I see cheerleaders

August 30, 2005

View from Silicon Valley- Subject to interpretation

Harking back to, "The Sixth Sense," where a boy says, "I see dead people," and is eventually discovered to be telling the truth, an EETimes article today caused me to spontaneously blurt out, "I see cheerleaders." Time will tell if I am just hallucinating.

This week's release from DRAMeXchange (free login required) reported nothing new. They continue to expect DRAM unit prices to rise 2H05 but also continue to only show very thin evidence. Our own interpretation was this week's release was particularly full of weasel words and equivocations, perhaps laying the groundwork to back off the oft-repeated 2H05 recovery prediction in the next few weeks.

Such suspicion seems reasonable after last week's DRAM eXchange report titled, "DRAM spot prices continue slide upon positive outlook," and the week before was, "DDR 256Mb spot prices drifted lower with sluggish demand." (Alert readers quickly notice neither of these reports was highlighted by the press.)

Much to our surprise, EETimes read this week's DRAMeXchange press release and rushed out the headline, "DRAM prices set to rebound in 2H." Even if you didn't read the press release and just went through the references in the article, "you can't get there from here." To wit:

"demand bit growth should lack behind supply bit growth by 4.4 percentage points in the fourth quarter"

--i.e., bits produced will grow 4.4% faster than demand, which should be a drag on prices.

-and-

“Escalating 12-inch wafer output and better-than-expected 90-nm migration should constrain the price in the fourth quarter.”

--i.e., and here's why bit growth will be faster than demand growth

In isolation, these two quotes from DRAM eXchange suggest 2H05 prices will be weaker, not stronger. Apparently, EETimes must have felt these points were trumped by:

“Both spot and contract DRAM prices should escalate, buoyed by the back-to-school demand in September,”

--Hey, it's already August 30. If there is/was going to be a back-to-school demand spike, wouldn't we have seen some evidence (in DRAM spot prices) by now??

-and-

"Taiwan’s first-tier motherboard makers will report a sequential 8 percent shipment growth in September, while the top six notebook ODM makers on the island are expected to see a 9 percent jump"

--Let's stop and review: 1H05 PC shipments surprised everyone with large upsides. Since DRAM prices tanked in the face of this surprisingly-large 1H05 growth, why would an objective observer expect a similar, but now widely-forecast(!), 2H05 demand improvement to push up DRAM prices?

As an aside, another EETimes article released today predicts SIA's overall July numbers will be up, specifically due to PC growth. Our own studies suggest SIA is far from a bastion of impartiality and neutral reporting. We feel safe in predicting SIA will trumpet even a small month-over-month increase, even after two consecutive m-o-m decreases, as further "proof" of the much-advertised 2H05 recovery.

In case time slips by you as quickly as it does for us, it should be noted we're already in the ninth week of 2H05. As the first third of 2H05 expires, the average spot price for a 32Mx8 DDR400 is $2.47. Last week (August 23) it was $2.53. The week before (August 16) it was $2.65. You call that a recovery?

Some argue it's OK it prices are down since DRAM makers will make up the smaller margins on higher margin. First, it's not clear how new 12", multi-billion dollar fabs are funded by making a few cents per part. More importantly, as estimated by DRAMeXchange, the industry-average manufacturing cost of the above-referenced 32Mx8 DDR400 is $2.60. There is a old salesman's joke about how a company was losing a $1.00 per part but making it up in volume. (At the risk of dating myself, I think the first time I heard this joke it was referring to Mostek and SRAMs.)

Conclusion: I'm sorry if it's boring repetition but you have to really read these reports before deciding to believe the headlines. The headlines may or may not have anything to do with the underlying details. Evidence for a 2H05 demand recovery is still missing. We don't expect the SIA report due this week will impact our assessment.