|
|
 |
|
|
|
December 13, 2006
The NAND price recovery
(c)
copyright View from Silicon Valley, 2006. All rights reserved.
The health of the semiconductor business
used to be driven by the strength, or lack thereof, in DRAM prices. If DRAM prices were up, or even flat, the business
was obviously growing. Demand was growing as fast, or faster, than supply.
However, one our first missives
published in 2003 questioned if DRAM was really still the horse drawing the semiconductor cart. (Please see: "How Key
Is DRAM?", August 1, 2003, http://www.viewfromsiliconvalley.com/id5.html) We thought analysts predicting semiconductor sales strength or profits based on DRAM results were
missing a sea change in which DRAM unit prices would be less-critical to semiconductor health. Nothing in the intervening
years has developed to change this opinion.
What has changed in the last three years is the acceptance of NAND as the
new arbiter of health of the semiconductor business. It has been further assumed that NAND price declines are normal
and even expected, since bit growth is very high.
We do not disagree there is NAND bit growth, We do question,
however, if this bit growth is truly off-setting unit price declines. We also wonder how multi-billion dollar 300mm
fabs are justified based on the thin profit margins likely to result from such price declines.
The first step is easy.
Let's look at relative NAND prices this year. We first took a crack at NAND prices in March:
Jan01$ Mar27$ Mar27%
1Gb $8.95 $4.12 -54%
2Gb
$8.89 $5.60 -37%
4Gb $14.98 $9.44 -37%
8Gb $57.40
$24.68 -57%
16Gb $84.78 $39.00 -54%
The cheerleaders said not
to worry about prices falling 40% to 50% in less than 90 days. Just as price and demand strength were common in the
fourth quarter of each calendar year, price weakness in the first calendar quarter of any year was "normal," even "routine."
NAND prices were predicted to quickly snap back based on strong demand.
The price update in April, however, was mostly
worse.
Jan01$ Mar27$ Mar27% Apr25$ YTD%
1Gb
$8.95 $4.12 -54% $4.00 -55%
2Gb $8.89 $5.60
-37% $5.00 -44%
4Gb $14.98 $9.44 -37% $8.00
-47%
8Gb $57.40 $24.68 -57% $17.58 -69%
16Gb $84.78 $39.00
-54% $35.75 -58%
Still not a problem, said the cheerleaders. NAND prices will recover
for the summer "season" as manufacturers ramp up for back-to-school sales on top of new iPOD models.
By May, the rationalization
by Semico was, "Instead of a price collapse we expect to see prices drift lower to trigger additional demand, firming for
a while later in the year." (http://www.viewfromsiliconvalley.com/id227.html)
So let's look at July prices (http://www.viewfromsiliconvalley.com/id246.html ):
Jan01$ Apr25$ AprYTD% Jul31$ YTD%
1Gb
$8.95 $4.00 -55% $3.04 -66%
2Gb $8.89 $5.00
-44% $4.73 -47%
4Gb $14.98 $8.00 -47% $7.49 -50%
8Gb
$57.40 $17.58 -69% $12.40 -78%
16Gb $84.78 $35.75 -58%
$25.67 -70%
Hmm... Prices fell -24% at 1Gb and -29% at 16Gb densities just within 2Q06. if
this is "drift lower" wouldn't you have to see what somebody would admit was a collapse??
Then by October, we found:
Jan01$ Apr25$ AprYTD% Jul31$ YTD% Oct23$ YTD%
1Gb
$8.95 $4.00 -55% $3.04 -66% $2.76 -69%
2Gb
$8.89 $5.00 -44% $4.73 -47% $5.01 -44%
4Gb $14.98
$8.00 -47% $7.49 -50% $7.89 -47%
8Gb $57.40 $17.58
-69% $12.40 -78% $15.33 -73%
16Gb $84.78 $35.75 -58% $25.67
-70% $33.98 -60%
What do you know? October prices were actually higher than July! The
YTD losses recovered to "only" -44% /2Gb to -73% /8Gb.
Does this explain why the SOXX and the NASDAQ have been running
since ~late-June?
Now in December, we see:
Jan01$ Apr25$ Jul31$
Oct23$ YTD% Dec11$ YTD%
1Gb $8.95 $4.00 $3.04 $2.76
-69% $2.52 -72%
2Gb $8.89 $5.00 $4.73 $5.01 -44%
$4.17 -53%
4Gb $14.98 $8.00 $7.49 $7.89 -47% $6.89 -54%
8Gb
$57.40 $17.58 $12.40 $15.33 -73% $11.38 -80%
16Gb $84.78 $35.75 $25.67 $33.98
-60% $22.09 -74%
The numbers suggest October represented "firming" from July but by December all the gains
were given back? And then some?
At the point when prices are classically their highest, and demand strongest,
NAND prices have fallen to new CY06 lows?
It would be VERY interesting to go back and find out how many NAND IC's actually
changed hands during the price "recovery" leading up to October 31. Was there really a demand spike? Or were
brokers just bidding up prices while they loaded up on inventory in anticipation of a seasonal increase?
Conclusion:
Since
we now see 3Q06 price gains bled off, and then some, in only about 6 weeks, the case for broker demand, and lack of seasonal
strength, can be made.
The oft-promised NAND price "recovery," if you could call it that, peaked in October. NAND prices are today
lower than ever.
Could a similar result on SOXX and NASDAQ stocks be forthcoming?
*
* * * * *
The above is 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.)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|