SL: Semiconductors & Computer Hardware: Upgrading Semiconductor Group
07:05am EDT 27-Jun-02 Lehman Brothers (Niles, Daniel)
INVESTMENT CONCLUSION :
* After starting to become negative on Semiconductors & Computer Hardware
names in Q3:2000, we believe investors now realize that end-demand is not
improving much till 2003 at the earliest and more importantly are starting to
give valuations to a few names that are appropriate for this low growth
scenario by being near historic lows.
SUMMARY :
* As a result, we are moving from an underweight position to a market
weight/slight overweight position in semiconductors. Though semiconductor
revenues bottomed y/y and started improving last August, valuations have not
started to reach a compelling level until now, in our minds. We are upgrading
Micron and ICST to Strong Buy from Buy (the only Strong Buys we have) in the
computer hardware/related semiconductor sector. We are also upgrading
Fairchild and Conexant to Buy from Market Perform.
* We place the downside on the SOX at 350 or about 15% above the peak seen
in 1995 or about 10% lower than current levels of about 383 but upside to
about 450 or roughly up 20% by year-end.
* We anticipate adding more names to this list as the summer unfolds and
estimates come down further along with stock prices. We fundamentally
believe that the global macroeconomic picture is improving and that IT
spending will follow suit but only as we get into next year. On that vein, we
are also cutting our estimates for HPQ and Intel today based on a the still
difficult IT demand environment and deterioration in consumer PC demand
leading to high PC channel inventory.
We are upgrading our semiconductor weighting from underweight to market
weight/slight overweight. We are upgrading selected names in our universe
that we believe have both reasonable earning expectations for the near and
longer term as well as compelling valuations. These names are Conexant
Systems and Fairchild Semiconductor from Market Perform to Buy and Integrated
Circuit Systems and Micron Technology from Buy to Strong Buy. There are four
main reasons that we have upgraded our sector weighting. First,
semiconductor revenues, units and ASPs started to bottom on a year-over-year
basis starting in August but investors are now finally starting to adjust to
growth being slower than expected in 2002 with the best example being Intel`s
pre-announcement for Q2. Second, we are forecasting growth in 2H:02 and for
the full year 2003 for three key end-markets (PCs, wireless and enterprise
networking). Third, end-market inventory levels are reasonable though we
would say not low given the inventory builds seen in 1H:02. Finally, we
believe investors are finally giving valuations that are consistent with the
slower growth that we expect over the next several years for many of these
names.
After being negative starting in Q3:2000 and remaining negative, we believe
now is the time to become more constructive on semiconductors. Rather than
try and upgrade the group early with the assumption that `the worst was
behind us` and `a strong upcycle will solve all valuation problems`, we chose
to wait until we believed fundamentals and estimates stopped deteriorating
AND valuations were at reasonable levels to discount the risk. Though timing
wise, we believe there will be more estimate cuts when IT related tech
companies report results during the summer including Intel, Hewlett Packard
and IBM, we believe some names in our universe have already contemplated
those risks in their earnings projections with valuations to match. To be
clear, certain sectors within semiconductors such as PLDs, comm ICs and mixed
signal, we believe still have a fair amount of risk as estimates get lowered
given valuations still seem to anticipate unreasonably high growth
expectations.
We believe investors are finally arriving at valuations that are consistent
with the slower growth that we expect over the next several years for many of
these names. Currently, our semiconductor sector trade at 4.5x (vs a 5-year
median of 7.3x) LTM sales, 28x (vs. 46x) forward P/E and 2.8x (vs. 6.1x) book
value. While stock prices in our semiconductor universe have declined
significantly over the last several months, and now trade at a 25% discount
to their median 5-year P/E, we note that certain neighborhoods such as Comm
ICs (40x CY03), PLD/ASICs (28x CY03), and Wireless (30x CY03) remain
expensive. We would therefore focus on companies with reasonable growth
forecasts and reasonable valuations through year-end.
Figure 1: Company Valuation Comparable
Source: Baseline, FactSet, and Lehman Brothers
We also believe that much like in the mid 1980s or the early 1990s when
stocks collapsed after the recovery was slower than expected, we believe we
have gone through that same situation this year. We have been using the
charts below for more than a year as the main reason to remain negative. In
1990 and 1985 semiconductor stocks recovered at the bottom of the cycle and
actually started to collapse going to lower lows during the upcycle versus
the lows set during the worst part of the downcycle. As can be seen below,
we have gone through the same process this year. While no one (including us)
debated that business was improving, our belief was that it would not improve
as much as expected much like in those two prior time periods. We believed
that this would result in many of these stocks going to lower lows during the
upcycle as it became clear that though improving, revenues and profits would
not match expectations for the rate of improvement as in the mid 80s or early
90s. We believe the absolute bottom if it actually has not been reached
already will be seen in July or August during the depths of the summer but
not at much lower levels.
Figure 2: Semi Stocks vs NASDAQ vs Semi Revenue Growth (Jan 85 - Dec 87)
Source: SIA and Lehman Brothers.
Figure 3: Semi Stocks vs NASDAQ vs Semi Revenue Growth (Jan 88 - Dec 91)
Source: SIA and Lehman Brothers.
Figure 4: Semi Stocks vs NASDAQ vs Semi Revenue Growth (Jan 99 - Jun 02)
Source: SIA and Lehman Brothers.
We place the downside on the SOX at 350 or about 15% above the peak seen in
1995 or about 10% lower than current levels of about 383 but upside to about
450 or roughly up 20% by year-end. We would caution investors that certain
sectors within semiconductors such as: 1) pure-play communication IC players
such as AMCC, PMCS or VTSS (see our estimate reductions in 6/17/02 First
Call), 2) mixed signal players (i.e. LLTC and MXIM), and 3) PLD vendors such
as Altera and Xilinx (see our estimate reductions on 6/21/02) are likely to
have a tougher time seeing their stocks appreciate given tough fundamentals
and higher valuations. We would therefore focus on companies with reasonable
growth forecasts, interesting valuations, and solid franchises with stable
balance sheets through year-end. This caused us (at least for now) to leave
out companies that had compelling valuations such as AMD which is bleeding
cash right now and ONNN semiconductor due to their heavy debt burden. We
view the PLD (programmable logic device) vendors as somewhere in between
given estimates will be lowered substantially when they give guidance but
valuations are becoming more reasonable and they have solid franchises (see
estimate reductions in 6/21/02 First Call).
We believe that semiconductor revenues will improve from down 49% y/y in
August of 2001 to flat to down 5% in 2002 and up 18-20% in 2003. We continue
to believe that the August of 2001 month was the worst year-over-year
comparison with a minus 49% decline in revenue. As shown below, the
semiconductor business started to rapidly deteriorate in Q4:00 as inventory
exploded and demand collapsed making this an easy comparison. These figures
indicate that business has mathematically hit bottom in Q3 from a y/y revenue
standpoint. Going forward, we believe that revenues on a year-over-year
comparison will continue to improve towards positive revenue growth.
Figure 5: Semiconductor Revenues Versus Units Y/Y % Growth
Source: SIA and Lehman Brothers
Figure 6: Y/Y Revenue Growth Comparison (August To May)
Source: SIA
While unit growth finally turned positive in March of 2002, we believe it is
highly unlikely, if ever, over the next year that ASPs will see positive
comparisons. Our view on semiconductor pricing is always that it is set by
supply and demand and not by the suppliers. It is hard for us to see demand
exceeding supply over the next twelve months except in selected commodity
areas such as DRAMs. That combined with the cost reductions driven by the
implication of Moore`s Law of twice as much silicon in eighteen months for
the same price as today, we believe will put a damper on prices till late
2003 at the earliest. We also note the aggressive shrinks to 0.13u and 300mm
which will result in an acceleration in silicon produced over the next twelve
months.
Figure 7: ASP and Unit Y/Y % Growth Comparison
Source: SIA
On a positive note, we have seen some encouraging signs on a q/q basis with
regards to pricing however which should at least stabilize the declines.
While the most recent year-over-year numbers (April) for the entire
semiconductor industry do not appear to be attractive (units up 13%, and
revenues down 8%), when we take a look at the pricing trend over the past
twelve months, it has been bouncing along the bottom really since July. We
note that while units were up 3% on a month-over-month basis in December,
revenues actually increased 15% for example. Based upon SIA data, it appears
as if pricing for the overall semiconductor industry bottomed in August with
the average year-over-year decline of 24%. Since then the ASP decline has
improved to only down 19% in October. In the 1998 cycle, pricing declines
bottomed at roughly -20% on a year over year basis.
Figure 8: Semiconductor ASP Y/Y % Growth Comparison
Source: SIA and Lehman Brothers Estimates
Figure 8: Semiconductor Revenue Y/Y % Change
Source: SIA and Lehman Brothers Estimates |
