4 Reasons I Believe The Rally In Semis Is Almost Over

This is typical of much sell-side
research. I received this note earlier this week.

"Japan Semi Industry Utilization stats out for 2Q"

(SICAS) – our sell-side analyst has a great albeit big bullet point that
summarizes it well: Semiconductor International Capacity Statistics (SICAS) put
2Q 2005 semiconductor capacity utilization at 88.8%, up for the first time in a
year after bottoming at 84.8% in 1Q, although this had been already priced in by
the markets. Foundry capacity utilization was up sharply from 75.2% in 1Q to
83.0%, although TSMC and others have already announced their figures, on which
the bullish outlook for semiconductor equipment manufacturers’ in April-June
results was based. Demand from domestic digital consumer electronic makers
meanwhile remains sluggish, and a long-term slide in discrete transistor
inventories started to slow in June. The three major stepper makers are
forecasting total shipments of new IC stepper products to go from 592 units in
2005 to 450-480 units in 2005, and 400-500 units in 2006. But Nikon’s relatively
bullish 2006 forecast of 500 units (10% YoY) apparently will be reduced to level
or even a slight decline if some is brought forward to 2005 to lift the 2005
forecast up from 450-480 units. The recently announced SEMI July SPE B/B ratio
was up from 0.90 in June to 0.93. This has been driven by growth in demand for
tester/assembly, which was up 13% MoM (water process equipment fell 5% MoM).
While the rapid shift to 300mm and accelerated miniaturization is bringing down
yields, tester/assembly demand apparently is increasing to maintain volume.
Although our semiconductor manufacturing equipment team is maintaining its
Neutral stance on the sector, relatively volume-driven issues such as
test-makers will likely see the benefits in the short term."

Now, why would an article with "Japan SEMI" in the
headline reference only US semiconductor numbers with no mention of SEAJ
(South-east Asia and Japan)? Similarly, SICAS (Semiconductor International
Capacity Statistics) is a global number and there are no significant foundries
in Japan.

As you already know the analyst’s statement "The recently
announced SEMI July SPE (aka semiconductor production equipment) B/B ratio was
up from 0.90 in June to 0.93. This has been driven by growth in demand for
tester/assembly, which was up 13% MoM " is literally false.

The June B:B was announced at 0.93 but revised down in the
July report. Second, both (3-month averages for) bookings and billings declined
month-over-month in July but, since billings fell faster than bookings, the B:B
actually ticked up. Big whoop… 

I think this sharp rally/spike of +20 to 25% in semis is close to over, then I
think we get a major selloff again in 2006. In terms of industry outlook, hard
to quantify much above 10% unit growth in the areas that consume 70% of semi
content (PC’s and handsets), and ASPs are not going up, so best case industry
growth around 10% in my view. The reason I think we get a rally: inventory
snapback to normalized levels at OEMs, EMS, distributors for fourth quarter
build. Then, I think we’ll get a selloff in early 2006 — no material upward
revisions. Street numbers currently bake in near peak margins for 2006, so it’s
hard to see material upside to the sector, or to the stocks in the sector, once
we move into early 2006.

Melanie Hollands

melaniehollands@yahoo.com