SANTA CLARA, Calif. — Speaking at a International Society for Optical Engineering (SPIE) manufacturing conference here, James Hogan, a founder of Cadence Design Systems and now senior vice president for business development at Artisan Components, opened a session on IC cost and performance with a discussion of "silicon economics."
Directly challenging the notion that the future is software, Hogan observed that the IC business is rapidly maturing, and that as a maturing market it would increasingly focus on economic, rather than technical, issues. This, in turn, would create a new round of growth for the IC business, he claimed.
"When the cost of a wafer from Shanghai is half the cost of a wafer from Taiwan, we have to stop and rethink our whole approach to what goes into hardware and what goes into software," Hogan said. He predicted that plummeting wafer costs would leave only design cost and design risk as barriers to far more aggressive use of hardware to solve problems.
To address this issue, Hogan drilled a little deeper into the design process. He suggested a model for manufacturing-aware design in which the designer's intent flows all the way down the chain to manufacturing, and information on manufacturing capabilities and limitations flow all the way up to design planning.
To keep this latter information flow from forcing design managers to become process gurus, Hogan inserted an abstraction layer in his design flow model. This platform, as he called it, served as a barrier to abstract manufacturing details and design intent so that they were both useful to their respective recipients on the other side of the platform layer.
Hogan also predicted further aggregation in the EDA, IP and foundry markets, leading to a handful of suppliers in each area. While this aggregation would compress the design chain, he said, it would probably not lead to the re-emergence of integrated device manufacturers.
As the IC industry thus matures and streamlines, Hogan suggested, the principles and lessons learned in microelectronics would go forth seeking a new home, and would quickly come to rest in the gray area between microelectronics and biology. He observed that at the 10-nm node electronics would be manipulating features on the same order as proteins.
If quantitatively accurate modeling could be created for manipulating the proteins, that would permit the same sort of hierarchy of abstraction that has given so much leverage to IC design. That, in turn, would make possible the same sort of technology explosion in the bio-electronic world that has taken place in electronics.
