Another python swallowing the elephant type story ???
Semiconductor Business News, © 1999, CMP Media Inc.
February 15, 1999
Intel could jump-start 300-mm
By Jack Robertson and J. Robert Lineback
SANTA CLARA, Calif. -- The event that could launch the next wafer
generation may be finally about to happen.
Ten months after Intel Corp. postponed the decision to go forward with
its 300-mm development fab, the industry leader is apparently ready
now to restart the $1.5 billion project in Hillsboro, Ore. When the
announcement comes, most likely in March, the chip maker also plans
to detail a $10 billion program to add 12-inch wafer production at
several of its worldwide plants over the next decade, according to
sources inside and outside the company.
Intel won't confirm that it's going ahead with the delayed fab, but
managers involved with the project say the facility will be equipped
later this year to begin pilot runs with 12-inch wafers by 2000.
The apparent restart of the project comes after Intel's recent survey of
production equipment suppliers and the Sematech consortium indicated
that sufficient 300-mm tools were now available to begin equipping the
Hillsboro fab. It was concern over tool availability and general business
conditions that caused Intel last April to delay the purchase of 300-mm
gear.
"Everything is moving ahead now," states one Intel manager who asked
not to be identified. Final approval of the project still has to come from
Intel CEO Craig Barrett, but his sign-off appears certain now,
according to another company source.
Any decision to move ahead and equip the nearly completed fab
building in Hillsboro, should give the industry's stalled 300-mm transition
a much-needed boost, agree industry analysts and managers.
It couldn't come too fast for the equipment industry. After more than
five years of development and tool testing, the capital equipment
suppliers have already spent $5.4 billion on 300-mm R&D, estimates
analyst G. Dan Hutcheson, president of VLSI Research Inc. in San
Jose.
Total revenues to date from 300-mm prototypes and initial production
systems are less than $60 million, according to industry estimates. This
total came in surprisingly low despite the considerable efforts by trade
groups and chip manufacturing consortia to accelerate the transition
from 200-mm to 300-mm wafers.
Attempts to jump-start the 300-mm transition were nearly derailed last
summer when equipment suppliers and chip makers began blaming
each other for delays in prototype tools and pilot lines. But after the
public debate quieted down, the battered 300-mm movement gradually
began moving again in late 1998. And it was this slow, but steady
progress that led Intel to decide it was time to restart its project.
When Intel's 300-mm development fab starts up, it is expected to focus
on 0.13-micron process technologies. One major equipment vendor --
Applied Materials Inc. in Santa Clara, Calif. -- reportedly is planning to
re-institute its 300-mm tool program, based on a large order for systems
from Intel that are slated to go into the Hillsboro fab. Applied halted
much of its 300-mm efforts last year when chip makers began backing
away from pilot line projects.
Intel plans over the next couple of years to expand its 300-mm
development fab into a high-volume production facility. The timing of
this buildup will depend upon market demand and the success of early
12-inch manufacturing tools.
Intel's eagerly awaited move to 300-mm will help push the industry
transition, according to experts, but the global buildup of 12-inch wafer
capacity will be gradual. Volume production likely will not be feasible
until after 2002, predicts Clark J. Fuhs, director of the semiconductor
manufacturing analysis group at Dataquest.
"We think up to 8 or 10 pilot lines will be built by 2001 in some way,
shape or form," the analyst estimates. But "the economics of going to
300-mm are still up in the air," he cautions.
"Once you get to the 2002-2003 time frame, you realize that fewer
wafer starts will be needed in 300-mm fabs [to meet market demand].
It [may be] possible to buy less equipment and have fewer wafer starts
[in order] to produce the same amount of products in terms of square
inches of silicon," Fuhs says.
The key job of early pilot lines will be to figure out the economics of the
300-mm fabs as well as to integrate the new sets of tools and
processes.
Dataquest estimates that 5% of the industry's total spending on fab
equipment will go for 300-mm tools this year, or only about $76 million.
But that's still up from $35.4 million in 1998, or about 2.3% of total
equipment purchases that year.
After a few years, sales of 300-mm systems should start climbing
rapidly. By 2004, they will reach $11 billion, or about 29% of total fab
equipment sales, Fuhs estimates.
"It ramps up pretty steeply after 2004," the San Jose-based analyst
says. His forecast is based on the assumption that two companies --
Intel and most likely Texas Instruments Inc. in Dallas -- will start
ordering 300-mm gear for pilot lines in 1999. Orders for the
new-generation gear also will continue at the world's only existing
300-mm pilot line, which was started in 1998 by Siemens AG and
Motorola Inc. in Dresden, Germany.
"We now assume that four more 300-mm pilot lines will be started in
2000, followed by two more in 2001," Fuhs says. "It will take each
company about two years from the start of a pilot line," he figures, "to
reach a level of maturity [where] 300-mm becomes economically
feasible."
Intel's decision undoubtedly will bring 300-mm back into the headlines,
but Fuhs and other industry observers still believe that most
semiconductor manufacturers will remain cautious about spending any
money on 12-inch fabs until they have recovered fully from the recent
three-year slump in chip revenues.
"What will persuade fence sitters to start up their programs again is
healthy finances," Fuhs figures. "We are projecting that the financial
conditions of chip companies will certainly get better in the year 2000,
and we expect by then to see some [300-mm] projects being
rejuvenated again." |
