邁威爾科技 (MRVL) 2016 Q1 法說會逐字稿

Good day, ladies and gentlemen, and welcome to the first-quarter 2016 Marvell Technology Group earnings conference call.

My name is Derek, and I'll be your operator for today.

(Operator Instructions)

As a reminder, this conference is being recorded for replay purposes.

I would now like to turn the conference over to Mr. John Ahn, Director of Investor Relations.

You may proceed.

Thank you, Derek, and good afternoon, everyone.

Welcome to Marvell Technology Group's first quarter of FY16 earnings call.

With me on the call today are Sehat Sutardja, Marvell's Chairman and CEO; Weili Dai, Marvell's President; and Sukhi Nagesh, Marvell's Interim CFO.

We will all be available during the Q&A portion of the call today.

If you have not attained a copy of our current press release, it can be found at our Company website under the Investor Relations section at Marvell.com.

We have also posted a slide deck summarizing our first-quarter FY16 results in the IR section of our website for investors.

Additionally, this call is being recorded, and will be available for replay from our website.

Please be reminded that today's discussion will include forward-looking statements that involve risks and uncertainties, and could cause our results to differ materially from Management's current expectations.

The risks and uncertainties include our expectations about our overall business; our R&D investment, product and market strategy; statements about design wins and market acceptance of our products; statements about general trends in end markets we serve, including future growth opportunities; statements about market share; and statements regarding our financial outlook in the second quarter of FY16.

To fully understand the risks and uncertainties that may cause results to differ from our expectations and outlook, please refer to today's earnings release, our latest annual report on Form 10-K, and subsequent SEC filings, for a detailed description of our business and associated risks.

Please be reminded that all of our statements are made as of today, and Marvell undertakes no obligation to revise or update publicly any forward-looking statements.

During our call today, we will make reference to certain non-GAAP financial measures, which excludes the effects of stock-based compensation, amortization of acquired intangible assets, acquisition-related costs, restructuring costs, litigation settlements, and certain one-time expenses and benefits that are driven primarily by discrete events that management does not consider to be directly related to our core operating performance.

Pursuant to Regulation G, we have provided reconciliations of the non-GAAP financial measures to the most directly comparable GAAP measures in our first-quarter earnings press release, which has been furnished to the SEC on Form 8-K, and is available on our website in the Investor Relations section.

With that, I would now like to turn the call over to Sehat.

Thanks, John, and good afternoon everyone.

I'd like to start today's call by thanking Mike Rashkin, who is retiring as our CFO.

Through his leadership, Mike has been extremely instrumental in improving financial discipline in the Company.

All of us at Marvell would like to thank Mike for his many years of service and dedication to the Company, and wish him a wonderful retirement.

The Board has also appointed our VP of Finance and Investor Relations, Sukhi Nagesh, as Interim CFO.

Many of you already know Sukhi quite well, and we expect this to be a smooth transition.

Now, moving on to the financial results for the first quarter of FY16.

We reported first-quarter revenue of $724 million, a sequential decline of 16%, which was in line with our revised guidance.

The sequential decline was mainly due to softer than expected demand from the storage end market, as well as from the emerging markets.

Our non-GAAP gross margin of approximately 52% was better than our original guidance, and we continue to effectively control our operating expenses, which also came in better than our original guidance.

As a result, our non-GAAP EPS came in at $0.13 per share.

We believe the weaker demand in Q1 was due to near-term macro economic conditions, as many of our customers and semiconductor peers experienced similar trends.

We expect this slower demand environment to persist in Q2, but expect growth to resume in the second half of the year.

Despite the near-term headwinds, we will continue to invest in the new and innovative technologies, as we believe these will allow Marvell to emerge stronger when economic situations improve.

Now, I would like to provide a brief update on each of our end markets.

First, on storage.

A revenue decline, 20% sequentially, due to weaker than expected HDD sales and seasonally soft SSD demand.

In HDDs, our revenue declined on weaker than expected HDD TAM during the quarter.

Despite the weakness in HDDs, we believe we maintained our market share.

Next, in SSDs, Q1 is a seasonally soft quarter, and revenue declined sequentially as expected.

We continued to maintain our technology leadership and focused on developing industry-leading solutions such as DRAM-less NVMe SSD products, which we expect to do well in the market.

For Q2, we expect our storage end markets to decline sequentially, on continued end market weakness.

At this point, I would like to take a moment and discuss a breakthrough technology that we believe will significantly benefit the storage industry.

This technology is related to the FLC technology that you have heard me talking about in the last few months.

For those of you who have not heard about it, FLC stands for Final-Level Cache, a technology that developed to solve the main memory problem in computer systems.

It is a Big Data caching technology, and today I want to talk to about the adoption of FLC into storage devices.

More specifically, using FLC, we will now be able to make an HDD perform like an SSD.

Why is this important?

The HDD industry is at a critical stage, where it is dealing with the competitive forces at play from SSDs, especially at the mid and high end of the market.

The HDD industry knows that the best way to deal with the competitive trend of SSDs is to adopt hybrid drive technology.

Unfortunately, first-generation hybrid drives do not adequately address the SSD challenge head-on, and as a result, market acceptance has been lukewarm.

More specifically, the performance of these early hybrids is closer to HDDs rather than to SSDs.

This is because the SSD caching algorithm that is used on the early generations, on the first-generation hybrid, has the hit rate of only approximately 50%.

In other words, these early hybrid devices behave more like an SSD -- behave like an SSD half of the time, and behave like an HDD the other half of the time.

The problem is, in mobile applications, the user expects the storage device to be asleep when not in use.

This, unfortunately, creates sluggishness whenever the drive needs to be awakened during that 50% of the time.

This latency time is clearly not visible to the user, and therefore user experience is poor.

We have now solved this problem by using the breakthrough FLC technology.

Using FLC, every hybrid drive will soon be able to behave like an SSD about 99.9% of the time.

Basically, practically, for practical purposes, 100% of the time.

In spite of the usage of the terms, the hybrid is now practically an SSD.

Using this technology, we can now build a 1-terabyte consumer hybrid drive with FLC cache for only $40 of [build] material, compared to a 1-terabyte SSD which would otherwise cost $300 of build material.

Similarly, a 4-terabyte enterprise hybrid drive with a 128-gigabyte FLC cache could soon be built with approximately $200 of build material versus peer 4-terabyte enterprise SSDs, which today would easily cost upward of 20 times or more.

As you can see, it is not a matter of if, but a matter of when the HDD industry adopts our FLC technology to power all their hard drives.

We have engineering samples now, and we are actively promoting our FLC-based hybrid technology to our customers.

We expect first production to our customers early next year.

Now next, to our mobile and wireless end market.

Q1 is typically a slower period.

Specifically, in Q1 we saw weaker than normal demand last quarter due to the overall slowdown in China smartphone builds.

As a result, our mobile and wireless revenue declined 13% sequentially.

In mobile, China Unicom recently launched the world's first RMB399 LTE smartphone using our 64-bit quad core 1908 mobile platform -- the same mobile platform used by China Mobile for their TD LTE devices.

A similar 64-bit LTE platform has also started shipping into global markets with Tier 1 customers, such as Samsung.

Additionally, our turnkey program is on track to launch at the end of Q2.

This will expedite our partners' adoption of our chipset and allow them to go to market sooner.

Moving to connectivity, revenue was slightly below expectations, primarily due to weaker mobile shipments in China and seasonality in gaming consoles.

While in connectivity, early this year, we announced the industry's highest-performance 4x4 11ac Wave 2 product, and have still broad market interest in adopting Wave 2 products this year.

Wave 2 provides multiple simultaneous data links over Wi-Fi, and increases network capacity for densely populated environments, thus extending wireless capabilities to a variety of new use cases, such as realtime video streaming to multiple devices.

Wave 2 is the future of Wi-Fi, and is the next growth driver in wireless connectivity.

We will be announcing additional Wave 2 products over the coming months, as we refresh our key connectivity products to support the Wave 2 ecosystem, to address all tiers of the market.

In the meantime, our existing mainstream 11ac products continued to gain adoption at multiple customers.

For example, recently, Linksys announced another flagship 2x2 ac router.

Now, moving to the mobile computing space, we are shipping our industry-leading 2x2 ac combo into mobile applications running Windows, as well as Chrome OS.

We expect to see more mobile computing products on the shelves using our Wi-Fi solutions later this year.

Moving next to IoT, we continue to gain design wins for our connectivity and microcontroller solutions for IoT platforms at multiple Tier 1 OEMs.

For example, Xiaomi has already announced a series of smartphone products powered by Marvell's wireless microcontroller IoT platforms.

Over the next decade, it is estimated that billions of IoT products will be used by consumers and businesses.

For this to happen, we believe there will be a fundamental change in how these products are engineered to allow them to realize the full potential of being always connected.

Historically, the vast majority of embedded devices have used simple 4-bit and at most 8-bit microcontrollers, running rudimentary software, and often with no operating system at all.

In order to put these products to fully participate in the Internet, products will need to incorporate much more sophisticated software, which would require more powerful 32-bit microprocessors.

The end-user expectations of connected products creates a requirement for significantly more sophisticated software, which calls for building embedded software in a new way.

We are addressing the software challenge with Kinoma JS, an application framework uniquely designed to address the application software needs for the IoT, by providing embedded programmers with the modern high-level scripting language, JavaScript, to power their products.

Today, JavaScript already power the web pages, mobile applications and web service.

As one of the most popular and productive computer languages, we believe it is poised to power embedded products, thereby accelerating the growth of the IoT.

We recently released our Kinoma JS software under an open-source license to encourage customer adoption.

Using Kinoma JS, Marvell's customers can create high-performance products across a wide variety of hardware platforms using common code across multiple chipsets, thus removing the barrier to adoptions and increasing scalability of their designs.

Moving next to our multimedia business.

In Q1, we launched our next-generation multimedia SoCs, the ARMADA 1500 Ultra, which features a quad core CPU, eight core GPUs, carrier-class security and state-of-the-art power management.

This product is designed to enable pay TV operators and set top box manufacturers to cost-effectively migrate all of their customers to 4k video capability.

As you know, 4k is the future of video.

For Q2, we expect our mobile and wireless end market to be flat to up slightly.

Turning next to networking, demand came in weaker than expected, and revenues declined 7% sequentially, mainly driven by muted enterprise networking demand.

However, we continue to make progress in design wins with our networking associates in areas such as network storage interconnect applications, and we're gaining share at telecommunications and access infrastructure customers.

In addition, we have secured more customers for our 10GBase-T, which is our 10-gigabit ethernet copper PHY solution.

Last year, 10GBase-T deployment doubled compared to the year earlier, to a total of 3.3 million ports.

Over 8 million ports are predicted for this year by some analysts.

Every major switch OEM has introduced the 10GBase-T interface option, and we are the leading supplier for the 10GBase-T solutions.

For Q2, we are expect our networking business to be flat sequentially on continued muted enterprise spending.

In summary, despite the near-term market uncertainty, we continue to focus on execution, and believe we are well-positioned to return to growth in the second half of the year.

In addition, we believe this year will be an exciting transition period for Marvell, as we incorporate FLC technology in many areas, including storage, as I mentioned earlier.

On top of this major drive of implementing FLC, we are also implementing MoChi, our new modular chip technology, into all of our products and solutions.

Once our MoChi technology is fully deployed, we will be able to drastically lower product development costs, and improve time to market.

More importantly, MoChi will allow our customers to develop new products that we haven't even anticipated yet, since they will be able to create virtual System-on-Chips to their liking, using any combinations of our MoChi devices.

I will speak about our progress in MoChi and FLC in the months to come, so stay tuned.

With that, I would like to turn the call over to Sukhi to go over our first-quarter results and our second-quarter outlook.

Thank you, Sehat, and good afternoon, everyone.

I'd also like to add my thanks to Mike, and congratulate him on his retirement.

Mike has been a great mentor to me, and I hope to offer the same level of integrity and principles in my new role going forward.

Moving to our financials, as Sehat mentioned, our first-quarter financial results reflected a muted demand environment, which is consistent with many of our customers and semiconductor peers.

While revenues were in line with our revised guidance, gross margin was slightly better than our original expectation, mainly due to mix.

We reported revenues of $724 million for the first quarter, which was a decline of 16%, driven by softer demand trends across most end markets.

As Sehat said earlier, we believe the demand weakness is temporary, and we expect to return to growth in the second half of this year.

Moving to the details on our various end markets, our storage revenue in the first quarter declined 20% sequentially, and represented 48% of total revenue.

Pay TV sales were lower, due to the well-documented weakness in the PC value chain, while SoCs declined in line with seasonality.

Our mobile and wireless revenue declined 13% from Q4 and represented 25% of total revenue.

Weaker LTE smartphone demand in China was partially offset by initial shipments into a Korean OEM's global smartphone platform.

Connectivity sales were also weaker, due to lower mobile and gaming seasonality.

In networking, our Q1 revenue was softer, declining 7% sequentially, and making up 21% of total revenue.

This was mainly due to muted demand from enterprise customers.

Moving next to margins and expenses, our non-GAAP gross margin for the first quarter was approximately 52%, or roughly flat from Q4, but better than anticipated due to favorable mix.

Non-GAAP operating expenses came in at $307 million, better than expected, due to continued operating discipline across all of our businesses.

This resulted in non-GAAP operating margin of 9% for the quarter.

Net interest and other income was about $5 million, and we recognized a non-GAAP tax expense of $1.2 million in the quarter.

This resulted in non-GAAP net income for the first quarter of $71 million, or $0.13 per diluted share.

The shares used to compute diluted non-GAAP EPS during the first quarter were 535 million.

Cash flow from operations for the first quarter was $59 million, and free cash was $44 million, or approximately 6% of sales.

Summarizing Q1 results on a GAAP basis, we generated GAAP net income of $14 million or $0.03 per diluted share.

The difference between GAAP non-GAAP results during the first quarter was mainly due to stock-based compensation expense of $33 million, and approximately $24 million related to amortization and write off of intangible assets, legal, restructuring, and a one-time cash compensation payment.

Now turning to the balance sheet -- cash, cash equivalents, and short-term investments as of the end of the first quarter was approximately $2.5 billion, a decrease of about $30 million from the previous quarter.

We used $22 million to buy back roughly 1.4 million shares of stock during the quarter.

We currently still have $420 million remaining in our authorized repurchase program, and we will continue to be opportunistic in our buybacks.

We also paid dividends of $31 million for the quarter, or $0.06 per share.

Net inventory at the end of the first quarter was approximately $340 million, an increase of $30 million from the previous quarter, in anticipation of new customer programs that are launching over the next few quarters.

Moving next to our outlook for the second-quarter of FY16, we currently project revenues to be in the range of $710 million to $740 million.

At the midpoint, this would equate to roughly flat to Q1.

We expect the storage business to decline sequentially, our mobile and wireless business to be flat to up slightly, and our network business to be flat to Q1.

We currently project non-GAAP gross margin of 50% plus or minus 100 basis points, and anticipate non-GAAP operating expenses to be approximately $305 million plus or minus $10 million.

We anticipate R&D expenses of approximately $253 million, and SG&A expenses of approximately $52 million.

At the midpoint of our projected guidance, this should translate to a non-GAAP operating margin of about 8%, plus or minus 100 basis points.

The combination of interest and other income should net out to roughly $2 million, and we expect tax expense to be approximately $2 million.

We currently expect diluted share count to be approximately 539 million shares.

In total, we currently project non-GAAP EPS to be $0.11 per diluted share, plus or minus $0.01.

On the balance sheet, we currently expect to generate about $75 million in free cash flow during the quarter.

We anticipate our cash balance to be about $2.6 billion, excluding any M&A activity, share buybacks, or other one-time items.

We currently expect our GAAP EPS to be lower than our non-GAAP EPS by about $0.09 per share.

With that, I would like to turn the call over to the operator to begin the Q&A portion of our call.

Derek?

(Operator Instructions)

Harlan Sur, JPMorgan.

HDD industry shipment TAM was down about 10%, 11% sequentially in Q1.

I think your HDD controller business was probably down more like 25% sequentially in the first quarter.

Here in Q2, I think HDD industry TAM is looking to be down 3% to 5%, and now you're guiding your HDD segment probably down in about that range, as well.

So the Marvell team is essentially undershipping consumption by 20% to 25% for two quarters in a row.

It seems like your customers are planning for some positive seasonality in Q3.

If that plays out, should we anticipate a return to growth than your HDD business that is greater than the TAM growth?

Just given how much you are undershipping consumption here these past two quarters?

Harlan, this is Sukhi.

You bring up a good point, but I think our story is that overall, we have never broken down the mix between hard drive and SSDs.

All we can say is our SSD business was down more than HD business in the quarter, for a [myriad] of different reasons, I think which is pretty well aware, people know about that in the market.

But in terms of the hard drive business for us versus the TAM, we have some customers who are actually seeing the pullback in their business, and we believe that our TAM over a multiple quarter period tracks at the overall TAM, our business tracks that of the overall TAM.

It's very hard for us to synchronize exactly every quarter.

But, I think, over a couple quarter period, we are pretty similar to what the TAM is.

Okay.

You did bring up a good point that the hard drive, the PC industry value chains are expecting to see the deployments of [indoor TAMs] over the second half of this year.

So, as a result, that is why we know that, by this time, it's widely known that it's a slowdown in the PC space due to the anticipation.

We do expect Sukhi, when we took TAMs at this launch, it would have good reception.

We all know the benefits from that uptick.

Harlan, if there is an uptick in the drive industry for Q2 or Q3, maybe we will see that benefit.

At this point, it's probably too early for us to comment on that.

Okay.

That's a fair point.

A question for Sehat.

So last call when I asked, you said you'd clearly be doing what's right for shareholders in terms of strategic options for mobile.

It seems like the pricing environment hasn't been getting any better, the competitive environment continues to be fierce.

If you exit mobile, I think you're left with the business that can grow low mid single digits top line and sort of be throwing off low mid 20% operating margins and free cash flow margins.

Given that your mobile business was down in Q1, and that's the fourth consecutive quarter, are you any closer, or is the team any closer to any strategic decision with mobile?

That's a good question.

As you know, especially with the deployments of FLCs and MoChi's technology into all our product lines, we will be even more competitive in the mobile space.

So, we as a Company, we have to focus on developing the best technology to differentiate ourselves.

So, mobile is no different.

However, as we said earlier, we have to be responsible for the shareholders' inputs, so, we continue to be open to any strategic opportunities that comes in front of us.

In the meantime, we will continue to build even better, even more advanced technology to make it even more attractive for our customers to use our products.

Thanks, Sehat.

Tim Arcuri, Cowen and Company.

My first question, Sukhi, maybe you could talk about the CFO transition.

Why did this happen now?

Maybe, from a top level, Sehat, maybe you can address whether or not the CFO transition might change how you think about any of the possible strategic decisions you might make with the mobile and wireless business?

Thanks.

I'll answer the second part, none at all, not even a bit.

There is no linkage to that at all, Tim.

As far as the transition now, Mike's been with the Company for 16 years.

He's been a fantastic leader for all of us, and a great colleague for many people who have been here.

It was a personal decision for him to retire, and so there is nothing more to that.

Okay.

I'm wondering how you might handicap, if I look at your SSD business, it looks like it's down more than I would expect, unless there was a decision by one of the large notebook manufacturers to maybe begin to use their own controller.

They recently did buy a controller company several years ago, and it looks like they might be doing that in their flagship notebook, now using their own part.

My question is, how does that impact your business?

How do you handicap the likelihood that customer or any other customer might use their own controller over yours?

Thanks.

Sukhi, do you want to answer that?

Tim, I think you're probably on the right track in some of your assessment there.

But we also have a very strong portfolio of products and indices.

As Sehat mentioned, about NVMe DRAM-less NVMe SSD products, and the PCI based products especially going to be very critical for this market going forward, and we are engaged with multiple other customers for this detailed product, as well.

We do believe that the SSD market is growing.

Given that there are certain near-term product transitions [in a] certain customer, we should be able to move beyond that fairly quickly.

I'll also add that there's always going to be a customer that might want to use their own solution, especially if the customer there has critical mass.

However, the vast majority of the customers do not have such critical mass, and they will still use third party suppliers, like from us.

As far as from the discussion, in terms of our differentiation, how do we begin differentiate ourselves in the long run compared against the competition?

The answer will be similar to the HDD space.

I said earlier, we are now deploying hybrid technology into our HDD portfolios.

To make our HDD solution to be very, very powerful, as powerful as an SSD solution, it will help increase the adoption of HDDs into the market, to reverse the trend for converting the HDD to SSDs.

Now, at the same time, using the same technology, we can also build an enterprise SSD, that has the cost more like an HDD.

So, this technology is applicable for HDD, as well as for SSDs.

So, we believe that when the dust settles, we will have the majority of the market share in this area.

Okay, Sehat, thank you.

Quinn Bolton, Needham & Company.

I just want to follow up on Harlan's question.

Obviously, you are undershipping the TAM here in the near-term.

Can you make any comments as to whether you are seeing your HDD customers holding lower inventory levels?

If so, do you think that's a permanent reflection of greater supply chain efficiencies?

Or, to the extent that demand conditions come back as Harlan suggested, maybe allows for a snap back in the second half of the year?

A good question, Quinn, and entirely possible.

We do know that some customers do very tightly manage their inventory in Q1.

And, their supply and their production towards the end of Q1.

How long that continues, or it switches back on, we don't have entire visibility into that, at this point in time.

If they do start to switch that back on, we may see a positive benefit.

Okay.

A follow-on question on the FLC technology.

Obviously, it sounds like it probably adds some kind of price premium over standard HDD controller.

Can you give us any sense as FLC starts to ship next year, what you think the penetration might be, and to the extent that technology is accepted, how much faster on a revenue basis could you grow rather than the overall HDD TAM?

Yes.

FLC -- the cost header -- the silicon cost header of deploying FLC into the chip is actually quite insignificant, especially if we're talking about the storage market.

This technology, we've been developing this technology over close to three years and over the years, we have improved -- we have mastered the implementation of this technology, to the point where this is becoming very low cost.

We do believe, because of the delta difference between the hybrid solution is so small, basically, it's practically an addition of a single device plus chip to the hard drive.

Once this thing is proven into the field the adoption rate will be very rapid.

There will be nobody in the world that wants to buy a standard hard drive once they see the significant improvement in the IOPS and response times of this FLC-based hybrid drive.

As I said earlier, it's not a matter of if, it is just when.

And, the timing is -- really is just a schedule of the time to port the software into the new device, the sophistication, validation and sophistication of the device, and showing to their customers.

That's typically is the order of about nine months or so, nine months, maybe at most 12 months.

So this would be the timing that we anticipate.

After that, I think everybody will want to demand FLC hybrid technology.

In addition to what Sehat said, FLC technology benefits storage, we are seeing the last few months, it's really across a different markets.

For example, mobile.

There's a huge hurdle everybody knows the battery, the power, is a big issue.

FLC, absolutely is solving this issue as well as the cost, the memory cost, for the platform.

We believe our FLC technology is going to help companies and differentiate in multiple markets, so therefore, we'll gain more business and new design wins.

Did you want to say something?

Quinn, did you have a follow-up?

Just a quick follow-up with FLC versus the first generation hybrid drives.

Does it use the same size flash chip, or do you need a significantly larger flash chip to implement the FLC?

Very good question.

For the same size of flash chips, the improvement in the hit rate from 50% to 99.9%, actually doing benchmark, it's actually 100%.

We say 99.9% because it's hard to believe that we could achieve 100% hit rate.

To give an idea, there will be once in a while occasion where the algorithm will miss, and we'll have to access the HDD.

But, that event is so rare with our FLC technology, versus the traditional caching algorithm about 50%, sometimes slightly less, sometimes slightly better.

It hits the SSD.

But the other 50% finally hits the HDD.

During that time, typically, the hard drive is sleeping.

During that 50% time, it has to go to HDD, the drive has to spin up first.

That takes about one second, and the data will be accessed then, we have to put it down to sleep again and then the next access happens to be in SSD, you are fine.

If HDD again, then you have to wake the HDD again.

In our case, 99.9% of the time it's SSD.

Highly unlikely that the HDDs -- if it's not on the HDD, takes one second, that anybody will notice it.

Because it's a very rare event.

So it's the same sort of footprint?

For the flash, for the same flash size, if it's 8 gigabytes, no problem, we can still use the gigabytes.

4 gigabytes, no problem.

We can still use 4 gigabytes.

Of course, nobody's building 4 gigabytes, 8 gigabytes flash chips, any longer.

The smallest flash chip that we can buy is 16 gigabytes.

But we have run a lot of benchmarks, a lot of simulations, whether it's 8 gigabytes or 16 gigabytes.

The performance is exactly the same.

Well, 16 gigabytes slightly better, but for practical purposes, it's the same performance.

Great.

Thank you for that clarification.

Sanjay Chaurasia, Nomura.

Yes, Sehat, another question on FLC.

There are multiple reasons SSD is being adopted in computing platforms.

A couple of things, better power consumption by 3 to 4 times, lower weight by 10 times.

SSD's enabling thinner form factors such as ultrathin notebooks.

Even if else FLC allows you to bridge the gap in the performance with SSDs, how much of the competitive dynamics do you think you could change, that could slow the decline in hard disc drives?

If you look in the laptops, the laptop markets, the difference between the laptops with the hard drive versus a flash, the difference in the weight is quite insignificant.

The biggest challenge of HDD in mobile laptop applications, especially at the high-end, where people are willing to pay $1,000 or more for their laptops.

It's really instant on, this microsecond or millisecond latency that people expect from the SSD.

With our FLC, there is no way HDD could be put into sleep mode and be able to wake up in time for people to start to notice it.

So, the market acceptance, we expect the market acceptance of the hybrid drives to be very, very good.

Now, in terms of the weight of the HDD, the thinnest hard drives in the world are 5 millimeters.

The vast majority of people selling right now is actually 7-millimeter hard drive.

We have been working with our partners to develop 4-millimeter extremely thin, thinner than a battery, extremely thin hard drives, which is going to be a full hybrid drive.

If we look at it, if you see a sample of the prototypes we have on hand; it's so beautiful, so light.

You wouldn't notice any difference by [the thickness] of the laptop.

Okay, this is very helpful.

Another question on mobile.

Last quarter you indicated that your turnkey solution would be available by end of Q1, and I believe I heard correctly you said Q2 this time.

Just wondering, what was the delay, and if you could provide some progress in that area, and part B in that question, do you see our mobile and wireless growing year on year in FY16?

If you look, responding to the first question, the turnkey, so the turnkey is progressing.

It's a lot of work that needs to be done, building the prototypes.

When I look at the prototypes, I could understand why it is a little bit longer.

It's very complicated.

You have to build it nicer, so if it doesn't look nice nobody wants it.

You need to get all the software running, validated, certified.

A lot of certification work has been going on with the carrier, especially with China Mobile.

We have to buy new equipment, test equipment, a lot more test equipment to get the certification process going on with China Mobile.

Those things have been progressing quite well.

The delay is minor compared to the amount of work that has to be done.

Now in terms of the second question, so the volume, we do expect the volumes to continue to grow.

With the second half of this year, we will be introducing our FLC-based cell phone, smartphone chip.

That would be one that would be like a huge hit because for the very first time in the history of mankind, there will be -- anybody will be able to build a phone with such a small amount of main memory, while behaving like a very high-end phone.

So, we can be able to compete with flagship phones that have 4 gigabytes of main memory with fractions of the main memory.

That is going to be the defining time of our success into this business.

In addition to what Sehat said, the China Mobile -- as you know that China Unicom, Marvell also is the leading provider in driving mass-market LTE phones.

There's a lot of effort there, and we have multiple OEMs are coming out with their phones in the next few quarters, as well.

More specifically, Sanjay, we think Q1 was probably the bottom for our mobile, and we should probably start to see an improvement from the unit and revenue going forward.

This is a pretty good quarter itself, I think we should see the double-digit growth in unit volume and for our LTE business across both in China, and in Korea.

Chris Caso, Susquehanna.

With respect to your revenue, it looks like the first half is running on the order of about down 25% year on year.

Could you characterize how much of that you consider to be more due to, what I would say, are persistent factors, things like pricing and market share, as compared to transient factors with just the industry conditions and perhaps what's happening with inventory?

Your answer to that probably gives some insight into your level of conviction for growth in the second half.

As we said earlier, the two factors that you just mentioned, really the biggest factors, there people waiting for Windows 10 in production.

People are not buying replacement for PCs and whatnot, there is a reduction in people buying, waiting for the Windows 10.

The other part is the slowdowns in the smartphone shipments in China, and which was apparently -- was already quite well known in the last few months.

As you see, the pricing of mobile DRAM has dropped quite a bit over the last few months.

Only just like the last few days that people I've been talking about, but sometime the next month or so, is expected that the DRAM pricing will go back up to be above $3 for 4 gigabits for the mobile DRAM for the smartphones and because people expect smartphone shipments to pick up again in the second half.

So, we do believe the market will pick up again as well as our continued traction in the customer space, the customer side on our mobile solutions.

We do expect we will get the bigger shipments.

Now, our customers also, we're getting some new customers, even with today's solutions, while they're waiting for our FLC-based technology, because as soon as they see the significant advantage they have from our future efforts in smartphones, they actually immediately can jump in and make decision for people to have and use our solution.

Quickly, they start working on our solution, because they know that once they are familiar with our solution, they could easily make benefit of the next generation solution, which is significantly lower cost, lower power, much lower power, and much longer battery standby time.

We are very bullish on this.

Okay.

My follow-up question is regarding the FLC technology, as well, and a lot has been said on that already.

Would you also potentially see licensing opportunities there, in either businesses adjacent to what you do, or perhaps totally different?

Is that something you guys would entertain?

I publish the FLC technology, because I know it will benefit mankind.

I'm open to licensing the technology to anybody in the world.

If it's good for the world, it's good for us, as well.

Great.

Thank you.

Something we definitely would be considering.

Thank you.

Hans Mosesmann, Raymond James.

Sehat, just another one on FLC.

So, you're basically incorporating the DRAM and a low-cost SSD.

The DRAM, who provides that?

Are all vendors and suppliers capable of supplying them?

For the very first generation, we're using standard off-the-shelf DRAM.

However, the true benefits of FLC will come when network DRAM, the future DRAMs are built differently.

More specifically, future DRAMs will be built -- if you look at my presentation at the ISSCC, future DRAMs should be built for performance, latency, lower power in mind, than for increased capacity.

Those things actually are -- if you look at the generic industry-wide organizations for memory, if you look at generics, our proposals in the new memory architecture, as has been shown in the last few months, those exactly follow our proposal for modifying DRAM to be truly, truly, truly optimized to get the maximum benefits of FLC.

For now, we use standard DRAM.

As a follow-up, what is the incremental dollars to Marvell from a conventional approach or hybrid today to going to the first generation FLC?

Is a 10%?

50% of your existing silicon?

We don't talk about there.

It's too early to talk about that.

Way too early to talk about that.

It's more important for customers, our customers at this point -- our customers actually never even asked that question.

Because they know the benefits way outweigh whatever the cost is to be, they have to incur, to use this technology.

The benefits, not just the money they will save from the less DRAM they have to buy.

They're saving also less battery, smaller battery, lighter phone, lighter mobile devices.

Or, they can use the same battery, and have much longer battery life.

As I mentioned in my speech at the ISSCC, we cannot build -- down the road we see this technology, we can build a smart watch with weeks of standby time.

That's beyond how much money we save to build this device, it's like if you want to spend all the money in the world, you cannot do it today because your battery life is so bad.

The other benefits, they are hard to measure from a dollar point of view.

Thank you.

Craig Ellis, B. Riley.

Congrats, Sukhi.

I'll just follow up with the recent line of inquiry on FLC.

Sehat, you already have a very strong share position in your core storage market, but you said FLC would be much more broadly applicable.

Where do you think that technology would be most meaningful, in terms of helping Marvell gain share?

Which applications?

Is it mobile and within mobile, would it be equally for smartphone and tablet, or would it be in other applications?

The answer is across the board.

The benefit in the mobile, the laptop applications is clear.

Just make things like really, really low power, so the drive is sleeping 99.9 of the time, is purely sleeping, consuming zero power so the benefit is very obvious.

The non-obvious would be the enterprise.

If you have, if we have a petabyte of hard drive with a terabyte of SSD FLC cache, a terabyte of FLC cache caching like 10 or 20 or 40 terabytes of hard drive, the performance of their solutions will be incredible.

You will notice zero difference between 10 terabytes of SSD, pure SSD, versus our petabyte hard drive has one terabyte of SSD.

In fact, you will notice a difference in power because now, the power dissipations of the SSD will be 1/10 compared to (inaudible) 10 terabytes of SSD.

While costing is 1/10 of the cost.

So, the benefit is that the optimization, the target optimization, the benefit, it's the same.

For the high-end we just have to have high-end capacity FLC cache for the individual users like the consumers, we could actually live with smaller cache, and everybody will see the same benefit.

Got it.

Thank you.

The follow-up is for you, Sukhi and more towards the middle of the income statement.

In the revenue outlook, not much change, pretty flattish but there is a much more meaningful change in gross margin.

Is that just inter-segment mix and with OpEx now at the $305 million level, is that a level that is sustainable, or as the business grows, as the Company is signaling later this year, is that going to move back up?

Good question, Craig.

The gross margin front, it is a function of mix, obviously.

If you look at the end market guidance across the three different areas, that should give you a fairly good explanation of why our guidance for gross margin is where it is for Q2.

As far as the OpEx, even at that $305 million, we are constantly looking at ways of improving ourselves.

Reducing our spending, being more efficient and the answer is, no, you shouldn't see our OpEx going back up in the second half of this year.

Thanks, guys.

We have time for one final question.

Ian Ing, MKM Partners.

A clarification on mobile and wireless.

It looks like guiding flat to up slightly double-digit growth in unit volume.

Does that mean the ASP declines are more benign than in recent quarters?

And then, say optical or pricing is based on some Deca-Core announcements recently from competitors?

Thanks.

As you know, in the mobile space, pricing is always competitive.

We want to make sure what pricing is going to do at any given quarter.

We do have a connectivity business as well as a mobile business, so we do expect to see growth in our mobile business in this quarter.

From the point of the customers, and maybe in the connectivity side, maybe less growth, if you will, or flattish outlook.

It's probably not appropriate for us to comment on pricing, per se, and we did notice some of our competitive copy was, Deca-Core.

We're not entirely sure how the market will adapt to Deca-Core.

Pricing is always an issue.

We all, in this industry, building similar stuff.

The beauty about FLC is once it's deployed, pricing will be a lot less of an issue, because it's not going to be about the cost of the chip any longer, it's the cost of the system that matters, all of a sudden.

Okay.

We will look for FLC.

Just a quick follow-up, over to MoChi, this modular development of silicon.

Are you going to implement this as build-to-order parts for customers, or can you incorporate this into your standard ASSP products?

It seems that you connect up some smaller silicon die, you can get a cheaper solution than a big monolithic die?

Should mention as well as to my primary speaking, as well as to these earnings call statements, we were building, basically, anything that's reasonably -- anything that we build, almost anything that makes sense, almost anything that makes sense would be built using MoChi interfaces.

Meaning like practically almost every application processors will have MoChi interfaces.

We're also building MoChi devices, many different kinds.

R&D are busy building different types of this function, so that our customers -- once we're done with this, our customers don't have to specify what chips need to be built.

All they have to buy basically, is like an a la carte menu.

They will take one of the eight application processors, gives them increased function, gives them modems and they can build whatever they want, and they can decide how much money they want to spend, how much functionality they want to have, and it's all under their control.

They don't need certain functions, they can save money automatically by not buying those functionalities.

This is going to be -- it's going to be so helpful for us.

It will improve our decision-making process, it will improved time to market, it will get things done sooner, because chips can be a lot simpler from now on.

Thank you, Sehat.

I think that's it.

I'd like to thank everyone for their time today and continued interest in Marvell.

We look forward to begin with you in the coming months.

Thank you, and goodbye.

Ladies and gentlemen, that concludes today's conference.

We thank you for your participation.

You may now disconnect.

Have a great day.