November 11, 2004

View from Silicon Valley- "It's over..."

(c) Copyright View from Silicon Valley, 2004.  All rights reserved.


Is the long-suffering Silicon Valley economy about to come roaring back?  Or is the semiconductor recovery on its last legs and about to roll over?  I am confused.  (I know, I know, you are not surprised...)  Maybe there is some evidence this week to help clear up the confusion.

It's widely reported the US lost 2M jobs over the last three years, with 200Ku lost in Silicon Valley alone.  Less well-known is only 6.4Ku (<0.8%) jobs have been added in Santa Clara County YTD2004 (net 300 since March) and the county workforce declined 10Ku.  WSTS, SIA and Craig Barrett are arguing over -2%, 0% or maybe +5% growth for 2005 worldwide semiconductor sales.  SEMI reports a declining book-to-bill for five months in a row with the B:B now below 1.00, indicating contraction.

Despite these numbers, empty R&D office space in Santa Clara county has declined eight months in a row and housing prices are up nearly 17% since October, 2003. (Updated county statistics will be posted on the web site in the next few days.)  The closely-watched NASDAQ is trying to confirm continued sales growth as it gropes toward its 2004 high.

One explanation here is "optimism."  Many people are convinced the worst of the tech industry downturn is behind us.  If you didn't lose your job in the last downturn, you won't lose it in the next.  If you haven't been off-shored, you won't be.  The state, county and municipal budget issues will all be painlessly resolved.

Clouding an accurate view is what can best be described as disinformation.  Mary Meeker, Harry Dent and Joseph Battapaglia are somehow all still all employed on Wall Street and still touting stocks.  (We should trust them, "Because..."?)  A couple reports claim any downturn in the semiconductor business can't be too bad since there is "only $1B in excess inventory" today.  (Didn't Intel just fess up to nearly $3B excess inventory of their own?)

Despite this "optimism," something still feels "wrong."  I hear the R&D office space numbers but see newly-emptied office buildings every week.  I read about housing strength but mostly hear from people who refuse to, or can't, buy at these levels.

Amplifying the dissonance, a radio ad today described an unemployed engineer who refinanced his mortgage, consolidating credit card debt onto said mortgage, "so that I could focus on finding a job."  If local mortgage companies now see this approach as a mainstream advertising strategy, they must believe there is a big audience around here receptive to such a "solution."

A great deal of confusion was straightened out this week when I came across, "Memory Industry Update," by Denali Software, Inc., in the packages given out at Memcon this week in San Jose.

Memcon is a semiconductor industry tradeshow focused on the memory market.  Two days of presentations give attendees insight into technology and market trends, for semiconductor memory products. 

As an aside, why Memcon's keynote speaker was from Xilinx is not clear.  My own inference is Xilinx had enough marketing budget to outbid the memory suppliers for the time slot.  Apparently, Xilinx's rationalized their claim, in the 44th minute of their 45-minute time slot, that building FPGA's is just like building memory would excuse the commercial to which we were all subjected.

View from Silicon Valley has been writing about many of the issues addressed by Denali for months.  It is always gratifying to see other independent research coming to similar conclusions.  (For the complete article, visit www.denali.com or the "stats" page at View from Silicon Valley.)  Highlights of Denali's "Memory Industry Update" include:

"More signs of weakness in the chip business have come in over the five weeks since the last report."  (i.e., all the bad news is not priced into the stock market.)

"Fourth quarter capacity utilization will again be reduced from the preceding quarter, signaling the reemergence of capacity excess, even at the leading edge.   The measures of inventory buildup, at chipmakers and at end users, which have been with us for more than six months, are more persistent today." (i.e. estimates of $3B inventory, let alone $1B, are too low.)

"Either our expectations were too high, or the 'quick-response industry capacity' was too great.  Capital investment, though paused for a few months late in 2003 and early 2004, quickly sobered up and began apace, chasing the demand surge.  Most of this new capacity remains to come into production, but we ’re already turning from capacity shortage to excesses.  If we ’re already caught up, what will that mean in 2-3 quarters down the road?" (-- i.e., this is not just a "soft patch.")

"The industry rarely doddles with growth or declines of a few percent, unless it is scraping along the bottom, as in the early 1980s or early 1990s...  2004...is unlikely to be followed by a few percentage change in either direction in 2005.  It will be more, if there’s another strong growth spurt in demand.  Or it will be less, or much less, as it now appears to be faced with a weakening demand and more and more capacity coming in to service it." (--- i.e., SIA, WSTS and Craig Barrett are all too optimistic.)

"Increasingly, this enjoyable and profitable market looks like the 1999-2000 uptick, only perhaps even more truncated.  So deep was the reversal of 2001-03 that the industry spent three years of growth (2002/01 = +1.3 percent; 2003/02 = +18.3 percent, and 2004, about 26-29 percent), just getting back to where it started in 2000, with only one truly profitable year, 2004, in which the industry will reap $15-25B in profits."  (What do soft prices do to P/E ratios?)

"We are late to a conclusion that others have embraced for some time...  To wit: The industry is so large and diverse, and growth has slowed so much, that almost any imbalance in supply and demand, which imbalance is critical to the financial health of the industry, is almost immediately extinguished by yield and productivity improvements..."

"(M)uch of the industry has been investing in R&D and capital with the expectation that revenues (and profits) were destined to grow either at the historical rate of 15-17 percent, which has been steadily trending down to 12-13 percent recently, and now being revised again down to 10 percent.  We believe now that even this is too high: the 1995 peak was $144B,and that was a decade ago —so it ’s been growing under 5 percent, year to year, for a decade."  (--a market growing only 5% demands mergers and consolidations.)

"Much of the investment in the 2001-03 and 2004 period can be expected to never return a nickel, so quick was the uptick terminated, with only a few vendors able to gain profitability for adequate ROIs, except in a sustainably short market."  (--i.e., Most semi companies need a few quarters of "allocation" pricing to make enough money to live through the ensuing down cycle.) 

"(U)nless there is a 'December surprise' in demand awaiting us under the Christmas tree, 4Q04 will not be that good, either.  This will leave us to face 2005 with a weak order book, increasing output, and more price competition."

Denali's accompanying power point presentation started with the line, "It's over..."  Barring a miracle surge in end-product demand, Denali predicts the current semiconductor industry recovery is...