At the 2005 International Solid State Circuits Conference (ISSCC), a panel looked beyond the horizon to consider what the hot topics might be in 2010. Did anyone get it right? Let's see based on the panelists' statements that I have quoted completely out of context.
Werner Weber of Infineon: "Expect new memory concepts to arise and mature."
We seem forever on the precipice without ever taking the plunge. A session at this year's ISSCC on nonvolatile memory offers a mix of the emerging (Unity's CMOx paper) and several on maturing technologies, including two Numonyx papers on PCRAM. The definition of maturity may be determined by whether Numonyx ships 1G samples this quarter as predicted.
Dennis Monticelli, National Semiconductor: "The most interesting advances in 2010 will come from CMOS/BiCMOS processes optimized expressly for mixed-signal circuits."
An analog techniques session boasts ten out of ten circuits built on CMOS platforms.
Bill Redman White: "Connectivity in general stays [as] a big driver with the hot wire/less standard of the day getting the limelight, while bio/medical will grow."
This was a safe bet since we always seem to have more data than the pipes can handle. Several panels this year could be traced back to this prediction (at least I would if I had made it five years ago). A Feb. 8 evening panel addresses energy efficient, high speed interfaces while a Feb. 9 panel on the future of bionics seems to suggest a steadily growing capability in that field.
The 2005 panelists deserve a passing grade for their predictions. To put them into some historical context, we can compare their predictions with what other experts predicted five years ago. "Is There Life Beyond CMOS?" is the question S.M. Solomon of IBM attempted to answer with his presentation. It was a time when the challenges to scaling planar CMOS technology were starting to appear insurmountable. With the benefit of hindsight, strained silicon channels provided the drive current boost that dimensional scaling could not.
This was followed by the introduction of metal gates and high-K dielectrics to avoid bouncing off the electrostatics brick wall for control of the FET channel. MugFETs and vertical channels have not matured beyond the lab but continue to attract a lot of research attention. FinFETs or something similar are probably still another decade away if III-V channel materials will arrive first.