Apologies if we sound like a broken record, but we had to send this out.

The happy noise emanating from the semi business lately has been deafening.  Just in time for year-end bonuses and stock grants, the SOXX broke 500 a few days ago.  Surely this proves everything is wonderful here in chip land, right?

If you've been reading us for awhile now, you already know what we're going to say.  Repeat after me, "Let's look beneath the headlines."

Intel's announcement today narrowed their revenue range to the middle of their previous forecast.  This should be "no news," right?  I mean they're just hitting their numbers.  No harm, no foul.

The "problem" with this news is the stock has run up based on rumors floating around about shortages.  If true, this would normally imply Intel could raise prices.  Higher prices mean higher sales and higher profits.  Since Intel didn't raise their numbers, the stock was hit in after-hours trading.

In our thinking, even this reaction understates the "problem."  If Intel can't raise their forecast when parts unexpectedly get hard to find, what happens when this "shortage" (if there ever was one) is alleviated?  Hmmm...

The happy noise in semis has been particularly shrill from memory guys.  A couple days ago a reader sent a blurb on Micron touting increased margins in the face of declining DRAM prices.  Micron's move into NAND and low-power DRAM were reported, for the first time in history, as more than compensating for falling DRAM margins.  We have some bones to pick with this article, only a few of which are:

(+)Micron claims "only" 50% of sales are now in DRAMs.
(-)OK, so they're "only" getting clobbered on half their production, down from 80%.  How does that make them a good investment?

(+)Micron claims they are gaining share in NAND
(-)Their 2004 share was effectively zero, so what does this really mean?
(-)Micron only offers 2Gb.  Outside of the Apple deal or by cutting prices (and margins), how will they gain share against some of the most aggressive and successful competitors in the world?  Competitors who also offer higher single-chip densities (4Gb, 4Gb, 16Gb, ...)

(+)"Creating a joint venture company with Intel was a wise way to increase the company's ability to produce NAND flash memory and increase Micron's market share."
(-) Even at this stage of the growth cycle?  Intel has never built or shipped NAND.  What do they really bring to the table besides money?  (Not that I will be turning down "more money" any time soon.)

(+)"(B)oth companies are investing $5.2 billion in a new joint-venture company to produce NAND flash memory"
(-) $5.2B from Intel is, maybe, not a huge deal but where is Micron going to get $5.2B?  Will they issue debt? Or more shares?  Open a new line of credit?  Or all three?  Why would you rush out to own this stock before this info was revealed in detail?

(+)"The total size of the flash market is now approaching the size of the DRAM market."
(-)According to Denali (http://www.viewfromsiliconvalley.com/id185.html), the DRAM market was 2.5x the NAND market as of June, 2005.
(-)Even within the article, they admit NAND will "total $21.5 billion... compared to $25.8 billion," for DRAM by end-2006.  I don't know about you, but -$4.3B (or 20%) is not the "same" size.
(-)Even these numbers assumes NAND bits quadruple next year in the face of a 50% fall in price per bit.  It won't take much of a miss on the demand side to upset that "balance."

But the ultimate missing ingredient from this article is mention of CMOS Image Sensors (CIS).  Why aren't Micron CIS sales mentioned?  Did the reporter not know about this product?  Or did Micron not bring it up, perhaps due to lack of profits?  Hmmm...

Rounding out today's trifecta is news from DRAM eXchange(**) whose "noise" is not so happy:

In our never-ending search for consistency, we recall the long-touted 2H05 DRAM price recovery.  Rather than recover, however, DRAM prices continue to fall.

DRAM eXchange reports, even with Hynix dropping 32Mx8 DDR production, a number of alarmingly negative data points:

(-)"a ramp up of DDR 512Mb supply on the spot market."
(--) As we have warned, production is growing faster than demand, even in DRAMs (where suppliers were allegedly cutting back to engineer higher prices).

(-)"Micron is trying to bundle DRAM with their NAND."
(--) Alert readers quickly realize, unlike Samsung or SanDisk, Micron NAND is not sold out if they are selling it in the spot market.

(-)"Spot price of (DDR) 512Mb 64Mbx8 (dropped) 8.5% further from US$4.23 to US$3.90."
(--)i.e., 512Mb DDR has gone from commanding a premium vs. 2x 256Mb chips to being sold at a discount.)

(-)"Mainstream DDR 256Mb 32Mbx8 slipped 2.5% from US$2.05 to US$2" 
(--) a year ago, these parts were ~$4.20.  The collapse was initially pooh-poohed as temporary.  Then minimized in importance since "everybody" was converting to DDR2.  As per the Micron article, the latest tack is to claim DRAM pricing "doesn't matter."  Hmmm...

And, finally:
(-) "The limited performance leap of DDR2 533MHz and the relatively high platform cost structure discourage consumers to upgrade their system."
(--) "Intel... Delay(ed) DDR2 transition schedule."
(---) i.e., DRAM pricing is going to get even worse before it gets better.

And the NAND market is not immune:
"Demand is scare in the NAND Flash spot market this week. Shattered demand coming from China memory card makers and US MP3 makers showed up. Comparing to last year's Christmas season, this year's demand is relatively slow."

Raising the alarm level from even a few weeks ago, "(NAND) traders who were previous bullish about the spot market are turning bearish and have started to dump their inventory at all cost to avoid further profit loss."

Conclusion:   We would much rather just sit back and be happy.  We would probably live longer.  Preferring not to be older and poorer, we instead try to objectively evaluate actual numbers.