Are inmates now
in charge at asylum?

I've just came up with another one of my famous rules of thumb. This one says: No expert really has a clue on when the current chip recession will end. Not when prices on semiconductor stocks start moving up again--that already may have happened. But when orders bottom out and start moving up again. (Personally, I continue to believe that the real bottom, as far as chip orders are concerned, won't happen until late this year at the earliest.)

In this week's column, I'll kick around the outlook as seen by three "experts"--one who says the downturn is over and that the chip business will climb 10% this year, another who says the global market will be down nearly 20% this year, and finally one that expects equipment sales to fall to 27% in 2001. Now those are spreads!

Are misguided analysts and
trade groups misleading us?

Forget what you've read about the chip market cratering this year. A new forecast calls for 10% growth in global chip revenues this year, claiming a semiconductor recovery already is underway but that it's being clouded by misguided analysts and trade groups that focus on sequential drops in monthly sales.

The chip business already is up 4.4% for the first quarter over last year, declares Robert Castellano, who runs The Information Network. "Few people realize this because the Semiconductor Industry Association has been doing the industry a disservice by pointing to negative sequential monthly growth in every press release."

While nearly all analysts are looking for a big drop in chip revenues this year, Castellano disagrees. "The analysts are focusing on the PC and handset market and what's happening to Cisco. They should be focusing instead," he says, "on the fact that more feature-laden products are requiring more chips even though the overall growth of the end product may have slowed."

Take flash memory, for example, he says. Cell phones this year require twice as much flash memory as last year's models, Castellano says. Other products needing more chips, he adds, include digital cameras, MP3 audio players, set-top-boxes, and antilock braking systems.

Here are a couple of interesting predictions. The integrated device manufacturers should bounce back before the troubled foundries, which will have to wait until late 2001 before recovering, Castellano predicts. And he says the production equipment business racked up a 12% growth rate in the first quarter, which might come as a shock to some vendors. This market should start seeing a boost in capital investments by chip makers in late September or October, he says.

(See May 22 story.)

Chip gear makers fall to
record low book-to-bill . . .

SEMI's CEO, Stanley Myers, bit the bullet this week. "The severity and depth of this industry correction is unprecedented," he admits. Other folks call it the worst recession in the history of the chip business. I dunno--most of you probably don't remember the early 1960s.

Nevertheless, this year is a pistol. North American-based suppliers of chip production systems posted a record low book-to-bill of 0.42 in April, the SEMI trade group reports. That means the industry ended up signing orders for only $42 worth of products for every $100 worth shipped.

Until now, the worst book-to-bill came in September, 1998, when SEMI reported a ratio of 0.57 at the end of the last industry recession. This time around, he says, "cancellations of previously reported orders for semiconductor manufacturing equipment were a significant contributor to the monthly bookings decline."

Worldwide bookings for North American-based tool suppliers fell to 712 million in April, 41% below the March's $1.2 billion and a 74% drop from the $2.7 billion in April a year ago. Shipments fell to $1.7 billion in April, 17% below March and 15% less than April a year-ago.

(See May 22 story.)

. . . so SEMI slashes
its forecast for 2001

It now seems like a long time ago on a planet far away when SEMI predicted the global chip equipment business would shoot up by more than 22% to $57.2 billion this year. But this week--just a half year later--the trade group reversed itself, predicting that 2001 shipments will decline by 27% to $35 billion.

For the first quarter, global chip-equipment shipments totaled $11.2 billion, an increase of 10.9% over last year but 12.7% below the fourth quarter of 2000. Orders received, however, were a disaster. SEMI says that orders amounted to $6.5 billion in the first quarter, a 50% fall from the same quarter last year and 53% below orders for the previous quarter.

But despite these awful numbers, Elizabeth Schumann, market tracker for SEMI, was optimistic. "While the rapidity of this downturn has been severe, we do expect a recovery in orders late in the second half," she says. This should contribute to a "more stable 2002 with flat or single-digit growth across the industry," she predicts. SEMI's new forecast shows only a modest 3% growth next year, but it expects the market to shoot up 22% in 2003.

(See May 23 story.)

El Segundo researcher checks-in
with most pessimistic forecast yet

This may be the most pessimistic forecast to date for this year's chip market. Our old friends--inventory gluts and excess capacity--will cause the worldwide semiconductor market to fall by 19% in 2001, predicts iSuppli in a new report this week.

"Bloated inventories continue to put a damper on unit ordering and prices," says vice president Greg Sheppard. "When order and revenue growth does begin to pick up in the third quarter, it will be at a slow rate due to soft demand from the main market drivers," he comments. "Add this soft demand to the new fab capacity coming online and downward price pressure for many semiconductor commodities--particularly memory and many ASSPs--is unavoidable," he adds.

The iSuppli outlook is not all bad. The PC sector has reached inventory equilibrium--even though unit growth rates are nearly half of what they were a year ago. It also holds out hope for communications chip markets with the wireless market expected to return to replenishment ordering by June, Sheppard predicts.

"But even after the huge inventory write-downs taken by Cisco and other major players, the overall networking sector will remain moribund until the fourth quarter," he predicts. "In fact, the FPGA/PLD suppliers serving this market are still holding six months of inventory and are clearly in the worst shape of any group of semiconductor manufacturers," he points out.

(See May 22 story.)

Avanti pleads no contest, but
financial woes have just begun

While Avanti executives may have ended their long-running criminal case by pleading no contest this week, the company's financial woes most likely are just beginning. Cadence Design Systems says it will seek hundreds of millions of dollars in restitution and will ask for a permanent injunction against Avanti's flagship Apollo CAE product.

A June 4 hearing in the criminal case will determine how much Avanti should pay Cadence for losses suffered as a result of Avant''s now acknowledged misappropriation of Cadence source code. Cadence general counsel Smith McKeithen expects the total to run in the "hundreds of millions of dollars."

Then there's Cadence's long-delayed civil case against Avanti Here, McKeithen says, Cadence may be seeking more than a billion dollars in exemplary, or punitive, damages over and above direct damages.

And Cadence "absolutely" will seek a permanent injunction against the Apollo placement and routing system because it contains Cadence trade secrets, he adds. The civil case could take as long as a year--time enough for Avanti to replace Apollo installations with its next-generation Astro product.

Earlier this week, Avanti, its chairman and six current or former employees surprised the design-automation community by pleading no contest to charges in the five-year-old trade secrets case. The defendants were accused of stealing place-and-route software code from Cadence and using it to build a competing router. The defendants will pay $35 million in fines, and five of the defendants will likely receive jail time. Avanti CEO Gerald Hsu won't have to go to jail but he has to pay $2.7 million in fines.

(See May 23 story.)

Analysts figure book-to-bill
will fall even more in May

So how just far is the book-to-bill ratio going to fall for North American-based suppliers of chip production systems? Analysts were prepared to watch April's book-to-bill ratio fall, but the record low of 0.42 reported this week caught them by surprise.

Now a poll of analysts and other observers by SBN indicates they believe that SEMI's book-to-bill ratio will break the record low at least once more--perhaps falling to below 0.40 in May before flattening out this summer. But they aren't sure whether tool orders will snap back up in the fourth quarter.

"This is the darkest moment yet," worries VLSI Research's Risto Puhakka. But, he adds, "it also reminds me of July 1998, when we were all at Semicon West and people wondered if the business would ever recover. That was the lowest point, until now in new tool orders, and then the recovery began," he recalls.

Despite the current market collapse, chip equipment vendors are still running at revenue and order levels far above those recorded in the 1998 recession. In September 1998, new orders ran $476.4 million, one-third less than new orders in April. And April shipments were more than double sales in September 1998.

(See May 23 story.)

Chip business ain't
no stinking auto industry

More than 30 years ago, the big question in Silicon Valley was: How much consolidation would the "maturing" semiconductor industry go through. I know, I was there. Many smart people were predicting that like the auto industry, the Top 10 chip makers (most of them were U.S.-based) would soon end up with 80-or-90% of the world market and startups would virtually disappear.

It soon became clear that these experts couldn't have been more incorrect, and over the next couple of decades whole new classes of semiconductor companies such as fabless and foundries flooded the competitive scene. Now believe it or not, a new generation of "experts" is predicting consolidation again.

One of my favorite "take no enemies" CEO is T.J. Rodgers of Cypress Semiconductor and I agree with his comments this week in EE Times, when he saw little or no consolidation coming in the chip business. Back when Cypress was founded in 1983, it was last on a list of 60 or so U.S. chip makers, he recalls. "That was the year that Intel and AMD said: Here comes consolidation and it's going to be the Big Three like Detroit and that's going to be it." Today, he notes, there are still "tons of chip companies out there."

Today, T.J. says, three factors are making the semiconductor industry "more diverse, not less." Tremendous specialization is going on" he notes, "and no company can possess all the intellectual property to compete in all areas." And he adds: "You're not going to get the world's best content-addressable memory guy to work in a division of a big company when he can start a company and make money and have control over his own destiny."

The second reason for more diversity, T.J. says, "is the advent of the fabless model. That overthrows the barrier-to-entry issue." And three, chip makers are encroaching into the systems world. To make the microprocessor, you have to be a systems expert. "And as the semiconductor industry gets smarter," he says, "it is encroaching into areas that used to be dominated by systems companies."

"That's why there can't be consolidation," T.J. says. "Silicon is working its way up into the food chain and not just as a special component. The semiconductor guys will take over on the systems side and not ice versa. The systems guys have cost structures that aren't competitive." Right on, T.J.!

(See May 23 story.)

Now's the time to get
that 300-mm fab going!

My old man always told me: The time to open a store was before construction began on nearby tract houses. He also said to build an apartment building when vacancies were highest. The point is, timing was everything--and that couldn't be more true than it is in the chip business.

Take the new generation of 300-mm wafer fabs. It may be too late now to start one of these giants to catch the next wave in the global chip business.

Showing what could be excellent timing is a Singapore joint venture among Philips Semiconductors, Taiwan Semiconductor Manufacturing (TSMC), and the Economic Development Board Investments of Singapore. The partners have just opened a $1.2 billion fab for their Systems on Silicon Manufacturing Co. (SSMC), despite the meltdown of global electronics markets and the worsening forecasts for the second half.

The downturn will allow the Singapore fab to increase its capacity and push its technology in time for a market rebound, says TSMC president F.C. Tseng. The new fab produced its first yielding silicon just 90 days after the process equipment was installed. Production cycle time and yields already are comparable with the best achieved by Philips and TSMC fabs, they claim.

(See May 23 story.)

Who me worry? Japan takes
back lead in buying chip gear

Are the Japanese smarter than the rest of the world regarding market timing? Could be. Guess what region is the hottest market for semiconductor production equipment in the first quarter? Yep, Japan, whose industry spent a third more money to buy chip production tools in the first three months than it did in the fourth quarter of last year.

Japan's $3.37 billion in purchases makes it the largest regional market, according to the SEMI trade group. Losing the top spot was North America, which showed a 24% sequential decline in tool purchases to $2.97 billion.

Last time that Japan was the leader was the third quarter of 1996, says SEMI analyst Elizabeth Schumann. That was the year that North America started pulling away to become the largest chip gear market, she says. "During most of the early and mid-1990s, North America and Japan markets were running neck-and-neck as the top equipment market in the world," she adds.

Only Japan and South Korea showed sequential gains in equipment purchases in the first quarter of 2001. Chip makers in both these countries lagged Taiwan, North America and Europe in buying chip gear during the boom of 1999 and 2000.

Taiwan was the big loser in the first quarter. Its consumption of chip making gear nosedived 45% to $1.13 billion from $2.04 billion spent in the fourth quarter. Taiwan, hottest market in the 1999-2000 boom cycle, fell behind Europe, North America, and even the "rest of world" in equipment sales during the first quarter, SEMI reports.

(See May 23 story.)

Picture keeps getting
worse for Chartered . . .

Business is falling far faster than Chartered Semiconductor had expected just a few weeks ago. The Singapore foundry has now doubled the size of the drop in its sales outlook for the second quarter. Sales are expected to drop 48% from the first quarter to $107 million--the wafer vendor previously had predicted a 25% decline.

What's more surprising, the company's average fab utilization rate is expected to fall into the "low 30%" range during the second quarter. Fab utilization had already slipped to the 40% range in the first quarter.

And there is still no light at the end of the tunnel, according to Chartered. "Given the current turbulent market environment, it is difficult to predict revenue trends in the coming quarters for both the industry and Chartered," declares CEO Barry Waite.

But the world's third largest pure-play silicon foundry "still believes that the second quarter will be the low point of this cycle," says Waite. "We say this largely because of the swiftness and severity of the falloff in the first half for all of us in this industry."

"Based on historical patterns and on the severity of the decline seen in the first two quarters of 2001," Waite figures that Chartered's revenues will increase sequentially in both the third quarter and the fourth quarter.

(See May 21 story.)

. . . but it still expects to invest
$1 billion for capital spending

Despite its disintegrating business, Chartered Semiconductor still expects to invest $1 billion in capital spending this year, and it's not likely to delay the start of 300-mm wafer processing next year.

"The continuing sharp decline in business indicates that revenues need to come up just to meet the actual consumption levels, even if there is only the modest growth in the end markets," Waite comments.

Networking ICs and wire-line devices continue to suffer the most in the current downturn, while PC-related products are holding up better in the foundry market, Chartered says. When foundry orders fell off the cliff in the first quarter, the Singapore foundry turned to "fab fillers"--mostly memory devices for existing customers--to keep its factory utilization rate just above 60% in the quarter March 31. But even the fab fillers are now running out of gas," Waite complains.

Foundry orders from major chip house--or IDMs--were down at about the same rate as bookings from fabless semiconductor companies, Waite estimates. But those IDMs that have not completely adopted a foundry strategy yet have pulled their products back to their own fabs and terminated foundry agreements.

Chartered's capital spending budget will run 9% over 2000's $911 million investment. It had cut back its original plan of $1.2 billion capital spending earlier this year. Waite indicated that 300-mm development continues to be a top priority.

(See May 22 story.)

Big CDMA design move
by Agilent Technologies

Agilent Technologies took a big step this week toward expanding its business in third-generation (3G) cell phone and wireless applications, including new hybrid personal digital assistants that also will serve as cellular telephones.

It acquired Sirius Communications, spun out in 1996 out of Belgium's IMEC independent research organization. The Brussels-based developer of ASICs for CDMA cell phones and satellite com systems was set up to develop CDMA chips using direct sequence spread spectrum technology. Sirius, which employs 19, has been licensing its CDMAx wideband (W-CDMA) transceiver technology to device makers and telecom companies.

"Our plan is to integrate Sirius technology with Agilent's existing data processing capability, and then combine it with our RF integrated circuits to form a complete wireless PDA solution," says Agilent vice president Steve Hoffmann. "We also plan to merge Agilent's CMOS digital camera sensors, optical navigation sensors and infrared transceivers to deliver a whole new generation of mobile information appliances."

Picking up the Sirius design team will give Agilent a 3G design center in the heart of Europe and put it at the leading edge of universal mobile telecommunications systems developments, says Sirius CEO Lieven Philips.

(See May 21 story.)

Will Intel feel more heat
from AMD-Transmeta deal?

Advanced Micro Devices continues to reinforce its positions in the microprocessor "battle of the century." This week the No. 2 processor maker brought in an ally to compete with Intel in next-generation processors.

AMD licensed its 64-bit x86-64 technology and HyperTransport chip-interconnect format to Transmeta. The low-power processor startup plans to use the AMD technologies in future x86-compatible processors to extend its product line from 32-bit to 64-bit computing.

The two processor makers called their licensing agreement an alliance to promote next-generation microprocessor standards. "AMD's x86-64 instruction extensions provide the best upward compatible path for adding 64-bit address capabilities to the x86 instruction set for the PC industry," maintains David Ditzel, Transmeta's chief technology officer.

The highly publicized startup, which launched its power-stingy Crusoe processor last year, also is moving to AMD's HyperTransport high-speed I/O bus technology to offer faster links between chips inside computers, networking gear, and communications systems.

AMD claims its HyperTransport bus will provide an I/O interconnect format that speeds communications between ICs by up to 48 times over existing bus systems. The big chip maker is currently lining up companies to form a HyperTransport bus consortium, which it hopes to launch later this year to derail Intel's 3GIO bus format for next-generation systems.

(See May 25 story.)

157-nm looks like it can
keep optical litho going

The big question in chip manufacturing continues to be whether optical lithography can be extended with 157-nm technology to handle 0.065-micron (65-nm) processes by the middle of this decade. If it can't, next-generation lithography (NGL)-either extreme ultraviolet (EUV) or electron projection lithography (EPL)-will be needed sooner than expected during this decade.

But the outlook for 157-nm technology was looking better than ever this week at Sematech's annual symposium on the subject at Dana Point, Calif. Lithography experts seem to be more optimistic about the use of 157-nm exposure tools in 0.065-micron process technologies.

"The confidence level in 157-nm technology continues to rise," declares Rich Harbison, Sematech's 157-nm program manager. "While the birefringence of CaF2-the material of choice for 157nm lens--presents a new engineering challenge," he admits, "the optical designers have been working on an approach to compensate for it."

Substantial process was reported on a number of fronts, he reports, including 157-nm resist overlay capabilities, reticle-protecting pellicles, new optical designs using calcium-fluoride (CaF2) lens materials, and other infrastructure technologies.

Laboratory systems based on 157-nm wavelength light already have demonstrated isolated lines with dimensions of 50 nm (0.05 micron) using alternating phase-shift photomasks and a 0.6 numerical aperture (NA) exposure system, according to Sematech, which organized the meeting along with the IMEC research group in Belgium and the Selete industry consortium in Japan.

The current workhorse in advanced wafer fabs--248-mn lithography--is expected to support 0.13-micron (130-nm) technology in volume production, but the new 193-nm scanners exposure tools are expected shortly to take on the critical dimensions in 100-to-90-nm processes.

(See May 24 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!.)