ML:Mid-cycle correction over,rec'd AMD ATML NSM,upward forcast due to ASP
Excerpts fr Merrill Lynch Research 8/18/00
Highlights:
• Today we are releasing our quarterly industry update.
• The concerns that underpinned our mid-cycle correction thesis are beginning to dissipate, and semiconductor stocks globally are well below their March highs. We believe that the mid-cycle correction has ended – we would advise investors to begin buying semiconductor more aggressively.
• Names in the communications-oriented and foundry businesses are of particular interest to us in the current environment. We would focus investors on names including Texas Instruments, ST Microelectronics,
Cypress Semiconductor, Atmel, National Semiconductor, Analog Devices, Taiwan Semiconductor Manufacturing Corp and Chartered Semiconductor.
• We have also revised our estimate for global revenue growth upwards, from 32% to 40% for 2000 and from 21% to 26% for 2001.
• Improving pricing underlies our new forecasts. We believe Capital spending relative to forecast revenue and volume in 2000 remains at reasonable to low levels, and we expect a period of sustained pricing strength in the latter half of the year as capacity restrictions really begin to bite.
The worst of this industry’s mid-cycle
problems appears to be behind us
In our February 2000 quarterly we first raised the
possibility of a mid-cycle correction, when we pointed out
that the rate of acceleration in revenue growth was likely
to slow through the spring. In our May 2000 quarterly we
discussed the mid-cycle correction in more detail, pointing
out that the mid-May weakness in semiconductor stock
prices was likely to continue as investor concern about a
deceleration of growth in global semiconductor revenue
intensified. We weren’t entirely right – the sector rallied,
and the Philadelphia Semiconductor Index rallied from its
May low of 820 back to nearly 1300 by late June, before
falling back below 900 at the beginning of August. The
SOX is still below its March high, though, and concerns
about whether the semiconductor cycle is healthy or not
have dominated conversation since March.
Chart 1: A Healthy Cycle So Far (Semiconductor Billings
Growth)
Source: Semiconductor Industry Association
Now, however, with the pieces for a strong second half
clearly in place and the SOX 18% off its March high, it’s
time to move on. In our opinion, the issues underpinning
our midcycle correction scenario are dissipating. We think
it’s clear that the industry simply cannot get enough
additional capacity on the ground during 2000 to create
any problems. The outgrowth of that should be a sustained
period of rising average selling prices in the latter half of
2000 – the current upturn had been notable until recently
for the lack of pricing power exercised by device makers.
All of our checks indicate that visibility at semiconductor
makers across the board continues to be excellent, and it is
now clear that trying to call turning points based on thin,
unreliable spot market pricing data is not analytically
sound. Finally, the messy cellular telephone channel is
settling down, and stock prices are reflecting a more
reasonable set of expectations for handset shipments.
Our top picks focus on communications and
foundry
We are therefore calling an end to the mid cycle correction
– we think that it is time to begin buying stocks in the
sector aggressively, and we expect to exit 2000 with
semiconductor stocks at new highs. In particular, we
would recommend buying oversold names in the wireless
communications industry, where any problems that might
exist have been more than discounted. Texas Instruments,
ST Microelectronics, Analog Devices, National
Semiconductor, Cypress and Atmel are all names that we
would focus on. Advanced Micro Devices, despite its
higher exposure to the PC market, has been seeing stock
price pressure as a result of its flash memory business, and
in light of our positive stance on the flash memory market
we reiterate our positive stance on that name as well.
Our top wireline communications picks, which include
AMCC, Broadcom, PMC Sierra, Transwitch and Vitesse
Semiconductor, have all performed better than the
wireless- geared stocks during the recent downturn. We
are nonetheless reiterating our buy recommendations on
those names as well – we see no slowdown in the demand
for merchant semiconductor products from equipment
makers, or in the consolidation that has made the above
group of companies dominant in the sector.
We also reiterate our positive stance on the major foundry
operators that we cover, specifically Taiwan
Semiconductor Manufacturing Corp. and Chartered
Semiconductor. Benefits associated with rising ASPs are
starting to kick in, and both companies are positioned to
benefit from communications-led growth as well – most of
the companies mentioned in the previous paragraph are
fabless.
PC-related names will do well also, but we are
more conservative here
We believe the market’s expectations for semiconductor
sales into the PC end market can only be characterized as
wildly bullish, as Intel’s relatively strong performance
during the mid-cycle correction has illustrated. We expect
Intel to benefit from falling deprecation and the continued
move to 0.18 micron manufacturing, but we have found
remarkably little evidence of the unusually strong demand
environment that seems to be taken for granted. Intel’s
status as the industry’s bellwether stock should insure
good stock price performance for the rest of the year even
if the company’s top line does not measure up to current
expectations, though. We think that at current levels, AMD
offers a more interesting opportunity based on continued
market share gain.
Our positive stance on Micron Technology and Samsung
has been based on our recently updated supply analysis,
which still indicates insufficient manufacturing capacity to
meet demand as the year progresses. However, we are less
vocal on the DRAM story for the next few weeks, as our
checks indicate some intermediate-term risk from
inventory that has accumulated in anticipation of higher
prices.
The revenue-stock price correlation continues
to be strong
Relative semiconductor stock price performance has
always correlated closely with global semiconductor
revenue growth, as the accompanying figure illustrates.
The trick comes with trying to predict revenue growth,
being as stocks always react real-time – by the time SIA
data have been published it’s too late to react. Our
positive case, then, must clearly be based on our
expectation that semiconductor revenue growth in the
latter part of the year should re-accelerate, and it is
reasonable to ask what supports our position.
Chart 2: Strongly Correlated (Semiconductor Relative
Performance Changes and Global Billings Growth)
Capital spending still reasonable relative to
revenue . . .
One important thing to focus on is the still-reasonable
relationship between capital spending and industry
revenue, which has always been a good indicator as to the
state of available manufacturing capacity. Our capital
equipment analyst Brett Hodess forecasts capital spending
of $52.5 billion in 2000, up sharply from 1999’s figure of
$33.9 billion. However, relative to our forecast revenue,
capital spending comes in at a still-reasonable 25% of
revenue, as compared to a peak level of 31% in 1996.
Although the dollar amount might increase further, the
industry’s ability to boost wafer starts in 2000 is extremely
limited – equipment backlogs on key tools, most notably
lithography, are too long. The higher than expected growth
in semiconductor revenues leads us to believe that
additional semiconductor capacity is required relative to
semiconductor demand. Thus, we remain comfortable that
supply will continue to tighten in the normal seasonally
strong fall/winter period for semiconductors, providing a
catalyst for stock appreciation in the group.
Our current forecasts for 2001 show capital spending at a
slightly higher 26% of revenue, although that figure is
likely to increase as capital spending plans for 2001
become clearer.
Chart 3: Still Just Barely Keeping Up (Capital Spending Relative
to Revenue)
. . .which should support further pricing
firmness later in the year
The outgrowth of a low capital spending to revenue ratio
has always been higher average pricing, and we have little
reason to expect anything different in the latter half of
2000. Barring a major economic downturn we expect
pricing to emerge as the driving factor behind
reaccelerating semiconductor revenue growth in the latter
part of the year. As the figure below illustrates, pricing
power is a relatively recent phenomenon in the current
cycle.
Chart 4: More Pricing Strength to Come (YoY Changes in Unit
Shipments and Average Selling Price)
Source: Semiconductor Industry Association
We are boosting our industry forecasts again
Our previous forecasts, published in May of this year, had
called for global semiconductor revenue to expand by 32%
YoY to $196 billion. Given the strength that the industry
has seen and that we expect for the remainder of the year,
that number is clearly now too low. We are raising our
forecast for shipment growth this year to 40%, which
implies 2000 revenue of $209 billion. We have also
revised our 2001 revenue estimate upwards, and we are
now looking for global shipment value of $263 billion, up
26% YoY. We point out that this is a number to build on –
visibility beyond 9-12 months is always limited in the
semiconductor industry and no observer can make firm
forecasts that far out. In any case, we believe market
attention will be focussed on the nearer-term performance
of the industry. Anyone doubting whether the
semiconductor industry is a growth business or not might
note that the $60 billion in incremental revenue we are
forecasting for 2000 exceeds the industry’s entire 1992
billings.
Interestingly, the bulk of the upward revision to our
forecast comes not from increasing unit shipments, but
from increasing average selling prices. Based on our
capacity analysis we believe that it will be difficult for unit
volume in the semiconductor industry to grow by much
more than 30% in 2000, our current forecasts. The
remainder of the forecast revenue growth comes rising
ASPs. |
