Applied sees slower growth for
chip production gear next year

The world's leading chip production equipment maker has suddenly turned cautious, not only about the next year but the next quarter. Applied Materials has cut its growth projections for chip industry capital spending in 2001 to 20%, down from its previous forecast of 30%.

Applied is concerned about nearly all segments of semiconductor capital spending next year. Its reasons include the undecided U.S. presidential election, the world's economic growth outlook, the still-troubled DRAM segment, and slowing in demand for telecommunications ICs. "I think it's a time to be careful and cautious--and we want to be able to put something up that we think we can do," says Joseph R. Bronson, CFO for the Santa Clara company.

This week Applied's top managers were cautious in talking about the outlook for business in their current fiscal quarter ending in late February as well as overall market conditions for 2001. Bronson and CEO Jim Morgan both hedged about the overall outlook for wafer fab spending next year, the fickle DRAM segment, which continues to be hit by price erosion, and the silicon foundry business, which may be about to slow down due to inventory adjustments and slipping profits in key telecom areas.

"I think it's clear the impact of the telecommunications segment has slowed Taiwan and that part of the world," Bronson says. "There is no question that there is some easing because the slower demand for telecommunications and earnings shortfalls. That is going to have some near-term impact on the backlogs of Taiwan companies," he notes. "My sense is that their rate of capital spending will slow a bit."

300-mm wafer gear will
take off in 2001--finally

Despite its increasing caution over next year, Applied remains extremely confident about 300-mm wafer gear, which has been coming ever since the late '90s. About $6-to-8 billion of the global revenues from wafer fab systems next year will come from 300-mm tools, the company estimates. And company managers did not dispute one Street projection that had Applied pulling in about $1.8-to-$2.5 billion in 300-mm revenues in 2001. In comparison, Applied's 300-mm revenues in its just-ended fiscal fourth quarter was just about "nil," reports Bronson.

Applied's 300-mm revenues will slowly begin to take off in the current fiscal quarter. But Applied Materials sees 300-mm becoming the main driver for capital equipment sales later in 2001. "The strongest portion of growth, we can see next year, is significant growth in 300-mm.... We think copper and materials--low-k dielectrics--will also increase," he says.

(See Nov. 16 story.)

Applied's 4th quarter net
doesn't disappoint Street

Applied Materials' fourth quarter ended Oct. 29 was not a disappointment to investors, with earnings beating Wall Street predictions by a penny a share. Net income hit a record $664 million, up from $303 million last year. The equipment giant reported an 81% increase in net sales and a record $2.92 billion, up from $1.61 billion in the same period last year. New orders in the quarter also hit a record, rising 10% from the previous quarter to $3.6 billion.

For the fiscal year just ended, Applied had sales of $9.56 billion, an increase of 88% from fiscal 1999. Net income excluding one-time charges was $2.05 billion vs. $757 million in the 1999.

CFO Joseph R Bronson says his "official guidance" for the first quarter of fiscal 2001 was sales in the $2.9 billion-to-$2.95 billion range and earnings-per-share between 75 to 78 cents--essentially flat with the just-ended quarter.

But he adds that Applied's book-to-bill ratio in the current fiscal quarter would likely come in at slightly more than 1.1--which cause some analysts to predict at least a 10% sequential decline in tool orders during the period from Applied's record bookings in the fourth fiscal quarter just ended.

(See Nov. 15 story.)

Microchip gambles
on new class of chips

Microchip Technology is taking the plunge and is making a major investment to develop a brand-new class of chips that it calls digital-signal controllers (DSCs). It is now forming a new division to handle the new product, with sampling expected to begin at the end of this year and volume production by early 2002.

The Chandler, Ariz., company's first DSC will be a single-chip solution combining the advantages of 16-bit microcontrollers and digital signal processors (DSPs). The new part "provides DSP functionality in the company's microcontroller design environment and offers an easy-to-implement solution to embedded engineers contemplating using DSPs," points out Sumit Mitra, who heads the new division.

Microchip estimates the digital signal control market will grow from $100 million today to nearly $2 billion by 2006.

Analysts are bullish about the new part. "Microchip has coined a unique and descriptive term for a burgeoning market segment," says Will Strauss, president of Forward Concepts, Tempe, Ariz., market researcher. "The early adopters . . . are variable speed drives driven by power efficiency requirements in Europe and noise reduction requirements in Japan. This market will grow eight-fold in 2000 over 1999," he predicts.

Applications for Microchip's new parts, which are expected to cost from $3 to $9 each in quantities of 10,000, range from industrial motors, appliances, and Internet-connected appliances to automotive systems, digital answering machines, terminals, and vending machines.

(See Nov. 16 story.)

Infineon will slim down
and refocus operations

Infineon Technologies will do some additional refocusing and fine-tuning of its operations.

After disposing its consumer-oriented audio-visual chip business to Micronas in Zurich, the Munich chip maker has decided to sell off two of its smaller non-core businesses-its optocoupler and optoelectronic semiconductor operations.

"Our optocouplers business has nothing to do with our strategic target segments," explains CEO Ulrich Schumacher. "We also have 49% of the opto joint venture with Osram for LEDs. We will divest that. The negotiations are running. We hope to close in the next quarter and definitely by next fiscal year," he adds.

By shedding the units, Infineon will lose about $680 million to $770 million in revenue, "but we will raise an attractive amount of capital," Schumacher says. Infineon has already committed 30% of its revenues to capital expenditure as it brings up additional capacity, particularly for logic products.

Infineon has just finished a good year, reporting a 72% increase in revenue to $6.3 billion for the year ended Sept. 30, thanks largely to a growing memory and communications IC business.

(See Nov. 16 story.)

Nikon leverages skills
to enter CMP market

Nikon thinks it has a better idea for chemical mechanical planarization (CMP) equipment.

Smoother planar surfaces will be needed soon because of the industry shift to chip making below 0.13 micron. Because present CMP equipment "does not necessarily satisfy users' requirements," says Yosuke Takahashi, head of the company's new CMP business, Nikon saw an opportunity to leverage its strong position in steppers and in lens-polishing technology to go after the CMP equipment market.

The Japanese company is now entering the CMP market with a technology that literally turns the CMP process upside down. Applying a technique used in lens polishing, Nikon has come up with a system that polishes wafers with the polishing pad held over the wafers. Conventional CMP systems hold wafers face down and rub them on a polishing pad. Canon claims the inverted setup has many advantages.

Because supplier rankings in the CMP market shift frequently, Takahashi says that "this means users are not fully satisfied with existing CMP equipment and may be more likely to give a newcomer a chance." The two major CMP suppliers are Applied Materials and Ebara. But current market practices may make it tough for Canon to penetrate this market. "Users tend to request CMP performance as a part of a total process," notes Takashi Ogawa, Dataquest analyst. Equipment suppliers like Applied Materials "have an advantage here," he says. "It may be difficult to enter the CMP market with CMP equipment alone."

Nikon will begin shipments next spring, initially to Japanese clients. Exports will begin in the following year. The company expects to build CMP sales to $280 million in 2005.

(See Nov. 17 story.)

Ericsson uses Qualcomm chip set
to enter CDMA cell-phone market

LM Ericsson has had a tough time developing a code-division multiple access (CDMA) cell-phone. But it finally entered the market this week with full-featured, lightweight handset based on a MSM5100 chip set from Qualcomm. The Swedish giant is late to market, reportedly because of ongoing problems with its component suppliers.

Ericsson had been developing a CDMA handset in the late '90s based on a chip set from VLSI Technology. At that time, VLSI Technology was supplying chip sets for Ericsson's line of cellular phones based on the global system for mobile communications (GSM) standard. But VLSI Technology reportedly was having problems getting its CDMA-based chip set out the door. Then late last year, Qualcomm terminated its CDMA licensing agreement with the VLSI Technology, which had been acquired by Philips, putting Ericsson's CDMA handset in limbo.

Then Qualcomm and Ericsson ended a bitter suit over CDMA technology by agreeing to cross-license their respective technologies. And this week Ericsson acknowledged that its new phone was based on a Qualcomm chip set. It certainly is full-featured. It operates in the 800-megahertz band for both CDMA and analog networks and has a wireless application protocol micro-browser for linking up with Internet connections over a wireless network. It also has a full-graphic screen, a 200-entry phonebook, a choice of 25 ring types, and speed dialing.

(See Nov. 15 story.)

Appleton predicts DRAM prices
won't turn up until first quarter

Don't expect plummeting DRAM prices to firm up any time soon.

Steve Appleton, leader of U.S. DRAM giant Micron Technology, just doesn't see DRAM prices turning around from current low levels for the rest of the year, and perhaps not until the first quarter of 2001.

OEM contract prices have dropped to $4-to-$5, he says, because of the large amounts of memory chip inventory still existing in the market. "About half of our OEM customers still have DRAM inventory on hand that they must bleed off. Until they get rid of this inventory," he says, "we won't see much change in price."

The current DRAM inventory buildup came about, he says, when other DRAM makers started talking up possible memory chip shortages for the fall. "Customers did what they always do," the Micron CEO points out. "They ordered heavily to have enough DRAMs on hand for their fall production. When the shortages didn't happen, they were left with large stocks that had to be worked down."

But Micron doesn't seem to be in that bad a shape. Half of Micron's OEM customers have worked off their DRAM inventories and are back in the market again. And because they can buy DRAMs now at a very low price, many of them are expanding the memory size of their PCs.

Micron still claims to be the lowest cost DRAM producer in the entire industry, Even with $4 prices, Appleton still claims to be making a profit. At the same time, he says, many of our competitors are losing money again on DRAMs. "That's why you see the Taiwanese pulling back to delay new fabs, and the Japanese continue retrenching in DRAMs," he maintains. Micron is now targeting about $2.3 billion for capital investment in its current fiscal year ending Aug. 21, 2001, or up by nearly half over last year.

(See Nov. 15 story.)

Here come the really
big wafers, en masse

The flood is beginning. Taiwan Semiconductor Manufacturing Co. has started processing 300-mm wafers for customers such as Altera and Brilliance Semiconductor, a Taiwanese supplier of SRAM for portable applications Don't feel too badly, I hadn't heard of them either. TSMC claims that it is the world's first silicon foundry to deliver the new generation of wafers to the market.

Chip makers hope that the 12-inch wafers, which have two-and-a-quarter times the surface area of the current 6-inch workhorse, will ultimately cause a significant drop in costs as yields go up. If they do, there's no telling what will happen to prices and margins in the next few years. It could get ugly.

TSMC's 300-mm wafers were made on its pilot line at Fab 6, which the company calls the world's largest IC manufacturing facility. When it begins hitting full speed next year, the line can produce about 4,500 wafers a month, or equal to almost 10,000 200-mm wafers.

TSMC currently is building two other 300-mm fabs--Fab 12 in the Hsinchu Science-based Industrial Park and Fab 14 in the Tainan Science-Based Industrial Park. And the company says that construction will begin on Fab 15, also a dedicated 300-mm manufacturing facility, before the end of the year. By 2001, TSMC is slated to have the capacity to turn out 23,000 300-mm wafers a month, more than 200,000 of them a month in 2002, and nearly 1 million of the big wafers by 2003, TSMC claims. Watch out!

(See Nov. 16 story.)

Intel will be tough to displace
as leader in low-end PC biz

Two new Celeron microprocessors that came out this week as Intel's main offering for low-end PCs during the year-end holidays should help Intel hold on to its lead in the low-end, or "value segment," of the PC microprocessor business.

The Santa Clara company already is shipping the 733- and 766-megahertz chips to such PC makers as Hewlett-Packard. The 766-megahertz chip costs $170 in lots of 1,000, while the 733-megahertz model is priced at $112 in the same quantity.

"Intel continues to be the leader in the value PC segment," declares an Intel spokesman, citing market data from NPD Intelect that indicates that Intel holds more than 90% share of the MPU market going into sub-$1,000 PCs sold in U.S. retail chains.

Advanced Micro Devices has been unable to mount much competition against the Celeron because a shortage of low-end chip sets has prevented AMD's Duron from shipping in substantial volumes. But AMD expects to get more competitive when Via Technologies starts shipping its KM133 chip set next.

(See Nov. 16 story.)

Willamette's lower pricetags
may shrink Intel's margins

Almost from the beginning of the semiconductor industry, chip makers have counted on getting their highest profits during the early stage of a new product. That was certainly true of Intel's long-awaited Willamette-class Pentium 4 processor, which is scheduled to launch next week.

But that may be no longer the case. The problem is the intense price competition that Intel is running into with Advanced Micro Devices. Given this current pricing environment, Intel has little choice but to lower its sights, declares Nathan Brookwood, a market analyst at Insight64. What was once described as a high-margin MPU will carry a very competitive price right out of the gate and will likely be slotted for $1,500-to-$2,000 desktop PCs.

And Intel has another problem here. "The Pentium 4 chip, at more than 200 square millimeters in size, takes more silicon than other Intel processors," Brookwood says. This makes the chip "more costly to produce," he adds. "All this cuts sharply into profit margins at the beginning of its product cycle," the analyst notes. "And this is the time when Intel should be expecting high margins from its highest-performance processor."

(See Nov. 13 story.)

Chip set will run voice,
ADSL over phone line

Legerity, the Austin communications chip spin off of Advanced Micro Devices, may be playing the "vaporware" game. The chip maker claims it has developed a "revolutionary" Watch out for that word! chip technology that will handle simultaneous voice and Asymmetric Digital Subscriber Line (ADSL) services over traditional telephone lines.

Not bad, but Legerity was vague about its new IC technology. "What we're announcing is a new technology," maintains Homer Lloyd, marketing director Who else would announce a new technology?. "We will announce products based on this technology within the next couple of months," he promises. Legerity says its new chip set technology could transform 1 billion voice-only telephone lines worldwide into a full-fledged broadband network that supports both voice and data services, according to the marketer.

The new technology will enable Class 5 switches to support ADSL services via a simple upgrade and permits digital subscriber line access multiplexors to handle voice traffic, Lloyd claims. As a result, the chip set will enable OEMs to deliver full-rate ADSL and carrier-class voice services simultaneously over the same twisted pair. The new chip also will reduce costs by eliminating the need for "splitters" on the network, thereby accelerating the deployment of ADSL services, he promises.

(See Nov. 13 story.)

China builds infrastructure
to support its chip industry

China seems determined to build a significant infrastructure to support what it is counting on to be a world-class local semiconductor industry. This week, the China State Council approved a $1 billion government investment to build the country's largest polysilicon wafer blank production plant.

The plant, to be built by Emei Semiconductor Material Factory in Emei Shan City, is designed to produce up to 1,000 tons of polysilicon material a year. Government officials said a domestic supply of polysilicon wafers is needed to reduce the country's dependence on foreign suppliers. China also is concerned that the nation's domestic chip industry could face severe problems in obtaining wafers in any polysilicon shortage.

China-based Joint-venture fabs, such as Shanghai Hua Hong NEC Electronics, or foreign-owned fabs such as Motorola's Tianjin chip plant, use their parent companies to obtain an adequate supply of wafer blanks. But the smaller domestic semiconductor fabs would have far less leverage in obtaining wafer substrates in any global shortage.

(See Nov. 14 story.)

Samsung pushes hard
on Rambus DRAMs

Samsung Electronics is still betting big on Rambus DRAMs, despite their cost problems. And indeed, the South Korean company expects to ship 53% of the world's Rambus DRAMs this year.

This week Samsung reported that it has started mass producing its "third generation" memory chips using a process technology with 0.17-micron design rules. The new process will not only lower production costs by going to smaller die size but also will raise the speeds of the memories by more than 30%. This will enable the company to build 1-gigahertz RDRAMs with this process, Samsung says.

Samsung expects to see strong demand for Rambus memories as Intel's Pentium 4 processors begin to hit the street this month. It predicts its RDRAM revenues will reach $900 million this year. It predicts that global market for all Rambus DRAMs will increase from $1.7 billion this year to $9.2 billion by 2002.

(See Nov. 14 story.)

Analog Devices sees
50% growth next year

Analog Devices checked in with good financial news on Wednesday, but it propped up faltering semiconductor stocks for only a day.

The Norwood, Mass., chip maker reported that in the year ended Oct. 28, it grew an impressive 78% over the previous year in revenues to $2.6 billion. Net income for fiscal 2000 nearly tripled from $205 million in fiscal 1999 to $589 million. These results easily beat Wall Street estimates.

"Analog products revenues increased 68%, while DSP products revenues rose 115%," says CEO Jerald G. Fishman. "Based on these results, we believe ADI has become the industry's fastest-growing analog and DSP IC supplier," he says.

Next year is looking good too at this point, according to the Analog CEO. "Given our high backlog and strong new orders during the fourth quarter, we believe we can achieve 7-to-10% sequential revenue growth in the first quarter of fiscal 2001," he predicts. For the year, Analog expects to grow 50% and hit $3.8 billion. "We believe our revenue growth is likely to be constrained by supply, not demand," Fishman adds.

(See Nov. 15 story.)

If you have any comments or questions, don't hesitate to E-mail us at bhenkel@aol.com. Have a great weekend!