GS US Semi & SPE weekly - If 2006 semi
capex is -4% y-o-y, which is what our
bottom-up model suggests, SPE orders and
shipments are likely to be flat for an
extended period based on current quarterly
run rates
1) Now that the dust has settled from Applied?s better than expected order guidance last
week, we thought it made sense to take a step back and compare our 2006 year-over-year
bottom-up semi capex model with Applied?s expected quarterly order and shipment run
rates exiting 2005. The analysis shows that either semi companies must increase their
admittedly very preliminary 2006 semi capex forecasts OR sequential orders and
shipments for the SPE companies are likely to remain flattish sequentially throughout
2006. A flattish 2006 sequential order/shipment environment for SPE would be in stark
contrast to both current sentiment and Street models which are calling for a strong cyclical
upturn, 2) SPE July book-to-bill review, 3) CMOS earnings preview, and 4) weekly
memory pricing monitor and stock price performance.
BASED ON OUR BOTTOM-UP 2006 SEMI CAPEX MODEL AND THE OCT/JAN
ORDER AND SHIPMENT RUN RATES GUIDED TO BY APPLIED MATERIALS,
SPE QUARTERLY ORDERS AND SHIPMENTS ARE LIKELY TO BE FLAT
SEQUENTIALLY THROUGHOUT 2006. The table below provides a breakdown of
Applied Materials quarterly orders and shipments for the first two calendar quarters of
2005, the company's forecast for the Oct. '05 quarter, and our forecast for Jan. '06 quarter,
based on the company's commentary that January orders will be higher than October.
While we certainly don't intend for this to be the only analysis investors need to do and
while the 2006 bottom-up semi capex model can be very dynamic this far out, comparing
the expected quarterly order and shipment run rates exiting the year to our bottom-up 2006
semi capex model (which suggests 2006 semi capex of -4% y-o-y), highlights that
sequential SPE orders and shipments could remain flattish for all of 2006. The reason we
say this is that at about $1,760m in Jan. '06 quarterly orders for Applied (we are basing
this on Applied management's commentary that January orders will be higher than
October orders), the annualized run rate for 2006 would be over $7,000m, which would
represent a 10% y-o-y increase vs. 2005. If 2006 semi capex is -4% y-o-y, why would
orders need to be up more than 10% y-o-y in 2006? The answer is they probably don't. So,
either semi customers will need to increase their initial (and admittedly very preliminary)
2006 semi capex forecasts or, SPE quarterly orders are likely to deteriorate as we move
through 2006.
The orders vs. shipments and revenue issue for the SPE industry can always be very dicey
because any orders the customers make toward the end of one year are typically for capex
in the following year. Therefore, it's important to do the same analysis for semi equipment
shipments (which track to capex more closely) that we did for orders above. To that end,
assuming a January quarter shipment number of $1,600m (that's our estimate; we believe the Street
is much higher and if we used the Street's number, our point would be even stronger), the implied
quarterly annualized run rate number for semi equipment shipments is $6,400m, or -4% y-o-y.
While this slight annual decline is consistent with our earnings models for our SPE companies,
recall that the Street is modeling a 46% increase in Applied's y-o-y earnings for 2006. Comparing
our bottom-up model for 2006 semi capex to run rate shipments once again suggests that sequential
shipments for SPE companies are likely to remain flat throughout 2006, in stark contrast to the
Street which is modeling a significant ramp.
LAST WEEK'S BOOK-TO-BILL RESULTS FOR THE MONTH OF JULY SHOWED
DETERIORATION IN BOTH FRONT-END SEMI EQUIPMENT ORDERS AND SHIPMENTS
WITH JUNE FRONT-END ORDERS REVISED 4% DOWNWARD. WE BELIEVE THAT
FRONT-END SEMI EQUIPMENT FUNDAMENTALS ARE INCONSISTENT WITH THE
STREE'TS EARNINGS MODELS DUE TO FRONT HALF LOADED 2005 SEMI CAPEX
BUDGETS AND MEMORY WEAKNESS. Semiconductor Equipment and Materials International
(SEMI) released its three-month rolling average US equipment manufacturers' book-to-bill of 0.93
for the month of July (vs. our 0.93 and the Street's 0.95 estimate) on Friday last week. Overall
orders of $1,021 million (-2% month-over-month, -36% year-over-year) were 3% below our
estimate and 5% below the Street estimate. Overall shipments of $1,097 million (-5%
month-over-month, -28% year-over-year) were 2% below our estimate and 4% below the Street
estimate. The front-end book-to-bill was 0.86 on a 5% month-over-month decrease in orders (6%
below our and 9% below the Street estimate) and shipments -7% month-over-month (4% below our
estimate and 5% below the Street estimate). The back-end book-to-bill was 1.29 on orders +13%
month-over-month (9% above our estimate and 7% above the Street estimate) and shipments +10%
month-over-month (6% above our estimate and 5% above the Street estimate). The June
book-to-bill was revised downward to 0.90 from 0.93 on lower orders and essentially flat
shipments.
We continue to expect both orders and shipments to decline further in the next several quarters
driven by, front-half loaded 2005 semi capex budgets and, what we expect will be, weakening
memory orders in H2'05. While memory orders have held up relatively well thus far, and many on
the Street, as well as company managements have been comforted by the recent stabilization in
DRAM pricing, we continue to believe that this stabilization is temporary and is primarily driven by
seasonal factors as well as some of the capacity shifts that have taken place away from DRAM to
NAND. We continue to expect DRAM prices to decline further in the coming quarters, and to
continue to negatively impact profitability for the DRAM makers (note that DRAM prices have
already declined approximately 10% from their recent highs in the beginning of August). As a
result, we expect memory orders to weaken in the coming quarters, as the DRAM industry is
currently losing money and we expect the industry to continue to lose money for an extended period
of time. Regarding foundries, we believe that some foundries have already begun to place new
orders (i.e. UMC and SMIC are already being aggressive despite their own choppy fundamentals),
with TSMC the only foundry to potentially place meaningful incremental orders in H2'05, which we
do not believe will be enough to drive a sustainable order ramp beyond one or two quarters.
Contrary to the front end, we believe that back-end fundamentals have likely bottomed, as back as
back-end orders and shipments have already declined 64% and 54%, respectively, from this cycle's
peak to the cycle trough. As a result, we continue to model an improvement in orders and shipments
for the back-end in August. While back-end fundamentals have clearly begun to recover, we remain
concerned with the pricing environment in the back end, which continues to impact the back end
companies' ability to generate earnings and cash flow over the course of the full cycle. Additionally,
valuations for the back-end companies remain very aggressive and are preventing us from getting
more positive on the back-end names.
We are estimating a flat overall book-to-bill ratio in August of 0.93 on a 2% m-o-m decline in
orders and a 2% m-o-m decline in shipments. We are modeling three-month rolling average overall
orders of $1,005 million (-2% m-o-m) and overall shipments of $1,075 million (-2% m-o-m). We
2 Goldman Sachs Global Investment Research
August 21, 2005 Analyst Comment
estimate front-end shipments of $885 million (-3% m-o-m) and front-end orders of $760 million
(-3% m-o-m), yielding an estimated front-end book-to-bill ratio of 0.86. We estimate back-end
shipments of $190 million (+3% m-o-m) and back-end orders of $245 million (+3% m-o-m),
yielding a back-end book-to-bill ratio of 1.29. We continue to emphasize that that the book-to-bill
should not be a significant trading event for the stocks as it is an unaudited and backward looking
metric.
WE EXPECT CREDENCE TO REPORT A SEQUENTIAL INCREASE IN CQ2 ORDERS OF
APPROXIMATELY 5% AS BACK END FUNDAMENTALS APPEAR TO HAVE BOTTOMED.
THAT SAID, WE CONTINUE TO RECOMMEND UNDERWEIGHT POSITIONS IN THE
STOCK AS WE REMAIN CONCERNED THAT THE HYPERCOMPETITIVE ENVIRONMENT
AT THE BACK END CONTINUES TO DRIVE SECULAR MARGIN EROSION AND FULL
CYCLE OPERATING LOSSES.
Credence Systems is reporting its CQ2 (July) earnings results on Thursday after the market close.
We model July-quarter revenues of $107 million (up 5% sequentially) with a loss per share of
$0.04, vs. the Street LPS estimate of -$0.02. We model July-quarter gross orders of $110 million, up
5% sequentially.
While we believe that fundamentals at the back end have likely bottomed and consider the flash
business Credence has recently won at Intel to be a positive, we remain concerned that the
hypercompetitive environment at the back-end continues to lead to secular margin erosion and full
cycle operating losses. Additionally, we believe that sentiment is already positive on CMOS, as the
stock remains one of the best performers in our coverage universe YTD (+8% YTD vs. the group at
3% YTD.)
With regards to valuation, while we prefer to use normalized earnings and free cash flow to value
cyclical stocks, Credence currently does not generate either earnings or free cash flow over the
course of a full cycle. As a result we are left valuing the company using book value and sales
multiples. To that end, CMOS is currently trading at 4.5x tangible book and 2.1x our estimate of
CY2005 sales vs. the semi equipment group at 2.9x P/S. While the relative P/S valuation may
appear reasonable, we would not recommend that investors become aggressive on the name, as the
company continues to generate full cycle operating losses and operates in a hypercompetitive
industry with secular margin erosion.
WEEKLY MEMORY MONITOR: NAND flash memory spot prices continued to decline w-o-w
(-2%) last week, but were up slightly m-o-m (+2%) with retail prices increasing 5% w-o-w (m-o-m
retail prices were down 7%). We would also highlight that August NAND contract prices were
released last week (please see table 4 for additional detail). On average, prices remained essentially
unchanged m-o-m, following approximately a 12% m-o-m decline in July. We believe that August
prices came in below expectations, which called for a slight increase in pricing. That said, we
believe that both Samsung and Toshiba were able to increase prices slightly in August. However,
the increase was not enough to contribute to an overall increase in contract prices in the NAND
space.
Each of the analysts named below hereby certifies.. |
