October 19, 2006

Fab Shakeout?

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


Wading through the constant cheerleading which in Silicon Valley substitutes for actual news reporting, we occasionally come across a gem.  Something which states in very matter-of-fact terms info which local industry icons cannot be trusted to admit.

We used to find hidden upsides and a steady stream of investment ideas due to this work.  We have termed these gems "edge" and have written about "edge" a few times, including:
"Looking for an Edge," "More Edge" and "Does wafer growth impact profits?"

These pieces were hotly disputed but, reviewing their contents today, we didn't miss out on any opportunities.

Today's candidate for "edge" comes from the cover of EETimes.  Even though it was a cover story, we find no MSM mention of the facts reported.

We encourage you to read the entire piece but here are some highlights from "Shakeout may be next for foundries"

"Simply put, the foundries are getting hit by a one-two punch. On the one hand, they are suffering the repercussions of an on-going IC inventory-correction problem. Vendors do not expect a significant pickup in demand until mid-2007 or later."
--What?  An inventory-driven correction?  No significant demand pick-up "until mid-2007 or later"???  Did we miss a memo?  Hmmm...

"(A)s the business slows, new competition has arisen as several big integrated device manufacturers--Fujitsu..., Infineon..., and Samsung... --have separately expanded their own foundry efforts... They tend to expand their foundry offerings when business is soft and they are desperate to fill their empty fabs..."
--Weren't the big guys supposedly rolling in dough (i.e., strong sales)?  But now they're "desperate" and so slow that they're effectively taking in laundry?  Could actual profits be so marginal that they're better off selling wafers than building and selling their own parts?  Doesn't this suggest most "fabs" can easily become a "foundry"?

"(I)t's a little-known fact, but only a handful of foundries are actually making money, ... Most vendors have never made a dime and continue to spill red ink, leaving some to wonder about the future of the weaker players. Soaring fab costs, mounting losses and the difficulty of finding and holding on to customers are among the challenges..."
--If only a handful of foundries actually make money, why is a new fab announced every other week?  Could it just be cheap money?  (As opposed to actual business plans which involves turning a profit?)

""The strong are getting stronger," said Mehdi Hosseini..., "Some of the second- and third-tier companies will have to consolidate. And you are not going to see any new players in the market, because the barriers to entry are too large.""
--Again, there are waaaay too may new fab announcements taking place for the "no new players" statement to be true.  (At least up until now.)  If it were true, what are the prospects for SEMI?  As we wrote September 1, 2006 in "SEMI-confusion,"  (http://www.viewfromsiliconvalley.com/id256.html) evidence suggests SEMI-company stock seem more likely to fall than rise from recent levels, regardless of new fabs.)

"It's still uncertain which vendor will become the next victim in the shakeout, but the economic indicators remain painfully clear: The soft business climate in the foundry sector is expected to extend into 2007 and perhaps even the beginning of 2008, analysts warned."
--Soft into 2008!  Has anybody heard this said out loud?  (Except us, of course, who claim top-line sales growth has masked deteriorating profitability for the last nine to 12 months...)

"After experiencing a boom in the first half of this year, silicon foundries began to feel a ripple effect in the summer, when their customers started to cut back on wafer orders amid what was then perceived as a minor and unforeseen IC inventory correction."
--Customers started cutting back wafer starts mid-2006?  Where was this publicized???

"Chartered ...gave a disappointing forecast ...due in part to soft sales of game-machine chips. Sources believe that Microsoft Corp. pulled back its orders for the Xbox 360."

"(F)oundries were hoping for a short-term correction and a rebound in the second half. But the IC glut has lasted much longer than many expected. Continuing to feel the inventory pinch from their cautious customers, foundry vendors experienced a disappointing third quarter, and analysts expect the fourth quarter to be a bust."
--Now the inventory correction is an "IC glut"?  Which already "lasted much longer than many expected"?   Who even admitted there was a glut in the first place?

"Gartner Dataquest lowered its foundry growth forecasts for both 2006 and 2007, due in part to the inventory correction. The foundry industry is projected to grow 16.6 percent in 2006, down from 18.8 percent in the original forecast... The business is expected to slow to 12.8 percent growth in 2007, down from an original projection of 15.5 percent..."
--Is this like the housing bubble?  A soft landing in the face of a huge glut?  Just like housing, the next soft landing in the semi business will be the first.

"(B)y this year's end, average fab utilization rates among the foundries are projected to fall from the low-90 percent range right now to mid- to high-80 percent levels."
--We maintain capacity utilization is only an indirect indicator of profitability.  Even so, given the cheerleaders' reliance on this metric, they might actually notice a problem --sometime AFTER it has become apparent to everyone else.  (Or at least to everyone who read down this far today.)

"Most other analysts also expect a slow fourth quarter, but some have a slightly different viewpoint about 2007. "We are going to see a dip in the first quarter," said Joanne Itow,...with Semico... "It will be a prolonged slowdown, but we expect to see things pick up by the end of '07.""
--Are "slow fourth quarter" or "prolonged slowdown" built into the Dow And NASDAQ flirting with 52-week, even multi-year, highs?

"Itow does not expect to see a price war in the foundry market, but she said that average selling prices for wafers are falling faster than previously expected..."
--Again, just like housing, excess capacity will magically be consumed with NO price decreases.  Maybe if I had an MBA, I would actually understand how this was likely --or even possible.

"The overall foundry market is expected to grow (dollars?) by 30 percent in 2006 and 17 percent in 2007, according to Semico Research. Foundry wafer demand is projected to grow by 26 percent in 2006, slowing to 17 percent in 2007, according to the firm."
--Even these fantasy top-line dollar figures suggest slowing profits.  (Fewer dollars per wafer start...)  Historically, when semiconductor profits start to fall, they do not slide gracefully but rather tank quickly as vendors race to be the first cut prices and move inventory.

"The first quarter of 2007 was originally expected to fall by 8 percent, but now the analyst suggests that the market could be "flat to up," thanks in part to demand for MP3 chips and broadband devices. Microsoft's Vista operating system software is expected to jump-start the industry by early 2007, he added."
--Oh yeah, Xbox 360 unexpectedly cut back but MP3's and Vista are startlingly new developments which will magically rescue the foundry business from over-capacity.  And I've got a worked-out potato farm in Idaho which is going to be the next great fab!

""We expect below-seasonal guidance for [TSMC] in the fourth quarter, affected by inventory work-down by TSMC customers," said FBR's Hosseini. But wafer shipment in next year's first quarter "could turn out better than seasonal [patterns]."
--TSMC and UMC unexpectedly missed recent sales and earning figures.  And they will be down in a the fourth quarter (historically the strongest calendar quarter).  This means they will next unexpectedly exceed future numbers.  Yeah, right.  That's the ticket!


Conclusion:
The people in the article seem to believe the "chip glut" and "slowdown" will be resticted to the fab portion of the semi business.  As though a "glut" will have no effect on the rest of the semi business.  We think that's the equivalent of finding an excess of oil in the middle east but it won't drive down prices in the USA.

"(A)ccording to an FBR report. "This is a cyclical business."'
--Finally something we agree with!  If the cycle started March, 2003 (or even October, 2002) what part of it do you suppose we are in today???