邁威爾科技 (MRVL) 0 Q0 法說會逐字稿

Please stand by for real-time transcript.

Ladies and gentlemen thanks for standing by.

Welcome to the fourth quarter fiscal year 2002 financial results conference call.

During the presentation all participants will be in a listen-only mode.

After the presentation you will be invited to participate in the question and answer session.

This conference call is being recorded on Thursday, February 27.

I will now turn the call over to cochairman and Chief Executive Officer of Marvell Technology group, Dr. Sehat Sutardja.

Dr. Sutardja, you may now begin your conference.

Thank you, Tiffany.

Welcome, everyone, to our fourth quarter fiscal year 2003 conference call.

Weili Dai, Executive Vice President of the Communications Business Group and George Hervey Vice President of Finance and Chief Financial Officer joining me in this call.

Those who have followed Marvell for the last several years know that the fourth quarter investor conference call is an especially important one.

Not just because Q4 results finally put to rest the results for the previous fiscal year, but more so because this is the quarter that we will provide guidance for the new fiscal year 2004.

As usual, we are extremely excited about the new incoming fiscal year.

However, we also realize that we are not immune to the political and economic factors surrounding us.

Just last quarter we have thought that the long awaited economy recovery was finally with us.

With the many problems the world is facing, including lower consumer spending, lower investor confidence levels and the possibility of going into a major war, we are now seeing a further delay in the economic recovery.

Looking back, the last fiscal year -- the last fiscal year has been very challenging to all of us.

Had we known the market was going to be so challenging for so long last year, we would have provided a more conservative guidance.

We have been extremely fortunate that for each of the quarters of last year, we were able to meet or beat our own projections, which, by now, in hindsight, were very aggressive projections.

We owe both our dedicated employees and our strong partnerships with our customers for the continued success of Marvell.

In this conference call I will review our business progress during the last quarter.

I will also provide you with a general overview of where we are going in the next several years and give you the background for some of the initiatives we are taking to build an even more successful semiconductor company.

Before I go over the quarter period report, I would like George to provide our Safe Harbor Statement and review of our Q4 financials.

Thank you, Sehat.

Good afternoon, ladies and gentlemen.

I'd like to remind all participants that the following dialogue will contain predictions, estimates and other forward-looking statements covering subjects such as data storage and communications market trends, competition, customers, suppliers, products and demand, revenue growth, gross margin expectations, operating expenses, other income accounts receivable and inventory.

Such statements will be preceded by the words like expects, anticipates, believes, should, will, may or other words with similar import.

The following factors, among others, could cause actual results to differ materially from those described in the forward-looking statements.

They include the inability to further identify, develop and achieve success for new products services and technologies, increase competition and its effect on pricing, spending, third-party relationships and revenues, as well as the inability to establish and maintain relationships with commerce, advertising, marketing and technology providers.

We direct your attention to our annual report on form 10-K, recent quarterly reports on form 10-Q, recent current reports on forms 8 K and other Securities and Exchange Commission filings, all of which discuss other important risk factors that may affect our business, results of operations and financial condition.

Please be reminded that we undertake no obligation to revise or update publicly any forward-looking statements for any reason.

Now, moving onto the Q4 financials.

Today we report Thad net revenue for the fourth quarter of fiscal 2003 was a record $150.8 million, an increase of 82% over the 82.8 million reported for the comparable quarter in fiscal 2002, and a sequential increase of 11% from the third quarter of fiscal 2003.

Pro forma net income which excludes the acquisition related expenses, amortization of stock based competition and a charge related to the write-off of an intangible asset in fiscal 2003 was 20.3 million, or 16 cents per share diluted for the fourth quarter of fiscal 2003, compared with pro forma net income of 7.4 million, or 6 cents per share diluted for the fourth quarter of fiscal 2002.

Shares used in computing pro forma earnings for the fourth quarter of fiscal 2003 were 129.3 million compared to 131 million shares for the fourth quarter of fiscal 2002.

Net revenue for the year ended February 1, 2003, was 505.3 million, an increase of 75% over the net revenue of 288.8 million for the year ended February 2, 2002.

Pro forma net income was 62.5 million, 48 cents per share, compared with pro forma net income of 18.7 million, or 15 cents per share for the prior year.

Now I'd like to turn the call back to Sehat for comments on our business outlook.

Thank you, George.

Despite the continued challenges facing our industry, Q4 was a good quarter for us.

Sales of our products continue to do well.

This is not surprising because a lot of our products are key to the success of our customers' own products.

Furthermore, despite some of the flatness that we have seen in certain areas of the PC market, we have been able to continue to grow thanks in part because of market share gains in our respected markets.

In terms of design wins, Q4 was no different than the other previous quarters.

Our design win pipelines continue to be full with many new design wins across all market segments.

Many of these design wins will start to contribute to our earnings several quarters down the road and some even a year or more down the road.

So despite the fact that we may not precisely pinpoint when some of those new revenues will materialize, we are confident that when they do -- they will significantly contribute to our future growth.

A key example are the design wins for the Prestera family of gigabit and 10 gigabit processors, fabric adapters and switch fabrics.

The Prestera design win pipeline has been in the process of filling for almost a year.

Yet only recently have we started to see product announcements based on the Prestera family from a major customer, Nortel with the new passport 1600 family of gigabit switching products.

Over the next several quarters, we hope to be able to disclose to you many more exciting products that our customers will introduce, utilizing the Prestera chipsets.

With the vast majority of the enterprise class of gigabit switching suppliers choosing the Prestera solution, we believe we will enjoy significant revenue growth as the enterprise infrastructure market migrates to gigabit Ethernet.

With a yearly opportunity of more than 200 million ports of connect enterprise switching to be replaced by gigabit Ethernet, the opportunity for Prestera solutions translates to several billion dollars per year.

This was the business reason behind our pending acquisition of Radlan.

As Prestera solution addressed, the enterprise class of switching higher layer software management functions become extremely critical to our customers.

Our experiences in the last several years supporting the early design wins of Prestera provided us with the technical justification behind this acquisition.

Since the announcement, we have gone out and discussed our plans with our customers.

Not only have we received strong approval from our customers, but more importantly, they, in turn, have given us even more commitment to our Prestera product line.

As they can trust us more to deliver complete hardware and software solutions to them.

With the software capability in house, we now believe there is not a single semiconductor company out there that can come close to us.

Now, moving to wireless LAN, as you're aware, we entered the 80211 B market last year as an underdog in the wireless LAN space.

With several major design wins just going into production, our customers can now truly see the advantage of our solutions versus the existing products in the market.

Just as we have done in the gigabit space, we are now coming on strong to displace the early players with better products, period.

With our latest 80211 G product announcements early this week, we now believe we have an even better position in the marketplace.

While we came later with the 11B we are right in line with the early players in the 11G.

We believe as soon as the 11G standard is approved in May of this year, we should have one of the very first products that can truly claim full compliance to the expected 11G standard.

Since this latest announcement and also several networking suppliers are already working with Marvell's 80211 B solutions, we have been receiving significant customer interest for both our 11B and 11G solutions.

Both of these solutions will continue to get design wins into fair different market segments.

The 11B will go into ultra low cost extensive consumer applications whereas the 11G will go into higher priced PCs, laptops as well as the Alpine video on-demand applications.

Either way, both have the potential to generate huge revenue opportunities as more of these devices go into cell phones and consumer audio and video products.

We believe the opportunity can go as high as several billion dollars in the next several -- in the next five years.

Now, next moving to storage.

We continue to execute our plan of providing storage electronics not just for the existing markets, but also for the emerging consumer markets.

For example, some of our customers are already sampling storage devices for the automotive market.

In a year or so most of the outer GPS systems are expected to be HDD based to give the user a much better user interface response time.

As digital video is becoming commonplace, a personal digital video recorder will be a standard feature in cable and satellite setup boxes.

Another example, video game machines are already incorporating an HDD device with more entertainment products expected to do the same.

With so many future consumer opportunities for storage, we continue to strengthen our enduring force and investments.

We are preparing to provide whatever our customers need for the enterprise server market, the PC market, the laptop market, or the futuristic cell phone market.

Now I'd like to go over some of the initiatives and changes that we would like to make in the next several years.

Looking back, the last two years of industry downturn did not deter us from the investing in the new technology areas.

This new fiscal year is no different.

Our experience has been that the success we've had in the past was always a result of the investments that we had made many years earlier.

While these investments greatly affects our near-term bottom line, we're confident that this is the right thing to do for the long run.

We will continue to invest organically and in select areas to require technology that will allow us to continue our historical growth for the years to come.

As we continue to grow our businesses, we're finding out that we are moving into new territory.

We have found that many of the technologies that we have developed have wide applications well beyond our original target market.

It is becoming increasingly difficult to call our products just storage and communications.

The term storage and communications are so often used that they become meaningless or at least confusing to many of our investors.

Because the devices that we built have found applications in many market segments, we feel that over time, it would be more meaningful for our investors to track the market segments that we play in.

These market segments include the PC market, which is, of course, highly mature, the business infrastructure market, which is undergoing a major change to gigabit connectivity, the consumer electronics market, which is driven by technology advancement as well as trendsetter factors which, by the way, is probably the biggest market opportunity for semiconductors, and the industrial electronics market, which is a big market by itself.

And definitely it's a market that we can use many of the technologies that we have developed for the other markets.

In the PC market, our success will be driven mainly from building better products as well as more integration, thus taking market share away from existing semiconductor suppliers.

Our products for this market include device or target storage electronics, gigabit mix, wireless LANs and other yet to be developed and introduced electronic components.

Our initiative in these areas is to leverage our mix signal and high-speed technology to serve the new requirements of the next generation PCs.

With PC express and seal connectivity as well as wireless becoming the standard features of PCs in the year 2004, we would like to play a more important role in the PC market.

In the consumer market, our success will be driven mainly from building new products that will make life more interesting, enjoyable and better for the end consumers.

Ubiquitous wireless connectivity and distributors for these devices will be important components for this market, yet most end users would not even know how these technologies work in their gadgets.

All they care about is reliability and low cost.

This is a huge market yet to be tapped by us.

As trendsetter factors play an important role in the growth of this market, this market will somewhat isolate us in the future from the ups and downs of the economic cycles of the PC business.

Our initiative in this area is to leverage many of our technologies into single chip components that can be manufactured at very low cost to spur new demand in the consumer market.

We should look at this market as a longer-term strategy as many of the products may take time to build and gain acceptance in the market.

Yet we see that the market is huge.

And we should never underestimate the spending power and disposable income of the end consumers.

In the business infrastructure market, our factors will be driven mainly by the wholesale change in the infrastructure connectivity standards as gigabit networking rapidly becomes the standard for business users, new installations of networking infrastructures will quickly move to gigabit standards.

As key factor, a key factor for this to happen is, of course, the cost of gigabit versus the decade old 100 mega bit solutions.

This is what Prestera is all about.

Our four years of investment in solutions are winning practically all the design opportunities out there.

This is a huge opportunity for us as the market is two to three times larger in port-com compared to the business PC, gigabit and LAN market.

The revenue opportunity for the infrastructure sites will definitely be significantly larger than that of the PC client site.

This was the driving factor as we proactively renegotiated our agreement with Intel to provide us the opportunity to bring our Yukon gigabit LAN product to the entire PC client market to further accelerate the deployment of gigabit to the desktop.

While we feel that we will eventually share the PC desktop client market with Intel, our continued partnership with Intel across all of PC client platforms as well as the strong position of Yukon in the market, we expect to see continued revenue growth from this segment.

The faster we can move the PC client market to gigabit, the sooner we can enjoy the several billion dollars of revenue opportunity from the network infrastructure.

Our initiative in this area is very straightforward, continue to develop the next generation higher end chipsets on one hand and to develop lower cost chipsets to address the entry-level market on the other hand.

Finally, our reserve the discussions on the initiative for the industrial market until next year when we are ready to talk more about it.

While we already the technology leader in many of the above markets, our goal for the next few years is to translate this technology leadership into market share leadership as well.

Even with the current products that we built for these market segments, a majority of market share in each one of those markets can translate into multibillion dollars of revenue potential to Marvell.

With the technological progress that we will provide to this market, we are hopeful that we will be able to achieve the Moore's law version of financial growth into our revenue targets.

I would like to turn the call back to George for additional comments regarding our financials and guidance for the next quarter and guidance for the new fiscal year.

Thank you, Sehat.

First I'd like to make some additional comments on our Q4 results.

Our Q4 revenue of 150.8 million was a new record for the company and an 82% increase from Q4 of last year.

Additionally, the 11% increase in revenue from Q3 to Q4 compares favorably to our guidance of a 10% sequential quarterly increase in revenue.

We also achieved a major milestone by crossing the $500 million revenue level for fiscal '03.

During Q4, both our storage and communication businesses performed well.

Resulting in increasing revenue for both businesses.

In Q4 the main growth drivers for our revenue increase remained to continue increase in adoption of gigabit Ethernet by the PC client market and increases in production ramps of our SOC solutions in the desktop storage segment.

During the quarter, we also saw an increase in gigabit Ethernet shipments for the networking infrastructure, system controllers for wide area network applications and early production shipments for our newest desktop storage opportunity.

For the quarter, storage products contributed slightly more than 50% of total revenue with communications products remitting the balance.

For Q4 our gross margin of 53.4% was consistent with our guidance of 53.5% plus or minus 25 basis points.

Product mix, which is the largest determiner of gross margin percentage, was consistent with our expectations.

Our manufacturing efficiency and benefits from our yield improvement programs continue to produce very cost-effective products, which is very important as high-volume PC client shipments continue to represent an increasing percentage of our revenue mix.

Operating expenses as a percentage of revenue in Q4 were 39.2% of the client down 100 basis points from Q3.

In Q4, we increased the rate of our R and D investments for the future and initiated development of more products on .13 micron technology.

We continue to control our other operating expenses and Q4 was the sixth consecutive quarter that operating expenses have declined as a percentage of revenue, and coupled with the solid gross margin performance, we have increased our operating income percentage over the same six quarters from 2% to 14.2%.

We are now less than 4% from our long-term model for operating income of 18%.

As we mentioned last year, we said it as an objective for fiscal 2003 to significantly improve our profitability versus fiscal 2002.

Our results for 2003 reflect a significant improvement in both pro forma net income and EPS.

Our pro forma net income for fiscal 2003 was 62.5 million, compared to 18.7 million for the last fiscal year, which is an increase of 234%.

EPS for fiscal 2003 was 48 cents per share versus 15 cents per share for last year.

Our Q4 balance sheet improved in several key areas.

Our cash and short-term investments increased by approximately $9 million to 265 million.

As a result of revenue on several new programs starting during the quarter, the linearity of our revenue in Q4 was slightly more in the second half of the quarter resulting in DSOs for Q4 increasing to 51 days from 48 days in Q3.

These number of days were in the range of our guidance of high 40s to low 50 days.

DSO are likely to remain at the high end of our range during these periods of rapid revenue growth.

During Q4 we reduced our inventory by approximately $9 million and our days of inventory from 70 days to 51 days.

Since we are fabless, our inventory dollars can fluctuate based on the amount of wafers at our foundries, which don't appear on the balance sheet until the completion of the wafer fab process.

During the quarter we increased our wafer starts on several products which were still in fab at quarter end.

These products, when completed and added to our inventory would put our days of inventory at 65 to 70 days, which is our long-term objective.

We'll continue to monitor our production levels going forward with the goal of putting us in the most favorable position to respond to increases in demand for our products.

Now turning to the future.

As we enter our fiscal year 2004 we'd like to take this opportunity to provide some guidance regarding the financial outlook for the company for fiscal 2004.

As we mentioned during the past year, the adoption of gigabit Ethernet by the PC client market and the continuing trend to SOCsfrom standalone rechannels in the storage market, provided the company with two significant revenue opportunities.

Our success in executing on those opportunities was reflected in our 75% year over year revenue growth.

While it's difficult to predict how overall economic conditions will impact fiscal '04, we remain optimistic.

Entering fiscal '04, we believe our strong product portfolio has increased the number of major revenue opportunities, and we are looking forward to another year of significant growth.

In our storage business, we expect our new desktop opportunity to reach volume production as we progress through the year.

And serial ATA to ramp more significantly in the second half.

In our communications business, our strong design win position with the Alaska FI and Prestera switching products should benefit us as gigabit Ethernet is more widely deployed in the network infrastructure.

Based on customer feedback and design win success, we believe that our Yukon gigabit Ethernet LAN and motherboard for the PC client market and our wireless LAN 802911 B and G products will provide us with significant revenue opportunities during this next year.

Based on all of these factors, we expect our revenue for fiscal '04 to range between 710 million to 740 million, which at the mid point is approximately a 44% growth from fiscal '03.

This level of revenue growth should allow us to continue making progress towards achieving our long-term operating income model of 18%.

We expect to continue the strong manufacturing performance we exhibited during the past year as well as our level of investment in our business to provide business and an organization able to sustain growth over a longer period of time.

For fiscal '04, we expect our pro forma EPS to range between 80 to 85 cents per share, which at the mid point is approximately a 70% growth from fiscal '03.

Now, moving more specifically to Q1, '04, we added Q1 with a significant momentum -- with significant momentum generated by our strong market position, broad product portfolio and increasing number of design wins for those products.

Historically, the first calendar quarter of the year experiences varying levels of seasonality, when compared to the fourth calendar quarter.

Seasonality is most evidence in the PC market and we expect some moderate level of seasonality in Q1 but still expect Q1 to be a good quarter.

We've targeted revenue to increase 7 to 9% from Q4.

Visibility into the quarter is good, as backlog for both of our businesses entering Q1, '04, remains at levels consistent with the past several quarters.

For Q1, we anticipate the product mix of shipments to our end markets to be consistent with Q4.

And normal ASP declines to be offset by continued reduction in our manufacturing costs.

This should result in Q1 gross margin of 53.5% plus or minus 25 basis points.

As we accelerate our new product investments, move more of our products to .13 micron and structure the company for continuing long-term growth, operating expenses in absolute dollars will increase in Q1 but decline 10 to 15 basis points as a percentage of revenue from Q4.

During these times of reduced interest rates, the return on our cash assets has decreased.

We anticipate Q1 interest and other income to decrease $400,000 from Q4.

In summary, looking at the various factors in our business, we anticipate Q1 EPS to be 17 cents per share, an increase of 1 cent per share from Q4.

Now I'd like to turn the call back to Sehat.

Thank you, George.

That completes our commentary, Tiffany.

Would you please poll for questions?

Certainly.

If you would like to ask a question during this time, simply press star, then the number 1 on your telephone keypad If you would like to withdraw your question, press star then the number 2 on your telephone keypad.

Our first question comes from Michael Masdia.

I'm sorry , Michael, will you repress star and then the number 1?

Thank you.

And Michael is from Credit Suisse First Boston.

Please, sir, go ahead with your question.

Yeah, thanks.

Congratulations on the quarter, guys, and great guidance.

You talked a little about Yukon contributing a little this year.

You didn't talk as much about growth this year being driven by the client side.

What are your expectations for client in terms of contribution this year, and what are your expectations from the Intel business, if you can give us any sort of granularity there.

Well, the client market is still expanding.

I mean, the penetration rate, you know, grew each quarter during the last year.

And our expectations here is that that will continue through each quarter this year.

The -- right now we're expecting Yukon is in production, and we have already begun action to ship that product to customers.

And are expecting an increasing revenue level, you know, for that product, you know, success increasing revenue levels as we go through the year.

As we mentioned to many of you over the past several months, relative to our Intel business, we have taken a very conservative view of that in our guidance, and, you know, we feel that, you know, we're in a very good position there.

Our partnership remains strong.

We're working on new versions of the chips with them.

As we indicated in -- as Sehat indicated, we ultimately will share the desktop part of the market with them, but we expect our Intel revenue this year to be higher than it was last year.

Great.

And on the technical side of the client side, you know, Intel's come out with a proprietary CSA bus.

How important is it to have a product.

And number two, how does PCIX change that, if at all?

Okay.

I'll take that one.

The interface is one of the things that change, you know, once every ten years or so.

If you look at the history, the interface had changed from the ISA bus to VISA bus ( to PCI bus.

And the next change, actually, will actually be the 3 GIO which is the PCI express.

PCI express is expected to be introduced sometime at the end of this year.

With deployment sometime next year.

So the CSA is the temporary interface standard which is proprietary to Intel.

I don't know what the market will -- how the market will accept a proprietary standard, especially the 3 GIO standard that's coming into the market in the next year and a half.

So I do believe that based on the historical data that I know of, that based on experience that I know of, the PCI solutions, which is the Yukon base, will be there for -- until we get it replaced with 3 GIO fully for probably near 2005 with initial -- the PCIX -- PC excess or 3 GIO sometimes called starting in middle of next year or so.

So it's probably a minor platforms to move to a proprietary standards.

Yeah.

With our customers that are engagement, we haven't run into any issue with addressing Marvell right now with the PCI for the desktop moving forward for the future PCI express.

So this really is the mainstream.

Okay.

But the answer to the PCIX, because you also asked about that.

PCIX is the standard -- it was -- it is the standard that was introduced several years ago and PCIX is already here for the server side of the market.

This is for the higher -- high-end work station and servers.

So we have bought five or so products on the PCIX side.

That market will stay with PCIX, again, until the 3 GIO is here to replace that.

Right.

On the infrastructure side you talked about Prestera starting to ramp some initially.

How does that ramp look throughout the year on your guidance especially as you get towards the high end of the guidance and customer side.

Are OEM bringing prices down enough on the equipment to start the infrastructure to be deployed?

We have already seen some movement in that area.

We believe there needs to be more to really accelerate it.

I think our experience on the client side, you know, proved how dramatically the market can move when the price points are appropriate.

You know, our guidance overall is a conservative guidance at this point, as we pointed out.

You know, a lot of the drivers that we have are all very, very exciting.

But they're all in the early, early parts of their ramps.

So you know, we need to get a little more visibility as we move through the year, and we expect, you know, Prestera, like the other growth drivers that I mentioned, to gain momentum every quarter moving through the year.

So right now it's at a modest level, you know, but a good, you know, a solid contributor to our you know, to the number.

But clearly has, you know, the potential of being a bigger contributor moving forward.

Great.

Thanks, guys.

Our next question comes from Jim Lange of Pacific growth Equities.

Hi.

A couple questions.

First of all, George, can you talk about the sequential growth rates for the storage communications and then gig-e business separately?

Jim, you know, I think as we've been kind of alerting you guys, too, there's so many moving parts of our business now and things are overlapping so much, especially with the new opportunities that we're not going to provide that specific, you know, breakdown any further.

We will, you know, the guidance is one guidance for the whole company for the quarter.

You know, we'll give commentary as we have about the strong, you know, market trends and, as I indicated, you know, there was growth across five or six areas in Q4.

But I think that's going to be the level of detail that we're going to go to going forward.

All right.

So going forward, you're not going to maybe even use a new classification standard to give revenue mix?

No.

We're going to stay in the storage.

When we report, as we did, you know here, we said, you know, storage revenue was slightly above 50% and com was the balance.

We'll continue with that.

We ultimately want to move to the end market breakout because we think that is the best way to measure us.

But we have got to work on that more before we get there.

Gotcha.

Next question, Sehat, can you talk a little bit more on the serial ATA side of the business?

Give us some more color as far as the opportunity for you guys this year, those in the PC applications and also in the enterprise storage applications?

Okay, Jim.

The serial ATA is actually moving quite -- quite strong.

In talking to all the drive guys this is one interface that they're moving to the fastest in the history.

If you look at the IDF, the chipset where serial ATA is coming, which is -- which we are glad to see, so serial ATA is going to be a very -- it's a very important technology for our SOCs as well as for other new products that we haven't talked about that someday down the road we can talk more about.

Also for the enterprise side you're asking about for what is the enterprise side, which would also be a very important technology for building, for example, near-line storage systems or video on demand servers where they can use low-cost desktop class or probably ruggedized desktop drives for the enterprise laptops of applications.

So we built many, many chips to spur the demand in the markets, but the biggest market opportunity for us would be for the target sites on SOCs as well as some markets for the HBA for the enterprise sites which were to introduce a product about a half a year or go or so which is getting a huge, huge interest from the -- from our customers.

Great.

Just last question, on the wireless LAN side, do you have any plans or if so any timing about introducing a combo solution?

The 11 G solution is a combo B and G solution.

Oh, it is?

Yes.

All right.

And this is a very important -- is a very important requirement because 100% today of the infrastructure today of 11B, so we do not want existing infrastructure to be, but our customers don't want it to either.

The industry would not want to see that either.

All right.

Thank you very much.

Thank you.

Our next question comes from Arnab Chanda from Lehman Brothers.

Hi.

Thanks, guys.

A couple of questions.

First of all, if you could talk a little bit about, you know,

going back to the Yukon question

again, around what time frame should we, given the way things

are going, see, you know, maybe

a revenue crossover versus your

Alaska to Intel?

The other question is that if you can talk about, if you think PCI express is going to be adopted,

you know, at the end of this year, does the CSA bust kind of go away as we go to next year?

And I do actually have a

follow-up.

Thank you.

I'll take the first part on

the Yukon ramp.

Arnab, again, we have very high

expectations for that product.

You know, it's initially been

targeted into the white box as we've told you before into the white box part of the market and

we've been very successful with a number of key design wins

there, which as I said already, we have started to ship, you know, the early, you know, ramps on those products.

You know, we have a very, very strong business with Intel, you know, with the Intel Marvell

partnership.

So it is going to take a few quarters for the -- and as we

said, the market is continuing to increase.

And, you know, we expect to get a lot of that market through Yukon.

We have a very solid business with our partner, Intel, which isn't changing.

So you know, it may still be a

few quarters before Yukon

ultimately, you know, goes past

that revenue.

But eventually it will.

Was the other question what

PCIS is to CSA ?

Yeah, with the PCI express,

PCI express is expected to be

introduced obviously over the next several years, starting at

the end of this year or so and

there will be more and more

platforms or chipsets or

chipsets or adoptive cards, you

know, many other functions that need to -- they'll be

introducing the PCI express.

So my feeling is, again, based on my experience in the past,

the industry will move when there's a standard that

everybody supports.

So I haven't seen anything otherwise.

So I cannot predict specifics -- I haven't seen in the past

specific proprietary standards

take off and win the

marketplace.

But, you know, I cannot predict,

that will be not the case or not.

But I don't expect it to be the case.

The other part is PCI express or 3 GIO, whatever you want to call it, is based on serial

connectivity.

Serial connectivity has a huge -- I mean long life

migration path.

The first generation of PCI

express went on a 2.5 gigabit

per connection.

I mean, I'm going to express that, per connection, per wire.

We already have chips that we

can demo to run at 6 gigabits per second per wire.

So it's not -- you should expect

basically in the next several

years PCI express to migrate to

PCI express by 2, 2 X, meaning 5

gigabit and then on the road to 10 gigabit per second per wire.

So it's going through the PCI express is the right thing to do

from future migration point of

view.

So I do expect that PCI express will be the de facto standard for high-speed connectivity as well as the USB two for lower speed connectivity.

And the rest will be on the sideline.

Thank you.

Appreciate the explanation.

And then one last question about the storage market.

You were talking, you know, a

little bit -- give us a sense of

the consumer drives and the opportunity there.

Maybe you could talk a little about the products specifically.

Are you thinking of integrating a PVR with your SOC type projects or thinking of coming

out with new platforms for the consumer drive markets because I don't think there's a

distinction between consumer drives or desktop drives right now, maybe if you could

shed some light on that.

Thank you.

Sure.

How things went to materialize, but the more important thing is

that every one of these consumer

devices will incorporate

storage.

Storage is something that nobody

can get away from.

There's no way to get around

storage devices.

So where they use standalone

SOCs or they will integrate other functions down the road, it depends on customer per customer cases.

There will be several customers who will want storage

functionalities with maybe lower

power or lower costs or with different interface for

consumers.

A lot of consumer electronics companies are already

standardizing on serial ATA.

So there's a standard committees

being formed in Japan to move serial ATA to consumer

electronics or integrating other functions including, you know, videos or

audios.

But, again, customer to customer specific.

Either way we don't really care because we have -- we can do --

we can do it one way or the other.

Thank you very much.

Your next question comes from Jeremy Bunting of Thomas Weisel.

Thanks very much.

First question is on the

Prestera products.

Which kind of markets do you expect that to play into,

high-end markets or low-end

markets?

And I'm thinking in terms of how high performance

switch fabrics currently differentiate themselves through software and through performance

on custom platforms.

Do you expect to take business

There?

Sure.

I'll have Weili answer this and then maybe I'd add later.

Thank you.

This is highly scaleable.

It covers all the way from metra enterprise desktops all the

way down to our managed platform.

So it is a complete solution.

Okay.

Thank you.

Regarding the switch fabrics,

the Prestera designs can be used with or without the fabrics on

the very high of end systems.

It can use fabrics, they can --

that can scale the system to

modified terra bits.

You have medium fabrics that can scale through half a terra bits, for example, which is I say

medium.

It's high-end today.

But when I say medium, maybe five years from now you call it

medium because it's half a bit, it's not a toy.

Or you can have chips without the

fabrics because they are

standalone systems.

So we have introduced fabrics, for example, up to 72 ports for

connections at 3.125 gigabit per second throughput on a single chip.

So modified chips -- modified

chip versions of the fabrics can be put together to build bigger

fabrics.

Okay.

That's very helpful.

On the serial ATA space, do you

expect a production ramp to kick in Q1 or Q2?

We've actually been shipping through ATA since Q3.

Okay.

Of last year.

We do anticipate that the rate of serial ATA adoption will be

stronger in the second half of this year.

So you know, it will be up

sequentially in Q1 versus Q4 and

Q2 versus Q1.

But the larger inflexion point will probably happen in the

second half of the year.

And especially as more and more people find out there's a serial ATA that should give you

better performance on the PC

benchmark as well as more

importantly as the next Intel

platform chipsets getting to

productions.

Spring Dale has native serial

ATA in it.

Thanks very much.

Your next question comes from Charlie Glavine of Think Equity Partners.

Thanks.

Can you guys briefly go over who the top five customers who are exiting the year both for the

quarter as well as the year?

And were there any material ships on that first?

Yeah, Charlie, we only go through the people who represent more than 10%.

Right.

Of our revenue.

And now there are only two, with

the rapid growth that we've

experienced , especially in the

communications business, we are spreading, you know, the revenue out across a lot larger number of customers.

But we still do have two greater than 10% customers and they are Intel and Samsung, and I would

say they remain on a percentage basis of our revenue consistent

to where they've been over the last couple quarters.

So Intel was I believe just over slightly 20% last quarter, so that would be the case.

Going forward, though, George, as a percentage of revenue, even with the additional revs, you

mentioned at the time of the agreement with Intel, you not only have new designs but early designs that still had the ramp with Intel, as a percentage of

revenue, do you expect both Sam

sung and Intel to remain fairly

constant , particularly since you have the newest westernage

products ramping up?

I'm not sure I totally understood that.

First of all, do you expect Intel to stay roughly in that?

No, we would expect, as we grow, that, you know, Intel will actually come down as a

percentage, stay very healthy in absolute dollars.

I don't want to imply that it's

not healthy in absolute dollars.

But, you know, in our 710 to 740

million for the year, that's a pretty, you know, substantial growth rate still going forward.

So therefore I would assume most likely Intel would come down and

realistically probably Samsung

would come down a little, although they still have some room for growth in that business

but, you know, there are other key things like, you know the

WD ramp that could generate them as a customer later this year.

Ok, sorry for belaboring the issue, but within the CSA not so much the bus and the interim solution before Intel gets to 90 nanometers when Intel

chipsets start to get to a 90

nanometer and can't

integrate the gigabit solutions absent the CSA, do you find that

the lack of a chipset technology

will either force you to resort

to doing joint ventures with

other chipset vendors or is there another solution that you guys are looking at when

eventually the control and the

trends, assuming Intel from Hyfa

and Organ can get it to work would be able to integrate that and, again, that would be the 90

nanometers next year.

Let me answer the question in a funny way.

If you give me a chipset technology today, I could tell you I will never integrate it.

It will take me a half an hour to explain why it would not make sense.

But you could easily ask the

motherboard guys why that doesn't make sense.

So but without going into all

the technical reasons, it is not the right thing to do, and it

will never work.

It didn't even work with fast Ethernet and you could integrate fast Ethernet today because they have the technology 5, 6, 7, 8 years ago, why didn't

they integrate it in the chipset?

So integrating a 1.5

million gigs of DSP and mix signal through the chipset will screw up your plan for your chipset.

But in terms of other applications, our work, we also work with every chipset companies in the business on our

gigabit with Yukon because they all see there's a need to have

gigabit connectivity for the

desktops.

So we're not -- we're not doing anything funny or anything that any of the companies should have done and that expect.

So we're very excited that the

Yukon has actually been getting very rapid adoption acceptance into the market.

Sehat, if I could, though,

within that, it's not saying that that would be the only option, Intel certainly made a

mistake within the graphics within the 815 when they did not allow a third party similar to what happened with the map code

processors, allowing a third

party, if that were the case within integrations who removes the CSA or opens up the PCI express for allowing third parties and

generally a better solution, is that the approach for the second half of calendar 2004 when Intel looks like they're going to integrate this into the

chipset that that would be the other approach, allow Intel to waste the square silicon and you would come up with the Yukon as the better performance to allow

motherboards vendors?

No, again, it would be impossible for me to explain why it wouldn't make sense, without

spending half an hour explaining why it doesn't make sense.

Why integration doesn't make

sense.

Even though I control the

chipset, because we would not do it.

If I -- the issue is not integration.

Integration is not the reason

that people -- putting

everything in one chip is not the reason why people do integration.

There are other reasons, if it's not because of performance or because of board layout issues or

whatever, you have other things you have to worry about.

I'm not going to be concerned --

Okay-- you should not be concerned that somebody will integrate gigabit five to the chipset or to the

CPU.

A microprocessor is different animal.

A microprocessor has to sit next to a CPU otherwise you cannot move 120 gets of data in and out at one gigahertz.

So it's a totally different reason to integrate.

Got it.

Thanks.

And I will take you up on that and take it offline.

Thanks.

I'll show you -- pick any motherboard.

Figure out how to put the

gigabit into the chipset.

Your next question comes from Cody Acree of Legg

Mason.

Thanks, guys.

Maybe I can follow up on that

just for a second.

Could you talk some about

maybe some of the partners that you've

already been working with the

current SpringDale architecture,

assuming that this is not going to be exclusively an Intel supply only relationship that you'll be working with as well with this architecture ?

Well, we have to be somewhat, you know, limited, Cody, in exactly, you know, what we can say.

But I think we'd be comfortable in saying that when we

introduced Yukon in October, the initial target market was the

white box market for that product.

And the white box market is looking to do spring Dale designs you know, once Spring Dale is available to the market here.

And we have a certain number of those where Yukon is going to be you know, the gigabit chip

interfacing to the SpringDale chipset.

To, you know , we can't be specific as to which customers those are yet , but it -- you won't have to wait too long.

Okay.

Western Digital has recently made some fairly aggressive statements about their plans for migration to some larger drives.

Can you talk about how much

expectations you've built into that '04 guidance for WD's ramp and just to give us some assumptions on to what kind of

dependence we're looking at for the drive site?

Well, first of all, we're very excited about W-D's plans to move to -- or move the majority if not all of their production ultimately to the 80

gigabyte flat form -- platforms because those are the platforms which we're working very closely with them on.

Their recent commentary, you know, and we need to be careful here.

We can't speak for WD and to how fast they're going to ramp.

That's their, you know, position to make those statements and then, of course, we'll respond to those.

But we have not, in our numbers,

done as aggressive in our

forecast as what is, you know, potentially being talked about for the ramp of that product.

So I think it's the appropriate tactic to take at the moment because you still have to see

the products be introduced and ramped in the market.

They have a great track record for that and they have a great

track record but I don't think

it's appropriate for to us get

out ahead of that curve.

But as soon as it becomes more,

I guess, real or stabilized,

whatever term you want to use,

we will be, you know, taking -- making the appropriate

adjustments to reflect that.

One last question.

Beyond WD, we haven't talked

much about IBM in a while, but

with Hitachi taking over here in January, can you talk about any

developments there that may have come about in the last few weeks?

Sure, I'll take that one.

So Hitachi has always been a competitor for the last four or

five years been a very close partner to Marvell.

So they're a very close customer as well as a partner.

So the acquisition of IBM by

Hitachi actually is a very good

opportunity for us down the road.

So we've been engaged with

obviously both sides, but we're not making any projections when

the IBM side will be able to

eventually move to the current solution that they're using to the solutions that we provide

to -- as our standard products, for example.

So it's probably a more longer

term opportunity, but we are

working -- our engineers are working closely with the engineering teams at IBM.

Both in the U.S. and from Japan.

So we're to the point where virtually we're in a very good

relationship with the engineering teams now across the board.

But that's as much as I can say

about that site.

Great.

Thanks, guys, and congrats on a good quarter.

Thank you.

Thank you.

Your next question comes from David Wu of Wedbush Morgan.

Yes, good afternoon.

I was wondering, could you comment on two things that you

mentioned a couple quarters ago?

The first one was that, you know

historic relationship between the lead and the lag on gigabit

Ethernet client and giga to Ethernet switches, when do you see, based on history, are we going to get to the sweet spot of the Prestera kind of a ramp

inflexion point, so to speak?

The second one is with

increasing 802911 B and G and

hopefully combo offering in the

second half, how would you characterize the kind of gross

margin in that business?

Because I guess it's not a surprise that a lot of B players right now,

not as many G players, but lots and lots of B players.

And including people from Taiwan

that don't seem to care much about gross margins?

Could you

hold 50% gross margin in the -- in that segment in the second half of the year?

I think maybe the last

part -- I'll take the last part first.

The -- every large businesses

will attract -- any large opportunity, this is opportunity will attract large number of

players.

Just because someone making a lower margin, that doesn't

necessarily mean we're making a

lower margin.

It all boils down to who can build a better product, number one.

Number two, who can build

smaller silicon die size and

other things, robustness you built, the wireless LAN that only works in for 20 meters

distance versus hundred, 150 meters distance.

There are so many factors there

that depend on margin and pricing.

And we believe, just like in

storage, where cost is an important part.

We never win or lose business because of costs.

It's really because performance.

This is no difference than any

critical functions that you

need.

It's a life line function.

If you want to deliver video

over your wireless LANs, you want it to work no matter what.

If it rains, it doesn't work,

that's not good.

All those things, okay, will

determine that our market share.

That's a long answer to that.

Let me maybe another comment on the gross margin part of it.

You know, the same technology

that produces the tremendous

products produces the very

cost-effective solutions, really

really low cost.

And I'm sure you've all heard us

say before, when we design a product, we target it, and this

is not going to be any different for the consumer space, when it

has to be in high volume production at the most

competitive cost point that it's going to have to get to, we still expect it to generate 50% gross margin.

And, you know, you've already

seen some examples over the past year as we've gotten more heavily into the PC market, you know, that our margins have come down a little, but all of a sudden they started to stabilize again because we are being very

cost-effective.

So I think we've a track record of being able to produce these

kinds of products, you know, to the targets that we set for ourselves.

It doesn't mean it's not easy.

It's very -- I'm not an engineer

but I know the engineers work very hard at achieving this.

But it is something that's

doable.

It is our business model.

And we think we'll do as well

as -- well as, you know, we move

into probably late this year

into next year with those kind of cost points are going to be out there.

So I'll address the Prestera question.

If you look at the gigabit

technology in the last one year from the client space, now

motherboard is going to convert

from fast Ethernet to gigabit

Ethernet.

And we believe, by the end of year, the corporate and business

PCs will use gigabit for the motherboard.

Now, from the infrastructure side in terms of technology for

the gigabit for the switching side is here today.

I think the challenge is the

cost per port.

And we believe, with the

Prestera family, the chipsets

and we can address that very

efficiently.

So based on the majority of corporate PC converter gigabit by the end of year, and we believe starting the end of this year moving to the first half of

next year the gigabit to the

infrastructure side will be very high volume.

So essentially a second half

calendar '03 kind of an inflexion point ?

Yeah.

Starting, yeah, towards the

end of year end and then, you know, running to a very healthy volume by first half of next year.

Thank you.

Your next question comes from Aalok K. Shah from

Pacific Crest.

Just a couple of questions.

Would you speak up?

We can't here.

Can you hear me now?

Yep.

Great.

On the gross margin side, can you go into a little discussion on how you think ASPs are

trending?

You indicated gross margins are going to be at the level of Q4.

I'm curious to see where you're

seeing the strength, if you're starting to shift towards more like the wireless LANs and some of the new products.

Right.

Well, first of all, we're in the

semiconductor business which,

you know, just in general has

decreasing prices over time.

So that is just something we have to deal with all the time.

We have seen, in our business, no extreme, you know, pricing

issues to deal with at this point.

And what we've been able to do, as I said earlier, is that the

normal price declines, which are

there every quarter, because of our very, you know, outstanding operations group and our designs our yields are very good.

We're able to reduce our costs.

So to the extent that we are experiencing, which we do, price

declines from a quarter to

quarter basis, we are able to

maintain the margin profile because we have our costs going down on that same curve.

And that's -- you know, that's really a big difference, I think for Marvell versus some other

companies because we have been able to do that.

And, you know, we've kind of, you know, for Q1, we kept our gross margin guidance essentially at the same level as we did in Q4, and I think that's an indication of how well we're doing from a manufacturing standpoint.

So primarily this is related to manufacturing or is this also a mix issue as we go forward?

I'd say in my earlier comments I said we expect the mix of revenue in Q1 to be consistent with the revenue mix in Q4.

Okay.

Great.

Thanks a lot.

No problem.

Our next question comes from Alex Gauna of UBS.

Thank you.

I was wondering if you could

clarify my understanding.

The Prestera is going to be in the market this quarter, but we won't start seeing its ramp until the July quarter, is that correct?

That would probably be, you know, earlier -- the earliest probably more the October quarter, Alex.

In terms of high volume.

Higher volume, yeah.

I mean, revenue starts, I

think -- let me think.

Nortel said they were going to

ship the platform 1600 in April, which, you know, which is the last month of our quarter which implies we have to -- we're

already shipping something to them to support that.

So that's the start.

And then other customers who haven't announced yet but as Sehat

mentioned, you will see that coming over the next period

here.

We'll just add to that as we

move forward.

But, again, this is the early start.

So these things take, you know,

a number of quarters to

ultimately hit their true, you know, volume potential.

Okay.

So does your Dell win that begins in the October quarter

and are there other Marquis names besides that that are start not guilty that time frame?

Oh, good try, Alex.

No.

We don't, you know, we'll wait for those customers to announce their products.

But we're very excited about a large number of name customers who have agreed to want to use our products and design

products.

So just stay tuned and you'll hear about them.

Okay.

And you talked about the price points that you're enabling.

Could you give us a range of

where that is as we get to the summer?

In terms of -- price per port.

Price per port.

It's -- it probably is not good for us to say how much our customers are going to price

their gigabits.

But my guess is it will be

comparable to the fast ethernet.

By the summer.

Whatever they've been pricing.

They've been pricing, so there's no single price that fits for everybody.

I don't think by summer, but

probably later than that.

Okay.

And you mentioned that, you know at some point in the future Intel will eventually get some LAN and motherboard wins with the gigabit Ethernet.

Can you give us an idea what you see in terms of that timing?

Can it happen in 2003 from your knowledge of what they're doing,

or is it more in 2004 effect ?

Yeah, Alex, I think that would be a good question for

Intel.

Yeah, no comments on that one.

Yeah.

Okay.

Well, I do ask them and they never seem to deliver on the timetables they always tell me, so I was wondering what your thoughts were.

And in terms of the new breakout that you're mentioning, the PC business infrastructure consumer and industrial, are you willing to give any color on what you expect those splits to be over the course of this year?

We haven't gotten -- we're still working that out.

We're going to -- for the time being -- stick with our two

breakouts that we currently have.

But that's where we wanting to.

So you know, we'll get there as

quickly as we can.

It does take a little bit of work on our part.

I'm sorry, Sehat.

No, that's what I was about to say.

Yeah.

Okay.

And so maybe the next conference

call we'll get those new splits, if I'm correct on that.

Yeah, we'd like to try to get there quickly.

Last question, in terms of you mentioned chips decline over time.

Where are we on the motherboard

side, or the client side including the cards in terms of

ASPs today.

What might we see over the course of the year, and if

you're willing to hazard a guess for the whole industry, not yourselves, but how many gig-E

client ports do you think we can ship in 2003.

Regarding price, in this

competitive market, I'd rather

not disclose pricings because,

number one, just given -- the most important thing they just give us a competitor an idea of how we're pricing things.

But we'll be very competitive.

We'll say that.

We'll stay competitive, and it will be good business for us.

Now, in terms of the number of

ports, we believe, this is our

belief.

Also, is what our customers saying, that the business PCs will go almost to gigabit.

By the end of the year.

By the end of the year.

The home PCs will stick with

wireless, with wireless LAN.

I mean, not stay, will go to

wireless LAN because they are not -- they don't have wireless LAN yet, they're the home PC.

The laptops are already going to

wireless, so laptops will have some of them will have gigabit plus wireless, some of them will

have fast Internet with wireless so they have to have both for

laptops, for any foreseeable

future.

So one way or the other, okay, it's

going to be big business for the

laptops or for business PCs.

But there's a migration path here.

Can I throw out a number?

Does

35 million client gigabit

Ethernet shipments make sense

for this year, or is that way too low?

There's been a lot of numbers

thrown around out there, so I don't think -- I guess the best way we can answer that is we certainly are not at that rate

yet.

But we expect further increases as we go through the year.

And if you put that number in

context of a total population,

that's not an unreasonable number.

But think we need to see, you know, the continued adoption both by ourselves, you know, with Yukon and then our

partnership with Intel.

So that's a number Alex we

probably should give more color to as we go through the year.

Okay.

Thank you.

And congratulations on the quarter.

Thank you.

Our next question comes from Clark Westmont of Salomon Smith Barney.

Hi, guys.

I'm wondering on desktops I believe you're starting off with PR Nell and then blending into

SSE.

Can you give us some idea of the shape of that and when do you expect to get fully over to SSE, and I have a follow-up.

Can you repeat it again?

Okay.

So for desktops, with Western Dig, especially, you're

starting off with a rechannel, I

believe, and then you're shifting over to SSE.

Can you give us an idea of how

you think that might transition just in general terms?

Oh, I see.

I see.

Yeah, in general, in general, all our customers in the past usually used our standalone device, which is the rechannel.

The rechannel happened to be the most complex device to anyone,

and it's built in a mixed signal,

very sophisticated high speed detailed analog circuit, digital circuits.

So people usually want to take

one step at a time.

So but eventually the SOC direction is the way to go because of other things we can do for integration, we can

improve performance, add more functionality and so on.

So the -- we expect that by next

year we will not sell standalone

rechannels in the desktop.

That's our projection.

Every chip that we sell will be

SOCs for the desktops.

And is that first half next

year or second half or just leave it?

It could be as early as first

half, but definitely by second

half.

Okay.

And follow-up, what -- do you have 100% market share at

the 80 gig platform at Western

Dig?

This is product --

Again, we don't want to get ahead of WD's introductions.

On the products that WD has introduced, yes, those are 100% ours.

And -- okay.

And but you don't want to make

any --

We can't speak for them.

Yeah , okay.

Okay.

We know the answer, Clark, but you have to wait.

I understand.

Okay.

Thanks a lot.

All right.

At this time there are no further questions.

Do you have any closing remarks?

Tiffany, thank you.

This completes our Q4 fiscal

2003 conference call.

I would like to thank every one of you for joining us and look forward to updating you next

quarter.

Thank you.

Thank you again for participating in this afternoon's conference.

You may now disconnect.