Predictions for 2009

Once a year editors take on the job of setting the following year’s agenda and mapping significant changes before they happen. The nice thing is that by the time the year is over, most readers don’t remember what we said. Frequently, even we don’t remember what we said, but that’s beside the point.

Here are a half-dozen predictions for the upcoming year, reflecting changes in both technology and business:

1.     Companies will finally admit the multicore approach works better as a virtualized platform for multiple applications than a multi-threaded platform for single applications. And they will be forced to admit that after 40 years of work on parallelization, there are limits to just how far threading can be extended. The result will be more standardized platforms with innovations in the software rather than innovative platforms with standardized software.

2.     Silicon-on-insulator will begin nudging out CMOS first because of current leakage and power efficiency reasons, and then for performance reasons. CMOS has a lot of life left in it, but not at the leading edge of Moore’s Law.

3.     The incoming administration in Washington will push technology as the solution to a more stable and responsive infrastructure, which will create a flood of projects and creative thought in the system-level design world. Energy conservation and intelligent machinery will be an important part of this reconstruction, but bottlenecks will remain in getting enough qualified help and enough bandwidth to do the job right.

4.     The downturn will continue to squeeze R&D budgets, and pure research will remain a to-do item for the incoming administration. With Bell Labs and Xerox PARC gone, and funding to schools already in trouble, this couldn’t happen at a worse time. No one in government has had their eyes on this ball for decades, though, which is why Lucent (replete with Bell Labs patents) was allowed to be sold to Alcatel, a French company.

5.     Investors will continue to undervalue chip companies, in part because they don’t understand the technology and in part because they understand the modus operandi of the people running them. Chip companies will continue to get less than is necessary for their technology, which will force other changes such as re-usable platforms and global development teams that can take advantage of lower-cost labor outside of the United States, Europe and Japan.

6.     Engineers will continue to do battle over the meaning of ESL. We just want them to know, at System-Level design, we’ll be watching.