邁威爾科技 (MRVL) 2019 Q4 法說會逐字稿

Good day, ladies and gentlemen, and thank you for your patience.

You have joined the Q4 2019 Marvell Technology Group Earnings Conference Call.

(Operator Instructions) As a reminder, this conference may be recorded.

I would now like to turn the call over to your host, VP of Investor Relations, Ashish Saran.

Sir, you may begin.

Thank you, and good afternoon, everyone.

Welcome to Marvell's fourth quarter and fiscal year 2019 earnings call.

Joining me today are Marvell's President and CEO, Matt Murphy; and Marvell's 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, and good afternoon to everyone on the call.

Let me start with a quick recap of Marvell's financial highlights for fiscal year 2019.

Our GAAP revenue was $2.9 billion, GAAP gross margin was 50.9% and GAAP loss per diluted share was $0.30.

On a non-GAAP basis, our gross margin was 63.9% and non-GAAP earnings per share was $1.19.

Fiscal 2019 marked another year of solid accomplishments.

We continued our drive towards operational excellence and scale, shifting our focus to the infrastructure market and delivering more value from our engineering efforts.

The result was an increase in profitability as we transform Marvell into a leading silicon supplier to the infrastructure market.

During the year, despite a tense and challenging geopolitical climate, we secured regulatory approval for the Cavium merger.

After the merger closed on July 6, we rapidly integrated the Cavium team and increased our prior synergy target to $200 million, which we are on track to achieve by the end of fiscal 2020 on a run-rate basis.

The combined Marvell and Cavium team quickly and successfully aligned our growth trajectory with the major market trends of 5G, cloud, AI, automotive and enterprise refresh cycle.

The result is a stronger and more diverse company with the scale and broad portfolio to deliver significant long-term value to customers, shareholders and employees.

The merger also accelerated Marvell's ongoing shift to the growing and lucrative infrastructure market.

Integrating Cavium gives us leading positions in processors, security and fiber channel storage, all which offer customers an unmatched end-to-end infrastructure portfolio.

On the operational front, we improved our year-on-year non-GAAP gross margin by 2.5 percentage points to 63.9% and increased our non-GAAP operating margin from 25.9% to 27.6%.

In addition, standalone Marvell delivered non-GAAP operating margin over 30% in the second quarter of fiscal 2019, 6 quarters earlier than projected.

Even as we improved operational efficiency, we continue to invest in R&D, the lifeblood of our future, which will further fuel long-term growth.

As we detailed in our recent 5G announcements, we are poised to become a leading semiconductor supplier for 5G infrastructure.

And when you add in our recent design-win momentum in the data center, automotive and enterprise and edge markets, we are well positioned to drive compelling long-term growth.

While we've made significant progress as a company in fiscal 2019, we did experience some end-market headwinds, particularly for storage, which impacted our fourth quarter results and our current projections for the first quarter of fiscal 2020.

Like many of our customers and peers, Marvell is not immune to the macroeconomic forces affecting today's market.

But I'm pleased about our new products in development, which we will continue to invest in, and about our design-win pipeline.

Infrastructure customers value innovation, and while design wins take time to ramp, they typically last many years.

This provides predictability and long-term growth, and that's why I'm so confident that Marvell is well on its way to becoming one of the world's leading infrastructure suppliers.

Moving on to our fourth quarter results.

Our revenue for the fourth quarter was $745 million, at the upper end of the preliminary estimate we announced on February 6. The majority of the weakness was in storage due to lower-than-expected demand in our storage controller business.

Networking revenue was also below expectations with the exception of embedded processors for networking and LTE and pre-5G wireless infrastructure, which met expectations.

Now turning to our core businesses of networking and storage.

Networking revenue was $387 million, down 3% sequentially.

This was lower than expectations, primarily due to the macroeconomic conditions mentioned earlier.

The one exception was embedded processor revenue, which grew by double digits sequentially, driven by stronger-than-expected demand from the wireless base station market.

We have now sampled the latest generation of our highly successful multi-core OCTEON embedded processor to several key customers in the enterprise market and to our lead 5G base station OEM.

This innovative multi-core design also provides the foundational architecture for our 5G Fusion baseband processor, which we'll start sampling next quarter.

Now as a reminder, our OCTEON embedded processors manage and -- control and data plane functions, and are responsible for the secure and efficient movement of data packets within networking infrastructure, such as base stations.

Our Fusion baseband products include advanced DSPs to process and code, decode and manage voice and data signals within base stations in the radio network.

Our design-win momentum continues in 5G, and we recently announced a significant long-term partnership with Samsung to deliver multiple generations of embedded processors and baseband processors for both LTE and 5G base stations.

We expect shipments of our 5G products to start to ramp towards the end of this fiscal year 2020 and continue to grow rapidly into fiscal 2021 and beyond.

In addition to supplying Samsung, we are also now working with an additional Tier 1 5G base station OEM.

We currently expect to sample a 5G Fusion baseband processor, targeted to their requirements, in early calendar year 2020.

We continue to engage with additional base-station customers for our 5G solutions.

As we work with our customers on their 5G road maps, it's becoming apparent that they want to implement artificial intelligence and machine learning to improve the performance and quality of 5G services.

In particular, areas where AI and ML have demonstrated clear value include both radio and edge functions, like channel estimation, intelligent beamforming and digital pre-distortion.

Marvell has already been developing solutions for the AI inference market and we are well positioned to integrate these capabilities into next-generation 5G processors.

AI has become a very critical building block technology and we believe that this capability can be deployed into additional Marvell product lines in the future.

Given our strong traction with 5G customers, we are expanding our processor capabilities, and we have recently opened a new R&D center in Raleigh, North Carolina.

At its core is a very experienced R&D group and we intend to add more resources, taking advantage of the area's rich talent pool.

Moving to Ethernet.

Marvell's Ethernet switch and PHY business delivered another quarter of double-digit year-on-year revenue growth driven by our refreshed product portfolio.

However, revenue declined on a sequential basis as our customers tightened inventory in their supply chains, we believe in response to macroeconomic uncertainty.

But our design win pipeline remains strong and we continue adding high-volume wins such as a recent access switch design for a leading networking OEM.

As we discussed at our Investor Day, we have been working on expanding our addressable market for Ethernet switches into the data center.

Next week, at the Open Compute Summit, we will be announcing a new high-bandwidth, feature-rich family of Ethernet switches.

This new product family extends our proven and widely deployed Prestera architecture to meet the distinctive data-centric requirements for edge computing.

These products can deliver multiple capacity points from a few terabits per second all the way up to 12.8 terabits.

This modular and scalable design maximizes our ROI to deliver a range of products from a single development effort.

Customers benefit from being able to select cost and power-optimized silicon tailored to their specific bandwidth requirements.

These products are not only very competitive on capacity, but also deliver a much richer set of features, enabling advanced workflow visibility, improved analytics and network simplification for our customers.

Early customer interest is high and I look forward to sharing our progress in the coming months.

Our Wi-Fi business will soon start ramping next-generation products built on the latest Wi-Fi 6 standard.

With a series of design wins already in place, we expect our Wi-Fi business to return to growth starting in the second quarter of fiscal 2020.

Now a quick update on some of our emerging investment areas.

ARM server processor evaluations with cloud and high-performance computing customers continue to make solid progress.

We are also making great progress on our next-generation ARM server processor, ThunderX3, which is being developed in 7-nanometer technology.

We currently expect to sample this product with customers in the second half of this year.

Similarly, our LiquidSecurity products continue to gain traction in the cloud market.

Earlier this week, we announced a collaboration with Oracle to deliver secure key storage using our HSM adaptors for Oracle cloud infrastructure customers.

Along with Amazon and Google, this constitutes another cloud customer that has adopted Marvell's solution.

And as we announced at the most recent RSA Conference, we are extending our portfolio to include LiquidSecurity appliances to provide private cloud and enterprise data center deployments, which significantly increases our addressable market.

Our automotive business continues to make progress as well.

Our new gigabit PHY transceiver will soon go into production, which is a companion solution to the secure gigabit switch we released last year.

These 2 parts have now been designed in at 16 car manufacturers, including top OEMs in North America Europe, China, Korea and Japan.

These devices are used in multiple domains in the car, including telematics, infotainment and gateway applications.

Moving now to our first quarter outlook for networking and fiscal 2020, we expect an approximate 10% sequential decline in revenue.

This projection reflects the residual impact from the tight inventory control we experienced from our customers in the prior quarter, seasonality and the continued cautious outlook from our China-based customers.

Now let me turn to our storage business.

Storage revenue for the quarter came in below our expectations at $317 million, declining 22% sequentially.

The majority of the shortfall was due to a reduction in demand for our storage controllers.

Our fiber channel business was slightly lower than expected too, but within a normal revenue range for that business.

We believe that demand for our storage controller products was impacted primarily by macroeconomic uncertainty, a reduction in cloud capital spending and PC CPU shortages.

In addition to a reduction in direct orders, we also experienced changes in customer demand for products consigned to third-party logistics hubs, commonly referred to as vendor-managed inventory arrangements.

Changes in consumption through these arrangements can occur without any prior notice, and in the fourth quarter, these were particularly pronounced and volatile.

As we noted at our Investor Day, we expected that the majority of the remaining notebook market using HDDs would shift to SSDs over the next few years creating a headwind for HDD storage controllers.

However, we expected to more than offset these headwinds with growth from our investment areas in near-line HDDs, preamps and enterprise and data center SSDs, including our new flash solutions.

We continue to believe that this is the right strategy for this business, but the sharp contraction in NAND pricing is accelerating the conversion to SSDs in the PC market faster than expected.

While this does not change our long-term strategy for this business, it does present a near-term challenge, and this is reflected in our fourth quarter results and guidance for the storage business.

In addition to the impact from NAND pricing, as we had indicated in our preliminary revenue update in February, demand weakness is carrying over from the fourth quarter into the first quarter.

Inventory levels in customer supply chains need time to rebalance and the first quarter results are seasonally weak for the storage end market.

As a result of these factors, we expect storage revenue in the first quarter to sequentially decline approximately in the mid-teens on a percentage basis.

However, we do expect the first quarter to be the bottom for our storage revenue in fiscal 2020.

Our customers have indicated that they expect demand to increase in the second half of this year.

We also believe that we started to undership the HDD end market starting in the fourth quarter, and project this undershipment to continue in the first quarter, which is a strong signal that inventory is starting to deplete in our customers' supply chains and our shipments will increase to catch up with end demand as we progress further into fiscal 2020.

We continue to grow our position in the data center market.

We have started shipping our controller and preamp solution and a recently announced 16-terabyte high-capacity HDD, which is targeted at both cloud and traditional data center applications.

This drive has an industry-leading multi-platter architecture that delivers significantly improved data storage capacity by using 2 Marvell preamps per drive and uses advanced signal processing enabled by Marvell's high-performance controller.

By closely coupling the controller and preamp design, we were able to develop a comprehensive chipset solution that delivers capacity leadership.

This win has increased our dollar content in this drive by approximately 75% from our prior nearline solution.

In closing, Marvell is entering an extremely strong new product cycle, driven by our design wins in the 5G, data center, automotive and enterprise markets.

These will increase demand for our embedded and baseband processors, Ethernet switches and PHYs, Wi-Fi, security and server processors.

We believe that our enterprise networking business will stabilize and keep growing over the long term.

Storage, which is a cyclical end market, is going through a downturn, but our storage products will remain very profitable even as they become a smaller percentage of our revenue as we diversify into higher-growth networking markets.

We are very encouraged by recent improvements in booking trends, which bode well for a recovery.

Thus, we currently expect the first quarter, which tends to be the softest seasonally, to be the bottom for revenue in fiscal 2020.

We anticipate resuming growth in the second quarter as end-market conditions improve and customers' inventory levels normalize.

Through these near-term headwinds, we will remain disciplined in managing operating and discretionary expenses.

As you may recall, we have already reduced our forecast for operating expenses, exiting fiscal 2020 to be approximately $10 million lower on a quarterly run-rate basis from the $290 million we had outlined at our Investor Day.

However, our focus remains on our long-term growth strategy, which is unaffected by current market conditions.

With this in mind, we intend to continue to invest in R&D through this period of market softness.

I want to close by thanking the more than 5,000 Marvell employees around the world for their hard work and performance during the year and for their continued commitment.

Without their contributions and effort, none of this would be possible.

With that, let me turn the call over to our CFO, Jean Hu, for more detail on the fourth quarter fiscal year 2019 performance and our outlook for the first quarter of fiscal 2020.

Thanks, Matt, and good afternoon, everyone.

I'll start with a review of our financial result for the fourth quarter and then provide our current outlook for the first quarter of fiscal 2020.

Revenue in the fourth quarter was $745 million.

Our core business of storage and networking accounted for 95% of revenue, networking accounted for 52% of revenue and storage accounted for 43% of revenue.

Other product revenue was $40 million and accounted for 5% of revenue.

GAAP gross margin was 43.2%, which included amortizing the remaining balance of $98 million Cavium inventory step-up cost.

As a reminder, we stepped up Cavium inventory as part of acquisition purchase price accounting.

Our non-GAAP gross margin was 64.5%, slightly lower than guidance, due to lower-than-expected revenue in the quarter.

GAAP operating expenses were $375 million.

Non-GAAP operating expenses were $286 million at a lower end of the guidance range provided in December.

We tightly managed OpEx in a difficult environment and are committed to doing so in the future.

Non-GAAP operating margin was 26%.

GAAP loss per diluted share was $0.40 and non-GAAP earnings per diluted share was $0.25.

Now turning to the balance sheet.

In the fourth quarter, we paid down $75 million of our long-term debt and returned $89 million to shareholders through $50 million in share repurchases and $39 million in dividend.

We exited the quarter with $582 million in cash and short-term investment.

Our long-term debt was $1.75 billion, and we expect to continue to pay down debt to achieve our target leverage ratio.

Let me now move on to our current outlook for the first quarter of fiscal 2020.

We expect our revenue to be $650 million, plus or minus 3%.

Our expected GAAP gross margin will be approximately 55% and our non-GAAP gross margin will be approximately 64%.

Note that we expect to complete our planned $50 million of cost-related integration synergies by the end of this fiscal year.

We expect our GAAP operating expenses to be between $378 million and $388 million and non-GAAP operating expense to be in the range of $295 million to $300 million.

This forecast reflects the seasonal increase in OpEx in the first quarter driven by employee payroll taxes, matching contributions and our annual merit increase.

We'll continue to tightly manage our operating expenses and we anticipate operating expenses to reduce rapidly as we progress into fiscal 2020.

And realize the remaining synergies to exit the year at a quarterly OpEx run rate of $280 million.

We expect our non-GAAP tax rate to be approximately 4.5% in fiscal 2020.

This forecast reflects the impact of the Cavium acquisition.

We expect net interest expense to be $19 million.

We anticipate a GAAP loss per diluted share in the range of $0.05 to $0.09 and non-GAAP income per diluted share in the range of $0.12 to $0.16.

We're now ready for your questions.

Operator, please open the line for questions.

(Operator Instructions) Our first question comes from the line of Vivek Arya of Bank of America.

I guess, Matt, the main question is, where we go from here and how you're looking at revenue growth for the rest of fiscal '20, you mentioned Q1 could be the bottom of the cycle for you.

But when do you think you can return to organic year-on-year growth later in the year?

Sure, great.

So yes, the way to think of it is we do see Q1 as the bottom.

And we're encouraged based on the booking trends rolling over the last several weeks and months, that this would lead to revenue growth in Q2.

When you look at our Q2 and Q3 normally, those are seasonally up quarters for us.

So we see that trend happening.

And then if you think about sort of by the time we get to the fourth quarter, we believe that at that -- by that point, end-customer inventory levels should have rebalanced and normalized.

And then on top of that, in the fourth quarter, you layer in the ramp that we project for our 5G shipments.

And so when I think about it on a year-over-year basis, and when do we return to growth, I think about it really as being Q4, for those reasons just given seasonality, inventory rebalancing and 5G starting to take off.

And just as a reminder, when we talk about our long-term growth rate, what we outlined at the last Analyst Day was in the range of 6% -- 6% to 8% on a year-over-year basis.

Got it.

And for my follow-up, on the gross margins side, I think before you had the reset for Q4, you were expecting to do over 65%, but that was at a $800-million-plus revenue run rate.

I am wondering, Jean, if there is a certain revenue level we should think about as to when you get back to 65%-plus and then hopefully, towards your longer-term target beyond that?

Yes, there are a few puts and takes that impact our gross margin.

So if you look at the Q4 result and the Q1 guide our gross margin was lower, as you point out, primarily due to a much lower revenue level.

The amount of leverage is significantly -- so when you have lower revenue, certainly, we lost 50 basis points on the gross margin side compared to the 65% guide.

But going forward, when you look into the fiscal '20, we are going to realize the $50 million costs synergies with Cavium acquisition.

So our gross margin actually is going to step up from this 64% level gradually to exiting the quarter with the -- exiting the Q4 fiscal '20 to about 65%.

And, of course, in Q4 and Q1, the mix also is less favorable.

So going forward, toward the second half of the year, certainly, mix is also going to help us to improve gross margin.

And as far as our long-term model, as Matt mentioned, once we return to the 6% to 8% organic revenue growth, our model will help us to leverage to get to the 65% to 66% gross margin in the longer term.

Our next question comes from Ross Seymore of Deutsche Bank.

Matt, I want to go back to storage side.

You gave some helpful details about that switch from HDDs, SSDs with NAND, et cetera.

If we took it to a slightly higher level and we just thought about your core business ex the fiber channel, and that was running north of $300 million in the first half of last year.

Talk about what you think it would take to get back to that level.

Do you expect to get above that level this year?

Or are some of those secular changes going to take a little bit longer to take advantage of as you switch your business over to the SSD side?

Sure.

So I'll make a few comments, and Jean can add as well.

So I think the way to think of it is, clearly, we're going through a pretty severe period of inventory correction of which the duration is not certain.

Although, based on our customer data points, people are hoping for a recovery in the second half.

I would state that it's pretty dynamic at this point, I think there is a number of factors that are going on.

We highlighted some of them, which is obviously the cost structure of that -- of NAND, driving some shift in adoption of drives.

In the notebook market, you've also got share shifts going on within the NAND player.

So it's a fairly dynamic environment is what I would say at this point.

But I don't -- I think that recovery back to that level is going to take some time to work its way through, probably more like the end of the year.

Got it.

And I guess for my follow-up question, switching over to the networking side.

It's good to hear you talk about another customer on the 5G side of things.

Could you give us a little more color on how you expect that to roll out between both your key customer, your first customer and the second?

You've talked about the content per base station.

But is there any way you can help us judge a little bit about how that per base station turns itself into revenue exiting the year?

Any sort of kind of goalposts around that?

Sure.

Yes, let me start with the content first, just to reframe it, and then we can talk about the range on timing.

So with our lead customer, and what we talked about at the Investor Day, what we call our 5G platform solution, that's really where the -- having multiple baseband processors, multiple control and data plan processors plus Ethernet switches and PHYs, that's the 4x increase in content we talk about.

So think of that as the TAM for us of the superset content that we could achieve, which we have in our lead customer.

Beyond that, we're not really in a position on any other customer to comment on content, although we've got a lot of engagement across a range of key OEMs who are going to play a very meaningful role in 5G.

With our lead customer, we expect them to ramp into production by the end of this fiscal year.

And just rough timing-wise, to the extent we're successful on our product development and our customer is successful at qualification, the normal things you go through, you should think about ramps beyond that probably being staged about a year behind where our lead customer is.

And I think -- but we're currently on track to sample that product, as we said, in the first part of next year.

Our next question comes from Timothy Arcuri of UBS.

(Operator Instructions) We'll go to the next question from John Pitzer of Crédit Suisse.

I apologize for the background noise, I'm at an airport.

I want to get a little more detail on your OpEx comments in your prepared comments, Matt.

I'm just -- it makes sense to me not to cut OpEx given the growth opportunities you had, especially if you think this revenue blip is kind of a 1 quarter issue.

I'm just wondering if you can help me understand where you prioritize your net OpEx now.

And I guess, importantly, as revenues growth starts to come back, how should we think about OpEx growth from these levels, and are you thinking about adding more customers on the 5G front and/or the AI and server opportunity.

Is that an incremental step up in OpEx or is that already sort of being accounted for in current run rates?

Sure.

How about we'll break into 2. I'll take the first part, which is how are we thinking about our spending right now in this environment, where are we prioritizing R&D, and then Jean will talk about when and if she will allow the OpEx to increase at some juncture.

No, but -- if you go back to what we're doing right now, I think we outlined in our Investor Day a pretty comprehensive product and end market strategy.

And if you look at where we're investing those dollars, it's very consistent with where we were kind of on the heels of integrating the Cavium team in.

So we're currently staying the course on our investments.

I'd say within that there's always some mix, as an example, given the strong traction we're seeing on the processor side of the business, that's why we're doing things like expanding our design capacity by opening up the Raleigh site.

So within our OpEx you'll see some level of moment, and we're pretty dynamic inside Marvell in terms of how we think about investing.

But the thesis that we outlined at our Investor Day is still very much intact and we're very committed to investing for the long term through this cycle.

Jean, maybe you want to talk about how we think about -- once we get through this, what OpEx looks like assuming we see the revenue growth?

Yes.

Matt -- as Matt said, right, we're funding all our programs because of all the opportunities ahead of us.

But we're striking a balance, also making a lot of cost control and aggressively managing our cost.

So I think for fiscal '20, we're doing both, in a sense, these -- we're continuing to increase the efficiency of operation to really streamline our business, to really manage our discretionary costs, at the same time, matching all those programs that Matt mentioned.

I think we feel pretty good about the balance that we have for the rest of the year, but certainly, with all the 5G opportunities that continue to ramp and the design wins that we continue to have, our team literally is, as every quarter goes, they win more designs.

Of course, when our revenues start to grow, we will increase the OpEx to fund additional investment opportunities.

I think if you go back to our long-term target model, we basically said hey, top line revenue is going to grow 6% to 8%, but we are going to fund the R&D at 24% to 25% of revenue; we're going to control SG&A to be around 6% to 7% of revenue.

So that's how you should think about beyond this year, we're going to just migrate toward our long-term target model.

That's helpful, you guys.

Then maybe for my follow-up, just on the 5G front.

Matt, as you think about the 2 potential large customers in Europe, you could add, clearly, I think both have had some fits and starts or some issues with -- around ASICs for base stations.

And I know you've said in the past that you don't want to commit resources to adding a new customer if you're just going to be a stopgap solution until their ASICs come out.

So I'm wondering if you can help me understand, where do you think that dynamic sits today.

And I guess help me understand what you think you can do with your silicon you normally embed in the networking side or the baseband side that's appealing to companies that have big ASIC efforts internally?

Sure.

Yes, John, that was a thoughtful but very detailed question.

I think at this juncture, I'd probably just keep it at a very high level, which is we're happy with the engagement.

We've got a lot of confidence in that engagement.

And our customer, I think, is happy at this point.

So I'll leave more details on this one for later as we make progress and as it's appropriate to discuss it in more detail.

Your next question comes from the line of Blayne Curtis of Barclays.

Matt, I was just wondering, I know at the Analyst Day you talked about, I think, a 7% exposure to mobile hard drive, sort of the faster adoption of SSD.

So with the April guidance, I wonder if you can just dial us in what that exposure is left.

And then I'm just kind of curious from a data center perspective, both across nearline and SSDs, what you're seeing there, and do you expect a recovery in the second half?

Sure.

I'll take the first part, and maybe the second, too.

But on the first one, just to clarify, the 7% we showed -- I think that's what you're referring to -- was really our estimate at the time of the Investor Day of what our total company exposure was to HDD controllers that sold into notebooks, right, and that was in October of last year.

Candidly, through this down cycle we're going through, I think trying to actually pinpoint exactly what percent exposure we have on a quarterly or monthly basis isn't all that helpful.

I think we really need to see these inventory levels and demand rebalance before we can comment on that again.

Otherwise, we're just in a very dynamic market where things have come down, and then based on, obviously, what we're seeing from a bookings perspective, they'll pick back up at some point.

Then that's probably a more relevant discussion about where do we end up when this is all done.

But we certainly continue to see that the networking part of the business is going to continue to grow faster than the storage part of the business, which will, over time, reduce our exposure to that segment.

And then even within storage, I think our progress, to your second part of the question on the data center and enterprise type of applications, both on just for our storage controllers, which span both types of media, will also continue to grow and diversify.

And we see that, but I think first things first is we've got to work through the supply-demand imbalance and get back to a normal state before we can comment on exposure.

Your next question comes from the line of Hans Mosesmann of Rosenblatt.

Matt, a question on the processors or multicore OCTEON processors in 5G base stations.

What's the competitive dynamic there relative to what it may have been in the past for these types of products?

Sure.

Yes, Hans, on the embedded processor side, which is -- and again, we're very pleased with the sampling and the performance of our latest generation OCTEON.

That product in the 5G market really has one main competitor, which is Intel.

And we're both battling for sockets in this cycle.

And we like our product.

We like our chances.

And certainly, the part is performing extremely well.

So that's the dynamic from a competitive standpoint.

Our next question comes from Joe Moore of Morgan Stanley.

Within solid-state drives, how are you thinking about the opportunity, and I guess, how are you seeing competition from custom solutions versus merchant, just the market share dynamics that you see over the course of the year?

And I guess, to the extent that that market is a little softer as well, is that a function of prices going down, so customers are reducing inventory?

Or what's happening there?

Sure.

Yes, thanks Joe.

So I think the more customized or what we call do-it-yourself solutions, are what continue to have -- where we have a lot of traction.

At the Analyst Day, we talked about that's really where -- that business model, I think, is were Marvell can add the most value, so that's where we're putting our effort and our energy.

And to the extent that on the merchant side, we can address the data center and enterprise applications, we're doing that, either with merchant parts or in some cases we're doing customer-specific products.

I think on the merchant controllers selling into retail and PC type of applications, that's one where we pivoted the R&D some time ago.

So we've got sockets there.

We're going to continue to ship what we have.

But in this cycle, and that's why I made the comments that as the remainder of this notebook market continues to convert over time, which make sense to SSDs, we're not going to really participate in that.

But we do see large opportunities, especially in the -- with the large cloud OEMs as well as in the enterprise segment for our solutions in SSD.

And we're continuing to invest there.

Our next question comes from Karl Ackerman of Cowen and Company.

I have a question on networking.

Now that you have integrated your ARMADA and XPliant switching roadmap, have you seen enhanced interest from new enterprise and service provider customers and/or meaningful design wins from a more streamlined product set?

Sure.

Yes, thanks for the questions.

So let me comment on them separately -- just get XPliant out of the way.

So XPliant was a product line that we decided to not continue to invest in.

So it will obviously take some of the advantages we got from a technology point of view, IT point of view and people point of view and we've combined that in our larger Prestera Ethernet switch group.

So that one is not a go-forward product line.

ARMADA, we're extremely pleased with the take-up of that, call it, our lower core count processors now that we've got the Cavium team on board.

Just as an example, our newest products that were inflight that were "ARMADA" products, we're going to actually release them as OCTEON products, so they're going to use the OCTEON SDK.

I'd say the Cavium marketing team has done an extremely good job of driving design wins for that product line, and even that new product that will come out as an OCTEON, primarily because in many of our applications, the customers want all the way down to single and dual-core, quad-core, all the way up to the highest-end OCTEON and everything in between.

And one of the gaps that Cavium always had was they didn't really have an offering at the lower end of the market.

So the fact that we're able to leverage all the effort and all the software capabilities of the OCTEON franchise onto ARMADA, I think, is going to be very compelling.

I would add on top of that that same team, along with our sales group, has done a great job of also really proposing solutions to our customers which also include our Ethernet switches and PHYs.

And so that's another area where I'm pleasantly surprised that we're finding the ability to actually cross-sell all those products together, and I think our team is doing an outstanding job there.

So I think having -- in short, I think having a larger processor home for the ARMADA lower core count family inside our company now will actually drive way more growth than we would ever have gotten on that product line had we not had OCTEON as the flagship to pull us through.

That's very helpful.

For my follow-up, if I may, I was hoping that you could update us on your design win progress across not just automotive connectivity but also automotive Ethernet?

I was -- should we expect revenues to accelerate meaningfully in 2019?

And perhaps when could automotive Ethernet reach your $100 million revenue target?

Okay.

A couple of things.

So just to be clear, in my prepared remarks, I made some commentary, so just for everybody on the line, the prepared comments were around automotive Ethernet, just to be clear.

So connectivity is where we started in automotive.

That continues to do well.

We've been in that business for some time.

We've got new products there.

That's going well.

But specifically, when I said we were designed in at 16 different car manufacturers across all the critical geographies, that was for automotive Ethernet, both on our Gigabit PHY as well as our switch.

Those design wins are things that we've locked in over the last few years.

But you should think of that as really being more than a year away.

So those are not a calendar '19 ramp.

They're really a ramp probably in more significant volumes at the end of calendar '20, which would be for model year '21.

But the progress has been very good there.

I think that trend of Ethernet being a critical backbone in almost all the new different car models is real.

And I think we've hit the sweet spot with our Gigabit products there.

So yes, just to be clear, we're going to have, I think, very strong offerings both in Wi-Fi as well as in automotive Ethernet in the model year '20 and '21 time frame, depending on if it's Wi-Fi or Ethernet.

Our next question comes from Harlan Sur of JPMorgan.

On the decline in the networking that you project for the April quarter, I guess first of all, is April quarter the bottom for networking as well?

And then you talked about the tight inventory management by customers in your enterprise business driving the weakness, what's happening with your service provider, cloud and wireless businesses in the April quarter?

Are they declining sequentially as well?

Yes, okay.

Yes, Harlan.

So -- yes, so let me answer that first.

So on -- the overall business is going to bottom -- overall networking will bottom in Q1.

The -- all those segments you mentioned, whether it was service provider, cloud or enterprise, they're all declining in Q1.

I'd say that the end-user data points are encouraging, if you just look at the -- our large OEM customers and what they're reporting in terms of their sales and their outlook, it looks relatively stable.

But clearly, as all of you that have been through these cycles know, when uncertainty gets injected into the equation, supply chain professionals around the world decide to tighten the belt and make sure that they don't get caught to the extent that there is a downturn.

So we've seen that, and that's clearly happened.

But we do see that working its way through starting in Q2.

The other thing I would note is that the service provider or carrier portion of our business is lumpy.

It's lumpy by design and our lead customer is even lumpier than a lumpy market.

So that does present its challenges from time to time.

And then the final data point I would give is that with respect to our Q1 bottoming, one of the factors that we also saw was that we believe in the second half of last year, we talked about this on the last call, although it's been difficult to quantify, that certain Chinese customers took product in advance of the tariffs kicking in.

And we don't know how to size that exactly, but we're certainly seeing the effect of that manifest in -- a little bit in Q4 and really into Q1 in terms of that inventory needing to work its way through.

So that's the breadth of what we see across all the moving pieces within our networking business.

And to add on that -- Harlan, to add to what Matt said is, our booking actually improved a lot after Chinese New Year.

So that really supports our networking will be the bottom in Q1, too.

Great.

Thanks for the insights there.

And then on the ARM-based ThunderX program, you've obviously had some great traction with the high-performance computing while you've got server OEMs locked up, like HP.

As you mentioned, Matt, you're qualifying with several of the cloud and hyperscale customers, given what you know of their timing, if all goes well with the qualification, do you anticipate them ramping ThunderX2 in the second half of this year?

Sure.

So yes, we're pleased with the traction.

As I mentioned, our next generation X, ThunderX3, which is in 7-nanometer, is going to be a very competitive product, and we're going to sample it this year.

With respect to X2, you're right, it's got very good traction in HPC.

We've put out a bunch of announcements about that.

And on that front, in addition to the cloud side, you should really think of that more being calendar '20 versus an end of '19, primarily because across both of those markets there's not insignificant qual cycle times that are involved with these things.

And we anticipate that that's going to consume a lot of the calendar '19 activity.

And so I would think of it more as a next year calendar '20 event.

Our next question comes from the line of Quinn Bolton of Needham & Company.

I just wanted to follow up on the 5G opportunity.

Looking back to Cavium's position in embedded processors and 4G, they sold OCTEON to 3 of the 5 leading base stations.

You guys have talked about your success with a lead customer in 5G taking your entire platform.

So just wondering if you could comment, how do you see your market share relevant to x86 converting your 4G OCTEON processor customers as they convert up to their 5G base station platforms?

Yes, thanks.

No, we feel good.

We feel good about that transition in the 5G cycle and our ability to have a very strong presence with OCTEON in 5G.

And again, we get a lot of leverage out of that development because the core architecture, we also use in Fusion.

And then we actually -- when we design our platforms, we actually design those products to interoperate and work better together.

So, yes, so we -- we're -- we like our position in 5G and the ability to transition from 4G successfully.

Then real quick, for Jean, just obviously with the lower revenue outlook in the near term, any updated thoughts on debt repayment outlook over fiscal '20?

Yes, I think we'll strike a balance.

We'll continue to pay down debt each quarter, probably less than the original plan of $100 million.

But we definitely will pay down the debt.

Because the business continue to generate a strong cash flow even with a much lower revenue, we're still very well positioned to generate a cash flow to both pay down the debt and also return cash to shareholders.

Our next question comes from C.J. Muse of Evercore.

I guess, the first question, Matt, as you think about geopolitics over the last few months and incorporate what you learned at MWC, what's your view today in terms of your lead customer and market share outlook over the next 12, 24 months?

Oh, I see.

Yes, so -- no, thanks for the question.

No, I think our lead -- yes, it's hard to comment on the geopolitical side.

I think that's playing out that we're all seeing that in the news every day.

I think in our -- when we look at -- a couple of things are encouraging for our lead customer.

One is, when you just look at their announced partnerships and deployments with new operators in new countries, I think that's very encouraging.

They're certainly, as usual, a very aggressive company and very bullish on their prospects.

So there's a view out there that they're going to do very well in 5G, and we're obviously hoping that they do.

But it's hard for me to handicap how that plays out exactly with some of the dynamics in some of these countries.

We just think that they're going to have a strong product offering and probably do incrementally well in the cycle.

That's helpful.

And I guess as my follow-up to you Matt, as well, you talked about bookings signaling April quarter as a bottom.

Can you give a little more color around that?

Is that specific to both storage and networking?

One or the other?

And just anything to help frame that would be helpful.

Yes, the booking, that's really one of the metrics we follow to judge our revenue forecast, right?

And we also get customer feedback, customer forecast.

On the bookings side, we do see the strength for both networking and storage.

And as Matt mentioned earlier, our storage, a lot of the revenue is sell-through VMI, and we can see, as a company, we're undershipping the market in the storage HDD side.

So typically, that means the inventory situation is getting reduced and fixed.

That certainly will help us to get through the bottom of the storage HDD cycle, basically.

Our next question comes from Srinivas Pajjuri of Macquarie.

Just first, a clarification, Matt, on the networking business in the quarter that you just reported, you said it tracked a little bit below expectations.

But at the same time, you said Ethernet was pretty solid, up double digits and also it looks like embedded did pretty well.

I'm just curious as to what was weak in the quarter that you reported in networking?

So probably I'll start, and Matt can add.

When you look at our networking business, certainly, if some of you recall, we do have a legacy piece of our networking business.

That certainly is a little bit below our expectation.

And the inventory conditions of some of our customers also tamped down some of our product lines.

Matt?

Yes.

No, I'd just say the way to phrase it is we -- when we guided, we had a certain expectation for that quarter sequentially in networking.

So for example, Ethernet switch and PHY, even though it grew year-over-year double digits, which was great, it came in lighter than we had thought.

So on the one hand, we're happy that it's still growing at a good rate through Q4.

On the other hand, it probably wasn't as high as we thought.

And again, some of that is due to the issues that I mentioned on supply chain tightening, China pull-in and some other factors.

So it was modestly below what we had guided.

But certainly, some of the business performed well if you look at it from a different point of view.

Got it.

And then Matt, on the same topic, I think you alluded to data center switching as well.

Obviously, a pretty large market, and you have the capability.

I'm just curious, in terms of your focus on that market, how should we think about it?

And both organic and potentially maybe through M&A?

If you can talk about how important that market is for you.

Sure.

So yes, we -- as I alluded to in my prepared remarks, and I don't want to steal the team's thunder too much because there's going to be a -- we're going to have a very strong presence next week at Open Compute, but we are going to be demonstrating and showcasing our high-end data center class switches.

The way to think of it is we've come at this from the angle that says we've got a very strong position in the enterprise market.

We've been feature-rich, strong operating system and software capability there.

Those customers, many of them, want to start migrating their way up into the data center applications, primarily on on-prem.

And we're able to actually migrate them upstream from that point of view.

So think of us as this product family is not going to be just feeds and speeds, streamlined, totally optimized like some of the other solutions that are out there just for hyper-scalers.

Think of this as being a much broader offering.

And we're very comfortable with our switch portfolio today.

We've got, I think -- we've got good technology all the way from SMB to enterprise to aggregation, all the way up to now, really, data center class switching.

And so next week, you'll be able to see some more information.

And hopefully, we'll talk about this product family in future quarters.

Our next question comes from Christopher Rolland of Susquehanna.

Just thinking back to the Analyst Day slide where you guys talked about your 4X base station content in 5G.

Can you guys help us think about what's contributing what there in terms of dollar value between OCTEON versus Fusion and versus networking?

Should we think of that as roughly 1/3, 1/3, 1/3?

Or is it (technical difficulty)?

Yes, without getting to -- so, Chris, you're breaking up a little bit, but I think the question was talk about how the content in our 5G platform is apportioned between the different technologies.

And the way you should just think about it is the bulk of this is in processors, and I would combine kind of Fusion and OCTEON together there.

The switch and PHY portion is meaningful, but it's much lower.

Within the processor business, we're not really in a position to break out exactly how much is from each portion.

And quite frankly, it's going to depend on also the different configurations that those ship in, what line cards they're on.

So it's not a real -- there's a lot of variability even within the content, so it's probably not -- just think of it as the bulk of it is going to be on the OCTEON and Fusion side.

Excellent.

And then, we haven't heard a lot about LiquidIO or smart NICs recently.

At one point, I know it was a 10% product for Cavium.

What's your outlook for the market there?

And is the problem that cloud guys have really migrated to doing it on their own?

Or is there still a market there?

Yes, I think it's a combination.

So, again, we've kind of -- it's certainly a product and a product line we have.

We're working on the next gen of it.

I think that market is clearly happening.

I think smart NIC offload is a real trend, and I think it's going to be a combination of merchant solutions as well as, I think, a bunch of it is also going to go ASIC.

It's not clear how that's all going to shake out.

So that's obviously a product line that's in our wheelhouse because we have the connectivity, we have the processor capability, so -- but I think that's one where we're going to continue to size our investment there relative to the opportunities and what our customers want us to build.

And certainly, one of the dynamics, as you mentioned, is, is the desire for some of the hyper-scale guys to really want their own very special customized ASICs for that application.

So how much will be merchant versus ASIC, I don't know.

But that's clearly a market trend that's out there, and we're paying attention to it.

And we can participate to the extent that our customers want us to.

Our last question comes from Timothy Arcuri of UBS.

So I'm still trying to understand what's going on within storage.

So I'm going by memory, but I think that the core storage business minus fiber channel was like $320 million last year, and which was, like, $100 million SSD and maybe $220 million HDD.

And it seems like it's down to maybe $200 million this -- for the guided quarter.

So it -- can you just maybe provide a little more disclosure than you normally would?

Because the business is just so bad, can you kind of help us understand exactly what's really happening year-over-year?

So Tim, this is Jean.

So I'll start, and Matt can add.

So when you think about it, we're really dealing with very broad slowdown in the overall storage market and the inventory condition, right, both -- across both HDD and SSD.

So that is the first thing is that's actually impacted both of our HDD and the SSD businesses.

So if you listen to our biggest customers, they all have seen significant issues and challenges.

And we continue to see more inventory in the channel.

And of course, the second thing, as Matt mentioned earlier, on the PC client side, HDD certainly declined more significantly, and the SSD side of our PC clients is certainly picking up quickly.

But that's the market that we have chosen not to participate.

So overall, you do see going through fiscal '20 before we really ramp up some of our customers' ASIC business and businesses and enterprises in the data center market, we're to see some headwinds in the overall storage market.

Yes.

And, Tim, I'd add, I think we've tried, candidly, to provide as much disclosure as possible given what we know.

And I think what's complicated -- well, so a couple of factors, right?

One is, we said it last quarter; I'll say it again this quarter, HDD and SSD both down almost uniformly across-the-board, every customer, every subsegment.

And that makes it hard to get into a lot of analysis because not only are all the different end markets going down, both segments are going down, there's all kinds of our reason codes as to why this is happening.

And so to parse through that and figure out, well, how much of it is because the cloud guys decelerated their spending; how much of it is because the leading microprocessor guy can't make enough processors to ship enough laptops; how much of it is because there was just a lot of inventory that was built up last year; et cetera.

So I think we'll have better visibility on all the moving pieces as we go through this process.

But right now, what we're communicating is both of those are down.

They're down significantly.

We -- and it looks like in the -- for the external third-party data points that we're able to get that we're undershipping the end market.

So that's as much as we can tell you right now.

We're just kind of reporting the numbers as they are and tell you what we see.

And we'll do that again next quarter, right?

But it certainly is -- it's been dramatic, right, in terms of the change and how fast, and I think it's been compounded by the fact that it's not just one particular end market or end customer or end dynamic.

It's got all the things I mentioned plus it's got an uncertain micro layered on top of it, in addition to some very specific things that are affecting the storage customers.

Got it.

Okay, Matt.

And then just the last thing from me.

So, just from a strategic perspective, obviously, the company's nowhere near as large as you thought it would be when you went into this merger, so does that sort of change the calculus in terms of how willing you are to support some of these very high development costs -- products like Thunder and things like that.

Just given the sheer size of the company at this point on an -- even on a combined basis, does it not make you begin to start to take a little different calculus in terms of what you're willing to fund and what you're not?

Yes, sure.

No, I don't -- it hasn't changed our calculus, I think for a couple of reasons.

One is because of our belief that the primary reason for the reduction in demand over the last 2 quarters was related to things that were outside of our control, i.e., macroeconomic conditions and excess of inventories across a wide range of customers and end markets for a wide range of reasons.

The way I'm thinking about it is, okay, when we work our way through this cycle, and when I look out to calendar '20, quite frankly, which is not that far away now, and we look at our growth drivers, in all the areas you mentioned, by the way, 5G, embedded processors, Ethernet, Thunder as well as, I think, storage coming back at that point, we have a very strong outlook for that year.

So the way -- when we (inaudible), right, and we sit down as a management team and we look at it, we aren't looking at it that Q1 was bad, so therefore, we should resize our R&D to fit the Q1, only to fit the Q1 envelope.

We're going to continue to run our -- to manage our expenses from an investment point of view relative to the future opportunity.

And so until something changes there, right, which says, hey, 5G is not going to happen for a long time or people don't -- ThunderX3 is going to take longer than we think, or some of these took the long-term view of the changes, then we're not going to adjust our investment.

That being said, as Jean mentioned, outside of the R&D envelope, we're going to tighten the belt.

And we've already done things to, as you saw -- I mean, even heading into this -- from our last call, we took out an additional $10 million of OpEx a quarter, roughly, on top of the already increased Cavium synergy target.

So we're going to manage expenses, especially from the discretionary point of view or things that we could push out a little bit that are not related to our investment profile.

But just to be super clear, we're going to size our R&D and our investment levels to the opportunity we see a year from now, let's call it, and just stay the course.

At this time, I'd like to turn the call back over to Ashish Saran for any closing remarks.

Sir?

Thank you, everyone, for joining us today.

And we look forward to talking to you again next quarter.

Thank you, and goodbye.

And that does conclude today's conference.

Thank you for your participation.

You may disconnect your lines at this time.

Have a wonderful day.