Simson is an acclaimed Canadian academic and engineer who moved over from Nortel Networks to become chief technology officer of Motorola's Semiconductor Product Sector in April 2003.

Simson pulled few punches as she characterized the semiconductor industry as one beset by both technical and financial challenges. Simson said the semiconductor industry has changed and matured fundamentally and will not return to the formative growth spurts of the 70s and 80s. But Simson also observed recent evidence of irrational exuberance within the industry.

She said that from 1995 to 2001 companies started introducing a new manufacturing process technology every two years instead of every three years, as had been done from 1965 to 1995. "The technology cycle was shortened to two years after 1995 without demonstrable benefit to semiconductor companies," Simson said critically.

Simson was not downcast about the challenges ahead but she seemed to relish the opportunities they would provide.

While speaking on Motorola's response to the challenges Simson pointed out that efforts need to be made at every level, from compiler work, through design and design reuse using programmable platforms, down to circuits and on to fundamental process changes.

Among those process developments Simson stressed the importance of getting embedded memory, preferably non-volatile, deployed in all of Motorola's target markets; mobile, networking, automotive and consumer electronics.

Simson said that by 2008 about 80 percent of an SoC die area would be memory, about 10 percent would be reused logic and about 10 percent would be newly designed logic. Therefore the computational and power efficiency of the memory would dwarf the significance of the merits of well- or poorly-synthesized logic. Motorola is pushing hard to introduce magnetic RAM non-volatile memory technology to meet that goal, Simson said.

Simson also spoke out strongly in favor of silicon-on-insulator (SOI) saying, "We really believe SOI is going to provide a lot of opportunities. They are getting the defect levels at the 300-mm wafer down."

SOI is already in use to improve performance—this can be used to raise performance and average selling prices&3151;but it should also be very beneficial in reducing leakage current. By 2005 designers should be able to make use of SOI-specific design ideas and novel process modules would be introduced into an SOI foundation process. As volume increases the price of raw wafers reduces creating a virtuous cycle, Simson said.

And in ten years not only would SOI be the foundation process of choice for the industry, with optional modules based around it, but it should provide the route to multilayer stacking of active circuits in SoC, said Simson.

The final message from Simson was the familiar one of partnership. While all of this technical work has to be done at exponentially increasing cost, customers' time to market and costs have to be reduced which is only possible through partnership and risk mitigation.

Motorola's partnership with Philips, STMicroelectronics and TSMC on process technology and plans for balanced internal and external manufacturing are therefore a third key part of Motorola's strategy going forward.

What was not discussed was the possibility that the challenges would not be overcome, or that partnership might be a temporary alternative to consolidation.