San Jose, Calif.-The semiconductor industry is bound to sputter as Moore's Law comes under strain, but it's unlikely that anything will replace silicon as the foundation for electronics any time soon, said noted Stanford University professor Mark Horowitz.
Speaking at TSMC's technology syposium here, Horowitz said there will come a time when chip vendors won't be able to double the number of transistors on a chip every two to three years as they have for the last forty years.
"Scaling will end, and the end is not that far away unfortunately. But it's not going to happen tomorrow," said Horowitz, who is known for his pioneering work in microprocessor design and as a founder of Rambus Inc.
Horowitz said many of the challenges that Gordon Moore foresaw when he penned his famous law are coming to pass. These include the increasing cost of designing chips, increasing power dissipation and deciding what to do with all the available functionality.
Still other problems have been exaggerated, like wiring delays. Though wire resistance may seem to increase as a chip is scaled down in size, in reality it stays about the same as the length of the wire connecting circuits gets shorter, Horowitz said.
"It's hard to think of (wires) as a real impediment to scaling, though it will change the way we approach design," Horowitz said.
The more pressing concern, he said, is how the wires relate to chip complexity. Chip scaling gave chip companies the ability to pile on more hardware to build faster microprocessors, but not without a price.
"Wire complexity issues are hindering our ability to execute instructions in parallel," he said.
To overcome these design issues, chip companies are leaning on process technology improvements. But this is coming under strain too as problems with direct tunnel current and soft errors come into play, he said.
Still, the electronics industry is inextricably tied to silicon because of the amount of money it generates. And some of the most fundamental aspects of electronics, such as binary logic, are based on silicon, he said.
Moreover, recent innovations such as organic transistors and nano wires won't be able to catch up to silicon for some time. "It's hard to make something work if you don't have a lot of money and it's hard to make a lot of money if you don't have a lot of customers," he said.
Silicon, he said, may not be the most exciting medium for the electronics industry in years to come but it will be no less important.
"Think concrete and steel. We still use a lot of that stuff even though the innovation cycle has been declining," Horowitz said.