Good day, ladies and gentlemen, welcome to the third quarter 2019 Marvell Technology Group Ltd. earnings conference call. At this time, all participants are in listen only mode. Later, we will conduct a question and answer session, and instructions will be given at that time. If anyone should require assistance during the program, please press star then zero on your touchtone telephone. As a reminder, today's program may be recorded. Now I'd like to introduce your host for today's program, Ashish Saran, Vice President of Investor Relations. Please go ahead.
Thank you. Good afternoon, everyone. Welcome to Marvell third quarter fiscal year 2019 earnings call. Joining me today are Marvell President and CEO, Matt Murphy, and Marvell CFO, Jean Hu. Before I turn the call over to Matt, I want to remind everyone that certain comments today may include forward-looking statements which are subject to significant risks and uncertainties and which could cause our actual results to differ materially from management's current expectations. Please review the cautionary statements and risk factors contained in our earnings press release, which we filed with the SEC today and posted on our website, as well as our most recent 10-K and 10-Q filings. We do not intend to update our forward-looking statements. During our call today, we will refer to certain non-GAAP financial measures. A reconciliation between our GAAP and non-GAAP financial measures is available on our website in the investor relations section.
With that, let me turn the call over to Marvell's President and CEO, Matt Murphy.
Great. Thank you, Ashish. Good afternoon to everyone joining us on the call. The third quarter of fiscal 2019 was the first full quarter that Marvell and Cavium operated as a combined company. We delivered solid results. Working together, we completed several key integration milestones ahead of schedule, including establishing common operating practices, converging our technology, integrating product roadmaps. During the quarter, we also hosted our investor day in New York City, which many of you attended on the call, where we shared our strategy and growth plans for the combined company as we continue our expansion into the infrastructure market. The breadth and depth of our combined portfolios will drive long-term growth in a broad number of areas, including 5G, Ethernet networking, enterprise and data center storage, Arm servers, cloud security, AI, and automotive.
During this call, we will provide you with an update on the progress we are making on a number of these growth initiatives and on our continued drive towards operational excellence. I'd like to cover the results of our third fiscal quarter. Our GAAP revenue was $851 million. GAAP gross margin was 45.1%, and GAAP loss per diluted share was $0.08. Switching to our non-GAAP results. Marvell's revenue for the third quarter came in above the midpoint of our guidance at $851 million, driven by strong performance from our networking business and solid performance from our storage business, which met our expectations. As you may recall, during last quarter's earnings call, we had identified approximately $20 million of excess inventory being held at Cavium customers, which was impacting our revenue expectations for the third quarter.
This excess inventory was depleted during the third quarter, and revenue from Cavium products came in slightly above our prior estimate of approximately $210 million. In the third quarter, we continued to improve our non-GAAP gross margin, reaching 64.6%, 10 basis points above the midpoint of our guidance, and we continue to expand gross margin over the coming quarters. Marvell's non-GAAP operating margin for the third quarter was 29.7%. Non-GAAP earnings per share was $0.33, just above the midpoint of guidance. Moving to our core businesses of storage and then networking. Our storage business, which includes fiber channel products and Marvell's HDD and flash storage controller products, performed well, meeting our expectations with revenue of $407 million. Storage controller revenue increased sequentially quarter-over-quarter with a seasonal increase from all storage markets.
On a year-over-year basis, revenue from storage controller shipping into the enterprise and data center market grew by approximately 30% year-over-year, which more than offset the year-over-year decline from the PC market. Our fiber channel business slightly exceeded expectations. Looking ahead to the fourth quarter, similar to what you have heard from many of our storage customers, we are also forecasting weak demand. There are several factors hurting this market, including PC CPU shortages, trade tensions, moderating cloud CapEx, and large inventory increases at our customers. Given these headwinds, we expect storage revenue in the fourth quarter to sequentially decline by about 10%. However, we remain bullish on the long-term prospects of our storage business. During the third quarter, we won several storage controller designs that will continue to diversify our storage business by adding significant future revenue streams outside the PC market.
These wins include three data center sockets and an edge video surveillance application. We expect these wins to start ramping late next year and into fiscal 2021 and help drive growth in our storage revenue. We also sampled our new flash platform solutions, which include NVMe aggregators, accelerators, and converters with several tier 1 server and storage OEMs. Aggregators increase SSD densities to maximize throughput for new emerging SSD form factors and architectures. Accelerators and converters reduce the need for additional servers while optimizing overall data storage utilization, performance, and scalability. Early customer reaction has been extremely positive, and these products once again demonstrate that Marvell is continuing to drive innovation in the storage market. The success of these initiatives will contribute to additional growth in our storage revenue from the enterprise and data center markets. Moving on to networking.
Our networking business includes Ethernet switches and PHYs, embedded processors, Wi-Fi connectivity, security products, Ethernet connectivity, Arm server processors. Altogether, we have a very broad and diverse set of networking products. Our networking business performed above expectations, delivering $398 million in revenue, driven by solid results from our OCTEON family of high-end embedded processors and strong demand from Marvell's Ethernet switch and PHY products. Embedded processors had a good quarter, with stable demand from the enterprise market. As expected, our service provider OEMs depleted excess processor inventory they had built up over the past few quarters. In addition, during the quarter, we started shipping embedded processors into trial 5G deployments, which are a strong precursor for larger roll-outs in the next year.
As we detailed at our Investor Day, we expect 5G base stations to be a significant long-term growth driver for Marvell, as our dollar content in these deployments is substantially higher than what we currently have in 4G. We anticipate our platform solution, which includes our 5G baseband processors, embedded processors, and Ethernet switches and PHYs, to start ramping late next year, as we have already secured a significant design win for our full 5G solution with a leading base station customer. In addition, since Investor Day, our momentum has continued in 5G engagements with additional tier 1 customers. Marvell's Ethernet switch and PHY business delivered another strong performance as new products continued to ramp in the enterprise market. Revenue grew by approximately 30% year-on-year. Our refresh portfolio continues to gain momentum with customers. We won several new sockets during the quarter.
These include a switch design with a major server OEM and several PHY wins with a top networking OEM. We believe our continued progress in the enterprise, combined with the 5G opportunity in the service provider market, will fuel significant long-term growth. As expected, our Wi-Fi revenue declined due to our planned transition away from older connectivity products. We expect our Wi-Fi business to seasonally bottom out in the fourth quarter. Start growing again as we ramp our next generation of Wi-Fi, 802.11ax, which is now called Wi-Fi 6. We have the industry's most complete and fully compliant set of Wi-Fi 6 products, which are gaining strong traction with leading customers. These include 8x8, 4x4, and 2x2 SoCs targeted at the enterprise and retail access points, carrier gateways, as well as the automotive and premium home markets.
Our Arm server CPU business continued to make progress with several cloud and high-performance computing customers who are qualifying our ThunderX2 family of processors. Given the complexity of introducing a new architecture, we expect this evaluation to continue over the next few quarters. In addition, Marvell was recently awarded funding from the Los Alamos National Laboratory to develop next-generation high-performance Arm processors optimized for exascale workloads. At the recently completed Supercomputing conference, there were multiple customer and partner announcements highlighting their adoption of ThunderX2. Meanwhile, Sandia, NRL, GW4, and NERSC released a number of benchmarks which demonstrate the compelling performance of ThunderX2 over alternative server processors. Notably, Sandia's Astra supercomputer, which is based on HPE's Apollo 70 and powered by ThunderX2, became the world's first Arm-based computer to enter the top 500 supercomputer list.
Finally, last week, Amazon Web Services announced an Arm-based public cloud offering based on their processor. This is a major endorsement of Arm-based servers by another tier 1 cloud vendor and meaningfully broadens the Arm ecosystem. Millions of end users will now have the opportunity to run their workloads hosted on an infrastructure powered by Arm, and we believe that this milestone will help accelerate future Arm server deployments. We are also making strong progress in our new initiatives, including automotive, AI inference, and cloud security, with a very robust design win funnel at several key customers. In summary, our networking business delivered a solid third quarter. Looking ahead, we expect low single-digit sequential revenue growth for this business in the fourth quarter, primarily from an increase in demand from the service provider market for our embedded and baseband processors.
Turning to our other products, which are our legacy products, revenue was $46 million, and we expect this category to decline by about 10% sequentially in the fourth quarter. For the fourth quarter, at the midpoint of guidance, we expect $810 million in consolidated company revenue and $0.32 in non-GAAP earnings per share. We also expect to increase non-GAAP gross margin to 65% and lower non-GAAP operating expenses to $287.5 million at the midpoint. In closing, while we need to navigate through some short-term market headwinds in the storage business, we continue to benefit from positive trends in our other businesses. First, the Ethernet switch and PHY business has maintained a double-digit year-on-year growth rate over several quarters, and we expect this to continue into the fourth quarter.
Second, we expect growth in the Cavium business, and at the midpoint of our guidance, we are projecting this business to grow double digits sequentially to approximately $240 million of revenue in the fourth quarter. We have now fully integrated Cavium, and going forward, we will not be delineating revenue between Marvell and Cavium. Third, 5G trials have started, which we believe will turn into larger deployments that will drive significant revenue growth from existing design wins and from ongoing engagements with additional tier 1 base station OEMs. We expect to be a leading supplier of merchant silicon for 5G base stations when they ramp to production based on our end-to-end portfolio. Fourth, despite anticipating lower revenue in the fourth quarter, we still expect to increase gross margins to 65%. We also project to continue increasing gross margins next year as we complete our planned $50 million of COGS-related integration synergies.
Throughout all of this, we will continue to tightly manage OpEx and drive operating margins while continuing to invest in the long-term growth initiatives we outlined at our investor day. Lastly, I'm pleased to report that we delivered free cash flow at 30% of revenue this quarter, which is a strong result. We are putting this excess cash to good use by reducing debt and returning it to shareholders through buybacks and dividends. Altogether, these achievements represent another solid quarter for Marvell. With the foundation and momentum we are building, combined with the opportunities in front of us, I am as excited as ever about our progress and long-term potential. Let me just close by recognizing and thanking Marvell's 5,000-plus employees around the world for their hard work and contributions during the quarter.
As a result of their efforts, we continue to make great progress towards building the world's next great semiconductor company. With that, let me turn the call over to our CFO, Jean Hu, for more detail on our third quarter performance and our outlook for the fourth quarter.
Thanks, Matt, good afternoon, everyone. I'll start with a review of our financial results for the third quarter of fiscal 2019. Then provide our current outlook for the fourth quarter of fiscal 2019. Please note our GAAP results reflect purchase price accounting, amortization of intangibles, acquisition and integration related to non-recurring expenses, as well as the stock-based compensation. Revenue in the third quarter was $851 million, about the middle point of the outlook provided in September. Our core business of storage and networking accounted for 95% of revenue. Storage accounted for 48% of revenue, in line with our expectations. Networking accounted for 47% of revenue, better than expectations. Other products accounted for 5% of revenue. GAAP gross margin was 45.1%, which included $103 million for amortization of Cavium inventory step-up cost.
As a reminder, we step up Cavium inventory cost as part of our acquisition purchase price accounting. We anticipate amortizing the remaining balance of $98 million by the end of the fourth quarter of fiscal 2019. Our non-GAAP gross margin was 64.6%, a 90 basis point improvement from the prior quarter. It reflects the operating mix of the combined company. GAAP operating expenses were $404 million. Non-GAAP operating expenses were $297 million, approximately $5 million below our middle point of the guidance provided in September. This was a result of the faster than expected realization of integration synergies and our continued focus on operational excellence in managing our core operating expense. Non-GAAP operating margin was 29.7%. GAAP loss per diluted share was $0.08. The non-GAAP earnings per diluted share was $0.33, above the middle point of our guidance range.
Now turning to our balance sheet and the cash flow statement. We generated $299 million in cash flow from operating activities in the third quarter. This result include the cash impact related to Cavium merger and the restructuring activities. Capital expenditures were $13 million. The depreciation was $39 million. We also spent $33 million on purchases and the payment for technology license obligations in the third quarter. Our resulting free cash flow for the third quarter was $254 million, or 30% of revenue. In the third quarter, we paid down $75 million of our long-term debt and returned $93 million to shareholders through $54 million in share repurchases and $39 million in dividends. We expect that quarter with $610 million in cash and short-term investments, an increase over $87 million from the end of the prior quarter.
Our long-term debt was $1.8 billion. Let me now move on to our current outlook for the fourth quarter of fiscal 2019. We expect our revenue to be in the range of $790 million-$830 million. Our expected GAAP gross margin will be approximately 46%, and our non-GAAP gross margin will be approximately 65%. We expect our GAAP operating expenses to be between $375 million and $385 million, and the non-GAAP operating expenses to be in the range of $285 million-$290 million.
At the middle point of this outlook, our non-GAAP operating expense would reflect achievement of $120 million of operating expense synergies and our continued focus on managing our core operating expense. Our current planning assumptions, we project operating expense exceeding fiscal 2020 to be approximately $10 million lower on a quarterly run rate basis from the $290 million we had provided during our Investor Day. As a reminder, our operating expenses have a certain amount of seasonality and tend to increase in the Q1 of our fiscal year, driven primarily by employee payroll taxes, matching contribution, and our annual merit process. We anticipate that these factors will drive our operating expense in the Q1 of fiscal 2020 to just above $300 million. We anticipate operating expense to then reduce as we progress into the rest of fiscal 2020 and exit the year at a quarterly run rate of $280 million.
We expect our non-GAAP tax rate to be approximately 4% and net interest expense to be $20 million. We expect GAAP loss per diluted share in the range of $0.05-$0.01 and non-GAAP income per diluted share in the range of $0.30-$0.34. We're now ready to take your questions. Operator, please open the line for questions.
Certainly. Ladies and gentlemen, if you have a question at this time, please press star then one on your touchtone telephone. If your question has been answered and you'd like to remove yourself from the queue, please press the pound key. We also ask that you please limit yourself to one question and one follow-up. Our first question comes from the line of John Pitzer from Credit Suisse. Your question, please.
Good afternoon, guys. Matt, congratulations on the solid results given the difficult environment. Matt, my first question is just on the storage guidance. You're guiding the overall storage business down about 10% sequentially into the fiscal fourth quarter. Curious if you could help us give us some color of how that breaks out between fiber channel, HDD and SSD. Are they all down about the same, or are you still seeing some secular growth drivers in the SSD business? That'd be helpful.
Great. Thanks, John. To answer your question, we see all of our sub-segments of storage down, fiber channel less, and then our storage controllers, which is classic Marvell, down more to equal to 10%.
Is there a big differentiation between HDD and SSD?
Both are down, and both are very weak when you look across the range of customers. I would say directionally, both are down the entire storage business, whether it's cold storage or hot storage, is being hit.
That's helpful. As my follow-up, just on the 5G opportunity, I like the fact that you're highlighting more than just one tier 1. If memory serves me correct, Cavium had a strong position at Samsung. I'm curious, as you look to broaden out the OEM exposure, how long does that process take? Again, can you help me understand the TAM opportunity here? Our math would suggest that you've got an opportunity to capture maybe $2,500 of content per base station and potentially a market that might have 1 million base stations per year. Does that feel like the right ballpark?
Sure. I'll break it into the 2 pieces. 1 is how's the progress going with the largest lead customer that we brought over from Cavium, second on the content. On the first 1, we continue to do very well with the lead customer, we're encouraged by their progress, in the end market, in terms of them expanding their footprint, not only in Asia, but also domestically here in the U.S. What I'd say is that the solution we have is very compelling because it's a platform solution, we continue to get a lot of interest in actually all aspects of the portfolio, the Fusion-M baseband, OCTEON for processing, then our Ethernet switches in PHY.
This is something we'll keep you updated on, but just directionally, we thought it would be helpful to share that we continue to build good momentum in terms of engagements on 5G. With respect to content, you're in the right ZIP code when in the type of content that you mentioned when we're all in with our full platform solution, that's certainly something that we expect with our large customer, that would be I think in the range of where you'd say, the TAM would be for our full solution at a base station.
Great, very helpful. Thanks.
Thanks.
Thank you. Our next question comes from the line of Ross Seymore from Deutsche Bank. Your question, please.
Thanks for letting me ask a question. I want to go back to the storage side, Matt. The answer to the prior question, the details are very helpful for the given quarter, I wanted to see what your thoughts were on the duration of this weakness. I know there's a lot of moving parts from it, how do you judge whether in aggregate your storage business or the different moving parts within it, how long this weakness is going to last? Maybe even what does it mean to seasonality in your April quarter, considering you're coming off such a weak January?
Sure. The commentary that we're hearing from our customers, I think you hear the same thing, is people are saying this is a, quote, "couple of quarters" in terms of the duration. It's very hard to call, in fact, where we sit in the supply chain, we're one step removed or one step behind. If you look at end OEMs that are consuming drives that have our controllers in them. We tend to be farther back in the supply chain. We're not as close as the customers to it, certainly their outlook, we tend to agree with, we're sort of following their lead. We'll prepare, obviously, to the extent that things come back earlier, we're planning for this to be several quarters.
Great. I guess as my follow-up question was, one of the comments you had in your preamble was on Graviton and the design of an internally developed Arm processor. I can see the side that you're saying it's great for the ecosystem, I could also see the side that they just removed a potential customer by servicing themselves. How do you think about the balance between those two and your commitment to ThunderX2?
Let me answer it in two ways. The first is, as we were in the final stages of closing Cavium, there was announcements from one of the other large competitors that they were pulling back in their efforts. What I'd say is in the short term that helped us because it helped solidify our design position in a number of ways. Also there was a lingering question, there's been a lingering question, is Marvell going to be the only last man standing with respect to Arm servers? On the Graviton announcement, all I'll say is, I think we've had a view for some time that that particular customer was going to do something on their own, therefore that's never been a part of our SAM assumption.
Again, from our point of view, we're happy to see them have made progress, get an initial start in terms of putting up their instance. I meant what I said, I do think this is going to be good for the ecosystem. When I look out to 2019 and our next generation chip, which is ThunderX3, which we'll be taping out and sampling, I think we'll be very well-positioned as Marvell and as Arm to compete very effectively with the next-generation processors from the usual suspects. That's the continuum of how I see this market evolving.
Perfect. Thank you.
Thanks.
Thank you. Our next question comes from the line of Vivek Arya from Bank of America Merrill Lynch. Your question please.
Thanks for taking my question, and congratulations on the consistently good execution. Matt, my first question on storage. In the past, you had kind of sized storage growing at a low single-digit kind of growth rate. Is that still a useful construct for fiscal 2019? I know it's early days, but could you just help us right-size where the business can trend? Because I understand that there is probably some excess inventory and CPU shortages on the hard disk drive side, but are there issues on the SSD side or other areas? Just overall, how do you think about storage from a fiscal 2019 perspective?
Sure. I think even with the weakness we see, if you look at our performance, storage is still hung in there pretty well, and I think it's probably flattish if you look at the performance in 2019. Looking forward into 2020, certainly understanding there's going to be this inventory correction we have to work through with the growth drivers that we've highlighted, especially the traction we see and the continued growth in the enterprise and data center portion of our storage business. We think that as we get through the inventory correction and we head back to a normal state, we'll continue to see growth in our storage segment.
I see. For my follow-up, how is all this noise around tariffs and the potential for them to go from 10% to 25%, or maybe they don't go to 25%, every day that news changes. Do you see your customers behaving a certain way? What are they telling you? Are they taking certain steps to kind of pull in inventory? Any change in behavior that you are noticing from their side or anything else that you are noticing in terms of cancellations or other things in your business?
Sure. First of all, you're right. The situation is noisy, and it's changing daily. It's been interesting preparing for this call, seeing what's happened between Sunday and today. Yes, a lot of noise. To kind of focus on your question, we certainly hear about this in the industry, about pull-ins in anticipation of the tariffs kicking in. I'd say in our own experience, it's very limited in terms of what we can actually point to. I'd say that we saw this in China through distribution for some of our networking products. You end up kind of going pretty specific in terms of where some of this may have occurred.
We look at it's speculation, quite frankly, because we don't have customers actively coming to us telling us, "Hey, we're planning on pulling in, and here's what's happening." We do hear about it, and the noisy environment certainly isn't helping the overall situation for us as well as our peers.
Okay. Thank you.
Thank you. Our next question comes from the line. Timothy Arcuri from UBS. Your question, please.
Thanks a lot. Matt, I had a question on the networking business. You're not giving us connectivity, it seems like the guidance assumes something a little more than $200 million for sort of the core Marvell business, which is up a good bit from sort of $150 here a couple of quarters ago. Can you talk about that? Is that real demand Or is that some of what you just talked about, some of the pull forward due to the tariffs? I'm just kind of wondering whether that's a reasonable baseline to sort of think of going forward. Thanks.
Hi, Tim. This is Jean. I'll give you some color about the Wi-Fi business. It became really below 10% of overall Marvell's business now. As Matt mentioned during his prepared remarks, this is the quarter we actually see the gaming product line completely going out. When you look at it year-over-year, the business has certainly declined significantly and kind of bottomed out in Q4. I would say, going forward, especially after we get out of Q4 next year, we actually see this revenue going to start to ramp up and growing in fiscal 2020. I think the run rate that you quoted is actually lower than what our actual revenue level. We do see this business continue to be going back, not going back to the level of with the gaming revenue, but certainly all the other business momentum continue to be good.
We are actually expecting the growth of Wi-Fi business in fiscal 2020.
Okay, Jean. Thanks so much. As a follow-up, I know that you were talking about the Cavium business, the baseline being like $230 million. The guidance is now $240 million. It's a bit better. The channel sounds like it's been pretty cleared out. Is there some restocking happening there yet? I'm just sort of wondering what we should expect for the trajectory of the Cavium business now that the channel's basically cleared out. Thanks so much.
Sure, yeah. You're right to point it out. We're actually very pleased with how Cavium's performing. It came in ahead of plan for Q3. We did lean in the last call, we said $230. Now we're projecting it to be $240. Both of those were very positive, and that's really due to growth in the OEM business. All the channel stuff is clean. Channel inventory's clean, our DSOs, everything's linearity, all that's very good. This is really a resumption of growth in that business, and it's performing well. Just to be clear, the growth quarter-to-quarter as well as the increase over what we had indicated before was really the OEM customer base coming back, primarily in the OCTEON and Fusion product lines.
Got it, Matt. Thanks so much.
Yep.
Thank you. Our next question comes from the line of Blayne Curtis from Barclays. Your question, please.
Hey, good afternoon. Thanks for taking my question. Just wanted to follow up on the SSD business. In the October quarter, enterprise data center was growing nicely year-over-year. Client was down a little. I'm just kind of curious your outlook into January. The whole segment sounds like it's down a bit. Can you just give us some color between those two segments?
Yeah. Blayne, yeah, the way to think of it is, if you look at the overall classic Marvell storage market, which is both our flash solutions products as well as our HDD controllers, that whole controller business for us is down from Q3 to Q4. It's both HDD as well as flash, SSD, and it's broad-based across virtually every customer. In fact, when I was preparing for this, we took a look at all of the customers above $1 million a quarter, and every single one is down sequentially. I'd characterize it as not related specifically to one or the other, but both are down in the fourth quarter.
Yeah, I think it's the inventory adjustment, right? We're really looking at the storage segment, especially the controller side. We see the inventory adjustment. It will take a few quarters, like Matt said earlier.
Thanks. Then maybe actually the kind of same question when you look at the HDD business. This is a business that has had cycles, if you look back over history, and clearly, your end customers are struggling. I'm just kind of curious your perspective of inventory of controllers when we look at your guidance for January. Is that just reflective of what the end customer are doing, or are you working down some inventory of controllers as well?
Yeah. Again, Yeah, sorry, Blayne, just to be clear, the last answer I gave on what you asked on SSD, I'm really answering for both together. The reason is because, one, when we look at this business, what's become apparent over the last couple of years, quite frankly, is that when we break out our business, looking at it by market is much more helpful. When we talk about our growth drivers, we talk about where our investments are, that whole commentary, we've been pretty consistent. We actually gave a lot of data at our investor day around the enterprise and data center segment. When you bring it back to inventory, as I mentioned, we see inventory issues both on the SSD side as well as the HDD side.
Both sides, from a storage perspective, have inventory, are going to be down in Q4, both are going to take some quarters to rebalance.
Got you. Helpful. Thank you.
Thank you. Our next question comes from the line of Mark Delaney from Goldman Sachs. Your question, please.
Yes, good afternoon. Thanks for taking the question. Question on the gross margin guidance for next quarter, which is coming in pretty nicely given the lower revenue. Can you give us a sense to what extent that's starting to benefit from COGS synergies related to Cavium? Or is it more about mix? And Jean, maybe you can remind us of the $50 million of annualized COGS synergies that you're expecting to realize. What sort of cadence should we have in mind for that to be achieved?
We are really pleased with our guidance over 65% gross margin. It actually has not included the $50 million cost of sales synergies that we outlined during our last earnings call. There are a few key drivers that I want to talk about. The first one is the mix of Cavium product, which when we include Cavium as a combined company, the mix improvement certainly coming from the Cavium product side. Secondly, we actually talk about before is when you look at the Marvell's networking product, when we release new product and ramp new product into production, those product lines actually have a higher gross margin. Matt mentioned our networking business, switch PHY, have been growing significantly. We are benefiting from that mix change, too, is networking product ramping up. Going forward, when our new product continue to ramp, we'll continue to see that benefit.
The third, of course, is operational excellence and efficiency. Our operational team continue to drive supply chain efficiency and improve overall cost of sales. This is really 65%. It's a starting point for the synergy achievement. Going into fiscal 2020, Matt mentioned when we achieve the $50 million cost of sales synergy, we are going to continue to expand our gross margin.
Yeah. I'll just add, I think, as Jean said, I think this is a very powerful setup for us for fiscal 2020. The fact that we're guiding 65% gross margins on the level of revenue that we're talking about, I think the mix improvements we've made and the new product progress we've made have been very good in terms of profitability. I would note, I think this is something like the 10th quarter in a row where sequentially we've managed to increase our gross margins a little bit each quarter for the last couple of years. We see that continuing into fiscal 2020 on the back of higher revenue, improved product mix, and our COGS synergies, which have yet to flow through the income statement. We're very optimistic on this for next year.
That's very helpful. My follow-up is on the broader topic, of margins and synergies. Jean, you mentioned some reduction in how much OpEx you expect the company to be spending next year compared to your prior view. To what extent is that just overall tight management of expenses given some of those inventory adjustments that you talked about? Or is there any change in how much synergies the company thinks it can achieve from the Cavium acquisition? Thank you.
Yeah, you're right. We raised our synergy to $200 million during our last earnings call. That's very well defined. Our team is executing ahead of achieving our $200 million synergy plan, especially on the operating expense side. We said at the middle point of guidance of Q4, we are achieving $120 million of the $150 million synergy. We still have $30 million to go, which we'll achieve in the course of fiscal 2020. The rest of the operating expense reduction is really because our team is doing great job really applying a disciplined resource allocation approach to a much larger scale operation. That's how we achieve the better operating expense.
Thank you.
Thank you. Our next question comes from the line of Harlan Sur from JPMorgan. Your question please.
Good afternoon. Nice job on the quarterly execution. Within the networking business, the enterprise and campus Ethernet switching and PHY, clearly 2018 we saw a strong upgrade cycle. How is the team thinking about the momentum of that upgrade cycle heading into calendar 2019? What we're hearing is that we're still early days in the upgrade cycle, given sort of the legacy infrastructure that hasn't been upgraded in a while. Also, these customers also have kind of solidified their plans in terms of what's going to remain on-prem versus what's going off-prem, wanted to get your views.
Yeah, our view is very consistent with that. We believe that that cycle will continue into calendar 2019. It's been a growth driver for us over the last 12 months. We expect that to continue. That's a good market trend. I would say on top of that, layering in our new product success will further help accelerate our growth, our continued growth in our wired networking segment, which is, as you mentioned, our Ethernet switches and PHY products, all of which are extremely competitive and are doing very well in the market.
Great. Thanks for the insights there. Looking at the better Cavium results in October and the stronger sequential growth from Cavium in Q4, it sort of implies that the core Marvell business is down about $65 million sequentially. Looks like $40 million of that is due to the storage weakness. Where is the other $20 million, $25 million of weakness coming from?
Hi, Harlan, this is Jean. Your math is right. The rest of it is coming from other product line. It's declining and also a little bit of Wi-Fi, because as we said, we don't have the gaming product anymore. And seasonality really impact that business, too.
Great. Thanks, Matt. Thanks, Jean.
Thanks, Harlan.
Thank you. Our next question comes from the line of Quinn Bolton from Needham & Company. Your question, please.
Hi, Matt. Just wanted to follow up on the storage business. Sounds like there are lots of moving parts. Some hit the client, some hit enterprise data center. Just wondering, do you see both segments down for a couple of quarters? Do you expect one to recover before the other? I've got a follow-up on the networking business.
Hi, Quinn. Just want to clarify, right? The enterprise and the data center of our storage controller business actually have been growing year-over-year and continue even in Q4, it's going to grow. Really the decline of guidance into the Q4, it's more related to other segments of the business. I think that's one thing to clarify, right? I think it's the inventory adjustment we're going through with the overall storage controller, both HDD and the flash side. We are trying to manage the headwinds for next one or two quarters.
Sorry, Jean. That's been mostly on the client side, the enterprise data center will continue to be strong?
Yeah. Enterprise data center actually are growing year-over-year. The decline, if you think about it, is really largely on the other and the client side.
Got it. Great. Then just going through, Harlan's been out there on the networking business with Cavium business up to $240 million. Does the core networking business, the core Marvell networking business, actually decline in the January quarter? Is that stable?
Marvell networking business, if you think about, Matt talked about both the switch and the PHY, they are actually growing year-over-year, double-digit again. I think Wi-Fi is the major seasonally down quarter, and with that gaming Wi-Fi revenue, that certainly impact the business on that side.
Great. Thank you, Jean.
Thank you. Our next question comes from the line of Craig Ellis from B. Riley FBR. Your question, please.
Yeah, thanks for taking the question. I'll ask a question about use of cash since the cash generation performance in the business was so strong and with good margins in the outlook. It was pretty balanced in the quarter with good share buyback and good debt reduction pacing. Is that a proxy for what we should expect both in the fiscal fourth quarter and as we look into fiscal 2020? With the stock at these prices, Matt, and at, I think, relative valuations to the S&P 500 that are at 10-year lows, would you be more biased to tick it up on the share buyback side?
Yeah. I think as you can see, right? We generated very strong cash flow in Q3. We're really pleased with that. Going forward, we do think with our business model, we'll continue to generate a strong cash flow. We have talked about we plan to pay down debt by close to $100 million each quarter to get to our target leverage ratio. I think because of how much cash we generate, you can see we can certainly have the strong flexibility to do buyback. We certainly will try to balance both different quarter. It could be different, but that's the flexibility we have going forward.
Yeah. Craig, I would just add, I think we're certainly very mindful of our valuation relative to those other metrics, as you mentioned. We are very bullish on our prospects, and given the cash flow generation of this business, clearly we'll be buying back stock as appropriate. I think we're going to do our debt pay down, as Jean mentioned, and we're going to be mindful of that. There's an opportunity here, and we're going to continue just to take advantage of the strong cash flow of the business.
That's helpful. The follow-up, Matt, is to you. You provided us with some very helpful color on the opportunity and in Base Station, both from a product standpoint and the way that can evolve from a customer standpoint. At Analyst Day, you and Jean identified two other upside growth drivers. Are you starting to get visibility that data center, and I think it was Automotive Ethernet, can come into view in FY 2020? Are those other two drivers really longer term with Base Station really being the principal driver of those three upside drivers for next year? Thank you.
From a total dollar amount, the 5G will clearly be head and shoulders the largest of those upside drivers, both of those three initiatives as well as actually for the whole company. That being said, we're happy about all of them. The other one you mentioned was Automotive Ethernet. That's coming off of a smaller base, but off a small base, that business has been growing quite rapidly, and we expect it to grow again significantly in FY 2020. Again, that's a much smaller base we're talking about. Design win pipeline is very strong there. Our new products, which have been in leadership positions both on our Gigabit Ethernet PHY as well as our secure switch, have significant design win pipeline, and those are starting to materialize into wins, which would translate into revenue ramps.
The third one was really around Arm server, if I recall, as the third upside lever. As I mentioned in my prepared remarks, I think our showing at Supercomputing '18 was very strong in terms of ThunderX2 performing well on all benchmarks, both for cloud and for high performance computing applications. That will also generate year-over-year revenue growth and contribution will be part of why we're going to grow in FY 2020 as well. I'd say the 5G, by far, is the one where we see the most upside potential.
Thank you.
Thank you. Our next question comes from the line of CJ Muse from Evercore. Your question, please.
Yeah, good afternoon. Thank you for taking me in, squeezing me in. I guess question on storage, and specifically SSDs. Can you walk through your thinking for calendar 2019? Obviously, there's an inventory correction going on, but bit growth still should be at least north of 35%, and you've got, hopefully, demand elasticity. Would love to hear your thoughts on how you see that playing out for that part of your business in calendar 2019.
Sure. Hey, CJ. Yeah. Again, I think the way to think about our growth coming out of the cycle next year in storage and controllers is really, I'd focus you more around the end markets and the end applications that we're going to grow in, which are, by the way, agnostic to whether it's hot storage or it's cold storage or it's flash or it's HDDs, and that is enterprise and data center. It's been growing very well for us. We believe it's going to grow next year. We believe that, once we're through the cycle, that'll continue to drive our growth. We've been less focused, and we talked about this at Analyst Day, on the client side, and in particular in the PC exposed segments. That one, we made our pivot to the cloud probably two years ago in terms of our R&D investments.
I think as we go forward, I think that's the more pertinent way to look at our business, because that's really where profitability is driven, that's where our R&D is focused, and that's where the growth is coming from and the future of our overall storage controller business.
Very helpful. If I could just follow up on that same theme, specific to NVMe enterprise controllers, can you walk through where you are in terms of customer design wins as well as what revenue opportunities could look like from just licensing IP, et cetera?
Okay. Yeah. Let me take the second one first. We don't have any plans to do any licensing of our IP. Our portfolio is rich, in particular, as you mentioned, in NVMe. I think NVMe and then NVMe over Fabrics as a technology is very important to our company, both because we own it at the drive level as well as now at the controller level upstream. I think what I'd say there is that'll be an important technology for us across a wide number of applications in the company, but I don't think I'm going to go one click lower into the design wins for our specific NVMe products for the enterprise. That's probably a bit too detailed for here, but we are certainly participating in that market.
I just say NVMe overall, as you saw at our Analyst Day, we talked about our new products, which are in the form of our aggregators and our accelerators. Those kind of products and those solutions, I think coupled with our SSD controllers, it provides a total solution for data center and emerging enterprise applications.
Very helpful. Thank you.
Thank you. Our next question comes from the line of Gary Mobley from The Benchmark Company. Your question, please.
Hey, everyone. Thanks for sneaking in my question. Matt, I want to start with a technical question covering your largest storage customer. I don't think it's been a secret that your largest storage customer has been a big supporter of RISC-V, as evident in the press release today, it's evident that they're developing their own SSD controller for enterprise and data center applications. With that specific customer in mind, I'm curious to hear your perspective on market share, then as well, overall market share for SSD controllers in captive versus merchant. Jean, as a follow-up, a housekeeping question. It looks like your share count came in below what I was expecting, just based on the transaction value of Cavium. As a baseline, could you give us what the share count assumption is for the fourth quarter?
Okay. Hey, Gary. I'll take the first two, then I'll let Jean cover the third. With respect to RISC-V, this is something that we're involved in also. We're a member. We're actively engaged and actually looking at a number of different processor architectures. Obviously, we have most of our products from the Marvell side based on Arm. From the Cavium side, they've made the MIPS to Arm transition. We're very open in terms of processor architecture out there, and we've had very good discussions, and we're very closely aligned with WD, as you can imagine, because they're our largest customer. I think we understand, and I think they've been very vocal, probably one of the most vocal OEMs with respect to implementing RISC-V in their own products, some of which they've announced.
I think we're aware of that, and that ties into your second question, which is the verticalization question and captive versus merchant in that market. What I'd say is similar to what I said at the Analyst Day, which is this trend has been happening for some time. Some of those products are already on the market today. The number of SKUs that are out there, the number of applications that are now present for us in the SSD market, in the flash solutions market, across a wide range of not only the guys that make the SSDs, but also the new emerging applications in terms of the DIY customer base in the data center and others, presents, I think that's really where the trend is going and where we see the most traction. We'll continue to participate in both.
I think both are going to be important parts of our business. No question on the verticalization, especially at the lower end and in the PC market, I think it's made sense for these companies to make their own solutions for cost reasons, and that probably will continue. The DIY model, though, I think is one where it really plays to our strengths, and that is because we have experience and knowledge of everyone else's NAND, we're able to go in and actually do almost an SoC, if you will, for different cloud customers, enterprise customers, and then there's even some new emerging applications to the future where we can enable our customers using our technology to take advantage of multiple NAND sources out there and get their own IP implemented in our products and get their own design that they need.
I'll turn it to Jean on the share count.
On share count, right. In Q3, we did buy back some of our shares, probably close to 3 million shares. The Q3, the share count, non-GAAP share count is about 666 million shares. I think for Q4, you can model similar, maybe one or two million shares more. That's what you can model for Q4.
All right. Thank you.
Thank you. Our next question comes from the line of Kevin Cassidy from Stifel. Your question please.
Hi, this is Nick Doyle on for Kevin Cassidy. Thanks for taking my question. Can you hear me okay?
Yep.
Right. Following up on the 5G, how should we be thinking of the timing for revenue ramping on 5G deployments? I have a follow-up.
Sure. Best as we can tell, and again, this is a new technology transition, so this is always subject to change by some quarters, but best as we can tell, it's about a year from now. It's Q4 of calendar. Well, it'd be our Q4 fiscal, which is a year from now.
Okay.
That's when we would start seeing I think some of the ramp for our products.
Okay, great. Thanks for the color. Have you seen any slowing in revenue from China, due to security concerns or even restrictions on buying products from any of their customers?
No. We haven't seen that. We have Chinese customers, and we have not seen that.
Okay. Thank you.
Thank you. Our final question for today comes from the line of Christopher Rolland from Susquehanna. Your question please.
Hey, guys. Thanks for letting me ask a question. I guess, Matt, now that we're a few quarters into Cavium, perhaps you can talk about the products that surprised you the most. What's been stronger than expected for you and what's been weaker?
Sure. Yeah. Hey, Chris. Great final question. Overall, we're very pleased with the set of technologies that we got. In fact, I think you nicknamed it in one of your reports some time ago as a roadmap in a box, I think we were very fortunate to get such a diverse set of product lines and technologies from them. I think in general, our feeling is from when we started the journey with them to now, the expectations have largely been met. I'd say that with 5G, if you go back to fall of 2017, okay, when we were announcing this, I think 5G at that point seemed like it was going to be sort of a 2021, 2022 type of thing. We hit Mobile World in February, All of a sudden everything changed, Since then, I think directionally we've seen those pull in.
I think the 5G opportunity on the basis of Fusion's success in 4G and then all the great work the team did to get the 5G technology ready, I think that's been a nice upside surprise to us. I think the rest of it, whether it's OCTEON, which is sort of what we would call one of the franchise businesses of Cavium, it's doing extremely well. We see that it's very complementary with our switches and our PHYs. When we go to market now, we're able to sell that as a total solution. Team's doing a great job there. I'd say as we also integrated our ArmADA, our processor line into the OCTEON team, I think they've also done a great job of taking their SDK and applying it to the ArmADA products.
In fact, we had some products in flight that were, quote, ArmADA that are actually going to be announced as OCTEON and they're going to be run like a Cavium product from day one. Lots of positive collaboration between the teams. The technologies are good. There's a whole laundry list of things that they have, and I probably won't go into every one in detail, but I'd say in aggregate, we're very happy with what we got, and I think the fact that their revenue has come back and it looks good for Q4, and we've got the outlook as we head into the second half of next year with some of the new ramps, we're very optimistic.
I think, again, the combination of the two companies, the ability to share the IP, the ability to sell a whole solution now with OCTEON as an example at the core, very positive on that outlook.
Thanks again and appreciate the color.
Yeah. Thanks, Chris.
Thank you. This does conclude the question and answer session of today's program. I'd like to hand the program back to Ashish for any further remarks.
Thank you, everyone, for joining us today, and we look forward to talking to you again next quarter. Thank you and goodbye.
Thank you, ladies and gentlemen, for your participation in today's conference. This does conclude the program. You may now disconnect. Good day.