Report on the sector outlook from SmithBarney after their
recent trip to Asia countries. Note SB comments on LRCX.
It is very bullish on LAM. The report is long but it is
worth going over it.
--SUMMARY:----Semiconductor Equipment Completed Asian trip visiting Japan, Korea, Taiwan. Strong
seasonal rebound has firmed up spending plans for 1999 and early intentions for 2000. Cap. Ex during
1999 up 6% (prior 3%) and 2000 up 14% (prior 12%). No longer have concerns about an order decline
at the end of 1999, but concerns about an order flattening (end-99/early-00) remain. Flat orders from
4Q99 levels leads to double digit growth during 2000. Most equipment stocks trading at 27-30x cal. 00
EPS. Focus on a case-by-case basis on companies in the photomask arena (ETEC and PLAB), tester
companies (CMOS) and situation specific cases such as LRCX.
--OPINION:------------------------------------------------------------------
Parts II-IV discuss the individual findings from our company visits in Japan, Korea and Taiwan, while
Part I analyzes the ramifications on the equipment sector.
Part I
Strong PC/Semiconductor Seasonality Instills Enthusiasm
With the strong seasonal PC/semiconductor market snapback, our Asian field trip abounded with
enthusiasm about the second half of 1999 and flat to up capital spending plans for 2000. As compared
to our prior trip in April, when there was tremendous concern about DRAM pricing and its effect on
capital spending, most Asian semiconductor companies believe that the strong seasonal snapback should
maintain pricing at the current levels in the second half of 1999, and as a result, capital spending levels
are flat-to-slightly up versus our forecasts (up 6% versus our prior estimate of 3%) and are expected to
trend steadily higher in 2000 (up 14% versus our prior estimate of 12%).
Our Concerns About An Order Decline Moved To The Back Burner
We had downgraded the top four companies in this sector (AMAT, LRCX, NVLS and KLAC) from a
Buy to an Outperform in May, due to our concerns of an order flattening/decline at the end of
1999/early 2000. Given the strong seasonal snap-back, the concerns of an order decline are behind us
and instead of mediocre order growth through the second half of 1999, we could very well see high
single digit to low double digit order growth in this time period.
Given the strong seasonal snap-back, we believe that the equipment sector can enjoy a tranquil 3-4
months, during which it would continue to mirror the behavior of the overall semiconductor sector.
Our Order Flattening Concerns At The End of 1999 Still Remain
However, our fundamental concerns of an order flattening at the end of 1999/early 2000 still remain.
Flattening orders/revenues from the 4Q99 levels throughout 2000 translates to a year-over-year growth
of mid double digits (11%-15%) during 2000, given the abnormally low equipment spending level in the
first half of calendar 1999. We still believe that these growth rates are prudent figures with little upsides,
given the SSB PC market forecast of mid-to-high single digit revenue growth during 2000, SSB
semiconductor forecast of 18%-20% growth and the current capital spending forecasts of most
semiconductor companies calling for 14% growth during 2000. A strong PC market growth forecast
during 2000 can create a case for owning these equipment companies, however, a more normal
forecast leads to a tough risk/reward over the next 3-4 months. We believe that companies in the
photomask area (Etec Systems and Photronics), which are yet to enjoy the 2000 discounting effect, the
tester sector (Credence Systems), and situation specific cases such as Lam Research can outperform
the overall sector over the next 3-6 months.
Japan Inc. Is Focusing On Product Differentiation Instead Of Brute Force Capital Spending.
While our Japan visit yielded capital spending figures equal to or slightly better than our model (20%
growth in 2000 versus our prior 17%), what was more interesting is the gradual evolution of the top
Japanese companies from commodity DRAM suppliers, to suppliers of differentiated logic and system
LSI products, that are attempting to minimize capital investment and maximize returns and profits.
NEC, Toshiba and Hitachi, which ranked #2, # 4 and #7 in worldwide semiconductor revenues in 1998
are currently spending 13%, 12.5%, 15%, respectively, of their 1999 revenues on capital spending. This
is exactly the same strategy followed by Intel (#1) and Texas Instruments (#5), which we estimate are
currently spending 10% and 14%, respectively, of their revenues on capital spending. We have always
believed that 20% of semiconductor revenues need to be reinvested back into capital spending to keep
the current facilities up to date. But in an era, where companies are focusing on technology buys and
controlled upgrade strategies, (which allow them to reuse existing fabs), we could very well be entering
a new cycle of a cost cutting recovery and not a full throttle revenue recovery like 1993-1995 or 1997.
The key takeaway from Japan was that having learned the hard lessons of competing in the capital
intensive DRAM market, Japan Inc., was focusing on system LSI and differentiated logic products in
an attempt to minimize investment and maximize return.
Aggressive Posture From Samsung And A More Controlled Approach From Hyundai-LG
Our Korean visits yielded capital spending figures that point to a 1999 capital spending growth of 70%
versus our prior model of 55%. The seasonal rebound in the PC business has given fresh hope to both
Samsung and Hyundai, which are both operating with lean inventories. Buoyed by a solid balance sheet,
and strong first half results, Samsung discussed its capital spending plans of 2.9 trillion won in 1999, up
significantly from 1.0 trillion won in 1998. The company reiterated its view that the winners and losers
in the DRAM business will be decided by who has the maximum sub 0.20 micron capacity. While
Samsung believes that 1999's accelerated capital spending should imply a slightly lower capital spending
in 2000, we believe that capital spending could very well be flat (our model) during 2000 given the plans
to begin Line 10, and continued technology buys in Lines 6-8. Hyundai-LG on the other hand guided us
to capital spending figures flat versus 1998 and in line with our forecasts.
Strong Foundry Trends And An Aggressive DRAM Posture From Taiwan.
While both the Taiwanese foundries and DRAM players guided us to capital spending figures roughly in
line to slightly better than our model during 1999, what was surprising was the aggressive posture of
several DRAM vendors with guidance for appreciably higher capital spending in 2000 (and the first
implementation of 300 mm R&D/pilot lines. Instead of our prior Taiwan estimate of a 21% capital
spending growth in 1999, we are now forecasting a 24% growth in capital spending. During 2000, we
had forecast a 2% decline, which now turns into a 6% growth.
Both TSMC and UMC are reporting 100% plus capacity utilization and good visibility over the next 6
months. While both companies talked about flattish 2000 capital spending, we believe that given the
strong outsourcing trends and the diverse product mix (50% of revenues come from non-PC
applications), we could very well see consistent 15% growth in foundry capital spending next year.
The Taiwanese DRAM vendors on the other hand (with the exception of Winbond) were excited about
their low cost manufacturing base, the high quality technology outsourced from Japan, and pointed to
year-over-year increases in capital spending in 2000. What was even more surprising was the view that
after the current empty shells are filled out the next round of fab construction will see the
commencement of 300 mm manufacturing. This is in stark contrast to the US, where the company
with the strongest balance sheet and the most robust business model, Intel, is pursuing 300 mm efforts.
ProMOS, Vanguard and Powerchip expect the current shells (constructed in 1997) to be the last 8"
fabs, with the next round of investments in 12" fabs expected to start production in 2001.
The Equipment Sector Is Still Predominantly PC Driven And Therein Lies Our Concern.
With DRAM vendors constituting 30%-40% of capital spending, Intel and AMD constituting 18% on a
combined basis, and Taiwan (PC dependence is roughly 50%) constituting 25% of capital spending, the
equipment sector is predominantly PC driven with close to 60% of the spending driven by PC market
dynamics. Salomon Smith Barney PC analyst Richard Gardner is forecasting PC unit and revenue
growth 24%-25% and l0%-12%, respectively, during 1999, and unit and revenue growth of 15%-20%
and 5%-8%, respectively, in 2000. We are forecasting the wafer fab equipment market to grow 16% in
1999 and 9% in 2000 driven by a 6% capital spending growth during 1999 and a 14% spending growth
during 2000.
Even Flattening Orders Lead To 11%-15% Wafer Fab Equipment Growth During 2000.
Using the data provided by Lam Research at their Semicon analyst meeting, the wafer fab equipment
market is expected to grow from $3.1 billion in 1Q99 to $4.1 billion in 4Q99 leading to a 1999 market
size of $14.8 billion (flat-to-up slightly from 1998). We believe that these are conservative estimates and
the takeoff ramp could quite possibly be faster with a quarterly wafer fab equipment market size of $4.5
billion-$4.7 billion by 4Q99. What is interesting is that even flattening wafer fab equipment revenues at
$4.3-4.5 billion per quarter through the four quarters of calendar 2000 lead to a very respectable
11%-15% growth rate during 2000, since the first two quarters of 1999 were very weak.
We Expect 2000 To Be A Steady Growth Year.
If the overall uptick in the PC/semiconductor business is secular and not just seasonal our new calendar
1999 and 2000 capital spending forecasts of 6% and 14% can be met with slight upsides to our 2000
growth assumptions. However, our KLAC and NVLS models already call for 30% calendar 2000
revenue growth, and 71% and 68% earnings growth. We believe that our revenue growth targets for
2000 are fairly aggressive with marginal upsides, however, given the excellent cost control shown by
the equipment companies, we could see slightly better earnings growth than our model. However, with
the stocks trading at or a slight premium to the market multiple (26 times calendar 2000 earnings for the
S&P 500), it is difficult to argue why these stocks should appreciate materially from the current levels
in the near term, unless one calls for a fundamental change in the PC business (full throttle revenue
growth recovery instead of a cost cutting recovery) and/or stronger growth in the non-PC areas such
as PLDs and consumer applications.
Case By Case Basis Analysis Is Necessary
In a tranquil 3-4 month period that we see the equipment market entering, we believe that it is necessary
to focus on a company by company basis at specific undervalued ideas. We like the photomask sector
with two main ideas, Etec Systems and Photronics. Both stocks are yet to price in the continued steady
recovery in the photomask business through 2000. Tester companies with almost no DRAM
dependence such as Credence Systems and situation specific cases such as Lam Research in the
front-end are companies that we would focus on. |
