GS US Semi Equip - Intel capex; 2001 all
over again
Intel guided its capex up about 35% y-y in 2005 to $5.1b, significantly higher than our
and the Street's expectation for a more moderate increase. While Intel's 2005 capex is
better than expected, global capex is likely to be -10% y-y in 2005, similar to 2001 when
overall capex declined meaningfully despite an increase from Intel. Note that Intel has
historically been a counter-cyclical spender and its capex/revenues was at a historic low in
2004. Given that we also expect Samsung to guide its 2005 capex up on Thursday night,
we expect the SPE stocks to experience a short- term rally post the group's
underperformance in Q4'04 and to-date in 2005. As we suggested in our note on Monday,
we would be sellers on strength as, Intel and Samsung notwithstanding, we believe 2005
is going to be a difficult year for the industry with foundry capex down as much as 50%
y-y and memory weakening throughout the year. We expect the weaker fundamental
environment will become more apparent beginning on next week's semi equipment
earnings calls.
INTEL'S 2005 CAPEX IS UP MORE THAN WE HAD EXPECTED AS THE
COMPANY CONTINUES TO INVEST COUNTER-CYCLICALLY IN CAPEX AND
CAPEX AS A % OF REVENUES WAS AT A HISTORIC LOW IN 2004. Intel reported
its Q4 earnings on Tuesday after the market close. The company reported 2004 capex of
$3.8 billion, up 6% year-over- year, in line with our estimate of $3.8 billion and the
mid-point of the guided range of $3.6 to $4.0 billion. More importantly for the stocks,
Intel announced its 2005 capex budget which is expected to be in a range of $4.9 billion
to $5.3 billion, up 29% to 40% year-over-year. We had expected Intel's capex to be
guided up in a range of 5% to 15% year-over-year or to $4.0 to $4.4 billion. Street
expectations were all over the map, including some calling for flat or down 2005 capex,
but we don't believe anyone expected the number to be quite so high. Management
indicated on the earnings call that the increase in spending is being driven by the
company's ramp up of its 65-nanometer processes as 90% of orders in 2005 will be for
65-nanometer capacity. Management also noted that most of the increase will go toward
the purchase of wafer fab equipment (as opposed to bricks and mortar), which obviously
benefits the equipment companies.
While the 2005 capex budget is up more than we had expected, we would reiterate that
Intel is a countercyclical capital spender and that the company was at a historic low in
terms of capex as a percentage of revenues in 2004. Intel has historically tended to be a
countercyclical spender of capital as the company tends to make technology transitions
even during weaker years in the market given its strong cash position. By investing early,
Intel derives the ability to get its costs down and provide cutting edge (and higher ASP)
chips more quickly than the broader market. Recall that in 2001 when Intel began its
300mm investment, the company maintained a significant capex budget of $7.3 billion,
which was up 10% year-over-year at the same time that industry-wide capex was down
more than 30% year-over- year. Intel's budget in 2004 was up only 6% year- over-year
while industry-wide capex was up more than 55% year-over-year. 2005 looks to be a
continuation of the company's pattern of investing in capacity when its competitors are
less able to do so.
In addition, as we noted coming into the call, capex as a percentage of sales in 2004 was at 11% vs.
an average over the past twelve years (including our 2005 estimate) of 17% (please see the table
below). With capex of approximately $5.1 billion in 2005 (the mid-point of the guided range), we
estimate that capex as a percentage of revenues will be approximately 14%, still well below its
historical twelve-year average. One driver of the lower capex as a percentage of revenues is the
efficiencies created by 300mm. Intel indicated on the earnings call that it would have spent an
additional $1 billion in capex in 2004 if it were investing in 200mm capacity rather than 300mm
capacity and that to-date, it has saved $4 billion in capital by investing in 300mm capacity.
Intel's capex as a % of revenues 1994 21% 1995 22% 1996 15% 1997 18% 1998 15% 1999 11%
2000 20% 2001 28% 2002 18% 2003 12% 2004 11% 2005E 14% Average 17% Source: Company
data, Goldman Sachs Research estimates.
SELL THE EQUIPMENT STOCKS ON A RALLY DRIVEN BY THE
BETTER-THAN-EXPECTED INTEL CAPEX NUMBER. We believe the better-than-expected
2005 capex number from Intel is likely to drive a short-term relief rally in the semi equipment
stocks post the significant underferformance for most of the stocks in CQ4'04 and to-date in 2005.
Additionally, we believe that Samsung (which is reporting earnings on Thursday night, please see
our most recent weekly for details on our Samsung 2005 capex estimate) is also likely to announce
a year-over-year increase in capital spending that could potentially fuel the aforementioned relief
rally. We would sell the semi equipment names on strength as, Intel and Samsung's capex
notwithstanding, the fundamental outlook for 2005 looks extremely challenging for the equipment
industry as memory orders (currently representing 50% of order books) will likely 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 (which our checks with
fabless customers confirmed last week). Our bottom-up capex model now shows a year-over-year
decline in capex of approximately -6% and we would expect this number to move lower throughout
the year (to down approximately -10% year-over-year) driven by capex cuts from the memory
makers as pricing weakens and cash flows deteriorate. We therefore 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 and flash
memory customers.
I, Jim Covello, hereby certify that all of the views expressed in this report accurately reflect my
personal views about th |
