Commentary & analysis of week's chip news
Greetings from Down-East Maine, where we still don't think the worst is over in the current economic slump. This was a weird week--the temp hit 90F two days here this week--a record for the coast--and investors even started bidding UP semiconductor stocks. They were lured by forecasts from the Street and some industry execs who see orders picking up---but mostly for seasonal reasons. I am still bearish and don't see a general chip industry recovery until third quarter next year. Sorry.
ASM Litho messes up
on Intel tool order . . .
One of the major reasons why ASM Lithography was so hot to buy Silicon Valley Group was that the acquisition would give the Dutch chip equipment company a major entre to the lithography business of world's biggest chip maker. Well, ASML ended up laying out $1.6 billion for SVG, but the Intel business is not looking all that good right now.
It seems that Silicon Valley Group messed up big time. Intel has now scrapped its initial plans to use SVG's 193-nm, argon-fluoride (ArF) lithography tools because of several equipment delivery delays. The decision, which was confirmed to SBN by an Intel official, may have killed the initial $100 million tool order--but Intel flacks deny it.
The deal is certainly in doubt. SVG "had a window of opportunity for 193-nm scanners at Intel," comments Mark Bohr, Intel's director of process architecture and integration. "That window has closed," he tells SBN.
As usual, Intel "spinmeisters" are trying to obfuscate the issue by saying the chip giant still has the big order on its books. But they wouldn't say whether Intel would deploy SVG's delayed 193-nm step-and-scan systems in its production fabs. The PR pros would only say the company was uncertain as to what it would do with the new SVG tools, once they are delivered. How's that for double talk?
Analysts now believe that SVG's 193-nm scanner order at Intel is in serious jeopardy and eventually will "fall by the "wayside." Still others are convinced that ASML's SVG unit is dead in the water at Intel.
. . . so Intel starts all over again
with 193-nm tool procurement . . .
Intel will now go back and start evaluating 193-nm scanners from several suppliers. It looked like this competition was all over 18 months ago when Intel reportedly placed an order for SVG's Micrascan 193 tools. But SVG ran into a series of production problems. It initially was scheduled to deliver the scanner in April, but then delayed it to July. Now this month, the shipment was delayed again for another three or four months.
The stakes here are high for equipment makers. Intel's total lithography business--including low- to high-end exposure systems--could be worth as much as $500 million over the next several years, based on estimates from analysts.
Intel could end up buying 200 to 400 exposure tools over the next several years and one-third of them will be for the more advanced 193- and 248-nm scanners, predicts Eric Ross, analyst at Thomas Weisel Partners.
"Intel will evaluate SVG's 193-nm technology," the New York analyst says. "But whether Intel buys all 25 units of SVG's 193-nm tools is still unclear," he adds. "My feeling is that SVG will fall by the wayside at Intel a little bit."
(See July 25 story.)
. . . As delays reportedly affect
Intel's technology roadmap
Despite constant denials by Intel, delays in getting the new generation of 193-nm litho tools have impacted its process technology roadmap, analysts say.
But the chip giant disputes reports that the delays have affected the ramp of its new 0.13-micron (130-nm) process technology. Intel says the Silicon Valley Group scanners had been targeted all along for 0.10-micron (100-nm) technology.
Nevertheless, the new SVG scanner was also intended to do critical device layers in the 0.13-micron process. Intel had hoped to use SVG's Microscan 193 to avoid extensive use of the costly phase-shifting photomasks that are needed to extend 248-nm scanners to next-generation technologies.
Analysts following Intel now believe it is paying a major cost penalty by using more expensive phase-shifting photomasks in ramping its 0.13-micron technology process. But Intel insists the phase-shifting masks are not raising its manufacturing costs.
While Intel waits for fall delivery of SVG's 193-nm tool, it reportedly also is evaluating ASML's own, internally-developed scanners. ASML Europe has just rolled out its 193-nm tool that it plans to ship by yearend.
The long shot for getting new scanner business at Intel is Canon, which has little or no track record at the chip giant. But the Japanese company claims to be one of the first to ship a 193-nm tool in volume.
(See July 25 story.)
Deepening downturn may be
forcing big layoffs at Applied
You'd have to give Applied Materials a grade of "A" for the way it has scaled down its operations during the current semiconductor recession. So far, the semiconductor equipment giant has avoided large layoffs by cutting salaries, delaying merit raises, shortening work schedules, and eliminating 1,000 jobs with voluntary severance packages.
Applied told analysts just a month ago that it was trying desperately to avoid "the hire-and-fire" cycle of past business cycles in the equipment industry. But the company increasingly worries that a second phase in the current industry slowdown could force it to resort to major layoffs.
So far, the company isn't letting out a peep, but rumors are swirling around the valley about potential layoffs of up to 20% of Applied's 20,000 global workforce. A story this week in the San Jose Mercury News reports that Applied has just laid off 100 employees.
The latest wave of capital spending cutbacks at chip makers is now causing Wall Street to expect more layoffs at Applied. Adds Morgan Stanley Dean Witter: "Applied did not lay off as many people as we would have expected earlier in the current downturn and now has to play catch up."
(See Aug. 2 story.)
European PC sales go south,
but it's the consumers' fault
Following the U.S. lead, Europe is showing a downturn in its PC market for the first time. But unlike the U.S., it isn't falling corporate demand that's doing it; In Europe, it's lagging consumer confidence.
Just over 6 million PCs were shipped in Western Europe in the second quarter, according to Dataquest, a decline of 4% from the same period last year.
"The collapse of the home PC market in Western Europe masks a cautious, but more optimistic outlook from business customers," comments Brian Gammage, U.K.-based analyst with Dataquest. "Large account customers are showing steady growth, but the overall picture is distorted by a general weakness of computer hardware in retail. The worsening economic outlook is clearly hitting consumer confidence."
On the other hand, PC shipments in Europe, Middle East, and Africa totaled 8.38 million units during the second quarter, a slight increase over the same period last year, Dataquest reports. But despite the modest growth, "there is plenty to worry PC vendors," Gammage says. "Many of the leading international vendors have seen shipments decline, some by a significant degree."
(See Aug. 3 story.)
New Avanti execs claim customers,
employees aren't abandoning ship
Avanti came out fighting Thursday, after the CAD company was rocked by July's sentencing and restitution hearings. As you'd expect, the new president, Paul Lo, painted an upbeat picture of a company under new management that's putting the past behind it.
Earlier this week, Avanti issued an apology to the EDA industry and to Cadence Design Systems. Avanti said it has paid $140 million to Cadence of the $195 million court-ordered restitution but needs an extension until January to pay the remainder. In late July, several of Avant!'s earliest employees got jail sentences for source-code theft.
Now analysts are worried about customer and employee defections, payment of the remaining restitution money, the reinvigoration of Cadence's civil suit, and Avanti's past refusal to provide analysts with full financial accounting and guidance. "It is still unclear how the nature of the business is going to change with the new management in place," says Garo Toomajanian, analyst at Dain Rausher Wessels.
Lo strongly denied that customers are abandoning Avanti. "We have visited and spoken to dozens of our customers, and we don't know of a single defection among them," he adds. And there is no mass exodus of Avanti employees, Lo claims. Of 264 "key" employees, only six--or 2.3%--have left during the past two months, he says.
(See Aug. 2 story.)
NEC makes massive
changes in chip ops
NEC is going through major convulsions now as it struggles with the global semiconductor recession. This week, after reporting a $140 million quarterly operating loss at its Electron Devices Group, NEC disclosed a flurry of major changes.
The Japanese giant is closing aging 6-inch wafer lines in Japan, cutting production in half at its Scotland fab, merging chip assembly operations, and cutting 4,000 jobs. It also is cutting back capital spending for semiconductor operations by 50%, to $975 million.
NEC now wants to be completely out of DRAMs in three years. It will shift control of DRAM production at its Hiroshima fab by 2004 to Elpida Memory, its joint venture with Hitachi.
Three of its Japanese chip assembly plants will be consolidated into a single operation called NEC Semiconductors Kyushu. And two other assembly plants will be merged in the second half. In addition, NEC says a planned expansion of production at its joint-venture fab in Shanghai will be stopped, at least for now. And output at Hua Hong NEC Electronics will continue at 20,000 wafers a month.
(See July 31 story.)
Here comes digital
Something I always thought was a neat idea was digital radio. But I didn't think enough people would pay for this service to make it profitable. It's been slow in coming, but here it comes--ready or not.
Two competing systems are due to switch on this year and they should make a tidy little market for chip sets--at least initially.
STMicroelectronics has orders for 170,000 chip sets from equipment makers that are building receivers for one U.S. digital radio service being launched by XM Satellite Radio late this summer for service in the Southwest and in November for the rest of the nation.
The European chip maker, sole supplier of the two-chip XM set for home and auto radios, will ship these sets this year to Sony, Pioneer, Alpine, and Delphi Automotive Systems. The XM Radio network uses two geostationary satellites--Rock and Roll--and terrestrial repeaters to cover the U.S. with 100 channels.
The second satellite-based digital radio service competing with XM will be put up by Sirius Satellite Radio. Its subscription-based radio service uses an eight-chip set from Agere Systems. Receivers go into production this fall, with service slated to begin by the end of 2001.
(See Aug. 1 story.)
TI qualifies fastest
64-bit CPU for Sun
It's several months late, but Texas Instruments has qualified a 900-megahertz, 64-bit RISC processor for volume shipments to Sun Microsystems, which will begin turning out workstations with the new part within the next 90 days. This CPU is the world's fastest commercially available 64-bit chip, the two companies claim, beating IBM Power4 processors and Intel's targets for its 64-bit Itanium chip.
The new UltraSparc III is TI's first all-copper, low-k processor and the latest step towards 64-bit RISC chips that will operate at speeds up to 1.5 gigahertz, according to a Sun roadmap, which last fall showed the 900-megahertz processors ready for workstations in the first quarter of 2001.
Even faster UltraSparc III processors already are being produced with the current CMOS process, which uses 0.10-micron channel-length transistors in a 0.15-micron class technology. The new processor is impressive: it has seven layers of copper metal and a low-k dielectric insulator of fluorinated silicate glass (FSG).
TI now plans to integrate copper dual damascene processes with low-k dielectrics in the form of organosilicate glasses (OSG) using CVD tools and processes from Novellus Systems. The next-generation 0.13-micron copper process will employ 193-nm lithography scanners. There are no plans to shift eventually to silicon-on-insulator (SOI) wafers, as some chip makers plan to do, as the two companies see little benefit.
(See July 31 story.)
Japan beats out U.S.
as biggest chip market
I'm not sure what to make of this, but it turns out that Japan in June consumed more chips than the Americas nearly all the U.S. did for the first month in more than a decade. Both regions, of course, have been losing equipment makers for a long time now to the Asia Pacific region, which for the second month in a row, saw its chip consumption lead the rest of the world.
The monthly report from the Semiconductor Industry Association underscores just how hard the current downturn is hitting the U.S. chip market. In June, revenues fell 12.9% from May to $2.91 billion, and ended up with a huge drop of 45.1% from the year-ago month.
June chip sales in Japan slipped 5.8% to $2.97 billion from May and were 20% lower than June 2000, the SIA says. Worldwide chip sales in June fell 8.8% to $11.6 billion and 30.7% below June of last year.
The Asia Pacific region slipped just 5.8% in June to $3.18 billion from May, and sales were 25.3% lower in June than in same month last year, according to the SIA. European chip sales were also hit hard by the downturn in June, falling 10.6% from May to $2.53 billion and dropping 26.8% from June last year.
And here's a surprise yeah sure. "We believe the industry will return to sequential growth in the fourth quarter of this year," predicts SIA president George Scalise, but only if inventories continue to be reduced in the third quarter as predicted.
(See Aug. 2 story.)
Yet another Bluetooth
booster--will it ever end?
Never have I seen so many wild forecasts for one emerging market as I have for our friend Bluetooth. Logging in this week with the latest Bluetooth excitement is Frost & Sullivan in Mountain View, Calif. And they are not to be believed.
Now there's no getting around the fact that Bluetooth technology has attracted a horde of companies to a market that can now be measured only in a few millions of dollars rather than billions. But a wide range of technical and market related issues continue to delay this technology from getting into products. These problems include device interoperability, interference from other radio technologies, and fears over the security of the technology. Bluetooth also suffers from market confusion because there are so many potential applications.
But listen to Frost & Sullivan researchers, who now claim there's been "major breakthroughs" in Bluetooth technology, including the development of the first true, single-chip Bluetooth solution. This technology, which is being developed to provide short-range (20-or-so meters) wireless connectivity between equipment, was originally conceived as a cable replacement but has now emerged as a far more potent and versatile solution.
Get this five year forecast! Shipments of Bluetooth-enabled products will jump from 4.2 million units this year (even 2001 is highly doubtful) to 1.01 billion (with a "b") by 2006. Actually, this forecast is pretty close to Cahners In-Stat Group's 2005 prediction of 955 million units that I've previously poked fun at.
And if you think that prediction is off the wall, get this one. Total revenues of Bluetooth-based products will soar in five years from $1.8 billion in 2001 to $330 billion by 2006! I don't know where the products are coming from this year--the most optimistic estimate I've seen is $16 million worth shipped in 2000. I hope my cynicism is wrong--god knows the chip business could use this kind of an emerging markets.
But Frost & Sullivan's Michael Wall figures the dam will burst now because of the single chip development. "Single-chip solutions have long been the holy grail of the industry and this development will be a major driver behind the falling cost of Bluetooth products," he says. "The single chip will bring greater silicon efficiency, smaller footprints for easier integration, and lower power consumption," Wall maintains.
(See July 30 story.)
Does Agere plan to turn
into fabless chip company?
Slashing head count and consolidating its fabs will not be enough to bring expenses in line with lower revenue at Agere Systems. At least that's the current buzz among analysts regarding the Allentown, Penn., chip maker that's facing mounting debt and a sales slump that has nearly idled its fabs.
What's to be done? A fabless ASIC strategy is not out of the question, says one Agere customer who doesn't want to be identified. He says Agere has told his company that it plans to discontinue CMOS chip manufacturing and will move internal production to offshore foundries. An Agere spokeswoman denies the firm is making that move, but does acknowledge that its long-term strategy is to increasingly outsource mainstream CMOS production.
What ever happens, analysts figure that Agere may have to shut down even more of its fabs to bring costs in line with business levels. The chip maker that was spun off from Lucent Technologies has emerged with a large infrastructure of factories and R&D that analysts say is way out of balance with rapidly declining revenue. And its R&D has been going for chip design rather than improving manufacturing technology, they say. As a result, the company is losing its production edge, some analysts say.
One problem has been around for a while. Agere never leveraged developments coming out of Bell Laboratories, the R&D arm of its former parent Lucent Technologies. "Agere has been underperforming since before the downturn," says Ambrish Srivastava, chip analyst at ABN Amro. "They're not getting the bang for their R&D buck."
"I don't look for them to survive or flourish by being on the leading edge of process technology and chip manufacturing," adds analyst Bill McClean of IC Insights. "It doesn't appear they have any interest, or now even the ability, to be strong in IC manufacturing and technology, like IBM."
(See July30 story.)
X-ray litho system
being built for DARPA
An engineering prototype, or "beta" system, of an integrated X-ray lithography system will be built by JMAR Technologies, with $7.8 million in new funding coming from the Pentagon's Defense Advanced Research Projects Agency.
The X-ray stepper will be designed for use in processing chips for optical networking, wireless, and military applications, says John S. Martinez, CEO of the San Diego company. It will be based on technology from Semiconductor Advanced Lithography (SAL), which was acquired in April by JMAR.
The new funds will "accelerate our efforts to complete, in 2002, a fully integrated engineering prototype X-ray lithography system capable of handling the first of what we believe will ultimately become a wide-range of high-performance semiconductor manufacturing applications," he says.
Before the end of this year, Martinez expects DARPA to award a separate new contract to JMAR to complete the purchase of a Model 5 stepper from SAL, install JMAR's X-ray source into the stepper, and perform the final installation and checkout of the fully-integrated system.
(See July31 story.)
Yep, VC investments
dropped too in quarter
I may be nave, but it seems to me that now is as good a time as any to invest in startups in our industry. I don't understand why venture capitalists are keeping their purses closed now even if business stinks. But they are.
Venture capital investments in U.S. chip and electronics startups during the second quarter dropped sequentially by 40% from $1.4 billion in the first quarter to $843 million, according to the National Venture Capital Association. Some 52 semiconductor and electronics-related startups received venture funding in the second quarter, down one-third from the 76 new companies funded in the first quarter.
In all industry sectors, venture capital dropped only 9% from $11.7 billion in the first quarter to $10.6 billion in the second quarter. Internet-related companies led all industry categories for VC funding in the second quarter with $3 billion invested in 301companies, according to the trade group. Computer software and services was second highest category with $2.2 billion invested in the second quarter.
(See Aug. 1 story.)
finally says yes to Vishay
I guess perseverance does pay off. This spring General Semiconductor rejected Vishay Intertechnology's offer to acquire it for $360 million in stock. But this week the Long Island company accepted an offer from Vishay that included $539 million in stock and assumption of $229 million in debt.
Maybe the semiconductor recession as well as the higher price had something to do with it. The acquisition will make General Semiconductor much stronger and "able to capitalize on the current difficult environment" for significant growth in the future, says CEO Ronald Ostertag.
The acquisition is aimed at extending Vishay's portfolio significantly in discrete semiconductors. General Semiconductor's earned $47 million in net income last year on sales of $494 million. Vishay had revenues of $2.5 billion last year.
"We have complementary product lines, opportunities for substantial savings and greater efficiencies, and the balance sheet to continue to expand opportunistically during the current industry downturn," declares Felix Zandman, CEO of Vishay.
"Consolidation is necessary in our industry," he maintains, and "Vishay is well positioned to remain at the forefront of this trend." A month ago, Vishay acquired the infrared components business of Germany's Infineon Technologies for $120 million. Vishay, originally a major supplier of passive components, moved into the discrete and power semiconductor business several years ago. It expanded its presence by purchasing a large portion of Temic Telefunken Microelectronics.
General Semiconductor was formed in 1997 when General Instruments spun off its chip operations. Vishay figures the acquisition will be immediately accretive to its earnings and yield "significant operational synergies and cost savings, expected to exceed $50 million annually when fully realized."
(See Aug. 1 story.)
Via is having tough time
selling Pentium-4 chip sets
Via Technologies is finding out that it is harder than expected to sell its Pentium 4-compatible chip set.
The major PC chipset rival of Intel is running into resistance from some important motherboard makers who are reluctant to use the company's Pentium 4-compatible chipset. The lack of an Intel license, has left the top tier Taiwan motherboard suppliers--Asustek Computer, Gigabyte Technology, and Microstar International--unwilling to use Via's new chipsets, casting a shadow on the outlook for Via for the rest of this year, analysts say.
Via's prospects are tied closely to its double-data-rate chipsets configured for the Pentium 4 processors. In June, Via began shipping some DDR chipsets to about 10 second-tier motherboard companies even though it hasn't received a license. The three top tier suppliers are shipping motherboards instead with Intel chip sets that support the P4.
(See Aug. 1 story.)
If you have any comments, criticisms, or questions, don't hesitate to E-mail us at firstname.lastname@example.org. Have a great weekend!
(Click here for last week's Semiconductor Alert!.)