drivers for us in 2017.
Joseph Moore
Thank you very much.
Operator
We will take our next question from Brian Alger with ROTH Capital Partners.
Brian Alger
Most of my questions have been asked and answered. Congrats, as I said.
Looking a bit more bigger picture, there seems to be a shift at least on the political landscape with regards to net neutrality. And as the implications from that work through the system, when might -- I presume it hasn't affected things yet.
But when might we see effects of that starting to work into your demand fall?
Syed Ali
That's a difficult one to answer because it really needs to figure out what happens there, number one. And based upon that, we will have to take a look and see what the impact is.
But just at a high-level takeaway, we don't think anything will be done that is going to seriously damage the U.S. business environment for networking.
So, there may be some puts and takes associated with it, or how costs are allocated between the service provider and the user; that may change. But overall, the key takeaway is the amount of data that is being consumed on a year-to-year basis is growing just at a phenomenal clip.
And whatever the rules and regulations are, people are going to consume this bandwidth. And if they consume this bandwidth, the markets will grow.
Brian Alger
Okay. I guess I perceive it, and maybe you could educate me a little bit here, I perceive it that if the network providers have the ability to slice and dice their bandwidth into -- charge for different levels of service, they are going to have to have a higher level of expertise, if you will, within their network and to be able to process those packets more efficiently.
Am I wrong in that understanding? Or is that what they just deal with every day, anyway?
Syed Ali
Obviously, the more complex you make parsing the traffic, prioritizing the traffic, the more work you have to do on every packet, that will be good for companies like ours. If you really need to look at every packet and say where is this coming from, what is this priority, where do I need to send it, at what quality am I to send it, that's a lot of work.
So whenever something happens in the market which requires the equipment to touch the packet more times than before is good for vendors like us.
Brian Alger
And then as we look, and I guess as a lead up to MWC in a couple weeks here, it seems that everybody is embracing that concept as they position and they run their tests and formulate for the next evolution of 5G. Shouldn't that drive a step function in terms of the value assigned to the deep packet and the high throughput that you guys provide?
Syed Ali
Yes. So one of the things I think really starting to talk about much more, we believe that 5G, whether it is for the wireless side of the equation, like telephones, or for wireless broadband, we believe that we will be one of the major vendors for that market.
And as you know, the competitive landscape in that is a grand total of maybe 2-1/2 people. So we feel very good about that market and what our position will be.
We think we will be more competitive in 5G than we were even at the 4G level. And in 4G, we were more competitive than 3G.
So overall, we feel very good about it. And at MWC, we will definitely be talking more about these new trends.
Operator
We'll take our next question from Matt Ramsay with Canaccord Genuity. Please go ahead.
Matt Ramsay
Syed, I was interested to see that there was a ThunderX II design win announcement with a nexus-scale project in Europe. And it may be a bit earlier than I had anticipated, given where you guys are with that product line and when things should lead to real revenue.
So maybe you could talk about that engagement and how they were able to interact with the project on ThunderX II and where you guys are along the sampling of the product and getting it out to customers and the ramp towards revenue on that second-generation product.
Syed Ali
Generally, that end customer already has deployed small volumes of our current ThunderX already. So they have a pretty good idea about the ThunderX architecture, the software, and everything else that goes with it.
So they are not coming in from the cold, number one. Number two, in this market, typically decisions are made almost like one or two years before because the gestation times are much longer.
So that's where that stands. They already have ThunderX servers.
They have a pretty good idea and they know what our oncoming, ongoing roadmap will be. So they feel pretty good about it.
Overall, this year we expect to release our next-generation products. And we will have a few SKUs on that and we will update you as we get closer to actually bringing out that product across the market application space.
Matt Ramsay
Got it. Thank you.
And it's just a follow-up to that. It has been my observation that one of the things that has been a headwind to the ARM server movement is just not necessarily having product on the end process node where Intel has been living.
You guys are obviously moving to FinFET and then have a roadmap beyond that. Qualcomm chose to go to 10 to kick off their program and some other folks are moving quickly.
So, I would be interested to see your long-term -- or to the extent that you can comment, how you guys plan to move down the node stack and stay on that leading edge on a go-forward basis.
Syed Ali
So far, more than what node it is, the software ecosystem readiness has probably had a bigger effect on the ramp of ARM servers than being on any particular node because, again, in servers, you have a lot of different performance points. Obviously, if you are trying to hit the highest performance points, then the node would be important.
But when you are going to the mid-range and lower end, there's still a healthy market, the node is not the most significant aspect of the equation. So the ecosystem expands nicely and we are seeing the ecosystem expand nicely.
It's starting to mature, and that is definitely going to drive adoption in this particular market. Having said that, we will probably be a little bit more aggressive as we move forward.
And as our products get more into the higher end, we will probably be more aggressive on the node side of the equation, too.
Matt Ramsay
Thank you, congrats on strong results.
Operator
It appears we have no further questions at this time.
Syed Ali
Okay, great in that case we're going to conclude this call, thank you everybody for joining us today.
Operator
And the audio replay will be available at 719-457-0820, with passcode 1944877. This does conclude today's program.
Thank you for your participation. You may now disconnect and have a great day.