View from Silicon Valley- Are we just perpetually bearish?
(c) copyright View from Silicon Valley, 2005. All rights reserved.
With the SOX zooming from 382 to 430+ over the past month, is
it time to re-examine our assumptions about why we expect semiconductors to dis-appoint in 2005? Are we just
perpetually bearish? Are semiconductor sales and profit growth really strong? What's likely to change current
conditions in the second half of 2005?
Starting with the last question first, the call for the semiconductor
business to enjoy a second half recovery seems universal. Despite consistently-lowered forecasts from a broad array
of semiconductor and electronics companies this past quarter, "analysts" continue to sound the all clear and assure the
public it is now safe to go wading back into the technology waters.
"Analysts" are mechanically banking on annual, back-to-school,
Christmas and Chinese New Year seasons to generate upsides. Even if these annual events are indeed as strong or stronger
than usual, and there is no guarantee they will be, "analysts" are downplaying an important variable -- fab capacity.
"(W)e have estimated the total of worldwide 12 inch
fab capacity ramp up from 246K per month in the first quarter to 408K per month till the fourth quarter." This is +65% 12" wafer capacity increase in CY05.
"According to Samsung's Q1 conference call, its NAND
flash capacity is about 700K/quarter in the first quarter of 2005 and will reach around 800K at the fourth quarter of 2005 [this omits the impact of announced die shrinks]. Hynix...
has switched over 50K DRAM capacity to NAND flash since early 2004 and may allocate another 50K to NAND flash depending on
market situation... Infineon and Micron have slower penetration in NAND flash
market, currently, their NAND flash wafer in is less than 10K per month." [Both have plenty of incremental capacity available to pour into NAND once
they start to get market penetration...]
In other words, there is a LOT of capacity pouring into NAND
flash. That should be OK, since there will also be a lot of new demand for NAND flash, according to the forecasts.
Less prominently trumpeted in those forecasts are the price declines which will continue to accompany the volume
increases. Gartner places the pricing changes at:
2004: -25%
2005: -51%
2006: -45%
As we said earlier, if NAND volume projections miss by
even a little, the price projections will be much worse.
Now the second, or middle, question, "Is semiconductor
sales growth really strong?" As is often true, the answer depends on what you measure. I continue
to see DRAM pricing as an important indicator. DRAMeXchange goes on to comment:
"As DRAM prices dropped below cost, we
believe most DRAM makers will allocate more capacity to NAND flash for higher profitability, therefore, the DRAM bit supply
growth will be constrained at the end of 2005." Has ANYbody
pointed out current DRAM prices are below cost???
"Though most PC OEMs still requested 5% further
down on 256MB U[Unbuffered]DIMM prices, DRAM
makers have been reluctant to settle the prices below $20as some buyers have increased
the procurement quantity up 20% to 30%, besides, $2.25 in terms of 256Mb chip price is below most DRAM makers' cost already.
We believe the DDR 256MB contract prices have reached near bottom in the short term and DRAM prices may trend up in line
with seasonal demand. However, we have not seen shortage concern for DRAM supply in the near future, we don't
expect DRAM prices to rebound sharply as V shape in pricing curve. We have forecast DDR 256Mb 32Mb average price slowly
up toward $3 near the end of the third quarter."
The data does show DRAM prices have not fallen further
this week, breaking what seems like a string of 12 -14 weeks where ASPs constantly declined. DRAMeXchange is expecting "seasonality" in DRAM demand but their data continues to show flat prices. PC
houses continue to press for further price cuts suggesting, even with demand increases, there is little-to-no pricing
power.
Even if 256Mb DDR DRAM prices do get back to $3.00 would
be a thin margin, or worse, for most suppliers. Alert readers have already noticed the hoped-for recovery to $3.00 would still leave
prices well below the $4.00 ASP in effect ~90 days ago.
The opening question is still unanswered, "Are we just perpetually
bearish?"
In our view, there is still zero hard evidence of the widely-alleged 2H05
recovery in semiconductor sales growth. Even if aggressive unit growth targets are achieved, profitability will
seriously lag 2004's rates, (by up to 70%, according to Denali, whose profit numbers already assume unit-growth
targets are achieved).
It is possible the stock market might continue to value sales
growth, regardless of profit growth, and continue to award ever higher P/E's to tech stocks. However, it seems prudent
to think twice before plunging back into the tech stocks based on the widely-advertised, but little-evidenced, 2H05 recovery
story.
* * * *
The above is strictly for entertainment purposes and should be construed
as advise to buy, hold or sell any security of financial product. Always consult a licensed investment professional
before making any investment decision.
