August 8, 2006

NAND prices vs. iPOD sales

(c) Copyright, View from Silicon Valley, 2006.  All rights reserved.



We last visited the NAND market April, 2006 , so let's stop back in and see how they're doing.

You may recall our warnings on excess fab capacity being created to build NAND flash starting about 18 months ago.  On top of new fabs being built left and right, all the DRAM makers are hastily converting DRAM capacity over to NAND.  NAND is universally seen to be a higher growth track and used in a wider variety of applications than DRAM, not least of which is the Apple iPOD.  Compared to DRAM's relative dependence on PC sales, everybody knows NAND demand is well-diversified.

Hopefully we all know by now that any market explained using "all", "universally" and "everybody knows" is ripe for re-examination.

Updating data through July, we find:

       Jan.01$  Apr.25$ Apr-YTD% Jul.31$  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%

Of course, senior managers at major semiconductor houses are not stupid.  Nobody invests $1B+ to build a commodity-oriented fab, regardless of how "hot" the commodity may be at the time, without expecting price volatility.  However, the hype has been so loud over the last ~18 months that it's hard to imagine Samsung or SanDisk/Toshiba or Micron or ANYbody expected price declines of more than -50%, let alone the >-70% declines in 8Gb and 16Gb parts over the first seven months of 2006.

There is also commentary from DRAM eXchange about Samsung holding back volume from the market to help hold up the price.  This gives cover to various brokers to dump parts but, at best, delays further price declines only momentarily.

As an aside, 8Gb and 16Gb chips were expressly touted as iPOD Nano solutions and they have the biggest price declines.  This strongly suggest 1GByte-and-up iPOD Nano sales are failing to meet forecast.  (Or perhaps Apple's mythical iPOD phone or next-generation Nano were built into these forecasts but have now slipped or been canceled?)

Of course, NAND prices declines bodes well for Nano margins, but declining unit sales may be deflating the anticipated profit gains:

4Q05  14.0M
1Q06   8.5M
2Q06   8.5M  (vs. 8.8M forecast)
3Q06   7.6M  (vs. 8.1M prior forecast)

Last month, forecasters expressed "increased confidence" MAC sales growth would offset iPOD deterioration.  Knowing all Apple's profits are in the iPOD was apparently insufficient to convince the market to ignore the just-in-time, feel-good stories.

This month MAC sales were admitted as slow (as we also predicted when the Intel conversion was announced) but the certainty of a holiday-inspired iPOD rebound was touted as making even this development irrelevant.

Some months back, the iPOD product line was ~60% NAND iPOD vs. ~40% disk drive-based.  With ~20% selling as low-end shuffles, only ~40% of ~40M, iPODs consume only ~16M Gbit-sized NAND chips per year.  If we arbitrarily assume the non-shuffle Nano's average 1GByte, Apple effectively consumes 1.33M 8Gbit chips per month.  This is less than Samsung alone can produce, not to mention the other umpteen suppliers vying for a piece of the NAND pie.  (Including Apple itself via their $500M investment in the Intel/Micron NAND venture.)

Apple selling 39.4M (current forecast) or the previously-forecast 43M or even 50M iPODs this year doesn't have much direct effect on the global market price for NAND.

As another aside, Nokia announced a music phone this week.  They expect to sell 40M in the first 12 months.  This strongly suggests Apple indeed does (or did) have a cell-phone based iPOD in the works.  After needing years to hit 40M/year standalone music players, Nokia forecasts their phone/music combo to hit 40M within the first 12 months. (This Nokia 3250 is iTunes-dock-able, by the way.)   Six (or even 26) other phone suppliers offering similar features are undoubtedly right behind Nokia.  Apple cannot be missing the volume advantage in cell phones, nor the threat to their iPOD money machine.

At risk of going a factoid too far, we'd like the see the (alleged) data behind Gartner's announcement today the world will be short "some 23 new 300-mm fabs by 2008."  There are already 30 or 40 in the pipe, according to Applied Materials.  Unit prices and even top-line SIA sales actually were down in last week's announcement.  Fellas, the "trick" is to identify when the trend will change. Any fool with a ruler and a $4.99 calculator from Wal-Mart can extrapolate from what is already reported.

Conclusion:
NAND prices are even more terrible than last quarter.  The highest-density parts crashed the hardest.  Yet a new NAND fab seems to be announced every other week (see "SV News").   Apple gets credit for driving NAND demand but the numbers suggest they are only a small player with their iPODs.  For the record, we choose to invest in neither.

We will continue to track figures for both types of device but hold no illusions that either dictates the conditions of the other.

* * * * *

The above is not intended as advice to buy, sell or hold any stock, bond, real estate nor any other financial product or service. Invest at your own risk.