View from Silicon Valley- Denali: Company Financials for 4Q FY 2004
* * * * *
Wall Street is out left and right upgrading semi and semi equipment forecasts based on assumptions both reasonable and
ridiculous. Seemingly reasonable, but un-dis-provable, assumptions all seem to revolve around a hoped-for
second half recovery. Until this materializes --or doesn't -- "analysts" are free to speculate. As long as they all speculate
along the same lines, none will get fired for being "wrong." There is safety, if not profit nor even preservation of capital,
in staying with the herd.
For better or worse, last week's "ridiculous" reason was Intel. CNBC openly expressed shock and dismay when Intel
raised the bottom of their quarterly revenue range and this "news" was not received by investors as a reason to rush
out and buy the stock. The Intel "news" can be seen as the profitless "buying the business" it really is when you read Denali's
dispassionate explanation below (originally published several weeks ago).
In short, Intel "wins" because AMD depends on profits from their flash business more than Intel. Due the their relative
sizes within Intel's finances, Intel can afford to subsidize the flash business with CPU profits almost indefinitely. That
is, perhaps, until AMD starts to win significant incremental CPU market share. While the reliability of AMD's execution can
be questioned, the recent anti-trust ruling in Japan on top of Turion64's roll-out may help AMD gain share over the next few
quarters. (Supporting execution concerns, AMD is suddenly out trolling for new 90nm and 65nm capacity to support hoped-for
CPU market share gains. Did they just now think of that small detail??)
As explained in some detail a few months ago by Fred Hickey, when Intel "buys" business, even at zero profit margin, it
keeps their fabs full. When Intel's fabs are thus filled, their accountants are allowed to claim incrementally higher gross
profit margins. This is "true" even when the additional production languishes in inventory, or Intel's distribution
channel, or they actually sell it but at zero profit. There is an old(!) salesman's joke about accepting a loss of $1
on very order because you can make it up in volume.
On a separate note, the spot price for 32Mx8 DDR SDRAM fell from ~$4.00 in early-February to ~$2.70 the last couple
weeks. The passing of Chinese New Year, and the subsequent demand decrease, is understood, if not widely reported, to
be the reason for the price decline.
It's important to examine your assumptions on an on-going basis. Even so, I have yet to see anything indicating
the SOX has genuine upside potential this year.
* * * * *
Company Financials for 4Q FY 2004
Overall, 2004 was a good year, with the chip market growing about 28 percent ($164B to $213B), and profits improved dramatically
from 2003. But it ended on a sour and uncertain note, which detracted from some of the cause for celebration. Memories grew
nearly 46 percent from 2003, and turned from loss to profit for most vendors.
But the course of the year was frighteningly like the 1999-2000 upturn, which crashed into bath of red ink soon after the
year closed; let’s hope the future looks better than what followed that strong cyclical market. Both markets built out
of a major downturn (1997-98 and 2001-03), both had barely recovered profit equilibrium before problems appeared (this time,
inventory buildup, and price pressure), and premonitions of tragedy from mid-year—2000 and 2004.
After the year closed, 2000 had the worst fears exceeded in the downturn of the following three years. The pessimists proved
to be optimists. This time, 2004’s rising market petered out in the fall, and opens 2005 with layoffs, CapEx uncertainties,
facility and manufacturing rationalization (just the beginning), and preemptive layoffs to reduce costs, and recover profitability.
The biggest difference is that 2005 promises to be a good global economy, while 2001-2 was a disaster amid the bursting
bubble economy of vast but still not fully appreciated consequences. Though the chip business marches to a somewhat different
drummer than the overall economy, it certainly benefits from strong demand, and decidedly from Americans who can buy foreign
goods with cheap dollars and have no apparent reluctance to go into debt and overspend their incomes, individually or collectively.
For memory makers, like the rest of the industry (though more consistent in their performance), 4Q04 was the time to say
goodbye to good times. In the largely communications segment, Cypress lost money and IDT also lost money. They both announced
layoffs.
In NOR flash, Intel’s aggressive flash pricing gained it market share but pulled AMD/Spansion into the red in 4Q04.
Intel has been red in flash for four years, and stayed there, though it has improved the profitability of the recently combined
Communications Division, in which flash is organized. SST lost money in NOR, Atmel substantially reduced its overall loss
from 3Q04, but we don’t know how well they did in their flash business unit. ST Micro, always tightly managed, made
money in memories overall, which are largely NOR flash.
Samsung was, as usual, unique, showing good profits in its memory division.
One analyst thinks Toshiba is more profitable in NAND flash than Samsung; as noted above, it is unlikely that any of the
NAND flash newcomers have been able to cost reduce as fast as prices have dropped. Sandisk noted a 30 percent price drop in
[flash] card prices in 4Q04, after a similarly large 22 percent in the summer quarter. Who can bring costs down that fast?
Taiwan DRAM makers, excluding ProMOS, which had not reported at press time, reported good profit results, while Micron
continued to struggle through its newly adopted "avoid commodity DRAMs" strategy, instead focusing on Cellular RAM, image
sensors (!), and legacy SDR DRAMs, and trying to get into the fast-growing
NAND flash market. They had reported their 1Q05 about a month earlier, as their fiscal year ends at the ends of August.
...
* * * * *
Data are taken from what are believed to be reliable sources and the author rejects any responsibility for their accuracy.
The above is intended for entertainment purposes and is not intended as advice to buy or sell any type of any stock, bond,
security, real estate or related product. Consult with a professional and invest and/or trade at your own risk.
