(c) copyright,
View from Silicon Valley, 2007. All rights reserved.,
Some of the worst investments we've collectively made
in our careers involved listening to the advice of someone else. Failing to understanding the risk was one
part of the problem but staying with something long after it started to fail was even worse. (I mean the guy
does this for a living and he sounds totally confident. Surely it's going to turn around tomorrow.)
For better or for worse, in our minds, something which is well-proven to have "failed" is any
assumption the little guy can rely on big Wall Street houses to give timely advice and/or preserve capital.
Despite this disdain, from time-to-time, we obtain research reports from Wall Street firms.
Today's rant stems from a Merrill Lynch (aka: "ML") report written by Tillman Ward, dated October 9, 2007.
Before
we run through the report, baseline data on the companies referenced seems appropriate.
Maybe it's just the sleepless fog from my head-cold, but extracting share prices and history data
on some well-known non-US companies turned out to be surprisingly complex. A cross-section of MSN Money,
UK MSN Money, Yahoo and the London Stock Exchange (hence the mish-mash dollars, Euros, Won and Yen currencies) yielded the
following:
The difficulty we experienced extracting these numbers highlights some obvious concerns:
1)
How do you know ML is giving you the true share price when you buy or sell on remote exchanges?
2) How do you
know if ML is giving you the true currency conversion rate?
2A) Would you like to wager whether or not ML is nicking
you for a currency conversion margin in addition to their trading commission and/or management fee?
3) After
you pay ML all these fees, wouldn't it be nice to get a bearish report BEFORE the average -22%
decline in these shares? (Or was the 20% drop just a mechanical bear market sell signal, after which ML changed
the "spin" on their data to reflect what the share price drop was telling them?)
And. most importantly:
4) If the semiconductor business, and stock market in general, are really so strong, why
are so many semi shares doing so badly?
Keep the above data in mind as you read ML's report (with our usual highlightsand/or comments embedded): * * * * * Contract
prices hit by spot weakness
Most memory chipmakers have started to admit that their October contract
prices for DRAM and NAND declined by more than what they had expected - about 20% for DRAM and about 15% NAND vs their previous
expectations of either flat or single digit fall from September levels. This is far weaker than normal seasonality and is
in contrast to solid sell-through of end products such as PCs (+12% YoY in 3Q) (PC unit growth the last
three years has yielded only flat revenue, so who cares?), handsets (+18%) (and who is making a
profit in cellphones these days?) digital cameras (+20%), etc. We attribute the supply glut and slow adoption of high
density memory to a weakening chip pricing environment (eg, slow migration to 2GB among low-end PCs, weak demand for 8GB NAND
from high end MP3 players and handsets).
Spot sentiment getting worse
Some
investors expect DRAM prices to recover soon given that the current spot prices are now close to cash cost levels for Tier
2 plays (i.e., sales but no profits). However, we believe a recovery, if any, will
just be a technical rebound and a short-lived phenomenon, as we witnessed this past summer - our analysis still
shows excess supply while channel inventories among spot distributors or memory module makers appear to be higher than our
previous survey (over one month as of early October vs 3-4 weeks a month ago). Furthermore, we are hearing of further price
pressure in NAND even at the current spot prices, which are down about 40% from the previous high of US$9.5 in August (8Gb
MLC). Since current NAND prices are still suggesting about 20% level of OP margins for the top 3 players - Samsung, Toshiba
and Hynix - we expect further price declines in NAND if suppliers remain aggressive on their NAND production vs DRAM while
NAND demand becomes weaker in 4Q, coupled with seasonal weakness in new orders. (NAND "price pressure" was
universally hailed as the precursor to geometric consumer demand increases. But now falling chip prices are suddenly
relevant? BTW, isn't this also a clear admission of slowing consumer electronics sales?)
3Q preview - Upbeat NAND margins were
not enough
We believe that memory chipmakers' 3Q profits will fall short of Street consensus or
their previous guidance due mainly to DRAM price weaknesses. (Didn't so many companies switch
from DRAMs to NAND that DRAM profits were assured?) According to our estimates, most Taiwanese DRAM plays
suffered bigger losses in September by posting negative 20-40% range of OP margins vs negative 10-20% in July-August. Even
Korean chipmakers' DRAM OP margins appear to be below 10% in September vs 15% level in July-August. (Wasn't
"global demand" supposed to ensure company profits in the face of weakening US consumers? If so, why are these
Asian suppliers doing so badly?) We also expect non-Asian plays Micron and Qimonda to record poor DRAM margins in a
negative 20-40% range for 3Q, whereas Inotera and Elpida should remain profitable, but only at a low single digit level (vs
our current estimates of high single digit levels) due to their internal transfer pricing (Inotera) and high exposure to non-commodity
DRAM (Elpida). We estimate that NAND margins were upbeat at the 20-30% level for the top three players in 3Q, but current
spot prices already indicate 5-10ppt margin erosions vs 3Q levels.
Conclusion: Despite the late-ness of this report, it may still
be useful if you own, or are considering owning, these shares of these types of companies. As always, trade (or don't
trade) at your own risk.
* * * * The above and any linked article,
website or advertisement are not intended as advice to buy, sell or hold any stock, bond, real estate nor any other financial
product or service. Buy and sell at your own risk (just like we do.)
