And I should mention that the weather has been glorious this week--lows in the 50s and highs in the 60s and low 70s. I could go on and on, but you get the picture. It sure beats the heck out of Broadway and the French restaurants along East 55th street. Really.

Excess inventories
just won't go away

The bad news just keeps getting worse. Another key, chip-market watcher has gone increasingly negative. This week, iSuppli revised downward its forecast for this year. The culprit is too much inventory, which is continuing to put a damper on reordering and prices.

Despite all-out efforts globally to cut chip inventories, the industry's supply chain is still awash with too many semiconductors--about $8 billion in excess inventory, according to the El Segundo, Calif., research firm. Getting rid of inventory has turned out to be a slower job than iSuppli expected and this has resulted in average selling prices continuing to drop.

It now expects global chip sales to drop 28% this year to $147 billion. "While there is a modicum of good news for the personal computer, computing platform and wireless market segments with replenishment ordering for components beginning, we are revising our forecast for worldwide semiconductor revenue this year downward to $147 billion -- an annual decline of 28%," said Greg Sheppard, vice president for market intelligence services.

"We now see the market decline hitting bottom in the third quarter," says VP Greg Sheppard. He sees revenues dropping another 2% from the second quarter, then turning up 4% in the fourth quarter, "thanks largely to a traditional seasonal boost from consumer purchases toward the end of the year."

More than three-fourths of the excess inventory resides on the shelves at semiconductor suppliers and contract systems manufacturers, the market researcher says. And 60% of the excess inventory held by contract manufacturers was in the networking and telecom sector. Sheppard warns this hangover "could well last until the second half of 2002 when capital spending upgrades and major infrastructure build-outs in China finally are expected to kick in." This puts him squarely in line with my six-month old forecast.

As far as prices are concerned, Sheppard expects commodity semiconductors to hit bottom this quarter or next and "linger there until the spring of next year." He advises component buyers to start locking in long-term price contracts by the end of the year "to take full advantage of the underlying glut of inventory and foundry capacity that is driving this trend."

(See Aug. 13 story.)

Things are serious--Moto Semi
will close 33-year-old Mesa site

Motorola continues this week to reinforce its crisis situation--at least in my head--by closing down two of its core chip making facilities.

The big chip maker is closing down its 33-year old Mesa, Ariz., site, which currently houses its three-decade-old Bipolar Manufacturing Center and its 20-year-old MOS-6 wafer fab.

The shutdown will affect 1,200 workers. Not all of them will be lose their jobs, as Motorola hopes to "transfer a majority of the Mesa employees to other jobs," says Christ Belden, vice president. "We plan to expand production at our MOS-12 facility in Chandler and CS-1 facility in Tempe," he says. Motorola's chip plants are now operating at 50-to-60% of installed capacity.

The bipolar line, which turns out chips for a variety of applications including automotive systems, will be closed down over the next 18 months.

Production will be phased out at the MOS-6 fab over the next 30 months. A few of the devices produced at this plant, which makes complex radio-frequency products for wireless communications and infrastructure customers, will be transferred to other plants.

Motorola opened the Mesa site in 1967 and began wafer fabrication at its original Bipolar-1 fab in 1969. Other fabrication lines were added to the site over the next 12 years. The MOS-6 fab began production in 1981, with the Bipolar Manufacturing Center organized in 1998 by the consolidation of three bipolar production lines into one facility.

(See Aug. 15 story.)

'Motorola's greatest strength is
ability to come back from dead'

I was taken to task gently this week for suggesting that Motorola was "eating its seed corn" by selling its defense and government systems business. While I and others consider this unit part of Motorola's core communications business, Stan Livingston, retired sales manager at Motorola Semi, maintains it always was a "non-core business," adding he was "very pleased" to see Motorola had sold it. "Sales growth was never a focus with this division although they have always been profitable."

The former Moto exec goes on to say his company got into its current troubles by not concentrating on its core communications business. "The idea for semiconductors was probably a good one, but management over the years strayed form the original reason for creating Semi," he says, which was "being a way to advance the state of the art in their communications business. Unfortunately," Livingston adds, "inter-division rivalry and blatant jealousy caused Motorola to focus on being the largest semi company in the world."

As a result, Moto Semi moved from being a "support group to a full-fledged business, which sucked up an astronomical amount of R&D money--money much better spent in the communications world," he says. Then, he points out, "while Chris Galvin, CEO was busy trying to fix the Semi division eight years ago, his communications business took their eye off the cell phone ball and continued developing analog phones while the rest of the world was focusing on digital phones."

But Livingston has not given up on his former company. "Motorola's greatest strength," he says, is their ability to come back from the dead. They have done it several times over the years. Unfortunately, the cruel business world makes such feats more difficult to do each time it is attempted. I hope Chris has one more comeback."

(See Aug. 9 story.)

Applied quarterly sales
were down by 51% . . .

The way Wall Street works these days, the news this week from the world's largest supplier of chip equipment was good--Applied Materials beat Street estimates for its fiscal third quarter by a penny-or-so per share.

But don't let that confuse you. Applied showed a 30% sequential drop in revenues to $1.33 billion from $1.91 billion in the previous three months. And these sales were 51% lower than the $2.73 billion reported in the same quarter last year. From where I sit that's a disaster, even in a boom-and-bust market like semiconductor production equipment.

And it gets worse. New order bookings fell a shocking 63% below the $3.28 billion written in the same quarter last year. Orders declined sequentially by 11% to $1.21 billion from $1.35 billion in the second fiscal quarter.

Gross margins could have been worse, dropping from 44.8% in the previous quarter and 50.9% last year to 40% for the quarter just ended. And Applied still made money in the quarter--$28.5 million net is better than a loss any day.

(See Aug. 14 story.)

. . . but chip tool business now
appears to be stabilizing . . .

Applied Materials management says that its business appears to be stabilizing and expects the production equipment maker to remain profitable in the current quarter.

As far as I can see, Applied has become somewhat more bullish. The equipment giant believes the sequential decline in new tool orders is ending at last and now expects to see "some modest improvement in early fiscal 2002" beginning in November, says CFO Joseph Bronson. He predicts that semiconductor capital spending will drop 27% to $42 billion worldwide this year from 2000.

The current quarter will be flat sequentially with revenues and new orders coming in at about the same rate as the quarter just-ended.

Applied has revised its outlook, however, with a "significant shift from capacity to technology purchases in 300-mm and copper" equipment, Bronson says. "Most customers have ceased ordering for production capacity as they begin purchasing equipment to implement advanced technology solutions for 0.13-micron and below with copper processing," he adds.

. . . as Applied's concerns ease
over lag in 300-mm tool orders

Applied's CFO Joseph Bronson and CEO James Morgan indicate their concerns over delays in 300-mm tool orders could be easing. They created stir in July when Applied suggested that the current downturn could be entering into to a second phase that potentially threatened 300-mm fabs. Now Applied says that 300-mm equipment orders are essentially holding steady with a few push-outs and only one cancellation of a $50 million order. And that one is likely to be rebooked in the fourth quarter, it adds.

The upturn in the production equipment business will be driven by 300-mm production fabs, Morgan says, and many of the leading chip makers are trying to hedge their bets by keeping their pilot line projects on schedule and holding off on capacity orders. Morgan expects to see a rush for equipment deliveries once chip makers believe the upturn has begun.

Applied is still trying hard to reduce its breakeven point and push down the 300-mm learning curve. "The more you can do that, the faster you can get your breakeven point reduced," Bronson says. And it continues to amass cash, reporting $4.7 billion in cash equivalents and short-term investments, an increase of $141 million from the prior quarter.

It also continues to invest $2.1 billion this year in new products, R&D, and manufacturing systems. More than 25 products are "in the R&D pipeline right now with additional projects underway," Bronson says.

"We have cut costs very, very hard in this company during the past nine months," says the CFO, who adds that Wall Street analysts have not understood Applied's challenges in reducing its breakeven point and still maintain a broad line of tools and advanced technologies.

(See Aug. 15 story.)

Cutbacks in DRAM output
won't help overcapacity . . .

Don't look for DRAM makers to shut down enough fabs to fix the current overcapacity problem. That's the feeling of industry execs and analysts who say that the recent spate of DRAM fab closings won't affect the supply side of the market enough to turn around collapsing prices any time soon.

"There's great reluctance by a lot of integrated devices manufacturers to shut down factories," says Len Jelinek, iSuppli analyst. "They're running factories high to cover costs."

Because of the glut of capacity that came on line last year and the additional capacity expected next year, Jelinek doesn't expect to see a recovery in fab utilization rates for memory products until 2003. That still means 128-megabit SDRAMs selling for $2 a piece.

"There is too much capacity," agrees Sherry Garber, Semico Research analyst. "The demand for 64-megabit DRAMs is just gone. That's one of the densities where there appears to be a lot of inventory, both at the DRAM vendors and some OEM sites."

The top five suppliers--Samsung, Micron, Hynix, NEC, and Toshiba--grabbed 80% of the $27 billion DRAM market in 2000, according to Semico's Graber. It would take a move by one of these companies to abandon the sector for the industry to see a price impact from the supply side, she predicts. But no one sees that happening right now.

(See Aug. 13 story.)

. . . but DRAM makers
in Taiwan want to try

Even though observers don't see current production cutbacks in Japan and South Korea reducing DRAM overcapacity, Taiwan memory producers are getting together to discuss the possibility of cutting their losses by slicing production.

Mosel Vitelic is talking to its DRAM rivals on the island about possibly reducing their output after the freefall in DRAM prices dragged their first-half results into the red. "Our preliminary agreement is to trim some production starting September," says Thomas Chang, VP at Mosel Vitelic, which owns a DRAM joint venture in Taiwan with Infineon Technologies.

Powerchip Semiconductor acknowledges that a reduction in DRAM output could be on the way. "People are expecting cuts in production," says Michael Tsai, Powerchip president. "Everyone is feeling the need of cutting production," agrees Charles Kau, executive vice president at Nanya Technology.

But analysts question whether a cutback in output by the Taiwanese would have any substantial impact on the worldwide DRAM market. Taiwan's suppliers have only about a 15% share. "Even if each player in Taiwan reduces its production, it would not be enough to reverse the slump in DRAM prices," says Connor Liu, analyst at SG securities in Taipei.

Right now, Taiwanese suppliers are fighting low DRAM prices by upgrading their process technologies to save costs and increase output. They're hoping to see the price of 128-megabit DRAMs rise to $3 each ahead of the traditional Christmas buying season. But that would still be less than the $3.50 it costs Taiwan's DRAM vendors to make this DRAM.

(See Aug. 14 story.)

Some things never change; LSI Logic
builds Minnesota center to tap locals

I remember back in the '70s and '80s just how difficult it was for Silicon Valley companies to recruit engineers out of Minnesota. Despite great offers, Minnesotans didn't want to relocate to La La land. They liked the winters and the easy living. It still must be so, as LSI Logic this week broke ground on a new semiconductor design center in Rochester, Minn., to tap into the state's engineering base.

The Silicon Valley chip maker will staff the new design facility with up to 100 employees that will work for LSI Logic and Mint Technology, its wholly-owned engineering services subsidiary. LSI Logic already employs 28 engineers in Rochester and the plan is for another 50 to move into the new center after it is completed in early 2002. LSI Logic already employs more than 180 workers at its engineering design center in Bloomington, Minn.

I can understand the unwillingness of Minnesotans to move to the Golden State After all, I grew up in nearby Nebraska.

(See Aug. 14 story.)

Chip business to hit new bunch
of records--all the wrong kind

This year will go down in the chip industry's record books in a big way. But the new records won't be kind to brag about, that's for sure. IC Insights came up with this list for the record books at mid-year.

1. For the first time ever, sales of worldwide electronic system are forecast to decline in 2001--they'll be off 4%. The lowest-growth year on record was 1998 when system sales rose a bare 1%, according to the Scottsdale researcher.

2. The worldwide semiconductor market is expected to show the largest decline in history--a fall of 26%. This expected drop would outpace the 17% decline recorded in 1985--until now, the worst year ever for the chip industry.

3. Integrated circuit unit shipments are expected to tie the worst decline ever, which was set in 1985 at minus 16%.

4. The "Perfect Storm" combination of global recession, overcapacity, and inventory burn will result in the current IC industry cycle (1999-through-2001) to experience an average annual growth rate of only 9%, nearly half that of the previous cycle's 16% (between 1993 and 1998).

5. The chip market for the Americas' which is almost all the U.S. will register its worst decline in history this year--a 39% decline, according to IC Insights. The previous record drop was a 30% decline back in 1985.

6.For the first time ever, the ROW (rest-of-world) will be the largest semiconductor market in the world, IC Insights figures, pushing the Americas out of the No.1 position. The ROW's lead will be due primarily to the strength of the Chinese market, IC Insights says. An aside: China will be the only major nation expected to have a growing semiconductor market. It should increase by 5% this year over 2000.

7. Semiconductor capital spending will decline by a record 30% in 2001, predicts IC Insights. This will surpass the previous record decline of 25% in 1998, it says.

8. The DRAM industry nearly hit a record. It is expected to show its second-largest decline ever with a 53% drop in dollar revenues, according to IC Insights. Only 1985's 56% collapse was worse. This year's DRAM market will be just one-third its size in 1995--$13.5 billion vs. $41 billion.)

9. The IC foundry business also hit a record number. IC Insights predicts that foundries will show their steepest decline ever with revenues falling 32% from last year, according to IC Insights. And the fab capacity utilization by foundries will average just 54% this year--an all-time low, which will be nearly half of the utilization rate last year.

IC Insights is hoping for a different set of number next year--and I predict they won't be records, up or down.

(See Aug. 14 story.)

General Semi gets ready for
Vishay merger by closing Irish fab

General Semiconductor, which has just agreed to merge with Vishay Intertechnology, was putting its house in order this week by closing its fab in Macroom, Ireland. This will "hasten our eventual integration with Vishay and better position the combined company for future growth," says CEO Ronald Ostertag.

The Irish plant, which produces transient voltage suppression diodes and rectifiers, is closing at the end of the year. Some 670 jobs will be eliminated, which amounts to 13% of the company's total workforce.

General Semi expects to realize annual cost savings of $25 million from the plant closing and will take pre-tax charges of up to $60 million to cover severance and other costs related to the shutdown. Most of the products turned out at Macroom will be transferred to General Semi's fabs in Taiwan and China.

The Long Island company also is making other cuts in operations worldwide and will outsource a number of its mature products to subcontractors and lay off an additional 3% of its work force. Some 23% of its global work force, which totaled 5,700 at the beginning of 2001, have been let go this year.

(See Aug. 17 story.)

Jack Gifford shows
how to run a chip maker

I'm no financial wizard, and an earnings report that involves a major writeoff and merger can be downright confusing to figure out. But you gotta give Jack Gifford a lot of credit. The Maxim Integrated Products CEO runs a tight ship.

Maxim has a history of managing effectively during industry downturns and emerging from such downturns earlier and stronger than most companies in our industry, Gifford noted a couple of months ago. And it's true. He issued his annual results this week and here are results that looked good to me, particularly in contrast to the flood of crummy financial reports pouring out in recent months.

For the year ended June 30, the Sunnyvale, Calif. chip maker showed earnings of $1.58 billion, up from $1.38 billion from fiscal 2000. That's UP 15%. And net income amounted to $335 million, down only 10.5% from the previous year. Just as interesting, gross margins hit 69.8% in the fourth quarter, UP from 63.8% in the year-ago quarter.

Maxim did have a 20% drop in 4th quarter revenues from the previous three months and posted a net loss of $16.2 million which was due to pretax charge of $163.4 million from its merger with Dallas Semiconductor. But the acquisition is "moving ahead of schedule and, as predicted, has already positively affected the financial performance of the combined company in the fourth quarter," Gifford says.

Gifford said the sharp decline in bookings--orders in the 4th quarter trailed the previous quarter by 15%-- appears to be slowing and order rates "may be stabilizing." But he still expects another sequential decline in revenues during the current period, the first fiscal 2000 quarter.

(See Aug. 16 story.)

China is definitely where
PC action is; look at Legend

Go west, young man or woman, go west! All the way to the Far East, that is. I keep preaching that China is the last frontier for the global chip business. Chances of making it big are certainly greatest here in this emerging nation, that is unless the Chinese succeed in dominating the production of chips, cell phones, and PCs like their 10-year plans calls for.

The numbers are exciting. China's state planners expect to have 70 million PC owners by 2005, up from only 19 million today. PC sales in China are expected to rise 25-to-30% this year. That's markedly different from the rest of the world, which is looking at declining sales this year.

China's largest PC maker, Legend Holdings, this week reported that quarterly net profits surged to $32 million, which was higher than expected. Unit PC sales hit 651,000 in the quarter, a 5% sequential increase and up more than 25% from the year-ago period. Legend sells one of every three computers in China, with IBM, Compaq, and Samsung distant competitors.

Legend, which is majority-owned by the government, is one of the biggest benefactors of the China's ambitious effort to upgrade its information technology sector and modernize public infrastructure. As part of this drive, China projects that IT will account for 7% of the country's gross domestic product by 2005, up from 4% today.

Although Legend may face more competition when the nation enters the World Trade Organization, it still expects to grow at an annual rate of 30% over the next three-to-five years, according to Dataquest.

(See Aug. 16 story.)

How bad can it get?
Check out Credence

This just may be a record sales decline in the current chip recession.

Credence Systems reported a 91% drop in sales in its fiscal third quarter ended July 31--from $204 million a year ago to just $18.8 million. The Fremont, Calif., producer of chip testers also had a bad sequential decline--dropping 57% from the $43.3 million in gear shipped in its second fiscal quarter.

"The semiconductor test business is clearly in the midst of an unprecedented downturn, the severity of which is even greater than that experienced in 1985," comments CEO Graham Siddall. "The low test equipment utilization rates being experienced by our customers have continued to result in a virtual cessation of orders or shipments of new production test systems."

With utilization rates as low as 30% at some of its major customers, the chip tester vendor doesn't expect a "significant upturn" in new orders until excess capacity has been absorbed. And this won't occur in this calendar year, Siddall predicts.

(See Aug. 15 story.)

Can Microsoft make up
delay in Xbox production?

Timing is everything right now for Microsoft's new Xbox game console. The software maker's latest, boldest attempt to sell consumer products is slated for launch for this year's holiday selling season. So missing this schedule even by just weeks could be fatal--at least for this year.

Thomas Weisel analysts jolted the market this week when they reported that an Intel motherboard design flaw may hold up production of the Xbox game console by three to four weeks. Production originally was scheduled to begin in mid-September.

Multiple sources throughout the Xbox supply chain have confirmed that the game machine will be delayed because of a flaw in Intel's motherboard design, says Eric Ross, analyst for the New York brokerage.

But all may not be lost. Ross tempers his warning by saying that Microsoft's original launch date may still be met. "Our sources suggest that because of the slack in the production schedule for Xbox, this three-week miss can be made up before its launch date," he says.

As of now, his sources tell Ross that "there is no reason to panic." And he balances his scary comments by noting that other sources say that such delays and subsequent schedule re-accelerations are common in the PC hardware space.

The Xbox motherboard is based on Intel's Pentium III processor using 400 megahertz DDR SDRAM.

(See Aug. 14 story.)

New process from Fujitsu, Ramtron
could boost FRAM share of market

This week's most interesting technology development comes from a Ramtron-Fujitsu joint venture. The big deal here is that the new technology could open up a lot more opportunities for ferroelectric RAM in the marketplace.

The two chip makers have developed an "embeddable" FRAM manufacturing process that should reduce the cost of standard FRAM memories as well as embed the nonvolatile memory in microcontrollers and ASICs. The new 0.35-micron, 3-volt process technology is now being installed at a Fujitsu fab in Iwate, Japan and should be ready to go sometime in the second half.

The multi-level metal FRAM process should "rapidly closing in on cost parity with EEPROM technology," predicts Ramtron VP Tom Davenport. Fujitsu and Ramtron first began working together on FRAM technology in 1996.

(See Aug. 13 story.)

We welcome your feedback, comments, criticisms, or questions. E-mail us at bhenkel@aol.com. Have a great weekend!

(Click here for last week's Semiconductor Alert!.)