Greetings from Down-East Maine, or vacationland as our license plates call it. At noon, the temp creped up into the 60s and the skies were clear--a great day!
But I admit it, I'm really getting bummed out by all the disturbing trends now battering the semiconductor industry. So I've been looking harder for a few rays of hope that may have been buried by the current landslide of layoffs, and lower sequential sales and profits.
And I found a handful--including one optimistic note from the hard-hit foundry business, of all places. And also from this week's comments from STMicro's Pasquale Pistorio, who repeated his earlier prediction in July that the current quarter will be the chip market's bottom.
It's still hard to believe this country is NOT in a recession--the semiconductor industry sure is. By the way, I still believe the chip recession won't turn into the chip upturn until later next year.
Bottom was reached
in spring, says TSMC
The top dog in the hard-hit silicon foundry business--Taiwan Semiconductor Manufacturing Co.--says that May and June appear to be the low point in the revenue slump and it now expects sequential growth for the third and fourth quarters.
The world's largest silicon foundry says its book-to-bill ratio has "hovered around 1 for four months since April. This is a reasonably solid sign that the bottom has been reached," the company says.
Indeed, its latest guidance, TSMC predicts that third quarter operating profit and net income should not be lower than it was in the second quarter.
But the second quarter was a killer. Sales dropped to $754 million, down 33.5% from the previous quarter, and net income power-dived to $8.9 million from $241 million. Capacity utilization fell to 44% from about 70% in the first quarter and will fall to the "upper 30% range" in the current quarter, TSMC says.
Despite the current slump, the Taiwan foundry giant does not plan to cut back its capital expenditures this year from its current plan of $2.2 billion. More than half of this investment will go toward building 300-mm wafer fabs and adding capacity in 2002 and beyond.
(See July 26 story.)
Third quarter is bottom,
insists STMicro's Pistorio
Here's someone else who sees the bottom in the current downturn--an industry leader I really admire but don't agree with here who has built one of the chip industry's leading companies.
"I think the bottom for the chip industry will be the third quarter, and we will see a bounce back in the fourth quarter," declares Pasquale Pistorio, CEO of STMicroelectronics, repeating his early-July prediction in Manhattan this week. But, he adds, "there still will be excess capacity in the industry, so price pressure will continue to be strong."
One reason why Pistorio feels this way is that customers are telling him that excess chip inventories are diminishing for such products as cellular phones. "The indications our customers are giving us is that they want more products in the fourth quarter than in the third quarter," he says. "And we are assuming they have reasonable visibility by now."
But the European chip maker is still working hard to trim costs until the upturn is confirmed. It closed a Singapore fab for three weeks in June and will soon shut a discrete semiconductor factory temporarily. It also trimmed 1,500 employees by attrition and layoffs, closed a fab in Ottawa, and cut its R&D budget.
"The secret in these things is to respond to the need to cut costs now without impacting long-term strengths," notes Pistorio. "It's a lot easier for a CEO to send out a memo about a flat 10% cut, but it's not the best thing to do for the future of the company."
(See July 25 story.)
Downturn shuffles Top 10;
Moto, Samsung drop 2 spots
The current chip recession has shaken up the rankings of the top 10 chip suppliers.
A 45% drop in global DRAM revenues in the first half clobbered Micron Technology, which a year ago was the 10th largest chip maker. As a result, the Idaho DRAM vendor dropped out of the Top 10 this year, according to IC Insights.
The bloodbath in chips for cellular phones and communications gear hit Texas Instruments and Motorola particularly hard. Among the 10 leading chip companies, Motorola showed the biggest drop in first-half sales off 30%, while TI was the second biggest percentage loser with a sales drop of 22%.
Both Motorola and Samsung dropped two spots in the list, Moto from sixth to eighth, and Samsung from fourth place to sixth.
There was some big upward movement too. STMicroelectronics vaulted from seventh to fourth place with its chip sales only falling 2% in the first half of 2001, according to IC Insights. Japan's Mitsubishi Electric broke into the top 10 ranking 10th vs. 13th last year, primarily because of other chip suppliers' big sales declines. And Hitachi moved up one notch from seventh to eighth place.
But some things never seem to change. The first three positions were the same as last year: Intel in win, Toshiba in place, and NEC in show.
(See July 24 story.)
SEMI not ready to call
turnaround--I don't blame them
So, is it good news or bad when incoming chip orders come in faster than shipments went out in June? I would always like to believe that the glass is half full--not half empty. But the SEMI trade group doesn't seem to be convinced yet that there's light at the end of the tunnel.
While SEMI's book-to-bill ratio for North American tool suppliers popped up slightly to a reading of 0.54 in June from a revised 0.48 ratio in May, the trade group was not ready to "declare a reversal in order trends yet," says SEMI researcher Elizabeth Schumann. I really don't blame her--her numbers mean that chip gear makers wrote only $54 in orders for every $100 worth of equipment they shipped last month.
"While the book-to-bill ratio improved slightly in June, total shipments and orders continued to decline," she points out. "Some specific equipment categories are showing modest improvement, which could be an indication that the industry is nearing the trough of the orders cycle."
June orders were down 1% from May and 75% from June a year ago. Shipments fell 12% from May and off 42% from last year. Not a very pretty picture, year-to-year, but typical for this boom-or-bust business.
(See July 23 story.)
Japan Inc. rides again
in major R&D launch
Japan is gambling it can replay the glory days of the '70s when a government-industry R&D team was able to push that country's semiconductor industry into a global power. The Japanese seem to be no less ambitious this time around and are launching a major R&D effort aimed at capturing one-quarter of the world's market for system-on-chip (SoC) products by around 2005.
The Japanese were talking up the new program this week, but to me it sounds like an overwhelming task for what appears to be just another government bureaucracy. Running the show will be Mirai, Japan's answer to IMEC, International Sematech, and the EUV LLC consortium all rolled into one.
Japan's new research juggernaut will develop SoC process technologies for the 70-nanometer node by 2004 and for 50-nm products by 2008. It will concentrate on five research areas: gate technology for high-K materials, wiring module technology with low-K, new transistor structures and measurement analysis technologies, lithography, and circuit system technology.
The need to meet the new targets is critical to the future of Japan's semiconductor industry, acknowledges director Masataka Hirose. He figures the world market for SoCs will pass the $200 billion mark in 2005 and $400 billion in 2008. And by 2011, the market will double again to pass $900 billion.
Industry, academia, and bureaucracy are all participating, which certainly sounds like Japan Inc. of the 1970s all over again. Mirai will kick off its giant effort this August with $31 million in funding from the Ministry of Economy, International Trade and Industry. Mirai also will build on the work of several other industry-academic-governmental research consortiums, including the Selete project and Asuka, Mirai's ongoing predecessor project. Are you still with me? Can you diagram the organizational structure?
Mirai will be controlled by the Advanced Semiconductor Research Center and the National Institute of Advanced Industrial Science and Technology. With a huge staff of 3,200, the center is a subdivision of a new research body called the Institute of Advanced Industrial Science and Technology, which is an integration of 15 research institutes. Pheew!
Despite this complex of organizations, the Japanese claim there will be neither turf wars nor paper shuffling. Mirai, they maintain, will be different from the standard government-backed research initiatives that have been long criticized because they are weighed down by bureaucracy.
To get around this kind of problem, the leaders of Mirai's five research projects will have direct responsibility and decision-making powers to pursue or to scrap ideas. The project supposedly will shelve Japan's notoriously hierarchical, even tribal, research culture and replace it with a flat structure.
(See July 24 story.)
China's chip requirements
to outpace its fab expansion
We all know China is the chip industry's final frontier and huge amounts of money are now being spent, much of it from the Chinese government, to add local chip capacity to meet the needs of this rapidly growing market. Well, one researcher has looked into this activity and concluded the nation's wafer fabs will still greatly lag the demand for ICs five years out.
A new study from The Information Network predicts that only 23% of the semiconductors needed in China will be produced by the country's wafer fabs in four years. A strong growth in chip demand will outpace China's fab expansion efforts, the New Tripoli, Penn., researcher concludes. Last year, China produced 27% of the chips it consumed--or about 5.9 billion devices, it estimates.
In the next five years--China's 10th five-year development plan--some 25 new advanced fabs are scheduled to be built, according to the market researcher. It estimates that China's semiconductor equipment market will grow nearly 30% per year through 2005 from less than $500 million in 2000.
(See July 25 story.)
SDRAMs should outsell
Rambus DRAMs next year . . .
The competitive outlook for Rambus DRAMs doesn't seem to be all that good. While Elpida Memory, for example, is still backing all major DRAM architectures, executives from the big Japanese chip maker believe that both its SDRAMs and double-data-rate SDRAMs will leave Rambus DRAMs in the dust next year.
DDR SDRAMs, and Rambus DRAMs each account now for 15% of its shipments, according to the joint DRAM venture between Hitachi and NEC.
But that is changing. By early next year, Elpida's unit shipments of DDR SDRAMs will pass those of RDRAMs, predicts Mike Despotes, CEO of its U.S. subsidiary. "Rambus is out and doing reasonably well," he says, "but these shipments will decline."
One reason for this shift is Intel's emerging support for DDR SDRAMs. At present, Intel's Pentium 4 processor line only supports RDRAMs. But the chip giant is now sampling a chip set that supports SDRAMs for the Pentium 4 line.
(See July 24 story.)
. . . and Intel plans to halt
Rambus discount program
More bad news for the Rambus DRAM came this week when Intel confirmed industry reports that it will phase out its discount program of bundling Direct Rambus DRAMs with its Pentium 4 processors to PC integrators and white box makers.
The price premium of Direct Rambus DRAM memory has dropped sharply and Intel no longer needs the bundling incentive to spur Pentium 4 sales, a spokesman says. And once Intel starts marketing its Brookdale 845 chipset supporting SDRAM memory, industry sources say, there no longer will be a pressing need to offer Direct Rambus discounts.
(See July 24 story.)
Elpida has big plans
for lousy DRAM market
Demand may be in the pits and prices may be falling through the floor, but Elpida Memory is going all out to increase its DRAM business. The Japanese joint venture between Hitachi and NEC aims to nearly double its market share and take 20% of the world's DRAM market by 2003. Now that's ambitious!
Already, the DRAM vendor is putting in the capacity to accomplish that growth. The company's first 300-mm wafer fab, located in Hiroshima, is scheduled to move into pilot production by the end of the year, with first silicon rolling out of the plant by April of next year.
But Elpida's current DRAM business, like its competition, is awful. "The DRAM market is alive and well, but we're in a tough spot right now," acknowledges Mike Despotes, CEO of Elpida's U.S. subsidiary. "We're not only seeing a dramatic decline in terms of ASPs average selling prices but also a decline in demand," he says.
Despotes predicts the worldwide DRAM market could range anywhere in size this year between $11-to-$18 billion. That would be down from $29 billion last year. "I don't know if the recovery will come sooner or later."
(See July 24 story.)
Would Ford ever buy
new engines from GM?
Analysts and competitors alike this week were quick to pooh-pooh Motorola Semiconductor's latest bright idea to expand its market for cellular phone chips and embedded systems technology--that is to sell its proprietary products to its competitors in the mobile phone market.
To my knowledge, no chip maker has ever been able to make that kind of strategy work. But Motorola execs claim this time will be different for a couple of reasons.
The cell phone industry is in a major transition, they say, with suppliers moving from doing both design-and-manufacturing to a new paradigm splitting product development and assembly of handsets from the business of selling units to consumers. Next-generation handsets and systems also are becoming a lot more difficult to develop.
After decades of developing parts only for its own systems businesses, Motorola is now making available all its proprietary technologies to mobile phone makers worldwide. This expanded market will be worth $35 billion by 2004, Moto figures.
Execs deny their internal customers Moto still owns 15% of global cell phone business will get parts any faster than outside users, which is one big reason why this strategy hasn't worked in the past.
Motorola has been working on this new strategy for a year, and already has transferred as many as 300 engineers from its personal communications business to semiconductor products. The entry point for the new, open Motorola semiconductor strategy will be the 2.5G phone, which is beginning to appear but is not yet in volume use. Motorola claims company watchers will know that its new strategy is working when the firm reveals key design wins in both 2.5G and 3G models in Europe and Asia by the end of this year. We'll be watching closely.
(See July 27 story.)
DRAM makers have slashed
capital spending by $9 billion
No one ever said the DRAM business was for the faint-hearted. No wonder there is only a handful of DRAM vendors surviving after hundreds of producers tried to make it in this giant semiconductor market.
Just look at the roller coaster ride the memory business is taking this year. In June alone, DRAM vendors slashed their capital expenditure plans by $1 billion. So far this year, this small group of chip makers have chopped their capital spending plans by more than $9 billion, according to David Parker, Micron Technology's investor relations manager.
Micron has cut its own capital spending from $1.8 billion in the year ending Aug. 31 to $1 billion in its upcoming fiscal year. The way things stand now, Parker estimates that Samsung has cut its spending from $3 billion to $1.8 billion, Hynix is slicing its spending in half to $520 million, and Nanya Technologies is reducing its expenditures by 33% to $358 million. Memory producers had no choice, he says, after DRAM prices in June fell below producer costs.
Micron is going ahead on its schedule to install a 300-mm wafer R&D fab in Boise, but has postponed indefinitely plans to equip its shell fab buildings in Lehi, Utah for 300-mm wafer production.
Despite the current state of the market, Micron is not planning any cutback in its own DRAM production, claiming it is one of the few memory firms not scaling back output this summer. So, is it shipping a dollar bill with each DRAM?
(See July 25 story.)
I worry about Hewlett-Packard
Today's Hewlett-Packard is not Dave Packard's or Bill Hewlett's company anymore. I worry about this great corporation. Over the years, I have watched it grow from a $20 million electronic instruments maker to a multi-billion dollar electronics giant, only to see it stumble in the past couple of years. That doesn't seem to be because of its products--I still own several pieces of HP gear and they don't come any better.
But what's happening? The company's espirit de corps, its morale, its style, its class all seem to be disappearing. Now this week, HP is slashing 6,000 jobs, or 6.5% of its workforce, and is warning that third-quarter sales will decline 14-to-16% over last year. And I can remember when HP didn't do layoffs!
(See July 26 story.)
Agere breaks record,
loses over $1 billion
I never thought I'd see the day when a chip maker loses more than a billion dollars in a year, much less in a quarter. But Agere Systems (formerly Lucent Micro) did it this week when it reported a net loss of $1.1 billion for the quarter ended June 29.
This awful figure included a larger-than-expected inventory write-down of $270 million and $440 million for restructuring charges associated with layoffs and plant closings. Revenues, which were slightly above revised estimates, came in at $927 million.
Now the company is trying to downsize its operations to lower its breakeven point but still continue investing in optoelectronics and ICs for the eventual upturn.
Agere is now forecasting a 30-to-35% sequential decline in revenues for the current quarter vs. last year.
(See July 25 story.)
Will this high-speed link
help AMD against Intel?
It had better, if Advanced Micro Devices wants to move up in the 64-bit world. The chip maker blew the trumpets this week to kick off the consortium to support its HyperTransport I/O Link, which was developed to compete against Intel's 64-bit microprocessor bus architecture. It will transfer data inside systems at speeds up to 12.8 gigabytes-per-second.
The HyperTransport Technology Consortium--made up of AMD, API Networks, Apple, Cisco, Nvidia, PMC-Sierra, Sun Microsystems, and Transmeta--will manage the development of the link's specs. Most of the members already have come out publicly in support of HyperTransport since its was launched by AMD five months ago as a point-to-point design to enable processors in PCs, as well as networking and communications devices to talk with each one another 24 times faster than current technologies.
"The consortium represents the commitment of its members to develop industry-wide adoption of HyperTransport technology and to drive this state-of-the-art technology into the market," declares Gabriele Sartori, president of the HyperTransport Technology Consortium.
(See July 24 story.)
Weak chip distribution business
forces Arrow to lay off 1,000 more
Like many companies in the semiconductor business now, Arrow Electronics is intensifying its efforts to resize and restructure the company to match its falling level of business. This week it revealed plans to lay off another 1,000 workers, close some facilities, and pare back other operating expenses as its electronic components distribution business continues to get worse.
While Arrow says its second-quarter net income exceeded expectations, it adds this was due to stronger sales of computer products and that its primary electronic components business "further deteriorated" during the period. This may mean that Arrow's third-quarter may be "somewhat worse than the second," according to CEO Francis Scricco.
The Melville, N.Y., company posted profits of $7 million in the three months ended June 30, down 92%, from $84 million in the year-ago quarter. Revenue slid 21%, to $2.5 billion from $3.2 billion in the second quarter of 2000.
The distributor continues to experience sequentially declining component sales as electronic equipment OEMs and contract manufacturers--"particularly in the communications and networking segments--are still ordering at greatly reduced rates in order to work off their own inventories of components and finished products," Scricco says.
Arrow will take a one-time reorganization charge during the third quarter, hoping that this will help reduce expenses by $100 million.
(See July 24 story.)
Fairchild manages
flat sales in quarter
Whenever I get the latest numbers from Fairchild Semiconductor, I always recall this was the chip business that National Semiconductor dumped a few years ago because it was serving a slow-growth, mature market.
Well, despite today's chip depression, the South Portland, Maine-based chip maker turned in second quarter results that were a lot better than those of most chip makers these days. Revenues slipped sequentially to $372 million, down only 3% from $385 million in the first quarter. And including all charges and other items, it posted a net loss of $8 million in the second quarter vs. a net profit of $60 million in the same period last year.
"We are conservatively guiding that our third-quarter revenues may be down as much as 15-to-20% from second quarter levels," comments CEO Kirk Pond. "We expect our third quarter gross margin to be in the 20-to-22% range, as we plan to continue to match production with near term demand and continue to drive our inventories lower."
(See July 24 story.)
TI's Engibous also sees
'signs of stabilization'
It was Texas Instruments' turn this week to report the same kind of grim second-quarter results that most other chip makers were racking up. In TI's case, a 19% sequential drop in revenue to $2 billion, a 12% drop in orders, and a net loss of $197 million.
But TI's CEO, Tom Engibous, like a small, but growing number of other industry leaders, is becoming more optimistic. "We now see some signs of stabilization," he says. "The rate of sequential decline for semiconductor orders has slowed, and it appears our semiconductor revenue is nearing a bottom."
In addition, Engibous notes that "wireless orders have increased, reflecting our customers' continued progress in reducing excess inventory as well as the impact of new programs. The downturn isn't over," he adds, "but we are beginning to shift our focus to recovery and growth."
TI is forecasting a revenue decline of 10-to-15% in the third quarter as some customers are still reducing inventories. As a result, it expects to report a third-quarter loss.
(See July 23 story.)
Infineon launches all-out
$1 billion cost-cutting drive
Infineon Technologies is taking draconian measures to lower its breakeven point. It has launched a major program that aims to lower its costs by a whopping $880 million over the next 12-to-18 months. Part of it includes eliminating 5,000 jobs, or 17% of its worldwide workforce.
Called "Impact," the huge cost-reduction program will also include shortened workweeks, streamlining logistics, procurement, and manufacturing operations. "The dramatic developments in the market and our business situation leave us no choice," comments CEO Ulrich Schumacher.
This effort is on top of other cost-cutting measures initiated in June, which included capital spending being chopped to $2 billion. It had also decided to cut its fiscal 2002 investment plans by $870 million to $1.3 billion and to institute an "almost complete hiring freeze."
Earlier this week, Infineon said that losses will continue in the current fiscal quarter, as it posted a 23% sequential drop in revenues to $1.1 billion for the quarter ended June 30. It has recorded a net loss of $323 million for the quarter.
The company seems shellshocked by the impact of the continuing downturn. "Our balanced product portfolio did not help to ease the current weakness in the memory market any longer," Schumacher says. "This dynamic development and the magnitude of the current downturn, especially in communications, was not expected."
(See July 23 story.)
Avant! managers go to jail
in source-code theft case
My pappy always told me that "crime doesn't pay." But I don't remember anything like this ever happening in the semiconductor industry.
The curtain rang down on the Avanti source-code-theft case and the CAD company will have to pay $195 million in restitution to Cadence Design Systems. Its co-founder Stephen Wuu received the harshest sentence and was taken away in handcuffs to serve two years at San Quentin because he took the Symbad database code from Cadence in 1991 while employed with that company. He was also fined $2.7 million.
Avanti co-founder Yuh-Zen Liao will spend one year in county jail and serve three years of probation. He will also have to pay $2.7 million in fines. Another co-founder, Eric Cho, will pay $108,000 in fines and spend one year in jail. Eric Cheng received 364 days in county jail and will have to pay a fine of $27,000. Mitch Igusa, was sentenced in late June to one year in county jail.
In addition, a class-action suit was filed on behalf of Avanti shareholders against the company's board of directors for paying the fines and legal fees of all defendants in the case. One Avanti defendant's attorney suggested that the class-action suit may result in the defendants paying their fines and attorneys out of their own pockets.
(See July 25 story.)
If you have any comments, criticisms, or questions, don't hesitate to E-mail us at bhenkel@aol.com. Have a great weekend!
(Click here for last week's Semiconductor Alert!.)