I have my doubts too. The concept itself is counter-intuitive. The cost per wafer may decline over time(as it has), but the cost to build fabs will still be enormous, and the profit potential will also be tremendous as IC's permeate nearly every aspect of our lives.
Brian
siliconstrategies.com
SBN Notebook: By 2025, fabs could cost $40 billion a pop!
Executive panel debates how chip making will change in 23 years
Semiconductor Business News
(01/08/02 15:16 p.m. EST)
PEBBLE BEACH, Calif. -- What will semiconductor production technology and wafer fabs look like in the year 2025? That was the brainstorming question tossed out at high-level executives with semiconductor equipment suppliers during the Industry Strategy Symposium (ISS) here.
As might be expected, the responses were entertaining as well as chilling. And, ISS panel discussion on Monday left many of those attending the session wondering how many chip makers and fab equipment suppliers will be around in 23 years.
Only the strong and rich will survive, according to the consensus from the panel discussion. For example, a single wafer fab could cost a staggering $40 billion by the 2025 timeframe, said Kenneth Levy, chairman of KLA-Tencor Corp. during the panel discussion.
In another ominous predication, Levy said he believes that lithography tools could cost as much as $300 million per system, based on current technology trends.
While the costs of wafer fabs and processing tools will continue to soar, the chips themselves will head in the opposite and historical direction, getting smaller, faster, and--of course--cheaper. The long-term question facing the chip industry is whether it will be possible to maintain historical growth rates as semiconductor plants become more expensive and affordable to fewer players.
Extrapolating from current technology trends and accelerating industry roadmaps, Levy predicted that chip makers will be processing wafers with device geometries of only "10-to-20 angstroms" in 2025. After the panel discussion, Levy told SBN that "for memories, we could see a terabit-on-a-chip."
There was another ominous predications during the ISS panel discussion: The IC industry may be maturing and will not grow at its current rate over the next 25 years.
The semiconductor industry has grown 35% over the last 25 years, from about $6 billion in 1975 to $210 billion in 2000, according to semiconductor analysts. Based on a 35% growth calculation, the chip industry should reach $7 trillion in revenues by 2025.
But is it possible that the chip industry will reach that level of worldwide sales in 23 years? It's highly unlikely, responded James W. Bagley, chairman and CEO of Lam Research Corp. in Fremount, Calif.
"If the IC industry grows to $7 trillion by 2025, then the semiconductor business could be 25 times larger than the electronics industry," Bagley noted. "I don't think that will happen."
It's also unclear just what will drive the overall IC industry during the next two decades. Will it be the PC? Or, communications? Biotechnology? Or even space exploration?
Bagley said he believes that the semiconductor equipment business may be the last to know what will drive semiconductor growth in the long term. "We are the most narrowly-focused people on the face of the earth--next to the Taliban," he quipped.
Most chip equipment executives agreed on one predication: It will be the survival of the fittest by 2025.
Only the strong IC and equipment companies will survive due to the soaring costs to stay in business, said Brad Mattson, former chairman and CEO of Mattson Technology Inc. of Fremont, Calif.
In fact, he suggested, the IC market will resemble the automotive industry with a wave of consolidation reducing the number of players. "I think the same thing will happen to us," said Mattson, during the panel discussion.
"The technical problems in the semiconductor industry will not cause Moore's Law to fail," Mattson added. "The economic problems in the semiconductor industry will cause Moore's Law to fail." |
