2-stage Power Amplifier 14.5GHz ultra-efficient Dual-Drive™ PA
April outlook: where IC markets are headed
April outlook: where IC markets are headed
By Robert Henkel, Semiconductor Business News
April 18, 2002 (11:49 a.m. EST)
URL: http://www.eetimes.com/story/OEG20020417S0057
Some analysts and OEMs are now conceding the forecasts they made earlier this year calling for the resumption of growth by the cellular phone in 2002 were too optimistic. As everyone crunches their numbers again, it looks like the main problem is China, where the expected high-double digit growth rate just isn't materializing. But most folks hold out for at least some growth [See stories below.] That isn't the only retreat by a market researcher. Dataquest chopped off $2.3 billion off its two-month-old forecast for wafer fab equipment sales in 2001. And it expects sales to drop another 19% this year. But I'm not giving up hope for the semiconductor industries this year. No way. Order rates are going up now for many chip makers--including the Taiwan foundries and some companies like Motorola and Microchip Technology. And some chip markets are looking good--DVD and wireless LAN, for example [See stories below]. Keep the faith, baby. These are not happy days for market researchers--even for the big ones like Dataquest. Their customers are looking hard for ways to cut costs during a downturn and market research is an easy target. Also, making accurate predictions in such volatile times is never easy for forecasters. Dataquest came out this week with its Top 10 ranking and revised revenue estimate for wafer fab equipment suppliers. Perhaps because it was stung on its earlier forecast, the market researcher didn't even make a forecast on how this business will do this year. San Jose-based Dataquest has finally decided that wafer fab equipment sales in 2001 plunged 29.8% to $23.3 billion from the record $33.1 billion in 2000. But back in January, it had predicted that fab equipment sales in 2002 would drop 19% from a preliminary estimate of $25.5 billion in sales in 2001--which means the research firm cut a big $2.2 billion chunk out of what it had estimated just two months earlier for last year's sales. There's two basic kinds of business in the chip gear business--those orders to add capacity and "greenfield" equipment, or those systems for going to new technology. "Capacity shipments virtually ceased during 2001 and the buying focus was on technology and productivity-enhancing equipment," notes Bob Johnson, Dataquest analyst. "The 2001 market was kinder to [production equipment vendors] in hot technology segments such as lithography, copper interconnect and process control." In Dataquest's new Top 10 vendor rankings, metrology systems maker KLA-Tencor and lithography gear supplier Canon were the only top-tier vendors to move up this year. KLA-Tencor jumped from No. 5 to No. 3 with its revenues growing 7.2% to $1.9 billion, while Canon climbed from No. 8 to No. 6 with a 7% increase in sales to nearly $1.2 bi llion in 2001, Dataquest estimates. Once again, Applied Materials led the pack, but its revenues tumbled 41.4%--the worst decline recorded among the Top 10 equipment vendors, Dataquest says. In the No. 2 position was Tokyo Electron with a drop of 28.7% in revenues. While analysts and OEMs are scaling back their cellular phone forecasts for this year because China is not growing as fast as expected, this market will still be up some in 2002, according to some analysts. "China has largely represented the declines for our handset expectations," concedes Jeffrey Schlesinger, analyst with UBS Warburg. But he still sees global handset shipments growing 10% this year to 430 million units and another 16% in 2003 to 500 million units. Expected to make up for disappointing Chinese business is South Korea. This cellular phone market has come roaring back at a much faster rate than analysts have anticipated. "Korea is doing exceptionally well," Schlesinger says. "The fast growth in South Korea is expected to account for tens of millions of unanticipated shipments." Handset shipments in Korea grew 18% in March over February to 1.59 million units, according to UBS Warburg. Based on the first quarter run rate, the domestic Korean handset market will hit nearly 17 million units this year, Schlesinger predicts, which would be up from his earlier estimate of 14.5 million units. "We believe this will offset weaker than expected results in other markets," he says. Better South Korean sales would certainly be good news for the CDMA handset business. "Better than expected results in Korea .. . gives us confidence that our total 2002 CDMA handset estimate of 76 million units is achievable," Schlesinger says. While silicon foundries may have been hit harder in the recent downturn than most chip companies, they still expect to outgrow the rest of the industry over the next several years. "The foundry industry will continue to grow faster than the overall average in the semiconductor industry," says Kenneth Kin, marketing vice president of Taiwan Semiconductor Manufacturing, the leading pure-play foundry. Foundries are expected to grow by more than fourfold by the end of this decade, he predicts. Kin expects the foundry industry to grow from $10 billion in global revenues this year to $45 billion by 2010. Kin may have an axe to grind in predicting such growth, but he lays out several reasons why he believes foundries will grow that fast. The industry will continue to be driven by the IC outsourcing trend, he says, especially among the integrated device manufacturers (IDMs) that are looking to reduce their costs. "More and more IDMs will become 'fab-lite' or fabless," he predicts. Another growth driver is the foundries' push to offer more and more leading-edge technologies, meaning they can handle the development of complex system-on-a-chip (SOC) and rela ted designs. "We are entering into a new era of SoCs," Kin says. Because of the rising growth rate, TSMC is "adding capacity as fast as we can," Kin says. The Taiwanese company reportedly has ordered $200 million worth of 200-mm wafer fab equipment from Applied Materials. Sources believe that TSMC is doing this in an effort to double its copper-interconnect foundry capacity. After leading for four years, Texas Instruments is no longer No. 1 in the analog chip business. That comes from a Reno, Nev., market research firm called Databeans. Dethroning the Dallas chip giant last year was Europe's STMicroelectronics. It is certainly a close race. STMicro's analog chip revenue was estimated to have hit $3.21 billion, just $10 million more than TI's $3.2 billion in analog IC sales last year, according to Databeans analyst Susie Inouye. STMicro "outperformed the industry in some of their key ASIC markets including consumer and computer," she says. Databeans had originally called TI the winner after estimating STMicro's analog chip sales at only $2.9 billion. But it then revised its numbers upward. IC Insights had also given TI the title, but the Scottsdale, Ariz., researcher is now rechecking its data and expects to give an update in May. STMicro's chief economist, Jean Philip Dauvin, has reported his firm's analog chip revenue as $3.79 billion in 2001. The battle for No. 3 in analog chip sales last year also was close, according to Databeans.. It had Philips Semiconductor as No. 3 with sales of $1.57 billion, while Analog Devices had analog sales of $1.56 billion-again just $10 million apart. Finishing up as No. 5 in analog IC vendors last year was Infineon Technologies with $1.25 billion in sales, the research firm estimates. While the market for personal digital assistants is expected to keep growing this year , it won't be long before a new competitor begins cutting into this business. Smart phones will probably overtake PDA shipments in late 2002 or early 2003, predicts Dataquest analyst Todd Kort. "We expect 2003 to be a strong year for growth in the PDA market, but from 2004 onward, sales of PDAs will be increasingly impacted by competition from next-generation smart phones," he says. This year PDA sales will grow at about the same rate as they did last year, according to a new Dataquest report. Worldwide shipments will total 15.5 million units, up 18% from sales of 13 million in 2001, the market researcher says. Growth last year was 18.3% after a booming 2000 saw PDA sales jump 114%. The good news this year for chip vendors to the PDA industry is that inventories were at "reasonable" levels entering this year, according to Dataquest. But weak economic conditions will impede PDA demand in the first half of this year, it says. But in the second half, Dataquest expects the PDA market to regain some of its strength and predicts that growth will be even stronger in 2003. "The increasing capabilities of PDAs and the growing availability of wireless technologies are beginning to stimulate large corporate purchases as solid productivity gains are realized from such applications as wireless e-mail or accessing corporate data bases from remote locations," Kort says. The market continues to shift toward PDAs with color displays, more powerful processors, and more storage capacity, Dataquest says. Wireless technology also being added to more handheld computers are Bluetooth, wireless local area networks, and wide area networks. The research firm estimates that end-user spending in PDA markets will increase more than 20% to $4.6 billion in 2002 from $3.8 billion in 2001. "At the same time, the build-out of packet-switched data networks such as GPRS and CDMA 1x will focus attention on wireless data devices and the growing competition between PDAs and smart phones," Kort says. You don't have to do any research to realize that DVD players have suddenly turned into the hottest consumer electronics product on the market. Just check out the newspaper ads or the DVD disk and player displays at your local Wal-Mart. Despite the speedy growth that new consumer electronic products often achieve, people sometimes forget that dollar growth rarely keeps up and even begins to flatten out. But here's an optimist that predicts DVD revenue will keep soaring. DVD player revenue jumped by more than 31% last year from about $4 billion in 2000, according to the market intelligence services group of iSuppli. This year, player sales will climb to $6.9 billion, then to a whopping $15 billion in 2006, the researcher predicts. I dunno. An estimated 30 million DVD players were shipped in 2001, iSuppli estimates, with unit sales shooting up to 43 million this year and to a cell-phone sized 125 million by 2006, the El Segundo researcher predicts. This kind of unit growth wouldn't surprise me. A major challenge, however, for player makers is a lack of standards, the researcher says. This has forced makers to build DVD players and DVD recorders that will play all three major formats--DVD RAM, DVD-RW, and DVD+RW. "Since there is no standard, backward compatibility is very important," says iSuppli analyst Jay Srivatsa. "Of the three leading formats, DVD RAM is the least likely to become the industry standard because it is not compatible with most DVD players on the market," he says. And in one of the major understatements of the year, iSuppli cautions that prices are expected to remain under pressure due to the fierce competition among Japanese and Korean players, which is certain to increase when suppliers from China enter the fray. The best way to maintain profitability is to keep cutting the chip count. This has put the pressure on chip makers to move quickly to systems-level chips. Vendor s already have integrated most of the DVD functions into a single chip, which includes central processing unit, MPEG-2 decoder, audio digital signal processor, NTSC/PAL encoder. It's becoming clear that growth in low-k dielectrics is stuck in low-gear. They won't admit it, but several chip makers reportedly are struggling to get low-k dielectrics into production. The problems boil down to low production yields and parts reliability. As a result, low-k dielectrics used in advanced interconnect IC structures will generate only a piddling $16 million in sales this year, according to a new study by Kline & Co. But this emerging thin-film materials business will eventually live up to earlier bullish projections once 0.10-micron production processes move into volume production, according to the study. "The real ramp-up should occur in 2003 and 2004, when production at 100-nm [0.10-micron] design rules start coming to commercialization," predicts John Davis, business manager at the Little Falls, N.J. company. And by 2006, low-k dielectric revenues should hit almost $400 million, he predicts. Chip makers have delayed their switch to low-k films, the Kline report says, out of fear that the materials are unreliable and unlikely to be processed at high yield. The 2001 semiconductor downturn also delayed low-k, according to the report. So the IC companies have worked around the problem by redesigning circuits to minimize resistance-capacitance delay and by extending fluorinated silicate glass technology into the 130-nm (0.13-micron) process node, the report says. Chip makers also are using trimethylsilane for the 130-nm node and have gone back to copper metal wiring. Despite the all the problems, chip makers are slowly but surely making progress with low-k dielectrics. Advances are being made in s pin-on stop-etch materials, says the Kline & Co. study. Spin-on precursors in low-k films require stop-etch materials to complement the interlevel dielectrics. Several low-k spin-on material suppliers are working on this problem, and they now say that using a silico-organic material is the solution. Dow Chemical, Dow Corning, Honeywell, and Shipley all say they are developing stop-etch products, according to the study. Progress is also being made in developing porous low-k films to address lower dielectric constant rankings, the report says. For example, Dow Chemical, which has established a "commercial beachhead" with a dense version of its SiLK polyphenylene, has now developed a porous version for 100-nm and below process nodes. "Nearly every developer of spin-on precursors" are developing porous dielectrics and expect to compete at this design rule, according to Kline. At the same time, chemical vapor deposition processes are being extended. Equipment vendors such as Applied Materials and Novell us have expanded the production life for CVD processes with technology that results in more highly strained lattices in oxide films, the study says. CVD technology will now compete at the 100-nm level, and probably the 70-nm level as well, according to the study. Trikon, developer of the Flowfill process, which deposits film in a CVD chamber as a liquid, is rumored to have reached the 2.2 level in its latest process design, said the report. The soaring chip market for communications gear went off the cliff last year and wasn't expected to rebound all that fast. But hope springs eternal as some market researchers now see real growth ahead in several communications markets. Take public ethernet services in North America, for example. This market will grow rapidly over the next few years, according to Dataquest. Revenue will reach more than $1 billion this year and then increase to more than $4 billion b y 2005, the Gartner unit predicts. "Public Ethernet is gaining popularity from its low cost, simplicity, and the widespread comfort commercial customers already have with Ethernet" in local area networks," points out Steve Koppman, Gartner analyst. "Carrier Public Ethernet services allow this same technology . . . to effectively extend their LANs out into a larger metropolitan area or, increasingly in coming years, into long-haul networks as well." Newly emerging varieties of "Gigabit" Ethernet, which promises bandwidth-on-demand-like capabilities, had been championed by small players. But the new public Ethernet will end up extending itself largely through the efforts of major carriers that had adopted the service to defend themselves against their new competitors, Koppman predicts. "The major carriers, building on existing legacy services, will seek to blunt their challengers' competitive appeal while preventing undue pressure on incumbent services and prices:" he says. Growth [initially] will be limited below the potential demand, however, by constraints on fiber connectivity and resulting service availability," Koppman says. But this should improve over the forecast period, he predicts. "Partnerships and joint arrangements will emerge over the next few years," Koppman says, "both to deliver public Ethernet more widely and better share last-mile Ethernet fiber access to buildings." Public Ethernet is any Layer 2 public network carrier service that extends Ethernet beyond the LAN. With the huge year-ago inventory overhang now history, the cellular phone market is beginning to look a little better to chip suppliers. But it's been a real roller coaster for vendors. After shooting up 71% to record sales in 2000, global handset shipments dropped 1% last year--first market decline ever for this young industry, according to the Strategis Group. Now the cell phone business is rebounding, a ccording to the Washington, D.C., researcher, and is predicted to grow 14% in 2002. Worldwide handset shipments will jump from 393.9 million units in 2001 to 458 million in 2002, Strategis predicts. It believes this growth will be driven largely by handset replacement sales. But the pressure will be on the handset makers to deliver products that are the most compelling next-generation models, the researcher comments. And the resulting scramble could result in another shakeup in the cell phone business. "We expect dramatic changes to take place in the mobile handset marketplace this year, as vendors adapt to this challenging new landscape," declares Strategis analyst Ozgur Aytar. Doing the best job to date here, he indicates, appears to be Nokia. The Finnish company, now leading the market with a 35% share, has aggressively taken on the challenge of delivering market-driven products, the research firm notes. Nokia has broken out nine consumer segments and now delivers unique products to all of the m. It plans to bring more than 20 new products to market in the first half of 2002. The top five handset makers currently control 72% of the global market, Strategis notes. Nokia led the market with a 35% share in 2001, it says. Tough times for researchers; Dataquest
chops $2.3B off two-month-old forecastTop 10 production equipment vendors
2001 sales in $ millions
2001 sales 1. Applied Materials $4,920 tr> 2. Tokyo Electron 2,639 3. KLA-Tencor 1,887 4. Nikon 1,525 5. ASML 1,273 6. Canon 1,196. 7. Lam Research 1,131 8. Novellus Systems 920 9. Dainippon 776 10. Hitachi 633 Source: Dataquest
Cell phones aren't looking as good;
but Korea may make up for ChinaFoundries say they will grow
more than fourfold by 2010STMicro knocks off TI as
No. 1 analog chip vendorSmart phones will soon
start cutting in on PDAsDVD player tur ns into hot product,
but competition puts heat on pricesLow-k dielectrics
stuck in low-gear . . .. . . but they're plowing ahead
to volume output by mid-decadeWhat? A chip market that's
growing in communications?Looks like cell phone business
will start growing again in '02
Looks like ASIC market will
turn around this year---barely
The ASIC market will definitely do better this year than last, but that ain't saying a whole heckuva lot. "Compared to the market disaster that was 2001, almost any scenario in 2002 represents an improvement," comments Jordan Selburn, iSuppli analyst who has just come out with a new ASIC study.
He isn't looking for much. "iSuppli expects the resumption of very modest growth in the coming months," he says, "driven by a second-half upturn in communications and continued strength in consumer applications." Sales should rise a scant 5%, Selburn estimates. But this should set the stage for l ong-term growth, he says.
Global sales of application specific ICs fell by 28% last year, he estimates, due mostly by the downturn in communications gear, which depressed demand for semi-custom chips, including gate-array and cell-based ASICs.
ASIC growth will kick into high gear next year, Selburn predicts, and the market will fully recover by 2004, when revenue will finally exceed the peak level of 2000.
But despite this positive outlook, Selburn figures that ASIC suppliers are facing new challenges that could diminish their sales growth rates over the next few years. The greatest challenge, he says, may come from the emerging fabless ASIC companies, which are expected to provide increasing competition to established ASIC players. Also, the increasing use of standard product solutions may cut into the growth of entrenched ASIC vendors.
One ASIC category that won't grow, Selburn says, is gate arrays, which are expected to continue their decline. He predicts these sales will fall from abo ut 10% of total ASIC revenues--or $1.34 billion in 2001--to less than 4% of ASIC revenues--or less than $1.2 billion--by 2006. This decline will happen, Selburn says, despite efforts by some gate array vendors to provide easy migration paths from more costly Field Programmable Gate Array devices.
Four out of Top 5 suppliers of ASIC/PLDs are U.S.-based
Global sales in $ billions
2001 | 2000 | ||
1. IBM Microelectronics | $2.76 | $2.72 | +12% |
2. Agere Systems | $1.31 | $2.32 | -43% |
3. LSI Logic | $1.24 | $2.01 | -38% |
4. NEC | $1.24 | $1.92 | -35% |
5 . Xilinx | $1.15 | $1.56 | -36% |
Source: Gartner Dataquest
IDC sees boom in WLAN gear,
sales will double in 5 years . . .
Another super-bullish report has come out on wireless local-area network gear. It seems like everyone is calling WLAN a growth market for semiconductors. It certainly looks like that to me too, but it might take longer for it to boom than these folks are predicting.
Worldwide revenue for WLAN equipment hit $1.45 billion last year, up 34.2% from the year before, according to International Data Corp. By 2006, it predi cts, this market will shoot up to $3.72 billion.
"Mobile connectivity is being driven by the acceptance of wireless devices and infrastructure both in and out of the enterprise," points out IDC analyst Jason Smolek. Entrenched data networking vendors, he says, will integrate WLANs as part of their overall networking strategies.
"But perhaps the largest opportunity will come from the SOHO and home networking space, where Asia/Pacific vendors are driving WLANs to the masses in all corners of the world," he points out.
Driving WLAN adoption, he says, is growth in such vertical markets as education, retail, airports, hotels, and SOHOs. Introduction of 5-gigahertz products will drive up spending in large enterprises, he adds.
. . . and In-Stat agrees--sees
WLAN chip sets up 50% in '02
Despite a slow economy, wireless LAN (WLAN) chips were a bright spot in an otherwise dismal 2001 worldwide chip market, according to In-Stat/MDR.
Unit sales of wireless L AN chip sets shot up by 23% to 8 million in 2001, a rare bright spot in a dismal worldwide chip market, the research firm says. But due to rapidly eroding WLAN chip set prices, revenues dropped 4% from the previous year to $216.9 million, according to In-Stat.
But this year could be better. Early indications indicate that chip set demand, already up sharply this year when compared with the end of 2001, will shoot up past 14 million units, an increase of 75% over last year. The best news is that WLAN chip set revenue is expected to increase to $324.5 million in 2002, a 50% increase over 2001, according to In-Stat.
For a long time, WLAN use was limited to a very few specialized vertical applications. But as new standards like 802.11a and 802.11g were developed for the home and the enterprise, the market is gaining even greater momentum, In-Stat says.
While the enterprise will consume the most wireless LAN chip sets this year, it's the home market that will experience the largest growth rate, the research firm predicts. Not only are home users adopting WLAN devices for their conventional computer connectivity needs, but other residential uses for WLAN chips are starting to become popular, especially in Asia.
For now, 802.11b accounts for almost all of the WLAN chipsets sold. But this will change very quickly. The enterprise has wanted a faster technology here for a long time, and they're about to get it in the form of 802.11a and 802.11g. Both standards offer data rates up to 54 million bits-per-second. In-Stat predicts that more than 13 million 802.11a chipsets will be sold into the home market in 2006.
(Click here for last month's SBN Outlook Column