1
Filed by Marvell Technology Group Ltd.
Filed pursuant to Rule 425 under the Securities Act of 1933
Subject Company: Galileo Technology Ltd.
Commission File No.: 333-50206
The following document was posted on the website of StreetFusion.com
on January 12, 2001 with a link from the website of Marvell Technology Group
Ltd. ("Marvell"). The document is an edited transcript of remarks made by Dr.
Sehat Sutardja, President and CEO of Marvell, at an investor presentation on
January 9, 2001. The remarks by Dr. Sutardja included a discussion of potential
product offerings as a result of the proposed merger of a subsidiary of Marvell
with Galileo Technology Ltd. ("Galileo"). An unedited transcript of the investor
presentation was previously posted on the website of StreetFusion.com on January
10, 2001 and filed under Rule 425 of the Securities Act of 1933, as amended, on
January 11, 2001.
================================================================================
The following edited transcript discusses Marvell Technology Group Ltd.
("Marvell"), its products and its industry, which includes a discussion of
potential product offerings as a result of the proposed merger of a Marvell
subsidiary with Galileo Technology Ltd. ("Galileo"). Investors and security
holders are advised to read the joint proxy statement/prospectus filed with the
Securities and Exchange Commission on December 12, 2000, and mailed to
shareholders of Marvell and Galileo on or about December 14, 2000. Security
holders may obtain a free copy of the proxy statement/prospectus and other
related documents filed by Marvell and Galileo at the SEC's website at
www.sec.gov or at the SEC's public reference room located at 450 Fifth Street,
NW, Washington D.C. 20549 or at one of the SEC's other public reference rooms
in New York, New York and Chicago, Illinois. Please call the SEC at
1-800-SEC-0330 for further information on the public reference rooms. The proxy
statement/prospectus may also be obtained free of charge by contacting Marvell,
Attention: Denise Franklin, Director of Investor Relations, 645 Almanor Ave.,
Sunnyvale, CA 94085, (408) 222-2551.
The following presentation may contain forward-looking statements within the
meaning of the "safe harbor" provisions of the Private Securities Litigation
Reform Act of 1995. These statements are based on the current expectations or
beliefs of Marvell and Galileo's management and are subject to a number of
factors and uncertainties that could cause actual results to differ materially
from those described in the forward-looking statements. The forward-looking
statements address the strategic business combination of Marvell and Galileo and
the future performance of Marvell. The following factors, among others, could
cause actual results to differ materially from those described in the
forward-looking statements; the risk that the Marvell and Galileo businesses
will not be integrated successfully; costs related to the business combination;
failure of the Marvell or Galileo shareholders to approve the business
combination; inability to obtain or meet conditions imposed for governmental
approvals for the merger, inability to further identify, develop and achieve
success for new products, services and technologies; increased 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.
MARVELL INVESTOR DAY
JANUARY 9TH, 2001
GEORGE HERVEY: First of all, good morning. I'm George Hervey. I know many of
you. I'm the CFO of Marvell Technology, and we'd like to welcome all of you here
to the first Marvell Investor Day. We hope this to be the first of many of these
types of events that we can put on for our investors, and have an ability and an
opportunity to tell you, and update you periodically with our, you know, with
the progress that we're making, and the strategy that we're looking to report.
So, what I'd like to do first is introduce a number of the Senior Management
from both Marvell and Galileo that are here this morning. First of all, to my
left here is Dr. Sehat Sutardja. He is the President, CEO, Chairman of the Board
of Marvell. Sitting over here at this table to the right is Weili Dai. She's the
Executive Vice-President of Marvell, and also the General Manager for the Data
2
Communications, or Communications part, group of Marvell. To her left is Manuel
Alba. Manuel is the President of Galileo Technology, and we're happy to have him
here with us this morning. Also in the back running the Web cast is Denise
Franklin. She is the Director of IR for Marvell. I'm sure many of you have
talked to her, and she was very instrumental in getting this day set up for us,
so we appreciate that very much.
As usual before we begin, I'll do the obligatory Safe Harbor, so please
bear with me for a moment. I'd like to remind all participants that the
following presentation may contain forward-looking statements within the meaning
of the Safe Harbor Provisions of the private securities Litigation Reform Act of
1995. These statements are based on the current expectations or beliefs of
Marvell and Galileo's management, and are subject to a number of factors and
uncertainties that could cause actual results to differ materially from those
described in the forward-looking statements. For other factors that could cause
the combined company's results to vary from expectations, please see the risk
factors section of Marvell's Joint Proxy Prospectus on form S4 relating to the
proposed business combination of Marvell and Galileo, Marvell's quarterly report
on form 10Q for the quarter ended October 31st, 2000, and Galileo's most recent
report on form 20F filed with the SEC. So with that, I'd like to turn the
presentation and the meeting over to Dr. Sutardja.
SEHAT SUTARDJA: Thank you, George. Good morning, ladies and gentlemen. I'd
like to welcome you to our first Investor Day meeting. Today what I'd like to do
is to provide to you with an update of our company, Marvell, as a combined
entity with Galileo in terms of our vision and our progress. Our vision is to be
the leading provider of high-performance, extreme broadband communications
technology. Together, we have developed true world-class technologies that are
applicable to many types of communication markets. Together, actually today, we
have combined about 700 employees, with more than 500 engineers working on
nothing than other communication types of technology, all the way from the
physical layer, mixing of DSP, all the way to the system levels, very complex
switching products. Now, I would like to talk about some of the businesses that
currently we are in, specifically the two businesses that we have at Marvell and
Galileo. The first one is our data communication business, and we'll follow on
with our data storage business.
Okay. Microsoft is very slow. Please bear with me. All right, I will start
with the communications aspect of our business. This is the business that we
have the most common synergy between Marvell and Galileo. As you are aware, the
in the communications space currently we are seeing a huge explosion on the
build-out of communications bandwidth and infrastructures, and the reason for
that is because of the huge demand that we're seeing in the market. This is the
same thing of what we have seen in the past several years, where the
availability of faster processors results in the introductions of new software
applications, which in turn results in the increased demand of faster
processors. In the communications we are seeing, the initial availability of
bandwidth to the end users results in new types of communications
applications,including the introductions of the infamous Internet types of
applications that you know of. And as a result today, we are seeing a huge
demand in increasing bandwidth for the access to the end users as well as the
core infrastructures of the Internet. Now obviously, there are many, many
challenges that we face in this area, specifically the system providers', the
equipment providers' challenge to build systems that can support much, much
higher data rates. While previously supporting maybe hundreds of megabits to
tens of gigabits types of solutions, now we are talking about devices that can
support, hundreds of gigabits storage, several terabits of types of throughput,
and at the same time, be able to build systems to interface with the different
types of communication infrastructure all the way from the local area networks
all the way to the core of the Internet. Now, the challenge for system IC
providers like us is to build integrated circuits that we can provide to our
customers so they can build their systems. So together, Marvell's and Galileo's
divisions will provide many of the important components that are needed into the
systems.
...
Now, what we seeing moving forward is that we're seeing the opportunities
to build silicon that can address the various types of the network
infrastructures, specifically as the industries are moving towards building
devices that can support gigabits and tens of gigabits connections, by building
silicons such as the physical
2
3
layer device, as well as the switching core products that can be used across the
board. While in the previous generations networks, where different types of
silicons are used in the different markets, we believe by providing silicons,
such as silicons that can interface with copper wiring as well as fiber optics
wirings, we can enable the convergence of the network sooner, and ergo also be
able to, you know, leverage the investments that we have made into one product
into the other products.
...
As you look at the build-out of the communication infrastructures, what we
are seeing today is that the emergence of new equipment being built that
supports gigabits-per-second types of throughput of connections, as well as the
demand for the next generation 10-gigabit per second types of interfaces. We
currently have developed silicon solutions for the gigabit, and we'll talk a bit
more about our development effort on the 10-gigabit space, both on the physical
layer as well as on the switching fabrics through the Galileo divisions. Now,
the silicon contents that are needed related to this build-out are tremendous --
approximately about more that $10 billion of silicon opportunity in about 3 to 4
years' timeframe. We believe that Marvell has a winning recipe for gaining
significant market shares out of this silicon opportunity, due to the facts,
first of all, the leading performance of silicons that we provide are in this
business. Specifically, we build devices that deliver the best performance in
terms of being able to build devices that can operate over longest distance,
over, you know, more robust communications links, over various types of media,
silicons that can deal with the imperfections that we've seen in the real life,
as well as to the, with the acquisition with Galileo, expertise we now have
several, many, many years of expertise in building very complex systems that can
support, you know, many hundreds of these interfaces into a single system. And
through the years of our investments in building high performance integrated
circuits, we also now have the expertise, the leading edge expertise in building
a true single system-on-a-chip types solutions that can incorporate analog,
digital, DSP and controller functions into a single piece of silicon. And we are
able to build this very complex product in a very short time, both in terms of
initial introductions to the market as well as be able to bring these products
into the market very quickly -- into production very quickly.
As a combined company now, we build many, in fact, most of the critical
components that are needed into the communications systems. What I've shown here
is the typical slides, the broad diagrams of what you see in a high-end
communications system such as LAN switching equipment or WAN router. Some of the
blocks that we have built, shown here in yellow color, and I will talk a little
bit more about these products in the next few slides.
The physical layer is the first one that I want to talk about. This is
probably the single-most critical functions in any communications systems. The
physical layer challenge today is to build devices that can perform at high data
rates, and as the data rates increase exponentially over the last several years,
the complexity of building such a product also increases exponentially -- in
particular, the challenge of building a very complex high-bandwidth analog
circuitry and complex digital signal processing technology into a single piece
of silicon. Today, we already have developed two types of technology to address
this type of the physical layer market. The first one is the copper interface;
the other one is the fiber interface. On the gigabit market segments, we have
introduced several product families, a generation of gigabit Ethernet PHY that
we call the Alaska gigabit. Starting from the single, quickly followed with a
dual in the middle of last year, and more recently, we introduced our quad
device gigabit Ethernet, as well as we're the only company in the industry that
can build such a device that incorporates both copper and fiber interface in the
same piece of silicon.
Today, I would like to use this opportunity to introduce our latest
generations of gigabit Ethernet transceivers. This is our Quad+ Alaska
transceiver. This is the only quad PHY, quad transceiver device available in the
industry. It is the only quad PHY interface that has both copper and fiber
optics. It is the most advanced device available in the market, it is the most
robust, it is the only device that can work over not just the standard category
five cable, but also the old, the older category three cables that are widely
3
4
deployed in the 10-megabit base. So now, we allow our customers to build devices
that can support gigabit data rate throughput while using the same
infrastructure that was originally developed to support 10-megabit data rates.
It is the lowest power device, delivering approximately about one watt per port
compared to our competitions in the markets, shipping today about 7 watts moving
to, towards 4 watts per port. So, a significantly low-powered solutions in the
market. This is our second-generation device -- .15-micron device geometry with
.13 microns gates devices. It is the smallest die size, as well as it's the
smallest footprint device in the market today, and we're only currently sampling
the device into our alpha customers.
Now, the next areas of expertise that are important, that are needed to
address the communications space, are the packet processing and the switching
fabric. Our expertise is derived from many, many years of experience at building
complex systems that are scalable to support, you know, many tens of gigabits,
and moving to, towards hundreds of gigabits type solutions in the future. Our
expertise is building devices that can support full wire-speed packet
processing, as well as building systems that can support high value-added
features such as quality of service for audio and video distributions, billing
option service to differentiate the demands, and the pricing strategy for
different customers requiring different bandwidth speed. In this market, we are
currently shipping products in the GalNet 2 and the 2+ family, as well as start
shipping production in GalNet 3, and we're currently working on our terabit
class of switching product families, the GalNet 4, that we will introduce
sometime by the end of the year.
Now, because of the fact that the systems are very complex in nature, it is
important that the system has to be managed, and a key critical component in
this communication is the system management expertise that we have developed
over the various, the last, you know, four years' experience building these
systems. So, an important component silicon that we provide in this business is
the system controller. The system controller performs all the functions needed
to interface all the critical data processing functions in the systems. Our
chips already support many of the leading-edge microprocessors, such as the
PowerPC's and the MIPS processors as well, supports many types of real-time
operating systems that are needed by different customers. Today, we have
developed many generations of products; in fact, about 70% of the high-end
systems in the market today already use our system controllers. More recently,
we have introduced our new generations of system controllers that we call the
Discovery and the Horizon family, and obviously we will continue to build even
more powerful system controllers that we will talk about in the future.
Now finally, another key point in expertise that we provide in this market
is our technology in terms of providing high-speed backplanes that allow various
subsystems to be connected together to build even bigger systems. One of key
critical components in this market, in this segment, is our technology in the
Fibre Channel SERDES technology that can deliver 3-gigabits per second
throughput on a single channel, while requiring a very small footprint in a
silicon, in expanded digital signal process. And also at the same time be able
to integrate many, many of these functions into a single piece of silicon. You
know, sometimes, and often integrating these functions in switching fabrics to
support terabits like plus types of solution that we will talk about sometime in
the next several quarters.
We're currently supplying our components to various customers in this
communications base. We have actually more than 500-plus customers, so what I've
listed here some of the top-tier customers. They are very important to us. We're
currently shipping products both for the gigabit Ethernet physical layer sites
as well as for the systems, switching fabrics systems controllers. Last quarter,
in our earning announcement, we have made an announcement about more than 30
design wins on our gigabit PHY, on the Marvell side, and on the Galileo side,
more than 100 design wins on their new products. So, we will give you an update
on additional design wins in our next quarterly announcement next month.
Now, the gigabit market is a very important market to address. We believe
that we now have absolutely the leading-edge position in this market. All the
high-end systems networking equipment in the market today that are being
4
5
designed are all using our products. So, we are very, very confident about our
progress in this business, but we also believe that it's important to address
the next-generation requirements in this market, namely the 10-gigabit market.
This is the market that we believe will allow the convergence of the LAN and the
WAN once for all. So one of the strategies that we at Marvell face -- is
striving in this business -- is to build a true single-channel 10-gigabit per
second interface technology using expanded CMOS process technology to drive both
the performance and the cost so that this device can be deployed across the LAN
and the WAN. By the way, we are currently in the active development of this
product, and we will make some announcement, we plan to make some announcement,
on the 10-gigabit solution sometime in the middle of this year.
Now, while it's important to address the build-out of the infrastructure in
supporting hundreds of gigabits or terabits types of performance levels, we
believe that it's also important, equally important, to address the end user
market opportunities, namely, the last 100 meters of the build-out of the
infrastructure. Our vision in this market is to build, to allow the end users to
have access to their local area network's database on wireless technology by
building, by allowing them to, by building a true, low-cost CMOS base 802.11
types of wireless LAN technology that can be connected to the DSL or the cable
for the last miles types of connections. And we believe this is the only way for
the DSL or cable modem to be deployed cost-effectively to the end users by
allowing many, many users to share the cost of this deployment of the Internet.
We are currently working on the 802.11B as well as the 11A development, working
on solutions that can be done in standard digital CMOS process.
We're planning to introduce this product sometime in the middle of this
year. This is a huge market opportunity. In fact, we believe this market is
going to be even bigger down the road as the cost structure goes into the level
such that it can be integrated as a standard product in your PC or your laptop.
Now, I'd like to talk about our data storage business. The data storage
business is an important business for us to deal with, and the reason is because
number one, this is a huge business opportunity for us; the fact that data
storage devices are used in practically all devices in the communication
infrastructure. But more importantly, the data storage market is the market
where we develop our core technology. This is the area where we develop our
mixed signal and DSP technology. And the reason for that is because data storage
device actually is a communications system just like your networking systems.
The main difference is that in the data storage, the communications systems is a
closed system, and therefore we can develop our unique new algorithms, new
technology that we can test it, prove it, you know, make it mature, and then
once the technology is available, we can utilize this for our, the added
business in the data communications side. And this is the reason why we are able
to build our gigabit Ethernet solution in less than nine months after we decided
to enter the business versus the competitors, you know, spending more than three
to four years while still not being able to achieve the performance that we have
introduced into the market. At the same time, this is a big market opportunity
for semiconductor to us. The read channel alone, the read channel as well as the
read channel-related types of solutions, silicon opportunity in this market
alone is about a billion dollars, you know, a year ago, and it's going to...
expected to move to around $2-1/2 billion dollars in three to four years
timeframe. So obviously, this is a huge market opportunity, attracting many,
many players into the business. Prior to 1998, there were more than two dozen
very large companies, all building solutions, or trying to build solutions to
address this market. We entered this business in 1997. We realized that in order
to win this business, we have to do something that's totally different than what
the other guys are doing, namely building silicon solutions that incorporate
advanced communication technology, building devices that can deliver much higher
data throughput. So we have been successful in our strategy over the last
several years, moving from devices that support several hundred megabits per
second throughput to the most recent solutions delivering more than a gigabit
per second throughput.
As a result, today we are one of the three players left that are able to
build such a product, and we're already the leader in the enterprise solutions
5
6
as well in the mobile market segments. Now, the next strategy for us in this
market is to change the market again by -- the strategy in addressing this
market again -- by providing a true system-on-a-chip solution. Now obviously,
there are several key components that are critical to make this happen, and I
will talk a little bit more about some of these components next. The physical
layer, just like in the networking market segment, is probably the most
important technology in this business. It is what we, you know, how we
differentiate our products against our competitor. Today, we already develop
technology, the technology that can deliver gigahertz types of DSP capability in
our silicon, as well as analog signal processing that can deal with gigahertz
types of signals. By the way, we also will be able to do this in the standard
CMOS process, which is critical in allowing us to build our highly integrated
silicon solutions. We also, over the last several years, have developed a rich
IP portfolio that will allow us to build a true single chip solution by working
both, by developing solutions internally as well as by working with our
customers to integrate their IP with our IP to build a true single-systems chip
solution. We're currently working, actively working with three big customers in
this market, building actually several generations of products, and we will make
some of those announcements in the next several months.
Now, the data storage business is a big business opportunity. However, it's
characterized by a limited number of customers, and as a result, each one of the
customers requires very, very large volume requirements. But this is what we
believe we also... by entering this market, we also have a significant advantage
compared to all our competitors in the communications space, namely to be able
to develop, deliver high-volume products, be able to ramp up the products in a
very short time. In fact, in the last 2-1/2 years alone, we have shipped more
than 64 million units of devices to these customers, not just with high-volume,
but also with high-quality levels.
Now, I would like to summarize what I have talked about this morning,
namely the Marvell advantage. First of all, Marvell is about mixed signal,
analog signal processing technology; our ability to...our expertise in
developing circuits that can process, you know, extremely high bandwidth
signals; our expertise in developing proprietary digital signal processing
algorithms; and our expertise in developing custom DSP engines that can process
the algorithms in real time; and our expertise in combining all this technology
into a single piece of silicon utilizing the most advanced standard yet expanded
digital CMOS processor technology. Now, with the combination with Galileo, we
also now have the expertise in building more complex, higher-level system
components, to be able to build systems that can deliver interfaces with both
gigabit and terabit types of throughput, as well as to be able to build systems
that can interface with different protocols, both for the LANs and the WAN
types. And as a result now, we can deliver devices, deliver to our users, the
best performance and the lowest cost structures, and earlier time to market.
Thank you. I'd like to make this... I'll open up for questions now. Any
questions?
MALE SPEAKER 1: Can you give us an update on the timing of the integrated
hard drive products? You said you're going to make announcements the next few
months. Where are you in terms of sampling and shipping the integrated products?
GEORGE HERVEY: We haven't announced those formally, but, you know, clearly
we have those products currently, you know, in house, and we are providing
samples to the relative customers that Sehat is making reference to, but it's
too early yet, you know, to specifically come out and, you know, identify, you
know, exactly what's in each of those chips, but the chips do exist, they are in
the process of being qualified, you know, by those customers. We did indicate
that probably in our first quarter, the first of the three programs will start
to have some volume that will be reflected in our revenue, and then it will grow
from there during the rest of the year. All three are in the sampling stage
right now, yes.
MALE SPEAKER 2: Can you give us some sense of the competitive environment
in the gigbit PHY business -- pricing, growth, any new competitors?
SEHAT SUTARDJA: Actually, it's interesting to notice that even though the
price is an important aspect, today the reason that we're winning the business
is
6
7
because of the performance. We are absolutely way ahead in terms of our
capability in the gigabit Ethernet. In fact, we have a demo systems running
here, showing our capability of delivered gigabit over standard category three
cable. This is just one of examples why, you know, people are choosing our
silicon. Now, we believe that because of nature of the competitive nature of
this business, price will be going down in the future. It's a little bit too
early to say, you know, what's the pricing today, but, you know, but you can
probably expect something in the lower 20s in the near future. I would like to
add that's on a per-port basis.
MALE SPEAKER 3: Enterprise spending seems to be moderating a little bit
from, you know, anecdotal things that are out there. I mean, have you guys seen
anything from your customers? Any thoughts or outlook on that?
SEHAT SUTARDJA: Sure. Actually, what is interesting to note is the dynamic
of how the communications business works. If you look at the -- in the last
several years, at different generation, different notes of the communications
build-out, you know, the customers, the system providers typically built
systems, and then once they're done, they move to the next generation, and in
order to compete in this business, the only way to do that is actually to build
higher-performance devices or equipment. So what we are seeing today is that all
the customers that we're engaging today are building, actually the majority of
them actually are not building the 100-megabit solution any more, any longer. So
all the new developments are the gigabit space, and by... in the next several
quarters, you'll be start seeing announcements from the majority of the
suppliers in the gigabit space, and those devices are expected to replace the
100-megabit solutions in a year or two.
MALE SPEAKER 4: [Inaudible]?
SEHAT SUTARDJA: What we have seen actually, there are some slowdowns in
the, on the ISP sides, on the core sides of the investments, infrastructure
investments, but we believe there is going to be a temporary slowdown, because
in order to continue to provide services, there's always a need for higher
performance infrastructure. Now, we've seen very little, if any, slowdowns on
the enterprise side of the business, because this is a business -- more of the
business users' - type of requirement.
MALE SPEAKER 5: Question about that. Can you talk a little bit about when
you expect unit crossovers, say 4-chip from gigabit Ethernet to exceed 100
megabits, around what timeframe?
SEHAT SUTARDJA: We believe that the gigabit bus Ethernet port ships is
going to out-ship the 100-megabits specifically for the enterprise sites very,
very rapidly, and the reason for that is because if you look at the cost of
building the systems, the majority of the costs today are the ASICS, the, you
know, the mechanical components, the boxes, the sales and marketing costs, the
power supplies, and so on which are not going to change much when you move from
100 megabits to gigabits. So in order to stay competitive in this business, the
enterprise solutions must move to gigabit as soon as possible.
MALE SPEAKER 4: Early next year?
SEHAT SUTARDJA: Maybe...
GEORGE HERVEY: I think it's out at least a year before you actually see the
crossover. But the ramp rate during this year should be quite, you know, quite
steep. But probably sometime next year before they actually crossover.
MALE SPEAKER 5: Question for Galileo, I think. What percentage of fourth
quarter system logics ship -- system controller shipments -- will be to Cisco,
and what, if any, indication is Cisco giving you with respect to their inventory
of those products?
GEORGE HERVEY: I'll answer that since it is now going to be a combined
company. We're going to update all of you about Galileo's fourth quarter
performance on January 22nd, which is the anticipated closing date of the
7
8
merger, and so Galileo will not be reporting their stand-alone financials, but
again we will give you some flavor about that, as well as, you know, some
commentary about the dynamics that they saw in their business at that point. But
until the 22nd, we won't be able to comment on that.
MALE SPEAKER 6: What is the pricing point right now between gigabit and
100, and what do you think that differential has to shrink to, to see that
crossover point, just getting back to that prior question?
SEHAT SUTARDJA: Sure. As I mentioned earlier, that the majority of the
costs needed to build the 100-megabit, or the gigabits actually, are not much
different, because the majority of the costs are actually the ASICS, and the
systems, the PC boards, you know, the boxes that go into the systems. As far as
the cost of the silicon is concerned, 100 megabits today costs probably around
$1 to, you know, somewhere around $1 plus or minus, you know, depending on the
features that you provide to market, or per-port basis, in a gigabits,
approximately around $20 something, and moving maybe in the future to slightly
below $30 on the per-port basis. As far as the product, as far as the end user
is concerned, today typically 100 megabits enterprise class solution is priced
approximately to $200 to, you know, $300 per-port basis. So as you can see,
there is not much difference in terms of the cost of building 100 megabits or
gigabits.
MALE SPEAKER 5: Are they likely to cut cost on the 100 megabit, you know,
stuff that's already shipping volume, or do they try to upgrade everyone to
gigabit by cutting prices on the new technology?
SEHAT SUTARDJA: Our belief is that, just as it happened in the past when
the industry moved from 10 megabits to 100 megabits, the industry will also move
from 100 megabits to gigabits to stay competitive.
MALE SPEAKER 6: [Inaudible] GBIC module for you, and along those lines,
when do you see the conversions over from fiber to copper on the gigabit side,
in terms of unit shipment?
SEHAT SUTARDJA: Sure. The GBIC module is the market there, actually only
Marvell can address today. We are the only company today that can build silicon,
the gigabit Ethernet PHY transceiver solution that can both support copper and
fiber optics. As a result, one of the market that we are the only one that can
address is the G-BIC module where it performs the conversions from copper to
fiber optic interface. This is a market that we believe will enable the
conversion from fiber optics to copper in the next several years. Our belief is,
you know in several years from now, the solution should be more of a type of
dual solutions where both copper and fiber can be supported (unclear) in the
systems, and this is what we introduced today in our Quad+ transceivers. Now,
our Quad+ transceivers can support both copper and fiber without doing the
conversion in the GBIC module. But the GBIC module will allow the transition
from fiber to copper to happen more seamlessly in the next several years.
MALE SPEAKER 7: [Inaudible]
SEHAT SUTARDJA: Actually, the cost saving is tremendous because the
majority of the cost of building fiber networks, or copper networks is really in
the deployment of the fiber optics and all the labor needed to install the fiber
solution. So as far as the module is concerned, probably they are going to cost
about the same.
MALE SPEAKER 8: I was wondering if I could clarify on the design win front.
On the gigabit single port, you have the Intel NIC Card that's out there and
understood, and you have the GBIC with Cisco? Are there other design wins that
you formally announced on the dual port or the quad yet?
GEORGE HERVEY: On our last conference call -- well, let me back up a
second. We first came to the market with a single port in May, introduced it in
May, to show the robustness of the technology. That product went into production
in July, okay? And our first, would be our second quarter conference call, we
had indicated that we had about a half a dozen design wins, including the NIC
product,
8
9
as well as some design wins with Cisco, which you see the GBIC module. In our
most recent quarterly report, which was, as for the quarter ended in October,
that half a dozen has now grown, had grown to over 30, including all of the
top-tier networking players. I mean, as Sehat said during this presentation, you
know, all of the gigabit systems that are being developed right now by any of
the major players are utilizing Marvell's technology. That's how dominant we're
becoming in this space, in multi-port configurations, okay? We will give you an
update in February when we report on the progress of how much, and there has
been significant progress above the 30 that we reported in last quarter, but
we'll give you the specific number of how that's, what that's grown to in
February.
MALE SPEAKER 8: And are these design wins entirely the single port at this
point?
GEORGE HERVEY: The ones that are in -- yes, at this point both single and
dual, okay. We have four customers that are alpha sites for the quad right now.
We started sampling the quad in the latter part of December, and we expect, you
know, and that's predominately for 24- and 48-port type of applications, so
those designs are going to be, I believe, you know, a lot of them will be
finalized over the next quarter or so, and again we're the only player right now
that's there with that technology. So I think our... we believe our lead that
we've established, you know, should continue moving forward.
MALE SPEAKER 8: And is it correct that for the current fiscal year, you're
anticipating about $80 million in communications product revenues, and if that
figure is correct, what percentage do you anticipate to be coming from these new
gigabit design wins?
GEORGE HERVEY: Well, the $80 million is the estimate that, for what would
be our fiscal 2002. We're a January year-end, so calendar 2001, fiscal 2002,
yes. Most of the models range between $70 to $80, $80 million for, you know,
communications products. The --
MALE SPEAKER 8: For the Marvell side?
GEORGE HERVEY: Yeah, for the Marvell side, right, yeah. It'll change
dramatically once the merger is completed. We feel very comfortable that --
well, two things are happening. Initially, I'd say most of those numbers had
about 70% to 80% of the number was going to be gigabit, okay? We still believe
that to be true, but what we are also experiencing is a increase in demand, even
at the Fast Ethernet level at this point. As customers are getting used to
working with our technology, we are getting exposed to a number of still new
Fast Ethernet opportunities that are there, and so we expect, you know, the Fast
Ethernet business next year to be also a very sizable part of our revenue.
MALE SPEAKER 9: Why are people choosing your 10/100?
GEORGE HERVEY: Well, I think it's -- do you want to --?
SEHAT SUTARDJA: Really, the reason for that is actually really, the reason
there is the performance, the robustness of our design. The fact that we're able
to build devices that can perform, that can communicate over longer distances,
devices that can deal with imperfections in the cable, it will be able to
compensate for low-quality insulations, and so on. So once, you know, obviously
the gigabit Ethernet, the gigabit is the one that, you know, opened the
opportunity for Marvell into the segments, but once our customers saw the
quality of products that we have delivered in the gigabit, in certain cases
where they still need Fast Ethernet solutions for some of the entry-level
products, they also, you know, are talking to us, you know, to give us the
opportunity to play in the 100 megabit solutions as well.
FEMALE SPEAKER 1: Would those be like second-course designs, design wins or
because -- I guess you had said earlier that a lot of the new projects that the
enterprise customers are on --
SEHAT SUTARDJA: Are probably gigabits, you know, types of solutions.
9
10
And the reason for that is because again, there is not much difference in the
cost of building gigabit systems or 100-megabit systems due to the cost of all
the other components needed in the systems. But on the entry-level systems such
as like, you know, home routers or small office routers, where 100 megabits are
still good solutions, there's still a lot of opportunities to address better
solutions for this market. In fact, maybe high-integrated solutions, including
integrating in a switching core with embedded memory with our physical layers
and 100 megabits, which we have introduced in the last couple months. So if you
want an example why, you know, some of our customers are using, are choosing our
products.
MALE SPEAKER 9: Question on the alpha customers for the quad. Are they, did
they design boards already around it, or what do you call an alpha customer? Is
it just sampling to those customers, or they actually have boards there that are
waiting for those products?
SEHAT SUTARDJA: Actually, the amazing thing is that these are the customers
that have built boards before we even tape out the silicon. So, we are very
excited about a lot of these opportunities because of the, not just because now
we have the silicon, but because, you know, the customers all are ready to
develop their systems, silicons in the actual systems, and these are the next-
generations, as George mentioned, systems that can support 24 ports or 48 ports
of gigabits on a per-line-card basis, meaning that when aggregated into either
single piece of equipment, these are systems that can support hundreds of
gigabits Ethernet ports in a single, in a small footprint of networking
equipment.
MALE SPEAKER 10: The specifications that you had promised and that they had
requested, these are essentially design wins?
SEHAT SUTARDJA: These will, pretty much will turn into design wins once the
full qualification is completed.
MALE SPEAKER 10: Okay, thanks.
MALE SPEAKER 11: I believe you're working with Intel to integrate the
gigabit MAC with the PHY. What are the details of that technology evolution? Is
there a cross-licensing? Will both Intel and Marvell market that chip
separately, or will Marvell own the technology?
SEHAT SUTARDJA: What we have announced actually last year in our S-1 filing
is the joint development effort with Intel in integrating our gigabit physical
layer technology together with Intel gigabit MAC technology to address the NIC
as well as the LAN on multiple market opportunities. The agreement basically
allows us to take the IPs from Intel and build the silicons, produce the
silicons, and okay, we sell the systems back to Intel.
GEORGE HERVEY: Maybe to just expand a bit. Currently, it has two parts,
actually what Sehat was mentioning. Currently, though, we're both in the market,
selling together our stand-alone PHY along with Intel's MAC. So there are, that
could be an Intel card you see out there has that particular configuration as
well as, you know, going to many of the other leading PC OEMs, and you know,
selling that solution to them. But once we do complete the integration of the
MAC into the PHY, you know, the end customer, its a revenue model for us, the
end customer is Intel, and then they will in turn sell the completed, the
combined solutions to the end customer.
SEHAT SUTARDJA: They have basically combined all the software, and built
the big solution, and sell it to the end users.
MALE SPEAKER 12: [Inaudible]
SEHAT SUTARDJA: Basically, I guess -- let me clarify it. Basically, we will
be the OEM to Intel, so we produce the silicons, and Intel will sell the
complete end products with all the softwares together with the silicon.
10
11
GEORGE HERVEY: Maybe just to answer your question, both companies continue
to own their own IP blocks, so there's no cross-license of Intel's MAC to
Marvell or cross-license of Marvell's PHY to Intel. We each continue to own, you
know, but obviously they had to give us, you know, some indication of what it is
so we could integrate it. But beyond that, there's really, there's no technology
cross-license between the two.
MALE SPEAKER 12: It will be under the Intel --
GEORGE HERVEY: It will be sold as an Intel product, yes.
MALE SPEAKER 12: And what is the target for introduction of this chip?
GEORGE HERVEY: In the first half of this year.
FEMALE SPEAKER 2: [Inaudible] some of the stuff that Galileo is doing on
the MAC side?
GEORGE HERVEY: This is a NIC Card for PC.
SEHAT SUTARDJA: Galileo's main business today is the infrastructure. And
when we talk about MAC, those are the MACs -- they're the same MAC. It's just
the technology. It's just the different ways of, you know, putting the Lego
blocks together. On the Galileo side, the solutions are targeted for more for
high-end systems, high-port COM systems.
MALE SPEAKER 13: [Inaudible]
SEHAT SUTARDJA: Okay. The question is do we have plans to build an octal
gig PHY. That depends on the, the question -- the short answer to that depends
on the market requirements. We already have the technology to build any ports,
any number of ports of gigabit PHY in a single piece of silicon, due to the fact
that our silicon is already extremely small in die size. We will always evaluate
that possibility when the time comes.
MALE SPEAKER 14: [Inaudible]
SEHAT SUTARDJA: The next question is what would be the turnaround time.
From the time when we decided to move from the stand-alone single ports to a
quad port, for example, on the latest quad port device that we just introduced
today, actually took us about less than four months to develop. We can do a .15,
but more likely the next generation device will be as small as process geometry
because as a company, we never look back into what is, what we have done in the
past. We always, you know, we always continue to develop the next-generation
technology. And I just want to add on top of that, okay, why we are so confident
that we can do this is because a lot of technology that we develop, we already
- -- therefore, we need in the networking space -- we actually already develop in
the data storage business. Not just building a test chip, but also already in,
you know, high-volume production.
MALE SPEAKER 15: Couple of things. I guess first of all, in terms of the
highest available density right now for 10/100 is 48 ports per rack unit, is
that right?
GEORGE HERVEY: Correct.
MALE SPEAKER 15: So, you think by mid-year you'll be enabling 48
10/100/1000 per rack unit, so you'll be able to get exact same configuration?
SEHAT SUTARDJA: Yes. The silicon -- the Quad+ gigabit PHY silicon that we
introduced today will allow our customers to build a 48-port count gigabit
per line card basis in a length of 14-inch space.
MALE SPEAKER 15: Okay. So, the other question is then, do you have idea in
terms of a split and design wins between stackable or six configuration switches
versus the bigger modular, you know, enterprise class?
11
12
SEHAT SUTARDJA: We don't have, we don't have the, we're not drawing any
difference between the line card business and the stackable solution, but we
probably pretty much assume that a lot of the systems that are going to be
introduced in the market in the near future are going to be stackable types of
solutions.
MALE SPEAKER 15: Okay. Is there any difference in technology needed due to
modular solutions?
SEHAT SUTARDJA: I mean, stackable and modular.
MALE SPEAKER 15: Oh, okay. So it's the same thing, basically?
SEHAT SUTARDJA: Yes. They pretty much require the same technology. The only
difference is on the modular solution tends to have internal backplane
capability.
MALE SPEAKER 15: Okay, thanks.
MALE SPEAKER 16: You specifically mentioned the wireless LAN as a segment
that you'll target within the next year or so. Can you talk about what type of
components or chips you'll develop for that market, and then, as well as if you
already have the IP expertise in-house to target that market?
SEHAT SUTARDJA: Sure. The wireless LAN is the area that actually we've been
working on the technology side for the last year and a half in terms of
developing the signal processing algorithms needed to build such a product.
We're currently in the development stage of developing the silicon, both
building the RF components needed to interface, as well as the analog and the
base net processors required to build the rest of the systems. So the first
products that we're going to introduce are based on the 80211B, and reason for
that is because this is the market, this is the area that we're seeing the
market is going to go initially with. This is the device they allow the end
users to communicate at 11 megabits per second to, you know, upward to a
distance of 100 meters. Our products is going to built based on standard digital
CMOS process technology. Compare that to the existing solutions out there that
are based on silicon germanium and BiCMOS process technology, as well as this
will be a two-chip solution instead of the five to six chips that normally are
found in the early generations devices out there.
MALE SPEAKER 15: And what's the timeframe that you expect to begin
sampling?
SEHAT SUTARDJA: We plan to introduce the product sometimes in the middle of
this year. All right. If there's no more questions, I'd like to thank you for
joining us here today. I hope to be able to see you again in the near
future.
12