GS US Semi Equip - Samsung CAPEX;
memory bad, TFT-LCD good
Summary: Samsung announced its 2005 capex guidance on Thurs night. While confusion
will abound, the bottom-line takeaway is that memory capex indications are bad while
TFT-LCD indications are good. While the company guided total capex +27% y-y in 2005,
that number is deceiving as: 1) they underspent their 2004 budget by $700m creating a
lower base, and 2) almost all of the increase in 2005 capex is coming from TFT-LCD,
which only AMAT benefits from among the larger companies. We would be strong sellers
of rallies as: 1) we continue to believe that fundamentals will deteriorate in 2005 driven by
a fall off in memory orders (supported by this announcement) that won't likely be offset by
an increase in foundry orders, and 2) SPE companies will likely be forced to admit weaker
fundamentals beginning on earnings calls next week. We maintain our Cautious coverage
view.
SAMSUNG'S 2004 SEMICONDUCTOR CAPEX WAS 5% LOWER THAN
EXPECTED AND SAMSUNG'S 2005 SEMICONDUCTOR CAPEX IS UP 9%
YEAR-OVER-YEAR; LOWER THAN OUR 12% YEAR- OVER- YEAR
SEMICONDUCTOR CAPEX GROWTH ESTIMATE FOR 2005 OFF OF A LOWER
BASE. Samsung reported its Q4 earnings on Thursday after the market close. The
company reported 2004 semiconductor capex (including only memory and system LSI
and excluding TFT-LCD) of $4.8 billion, $235 million or 5% below management's
previous estimate of $5.0 billion. More importantly for the stocks, Samsung announced
its 2005 semiconductor capex budget, which is expected to be $5.2 billion, up 9%
year-over-year. We had expected the company's semiconductor capex to be up about 12%
year-over-year, so the increase off of a lower base in 2004 is worse than expected. We
would note that spending for memory in 2005 is expected to decline 6% year-over-year
while spending on System LSI is expected to increase 107% year-over-year in 2005.
TFT-LCD CAPEX WAS SIGNIFICANTLY LOWER THAN EXPECTED IN 2004 BUT
WAS GUIDED UP SIGNIFICANTLY HIGHER THAN EXPECTED IN 2005. The
company had previously guided for TFT-LCD capex to be $1.8 billion in 2004 but
spending came in 28% lower than estimated at $1.3 billion. That said, 2005 TFT-LCD
capex was guided up 96% year-over-year to $2.5 billion, which, considering the $500m
that was shifted from 2004 to 2005, is essentially in-line with our estimate for flattish
TFT- LCD capex in 2005. Recall that we include spending for TFT-LCD in our
bottom-up capex model as both Applied Materials and Tokyo Electron sell equipment
used in flat panel manufacturing facilities. However, none of the other major companies
in our coverage universe benefit from the TFT-LCD capex.
Total capex (including semiconductors and TFT-LCD) was 11% lower than estimated in
2004 and was guided up 27% year-over-year in 2005, vs. our +12% year-over- year
growth expectation.
Please see the table below for comparisons of our 2004 capex estimates versus 2004
actual capex as well as the company's guidance for 2005 capex.
Samsung's capex US$ in millions; Exchange rate 1150
2004 planned 2004 actual % difference Semi $5,017 $4,783 -4.7% Memory $4,278 $4,096 -4.3%
System LSI $687 $643 -6.3% TFT-LCD $1,757 $1,270 -27.7% Total $6,722 $6,009 -10.6%
2005 capex guidance y-o-y % change Semi $5,226 9.3% Memory $3,835 -6.4% System LSI
$1,330 106.8% TFT-LCD $2,487 95.9% Total $7,652 27.4% Source: Goldman Sachs Research
estimates, company data.
WE WOULD EXPECT THERE TO BE A GREAT DEAL OF CONFUSION GENERATED BY
SAMSUNG'S CAPEX BUDGET BUT OUR BOTTOM LINE IS THAT THE RESULTS ARE A
BIT DISAPPOINTING WITH MEMORY SPENDING DECLINING IN 2005. There is likely to
be a great deal of confusion generated by Samsung's capex release for two primary reasons. First,
the company provides its capex budget in Korean Won so different analysts on the Street are likely
to be using different exchange rates in translating the budget to US dollars (note that in our analysis
above, we use an exchange rate of 1 USD for 1150 Korean Won). Second, different analysts are
also likely to choose to either include or exclude spending for TFT-LCD in their capex estimates.
Given the confusion around the release, we would expect the bulls to highlight the more substantial
expected year-over-year increase in total (including TFT- LCD) capex in 2005 (+27%
year-over-year vs. our +12% year- over-year estimate), while the bears are likely to highlight, both,
the fact that the company under- spent its budget in 2004 (thus making the year- over-year increase
exaggerated off of a lower than expected base) and the decline in memory spending.
We indicated in our weekly out earlier this week as well as in our note following Intel's earnings
report on Tuesday night that we expected both Intel and Samsung to guide for increases in capex in
2005 that would be viewed positively by the market. While Intel's 2005 capex budget was certainly
better than expected, we view Samsung's capex as a bit disappointing given that:
1) Much of the increase in spending is being driven by TFT-LCD capex with memory capex
expected to decline year-over-year in 2005. Most semi equipment companies are significantly
exposed to Samsung's memory spending while Applied Materials is the only US company that has
exposure to Samsung's TFT-LCD spending. And
2) A look at the data shows that Samsung under-spent its 2004 capex budget by approximately
$700 million. Therefore, while the 27% year-over-year increase in 2005 planned capex looks
exciting on the surface, if one backs out the $700 million that the company didn't spend in 2004,
the increase in 2005 spending over 2004 levels is closer to +14% vs. our +12% estimate. Again, the
incremental spending in 2005 is coming almost entirely from the TFT-LCD business as opposed to
the memory business.
SELL THE EQUIPMENT STOCKS IF CONFUSION AROUND SAMSUNG'S CAPEX DRIVES
A RALLY IN THE GROUP. We continue to believe that investors should be underweight the semi
equipment stocks as our view remains that memory orders (currently representing 50% of order
books) are likely to decline meaningfully beginning in CQ2 and foundry orders aren't likely to
rebound for several quarters due to low capacity utilization levels and a deteriorating pricing
environment. We expect most of the stocks to make new cycle lows as the weaker fundamental
environment becomes more apparent, likely beginning with semi equipment earnings season next
week, as we believe managements will be forced to guide to sequentially lower CQ1 orders given
the weakness at foundries in Japan and at flash memory customers. We therefore believe that
investors should use any strength as an opportunity to sell the stocks.
I, Jim Covello, hereby certify that all of the views expressed in this report accurately reflect my
personal views about the subject c |
