SAN JOSE, Calif. – A post-silicon era is coming quickly, a veteran semiconductor executive at IBM said after his company announced it will spend as much as $3 billion in the next five years on R&D programs to drive to and beyond the limits of silicon.
The spending does not represent an increase or any specific new programs at IBM Research. Rather, it is a statement of IBM's commitment to core hardware technology amid reports that the company is looking to sell its two chip fabs.
Analysts and insiders have said IBM may sell its East Fishkill, N.Y., and Burlington, Vt., fabs this year, most likely to Globalfoundries, but retain its research efforts. IBM is not commenting on the reports, but it did summarize its extensive research efforts across a broad array of chip-related programs.
IBM posted a graphic summarizing its programs.
"The reason we made the announcement is the post-silicon area is coming, and we're not giving up but recommitting," said Bernie Meyerson, chief innovation officer at IBM.
The investments include programs in areas such as III-V materials expected to be used around the 5nm node, which some say could be the last generation of silicon-based chip technology. They also include a broad set of programs ranging from 3D chip packaging to computer architectures such as quantum and neural processing and post-silicon chip materials such as carbon nanotubes and graphene.
"This is not about semiconductors per se but a broad statement about reinventing computing," Meyerson said.
It's widely agreed that today's silicon CMOS processes will hit atomic limits somewhere around the 5nm node in about 2020. Many of the IBM programs are about finding new techniques that will drive hardware performance beyond that point. "The real issue is silicon goes quantum mechanical at these dimensions. It no longer works. At this time, there is no consensus on what's next." However, "we still have a 5-10 year horizon" to find solutions.
For whatever underlying technology comes next, "you have to make a billion devices that work at the same time and same way," Meyerson said. "That's a tremendous challenge and requires significant efforts."
— Rick Merritt, Silicon Valley Bureau Chief, EE Times