Semiconductors: 2009 Outlook
SECTOR REVIEW
Depth Positive, Duration Negative - Be Patient
and Nimble
■
Bottom Line On 2009. The Semi Cycle is in a bottoming process. The depth of
the current downturn is promising relative to easy compares off the bottom -
unfortunately continued depressed end consumption, excess inventory and
accelerating pricing pressures could extend the duration of any sustained
recovery until late 2009/early 2010. We expect stocks to be in a tradable range
for the 1H09, with a sustained stock recovery in the 2H09 in anticipation of a
2010 upturn.
o We are lowering our 2009 industry revenue forecast from down 9% to down
26% - consensus is at down 13%; In 2001 revenue declined 32%.
o We would note that bottom's up 2009 consensus revenue estimates
assume an aggregate revenue decline of 14% – more numbers cuts on the
way.
o We believe that 1Q sequential declines could be as bad if not worse than
the 23% q/q decline projected for 4Q08.
o On the positive side we see coming inflections in utilization, earnings
estimates, and y/y unit growth – have some exposure.
o On the negative side we see continued slowing consumption, excess
inventory, and increasing pricing pressure – don't chase rallies.
o Longer term, the 2010 upturn could be more powerful than anticipated as
slowing supply growth post 300 mm transition drives a "proper" cycle.
Key Points
■
Our new industry revenue estimates. We are reducing our industry revenue
estimate from down 9% to down 26% (the industry was down 32% in 2001) to
reflect a worse than expected starting point off of depressed 4Q08 levels, our
view that 1Q09 sequential declines could outpace 4Q08, and a more muted
recovery off of the bottom. Our new 2009 revenue estimate assumes unit
declines of 18% and ASPs declines of 10% versus 21% and 14% respectively in
2001. Of note pricing declines increase as sequential growth in units resume.
Our new model assumes 1Q09 represents the worst sequential revenue decline
this cycle, 2Q09 the worst year over year decline and 1Q10 the first quarter of
year over year revenue growth. Of note, current street consensus revenue
estimates assume 2009 revenue decline of 14% implying 1200 bps downside if
our tops down estimate is correct. For 2010 our preliminary estimate is for
growth of 10% - the average recovery year since 1995 has been 8%.

Our view on utilization, estimate revisions, y/y unit growth. Its not all bad news -
we see several important inflections occurring in early 2009 which would argue for at
least some exposure to the group right now. We believe utilization which peaked in
1Q08 at 90.5%, will decline to 77.2% in 4Q08 and 70.7% in 1Q09; ex-memory
utilization peaked at 79.5% in 2Q07 will decline to 60.1% in 4Q08 and 55.1% in 1Q09 -
it is likely that this will represent a near bottom. In addition, the estimate revision cycle
is bottoming as well - we have seen 14 months of downward revisions in semi EPS
estimates with a magnitude of approximately 76% - this compares to the average of
cycle of 10 months and 51% respectively, and the 2001 downturn of 14 months and
97%. While 1Q09 estimates are still wrong and too high in our opinion, 4Q08 earnings
reports should correct that issue. Lastly, we believe that y/y unit growth will bottom in
2Q09, at down 28% year over year - this compares to a 2001 trough of down 30% y/y -
historically stocks have discounted unit bottoms by one to three months.
■
Our view on end demand, inventory and pricing. Its not all good news either - we
see several factors which would prevent a V-shaped recovery for semi fundamentals
arguing that a more sustained recovery in stock prices may not occur until 2H09. End
demand is of course a significant wild card - as weak as end demand seemed in
C4Q08 (PCs 1788 bps below normal seasonal q/q growth, cell phones 989 bps below
normal seasonal q/q growth) - demand benefited by both Xmas and Year-end Budget
Flush - both disappear into C1Q09 - potentially exacerbated by the prospect that heavy
discounting in 4Q08 could have stolen from 2009 demand. As goes end demand so
goes inventory - although chip companies have aggressively cut production levels - we
still estimate that 4Q inventory days and dollars will increase 17.7 days and 4.2% q/q
respectively, with the potential for an additional 17.2 day increase (flat on a dollars
basis q/q) in 1Q09. In addition, semiconductor fundamentals tend to be sluggish until
end demand begins to re-test last cycles peaks - in the 2001 downturn that took 24
months, this cycle we are estimating between 12-24 months - as initial increases in end
demand are usually muted by accelerating pricing pressure. In 2001 from unit peak to
unit trough (as measured by y/y change) ASPs were flat, from unit trough to ASP
trough, ASPs declined 17%.
■
Our view on supply growth - to the patient goes the spoils. Importantly - especially
for the patient investors - we do believe that the next sustained upturn (beginning late
2009 into 2010) could be more dramatic than most anticipate based upon our view that
supply growth post 300 mm should slow structurally, leading to a better pricing and
inventory cycle than current consensus anticipates. Specifically, the semi industry has
not enjoyed a utilization cycle since 2000. During the 1990s, the average utilization for
the industry (ex memory) was almost 92% - since 2001 it has only averaged 81%.
From 2000 to 2004 under utilization was a function of too much physical capex, from
2004 to 2008 a function of the transition to much more efficient 300 mm wafer
processing - capex to revenue declined from 20.9% to 13.1% during this transition but
supply growth accelerated from 10-11% in 200mm to 16-17% in 300mm. Fortunately,
the 300 mm transition is now behind the industry which should lead to slower supply
growth, higher capex intensity, better pricing and most importantly a better inventory
cycle - as the supply chain reacts to a growing scarcity value for chips - in 2000 semis
were 12% of total tech inventory, today approximately 23%.
■
Our view on stocks. We are arguing for a trading environment in 1H09 with a more
sustained recovery in stock prices in 2H09 into 2010. Inflection points coming up in
1H09 would argue for more than zero exposure to the group. The prospect of longer
than expected duration, would equally argue not to chase rallies too aggressively. Our
top picks on the long side would include: LLTC, TSMC, MU, ONNN and INTC
(especially after 1Q guidance). As far as stocks we would be reluctant to chase - NSM,
ISIL, and TXN. Although relative valuation might diminish this comparison, we would
remind investors that during the 2001 downturn fundamentals started to deteriorate in
4Q00, while stocks did not find a hard bottom until 3Q02 - albeit we saw more than a

few rallies in-between of 30 to 50%. Lastly, we would look to turn more bullish on
individual companies based upon cost cutting and more importantly inventory
strategies - we fervently believe the semi industry should take advantage of the current
slowing business environment to reduce inventory on their own balance sheets - days
which averaged 58.2 days in the late-1990s have averaged 74.4 since 2004. Lower
inventories will help to create scarcity value when the cycle resumes and will indicate
whether or not a company is truly confident with their competitive position.