GS US Semi & SPE weekly - Inventory
datapoints accumulating; thoughts on
ongoing SPE multiple compression
1) A flurry of datapoints last week suggesting that distributors are building inventory
heightened investor concern about an inventory driven correction in 2006 and the semi
and SPE stocks closed down -5% and -6% w-o- w respectively. While many continue to
argue that inventory remains in check, we believe that last week?s datapoints, on the heels
of many similar datapoints over the last several weeks make it harder to argue that
inventory won?t be an issue in 2006. 2) We present data suggesting that the market is
paying lower multiples for SPE order growth this cycle than in previous cycles. If we are
correct that the stocks have already peaked for this cycle, we believe the lower multiples
this cycle are the best evidence yet that the market continues to recognize slower
long-term growth in this sector and is unwilling to put peak multiples on peak earnings as
they have in the past.
COMMENTS BY SEVERAL SEMI SUPPLIERS LAST WEEK SUGGEST
DISTRIBUTORS ARE BUILDING INVENTORY, ADDING TO OUR CONCERNS
THAT THE SEMI INDUSTRY HAS BEEN OVERSHIPPING AND MAY FACE AN
OVERHANG AS WE MOVE THROUGH '06. We have been highlighting for several
months now our concerns with the potential for building inventories in the semiconductor
supply chain given that semiconductor shipments are 11% above their trendline, the
furthest above normalized levels since 2000. Several additional companies last week
commented that distributors are now building inventories, including Texas Instruments,
Microchip and National Semiconductor (note that ADI had made similar comments on
their earnings call last month). In particular, National Semiconductor noted that inventory
levels at distributors went from just under 10 weeks to just over 10 weeks, implying a
5-10% sequential increase. Most notable from a stock perspective was that NSM sold off
3% on the day the company reported earnings, despite the fact that the company solidly
beat top- and bottom- line estimates, with the sell-off at least partially due to the inventory
build at distributors, and a 12% Q/Q inventory build on National's balance sheet. The
market's reaction to NSM's quarter reinforces our expectation that in the Q1 earnings
season, inventory will be more important than the P&L, given the stocks already discount
strong P&Ls based on rich valuations and on bullish comments by the companies at
investor conferences over the last two weeks. Last week's comments on rising inventories
at distributors are the latest addition to the list, which includes (1) CQ4 inventories at
Marvell, Intel, and Texas Instruments up 40%, 11%, and 10% respectively, (2) inventories
at EMS companies up 5% Q/Q in Q4, (3) inventories at Taiwan fabless suppliers
including Sunplus, Mediatek and Novatek up 25% Q/Q in Q4, (4) inventories building in
the NAND flash supply chain such that NAND spot pricing for some configurations is
down 50% QTD. While many have argued that absolute levels are still low and that
companies/distributors are only building because inventory levels were too low last
quarter, our experience suggests that once inventory builds start, the semi industry
historically overshoots and winds up in excess situations. At the very least, the strong
results that the chipmakers (especially those that recognize revenue on sell-in) are posting
need to be taken in context considering that one of the reasons that the results are strong
in the first place is that the disti channel is building inventory so the chipmakers results
are at least partially inflated by the inventory build.
We remain cautious on the semiconductor cycle in 2006 given above-trendline shipments and
building inventories were leading indicators for the downturns in 2000 and 2004. We would
continue to avoid deeply cyclical stocks such as AMD (U/N), and Texas instruments (TXN, U/N),
and instead focus on product cycle stories such as Marvell (MRVL, OP/N) and Broadcom (BRCM,
IL/N), on defensive analog names such as Linear (LLTC, OP/N) and Maxim (MXIM, OP/N), and
on the comm. infrastructure cycle through names such as PMC-Sierra (PMCS, IL/N).
MULTIPLE COMPRESSION HAS BEEN QUITE EVIDENT IN THIS SPE CYCLE AND WE
EXPECT MULTIPLE COMPRESSION TO CONTINUE AS THE MARKET INCREASINGLY
RECOGNIZES THE SLOWING GROWTH RATE OF THE INDUSTRY. We have argued for
some time that the growth rate of the semi equipment industry has slowed over time and is currently
likely roughly in-line with nominal GDP growth. While there has been significant debate in the
investment community over the past several years as to whether SPE growth rates have in fact
become as low as we believe they are, it seems as if the stock market continues to recognize the
slowing growth rate and is no longer willing to pay the same multiples for companies with less
robust growth prospects.
To that end, we examined multiples on peak forward four-quarter consensus EPS estimates for
AMAT, KLAC, and NVLS during periods of robust order growth (note that we have excluded
LRCX and TER from our analysis, given that both companies have undergone a restructuring in the
last several years which makes multiple comparisons across cycles more difficult). As Exhibit 1
(below) shows, multiples have continued to compress over time despite the fact that absolute order
levels this cycle have reached approximately the same levels they reached during the last upturn in
2004. We would argue that this multiple compression has been the result of investors increasingly
recognizing the slowing growth rates in the SPE industry and re-evaluating what the appropriate
multiples are for companies operating in a slower growth industry.
In April of 2002, Applied Materials' stock price peaked at $28 in anticipation of an order peak,
which occurred in the July 2002 quarter. In late 2003, Applied's stock price peaked at $26 in
anticipation of an order peak, which occurred in the April 2003 quarter. We believe that Applied's
stock price has already peaked at only $21 in early 2006, in anticipation of an order peak later this
year.
Similarly, KLA's stock price peaked at ~$70 in March of 2002, anticipating an order peak that came
in the June quarter. In early 2004, KLA's stock price peaked at $63, anticipating an order peak,
which occurred in the June 2004 quarter. Similarly to AMAT, we believe that KLA's stock price has
already peaked in early 2006 at only $55, anticipating an order peak later in 2006.
Finally, Novellus' stock price peaked in May 2002 at $53, anticipating an order peak which
occurred in the June 2002 quarter. In late 2003, Novellus' stock price peaked at $46 dollars,
anticipating an order peak in 2004. We also believe that Novellus' stock price has already peaked in
early 2006 at only $30 in anticipation of an order peak later in 2006.
While many bullish investors will argue that the data presented here implies that the SPE stocks are
likely to move higher this cycle until multiples or peak stock prices are closer to those achieved in
previous upturns, we would argue otherwise. We believe that SPE fundamentals are likely to
weaken as 2006 progresses driven by an inventory build at semiconductor device makers, which is
likely to lead to declining utilization rates and a slowdown in SPE orders. As such, we believe the
stock market has already begun to price in likely weakening fundamentals later in 2006, given that
the stocks have traded down meaningfully off of their highs over the past few weeks.
RECENT DISCUSSION WITH ADVANCED ENERGY'S MANAGEMENT GIVES US
ENCOURAGEMENT REGARDING THE COMPANY'S ABILITY TO FURTHER EXPAND
ITS PRESENCE IN POWER MANAGEMENT AND FPD. We recently caught up with Advanced
Energy's management and had the opportunity to hear the company's thoughts regarding its growth
2 Goldman Sachs Global Investment Research
March 12, 2006 Analyst Comment
prospects as well as its progress with the transition to Asian-based manufacturing. While our
discussion did nothing to assuage our concerns regarding the semi equipment cycle or our view that
SPE fundamentals will likely remain challenging as 2006 progresses, we feel more encouraged
about Advanced Energy's business this cycle compared to previous cycles, as we believe AE is now
better positioned to penetrate key customers like Tokyo Electron and AKT in the power
management segment.
Recall that Advanced Energy supplies mass flow controllers and power management solutions to
OEM customers such as Applied Materials and Ulvac (AE's largest customers), which span a
number of different end markets including semiconductors, flat panel display, and industrial. In late
2001, AE acquired Aera Japan (a mass flow controller manufacturer) with an intention to further
penetrate the Japanese market, particularly Tokyo Electron with its power management solutions.
While AE has not been successful in penetrating this market since the Aera acquisition, we believe
that the company may finally be positioned to gain incremental traction here, as:
(1) There
is currently a new management team in place at AE. To that end, we are encouraged by the fact that
the management recognizes the hurdles AE was unable to overcome in order to penetrate the
Japanese market over the last several years and now has a renewed focus on the power business in
Japan.
(2) We believe that AE's increased presence in Asia makes the company better positioned to serve
its Asian customers. For example, AE now has engineers based in Japan that are able to service
Japanese customers, which we believe was not the case several years ago.
Importantly, we believe AE's ability to expand its franchise in power management will allow the
company to further penetrate the flat panel display market (note that flat panel display is one of the
many applications that employ AE's power management solutions), which AE views as one its most
significant growth opportunities going forward (currently FPD accounts for approximately 15% of
the company's sales).
While there is currently no change to our U/C rating on AEIS, we would look to be more positive
on the name when/if we see the company gaining traction at TEL and AKT in power management
(note that AE currently has evaluation tools at both customers that may lead to design in wins),
which would help AE grow its top-line in an otherwise flattish business environment in 2006.
Additionally, we believe AE may be able to recognize some additional margin improvements in
2006 and 2007 as it continues to streamline its cost structure by moving to Asian based
manufacturing (currently 85% of AE's production is done out of its facility in China).
WEEKLY MEMORY MONITOR: Memory retail prices recovered slightly this week after a near
5% W/W decline in the preceding week. Specifically, retail prices increased ~1% W/W and were
down ~2% M/M. NAND spot prices declined ~3% W/W and ~29% M/M. We continue to expect
prices to decline as we move throughout the quarter as memory makers continue to attempt to
stimulate demand through price reductions and as incremental NAND capacity continues to come
on-line. (Please see Exhibit 2 for additional detail)
Each of the analysts named below hereby certifies that, with respect to each subject company and
its securities for which the analyst is responsible.. |
