Goldman Sachs Report -
AMD (OP/N): MEETING WITH CEO
REINFORCE DRIVERS BEHIND
OUTPERFORM THESIS
We rate AMD stock OP/N. We remain upbeat following a meeting with CEO Hector Ruiz
as key drivers behind our thesis seem intact. We expect upside to consensus estimates
over the coming quarters could arise from higher MPU revenue on Athlon64 mix, and
better gross margins from higher proportions of AMD64 and MirrorBit NOR flash. The
risk is that if PC unit demand falls in September, AMD would likely miss numbers,
obscuring its secular improvement until end demand improved. 3 Other Points: 1) AMD
began shipment of AMD64 on 90nm this week and expects to reach 10% share in x86
servers by year-end. 2) Q3 is back end loaded, but appears to be growing seasonally
without unusually aggresive pricing by Intel in either MPU or NOR Flash. 3) Key to our
thesis on AMD is improvement in technology partners and enterprise opportunities. We
expect this could include announcement of a foundry partner for 2005 to increase MPU
capacity at a reasonable cost and high ROCE, formal entry into data flash with MirrorBit,
and more from companies like SUNW (IL/A) and BRCM (IL/N).
NOR BUSINESS TRACKING WELL, PROFITABILITY IMPROVING; POTENTIAL
ENTRY IN NAND FLASH USING MIRRORBIT ARCHITECTURE
AMD's NOR business, which accounts for half its reported revenue, appears to be
tracking well in the current quarter. The company indicated it has not yet experienced any
unusual price action as a result of Intel's renewed push to gain share or a slowdown in
handset units. Rather, it is seeing a normal and fairly linear price/bit erosion, which is
being more than offset by bit growth. The company is maintaining strong ties with its
tier-1 wireless handset customers, which we believe include Nokia, Motorola and
Samsung. As it continues to be fully utilized and on allocation for certain parts, AMD has
not aggressively pursued share in the embedded portion of the NOR market, which
accounts for about 15% of its flash sales. The embedded segment is where Intel has been
focusing experienced significant growth in Q2 (recall in Q2 Intel grew 41% Q/Q, vs 7%
for AMD) in products such as DVD players for the China market. Longer term, AMD is
still planning on addressing the data storage (NAND) flash market with MirrorBit, which
has the potential to be cost-competitive while offering better security and reliability. In
addition, we believe AMD may be working with an established NAND Flash Card maker
to on how to best enter that market in 2005.
AMD once again highlighted the low-cost structure of its MirrorBit architecture, which
has 40% fewer critical steps than Intel's competing StrataFlash product. We have already
seen the results of this cost structure in Q1 and Q2, when AMD posted an operating
profit in its NOR business unlike Intel. AMD continues to move up the profitability curve
in flash, or at least better insulate itself from some of Intel's potential price action, as it
increases MirrorBit as percent of total output form 10% currently to 50% by year-end,
and as it transitions to 110nm and 90nm feature sizes. The cost advantage from MirrorBit
is not just due to its 40% fewer critical steps, but also to very high yields (in the 90%s).
AMD and INTC are locked in an effective dead heat as far as process technology
roadmap in NOR: AMD is already manufacturing on 110nm and is ramping 90nm at
year-end, while Intel is still manufacturing on 130nm but currently sampling on 90nm (and skipping 110nm). In addition, AMD indicated it is ahead of plan on wringing synergies out of
the FASL JV integration, with 70% of the synergies already in place and the remainder expected by
year-end. AMD's longer-term (3-5 year) operating model for the NOR flash business is to achieve
gross margins of 30-35% (up from est. 25-30% today) and operating margins of 15-20% (up from
7% today). However, while AMD views NOR as a strategic business in its role as an 'N-1' fab filler
and conduit to wireless and communications customers, we would not exclude the possibility of a
spin as the MPU business remains AMD's primary focus.
MPU BUSINESS EXPERIENCING NORMAL SEASONAL TRENDS; LONG-TERM
ROADMAP APPEARS STRONG, THOUGH NEAR-TERM RISK FROM INTEL'S
INVENTORY REMAINS
AMD is currently experiencing normal seasonal trends in its microprocessor business, accounting
for about half the company's sales. However, AMD indicated the third quarter is typically back-end
loaded, with a 20-20-60% profile, given slow July sales in North America and slow August sales in
Europe.
AMD's forecasts from PC customers for the quarter have not changed materially since it gave
guidance; however these forecasts don't turn into orders until 2 weeks prior to shipment, thus
visibility on 60% of the quarter is still minimal. More importantly perhaps, AMD is not seeing any
unusual pricing pressure from Intel as a result of its excess inventory. AMD's own recent price cut
was part of a preexisting plan. We believe AMD's guidance for a "moderate" rather than "seasonal"
increase in Q3 sales was partly driven by concerns over pricing pressure that could result from
Intel's excess inventory, pressure which has yet to materialize. Gross margin in MPU's should be at
least in line as checks have shown better momentum in 64-bit products recently. We are modeling a
seasonal unit increase of 8% and flat ASPs in Q3.
AMD appears to be executing well on its AMD64 roadmap. AMD is one of the few companies on
90nm that does not seem to have had significant delays or defect issue. Revenue shipments of
AMD64 notebooks on 90nm started this week, well within the planned schedule for shipments
prior to the end of Q3. Desktop AMD64 shipments on 90nm will commence a month later,
followed by servers. An on-time transition to 90nm was one of the keys to our thinking in
upgrading AMD, as it will allow it to cut prices in line with Intel while aggressively shifting mix to
higher margin parts. AMD remains on track for reaching 50% of total MPU revenues on AMD64
by year-end, and crossing over in terms of units in Q205. Recall the AMD64 platform has
significant operating leverage, as blended ASPs are currently about 3x as high as AthlonXP. Over
time this premium will diminish, as AMD drives AMD64 into the lower end and broadens its
reach.
AMD's pricing could be a bit more stable going forward than it was in prior years when its MPUs
were socket-compatible with Intel's, and its only differentiation was price. AMD is now able to
obtain longer-term roadmap commitments from customers, as its MPUs now offer competitive
performance. AMD is tracking particularly well with Opteron. It is on target to reach 10% market
share by year-end, up from 4% today, based on design wins already in place. Given the 12-month
or more design cycle in servers, AMD has some visibility into 2005, when it expects to further
increase its server share, ultimately toward 20%. It has very strong customer support from Sun, HP
and IBM, and is continuing to work on securing a design win with Dell. The catalyst for the latter
would likely be market share growth to up to 20%.
AMD is also seeing solid growth in its AMD64 notebook products, while desktop adoption is still
relatively muted given the slight delay of the Windows 64-bit OS. AMD did remark that it would
not be surprised to see its total AMD 64 (including Opteron) shipments reach 20 million units in
2005, which is slightly ahead of our forecast of 19.3 million.
DUAL-CORE OFFICIALLY A TREND - AMD's roadmap in microprocessors includes launchinga dual-core processor next year, and potentially a quad-core processor in 2007 on the 65m/300mm
process. AMD has relatively high confidence in its dual-core product development, given its
AMD64 architecture was designed from the ground up to optimize multi-core configurations with
its integrated memory controller and Hypertransport features. Note that the die size (and thus likely
cost) for dual-core is likely to be well less that double that of a single-core, given the amount of
shared circuitry between the two cores.
The company also emphasized the profit potential of its MPU business. Even though its ASPs are
currently about 40-50% lower than Intel's (given its 9% revenue share despite an 18% unit share), it
is still able to reach gross margins of 52%, despite its much lower revenue base. AMD's
longer-term (3- 5 year) business plan for its MPU business is to reach gross margins of 55- 60%
and operating margins of 15-20% (up from 10% currently).
NEW CAPACITY AND CAPEX AMD remains on track for starting production at its new 300mm
fab (Fab 36 in Dresden) on 65nm in 1H06, with equipment arriving in Q3. Recall that AMD is
doing the 65nm technology development jointly with IBM, and estimates that it is about 70% done.
In addition, the joint process development and our checks have suggested that IBM and its SOI
(Silicon on Insulator) process, are the most likely foundry partners for AMD should it choose to go
more fully toward a fab-light model in 2005 and beyond
. AMD's capex plan remains at $1.5-1.6 billion for 2004, and likely somewhat less than that in
2005 (perhaps $1 to $1.5 billion).
RISKS - INTEL AND MACRO WEAKNESS ALL AROUND - In the short run much of the
secular improvement we have noted at AMD could be obscured by two factors: 1) Given its
bulging in inventory, Intel may begin to cut price aggressively in NOR flash and MPUs in Q3 if
revenue does not track to plan. AMD noted that it has not seen any unusually aggressive pricing
action from Intel in an effort to sell off its excess inventory. 2) Many macro indicators have turned
down and Q3 is a back end loaded quarter for AMD. As a result, if PC sell-through is poor in July
and August, the all important September month could be weak. Dell, HPQ and Taiwan
motherboard and notebook data will be important indicators here.
I, Andrew Root, hereby certify that all of the views expressed in this report accurately reflect my
personal views about the subject company or companies and its or their securities. I also certify that
no part of my compensation was, is, or will be, directly or indirectly, related to the specific
recommendations or views expressed in this report.
